Minutes of the Columbia University Seminar on Appetitive Behavior (#529)
Chairman: Harry R. Kissileff, Ph.D.
Rapporteur: Janet L. Guss
Date: Thursday, December 14th, 2000
Title: Is 'Food Addiction' an Accurate Concept and, if so, What are its Implications?
Speaker: Dr. Bart Hoebel, Princeton University, Princeton, N.J.
Presentation:
The neurochemical processes that mediate addictions are not well understood, but there is strong evidence that dopamine (DA) and opioid neurotransmitter systems are both involved. DA is believed to elicit 'incentive motivation', a phrase used to describe the ability of a reinforcer to arouse an animal such that its behavior is directed towards activities that are predictive of a reward (e.g. feeding or mating). Opioids of various kinds have a wide range of effects in different parts of the brain. At both ends of the DA system, in the ventral tegmental area (VTA) and nucleus accumbens (NAc), opioids that act at the mu-type receptor can elicit eating and also reinforce behavior.
The neurochemistry by which DA mediates incentive behavior is unknown, but recent research suggests that changes in the ratio of extracellular DA to acetylcholine (Ach) in the NAc may be involved in its control. Conditioned taste preference and aversion, and drug approach and withdrawal sensations that are linked to addiction, both appear to be related to changes in extracellular ratios of DA and Ach. Learned flavor preferences and drugs of abuse are both associated with increases in extracellular DA and decreases in Ach, whereas conditioned flavor aversions and drug withdrawal are both associated with declines in DA and increases in Ach. Substances and/or events that mimic the drug effects could therefore have the potential to be addictive.
Consumption of food, particularly that which is palatable, causes transient DA release and the build up of extracellular DA in the NAc, as does mating, intracranial self-stimulation, and stimulation-escape behavior. Since consumption of palatable foods stimulates DA, as well as opioid, neurotransmitter systems, it stands to reason that such foods may have an addictive component to them. Palatable food acts via 'normal routes' to the DA cell body regions, whereas drugs such as cocaine and amphetamine act pharmacologically to release DA or block its reuptake in the neuron's terminal fields. Repeated administration of such drugs causes prolonged changes in receptor binding as well as intracellular adaptations in second messenger production and gene expression that may underlie drug addiction. Saccharin, which rodents find extremely palatable, is potent enough to substitute for cocaine in the judgment of an addicted rat or monkey, under certain conditions.
One key feature of addiction is an increase in consumption of the substance over time, after it has been available on an intermittent-access schedule. With palatable foods, intermittent access can lead to excessive consumption bouts; behavior referred to as 'binge eating'. This pattern of bingeing and deprivation could create the neurochemical conditions for opioid-like withdrawal, since the behavior resembles that associated with addiction. Behavioral and neurochemical evidence of withdrawal-like symptoms following repeated access to highly palatable foods would also be indirect evidence that foods can have an addictive quality.
Addiction to palatable food would not be expected to share all the properties of addictive drugs, but evidence already exists from other laboratories that there are some similarities. Palatable food can suppress pain and potentiate the analgesic effects of morphine, and morphine-like substances can increase food intake. An opiate antagonist, such as naloxone, reduces eating that occurs following food deprivation, access to palatable food, or following hypothalamic injections of Neuropeptide Y. Interestingly, naloxone can block intake of palatable food while intake of less palatable food is unaffected. Thus, consumption of palatable food releases endogenous opioids, resulting in prolonged consumption bouts.
To test whether behavioral and neurochemical features of addiction occur in response to intermittent access to highly palatable foods, rats were tested for behavioral and neurochemical markers of addiction after being placed on a cyclic feeding schedule. Each day they had a 12 hr period of access to 25% glucose solution plus chow, and a 12 hr period of deprivation. After several weeks on this cycle, behavioral and neurochemical features consistent with addiction emerged. The rats engaged in excessive 'binge-like' consumption of the glucose solution, and showed changes in the pattern of DA receptor binding and up-regulation of opioid receptors- both of which were consistent with that observed in a cocaine-addicted animal.
After administration of the opiate antagonist naloxone, they showed signs of withdrawal, including teeth chattering and anxiety. They also developed naloxone-induced disruption of DA and Ach balance, such that extracellular Ach increased while DA decreased. This pattern mimics that seen with naloxone-precipitated morphine withdrawal. These data, taken together, suggest that a cycle of dietary deprivation alternated with excessive intake of highly-palatable food induces behavioral and neurochemical features similar to addiction.
Discussion:
Dr. Hoebel began the discussion by asking attendees to suggest criteria that ought to be met by his model before it should be referred to as one of sugar addiction. In response, it was suggested that he attempt to block the withdrawal symptoms he had precipitated with naloxone. If he can block the neurochemical symptoms that mimicked withdrawal by giving intravenous glucose (in the same cycling pattern), then most attendees felt that he will have satisfied the criteria for a model of sugar addiction.
Q: It seems that the temporal pattern, i.e. the cycling nature of the paradigm, is the key to this addiction model, not necessarily the palatability of the food- can you comment?
A: Yes, the pattern of access to the glucose was key, but we must not neglect the fact that highly palatable foods release DA in the NAc whether or not they've been given in this temporal pattern. Thus both the palatability and the temporal pattern may be necessary to cause bingeing.
Q: What about tolerance? How is it involved in food addiction?
A: This has not been tested.
Q: Is addiction an issue of tolerance and dependence, or one of sensitization? For example, intermittent drug administration leads to stronger addiction than consistent administration of the same amount of the same drug. But consistent administration leads to tolerance, while intermittent administration does not.
A: This is an area currently under investigation but there is no consensus on the issue at this time. Both tolerance and sensitization are probably involved at the receptor level. The evidence I presented was for sensitization in the form of D1 and mu-1 receptor up-regulation in the accumbens and what might be called behavioral sensitization in the form of progressively enhanced glucose intake.
Q: What is the relationship between the glucose addiction you were able to generate in your study and the glucose concentration you used? Does the concentration matter?
A: (Scalfani) Glucose preference studies have found that concentration-dependent increases in glucose preferences plateau once the solution exceeds a concentration of 8-10% glucose.
Q: Have you considered how glucose-responsive neurons might be involved in your paradigm?
Concentrated glucose solutions also stimulate dopamine in these neurons, so this may be a confound. Have you studied other palatable substances?
A: We have studied saccharin on the same intermittent schedule and found withdrawal symptoms in the form of somatic symptoms. This indicates the importance of taste. The reason to start with glucose was to get the benefit of conditioned dopamine release by association of the taste and the nutritious calories. The idea that glucose might affect dopamine cells directly is also very interesting.
Q: Can the effect you observed be due to conditioning?
A: Yes; in addition to conditioned dopamine release, which we reported using a taste paired with intragastric feeding, work by Ann Kelley's group with the mu-opioid receptor agonist DAMGO showed conditioned eating in response to local injections in the accumbens. Thus both dopamine and opioid conditioning could be involved.
Q: How do you know that your behavioral measure of addiction, the excessive consumption pattern that evolves several weeks after the intermittent exposure, is not a simple learning effect? If the rats only get periodic access to this palatable solution, wouldn't they learn to cram as much of it in as possible during the short time in which they have access?
A: It could be that the animal learns to 'compress' their intake because of the 12-hour deprivations. Nonetheless, the result of such an eating pattern will be that post-synaptic terminals will be deluged with the surge in DA that will be released during the eating bout, since it's been stored over the 12-hour fasting period. Such would not be the case under freely feeding situations because some DA would get released with each smaller eating bout and hence it wouldn't be stored up for long periods and then released in surges. The resulting DA 'rush' could be a powerful reinforcer, creating enhanced incentive-motivation and may foster post-synaptic changes involved in addiction.
Q: What happens under conditions of intraoral glucose intake?
A: (audience reply) In that paradigm, you can't get the same conditioned preference, or learning, from saccharin as you can from glucose. These different means of ingestion therefore seem to have different impacts on learning.
Q: What about cross-addictions?
A: Since we saw significant increases in both D1 and mu-opioid receptor binding, we might also see cross-addiction, or at the least cross-sensitization, in this model. We are doing this.
Comment: Your data imply that the pattern in which the highly palatable food was consumed (i.e. the feast and famine cycle) really formed the basis for the addictive features observed in your animals. It therefore seems essential that you not focus so much on the highly-palatable food per se, but the pattern in which the food gets consumed.
Q: How do the neurochemical changes observed in the current study compare with those measured after intermittent exposures to amphetamine or cocaine? Are the changes that occur with 'sugar addiction' quantitatively comparable to those observed with cocaine addiction?
A: We haven't directly compared the two, but we don't expect to see the same severity in receptor changes, for example following cocaine addiction, as we found with this study. But the types of adaptations we found were in the same direction as some of those seen with drugs of abuse.
Comment: Most animals eat intermittently in the wild anyway so your paradigm seems completely 'naturalistic'; experiments in which animals have ad libitum food access are the exception to the natural condition.
Comment: But there is no natural environment in which a food would contain 25% glucose, so it is impossible to say whether we are observing a mechanism that is part of natural reward, or one related more to addiction.
Comment: There are theories dichotomizing the neurotransmitter systems that underlie eating disorders, suggesting that patients with anorexia have an addiction to endorphins, and patients with bulimia have an addiction to dopamine.
Q: Can you separate out the role of NPY in driving the excessive eating in your cycled-access model?
A: (audience reply) We haven't, but one idea would be to study the dopamine knock-out animal model, by giving it intermittent NPY injections, or we could block D1 receptor action in the NPY knock-out mouse, and put it on this feeding schedule. Unfortunately, since addiction researchers don't interact much with feeding researchers, the neuroanatomical pathways linking feeding and addiction have not been thoroughly studied. Feeding researchers have carefully described neural circuits within the hypothalamus and addiction researchers have focused in on the nucleus accumbens and surrounding regions. The most fruitful approach will be to examine how all of these areas respond in addictive paradigms.
Q: If your rats developed features of opioid-like addiction, can you get any symptoms of withdrawal without giving an opioid antagonist?
A: We probably could, but this would take longer to observe and would be less overt to measure. Withdrawal in an addicted animal will be demonstrated by the difference in symptoms between what was observed pre-treatment vs following addiction to the substance in question. We are studying spontaneous withdrawal in rats pre-treated with the glucose/chow cycle and in rats that were on a saccharin cycle. There is an increase in somatic withdrawal symptoms, notably teeth chattering, during spontaneous withdrawal.
Comment: It is of great interest that this paradigm mimics the eating behavior of many patients with eating disorders. Naloxone and naltrexone have both been used to eliminate feelings of reward associated with binge eating, but they were not helpful, so despite the similarity of this model to compulsive eating disturbances, there must be additional disturbances that drive such behaviors.