Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Basic Biological Overview of Eating Disorders

Katherine A. Halmi



The eating disorders anorexia nervosa and bulimia nervosa are complex syndromes that most likely result from, and are sustained by, environmental, psychological, and biological factors. Although they are disorders of eating, very little is known about psychobiological phenomena of hunger, satiety, taste, and eating behavior in these disorders. Both the chronicity and relapse rate of aberrant eating behavior is a serious problem (33). Many of the biological aberrations present in anorexia nervosa and bulimia nervosa reflect the nutritional state of the patient and revert to normal with weight gain, cessation of bingeing and purging, and eating a normal diet. Biological mechanisms that influence eating behavior, hunger, and satiety also affect mood, activity level, and cognitive states, which are disturbed in anorexia and bulimia nervosa (28). Thus, it seems reasonable to assume that studies of the basic biology of eating behavior will provide some insight into the aberrant eating behavior present in anorexia nervosa and bulimia nervosa. The first section of this chapter will provide a guide to the behavioral pharmacology of eating. Eating behavior hypotheses and experimental methods used to test those hypotheses are presented. The role of central neurotransmitters, neuropeptides, and peripheral physiology and metabolism in eating is discussed. The second section presents the actual neuroendocrine aberrations measured in anorexia and bulimia nervosa, and the third section integrates the psychological domain of eating behavior with basic physiological mechanisms of eating behavior (see also Vassopresin and Oxytocin in the Central Nervous System, Neuropeptide Y and Related Peptides, Cholecystokinin, Sress, Animal models of Psychiatric Disorders, Neuropsychchiareic Manifistations of HIV-1 Infection and AIDS, and Potential Mechanisms of Neurological Disease in HIV Infection).



Eating behavior reflects an interaction between an organism's physiological state and environmental conditions. Blundell and Hill (4) proposed a model in which the salient physiological variables included the balance of various neuropeptides and neurotransmitters, metabolic state, metabolic rate, condition of the gastrointestinal tract, amount of storage tissue, and receptors for taste and smell. The environmental conditions included features of food such as taste, texture, novelty, accessibility, nutritional composition, and other external conditions such as ambient temperature, presence of other people, and stress. More succinctly, Blundell and Hill's formulation is that the capacity to control nutrient intake to meet bodily needs requires special mechanisms to harmonize physiological information in the internal milieu with nutritional information in the external environment. Hypothalamic eating centers are part of the broad complex of neuroregulator interactions that include a peripheral satiety system (gastrointestinal and pancreatic hormones released by food passing through the gastrointestinal tract) and a broad neural network affecting feeding within the brain. When an exogenous agent such as a drug or peptide is given to an animal or human, it not only activates a specific set of receptors that induce specific responses, but also intervenes in a complex transactional fabric (4). Perceptual capacities identify (a) the characteristics of food materials in the environment and (b) the mechanism to link the biochemical consequences of the ingested food with the consumed structured form. Thus, a control over selection of foods must involve both conditioned and unconditioned responses (5).

Booth (7) proposed a concept of nutritional hedonic conditioning that occurs from the integration of sensory characteristics of foods (including nutritional functioning), culturally derived attitudes, and satiation cues (gastrointestinal tract, peptide hormones, neurotransmitters). Thus, nutritional hedonic conditioning is the process whereby the nutritional functions of the food are related to its sensory characteristics and, thus, its conceptual identity. Immediate determinants of actual food intake include the influences of sensory input and somatic physiology (8).

Models developed to study feeding in animals have been successfully used with humans. Early test models in animals either (a) used food deprivation to induce eating or (b) observed the effects of hypothalamic lesions. Later, pharmacological agents, either agonists or antagonists to neurotransmitters present in the hypothalamus, were used to probe effects on eating behavior. A microstructural analysis of feeding behavior has been used that involves the simultaneous recording of many behaviors such as drinking, grooming, locomotor activity, resting, and the eating behavior of animals, within a short time frame. Blundell and Latham (6) used a microstructural analysis to obtain extensive data on serotonin manipulations and animal feeding behaviors.

The macroanalysis of feeding patterns is a measurement of long-term feeding patterns in free-feeding animals never subjected to food deprivation. This continuous monitoring procedure has improved the precision of measuring parameters of meal patterns such as meal size, meal duration, meal frequency, inter-meal intervals, and ratio of meal size to meal interval. It allows assessments to be made under normal physiological conditions.

The technique of using varied and palatable diets was used to produce experimental obesity in animals. This dietary self-selection model allows the study of pharmacological agents and exogenously administered hormones on macronutrient (fat, protein, and carbohydrate) consumption.

With a refined microstructural analysis technique, Rogers and Blundell (68) studied human eating behavior using videotaped records. This provided a means for studying intra-meal selection patterns. Using the microanalysis technique, amphetamine was found to inhibit the onset of eating and increase eating rate, whereas fenfluramine shortened the duration of the meal and markedly slowed the rate of eating. With the dietary self-selection method, Silverstone and Kyriakides (70) used an automated food dispenser to study the action of various anorectic drugs on eating profiles.

A sham feeding model has been used in animals (80) in order to determine whether satiety signals arise from oral, gastric, or intestinal sites. In this technique, cannulas are placed in the esophagus or stomach so that they can be temporarily opened during a test to allow drainage and recovery of an ingested food. During sham feeding (cannulas are open), all species overeat. This demonstrates that food stimuli in the mouth are not sufficient to exert a normal satiety reaction. Food infused directly into the intestine produces a dose-related suppression of sham feeding.

The effects of stress on eating has been studied on animals with a mild tail-pinching technique, immobilization, or exposure to a novel environment (64). Obviously, these stress models and the sham feeding model are more difficult to adapt for studying human behavior.

The experimental models described above have been used to study the role of central neurotransmitters, neuropeptides, and peripheral physiology and metabolism in eating behavior.

Norepinephrine produces an appetite stimulant effect within the paraventricular nucleus (PVN), through a2-noradrenergic receptors (55). It appears to regulate feeding by inhibiting an inhibitor within the PVN. Tricyclic antidepressants stimulate food intake by activating the noradrenergic system in the PVN. During food deprivation there is a decrease of a-adrenergic receptor binding in the PVN and an increase in a-adrenergic receptor binding in the lateral hypothalamus (42). Norepinephrine, when injected in the PVN, causes preferential ingestion of carbohydrate-rich foods (55). Adrenergic b2 receptors in the perifornical hypothalamus (PFH) inhibit feeding when stimulated (54).

Hoebel (39) has used PVN injection studies to demonstrate the utility of serotonin, an indolamine, in the facilitation of satiety. Serotonin injected peripherally and centrally into the PVN suppresses deprivation-induced and norepinephrine-induced eating (53). The Wurtmans (79) demonstrated a feedback mechanism of carbohydrate intake and increased serotonin synthesis in the brains of rats. They demonstrated that increased carbohydrate intake increased the ratio of tryptophan to large neutral amino acids in the blood; this facilitated the entry of tryptophan across the blood–brain barrier, which, in turn, facilitated increased serotonin synthesis because tryptophan is a precursor of serotonin.

Dopaminergic systems are necessary for self-administration behaviors and could be a major link in the role of food as a reinforcer. Low doses of dopamine agonists stimulate feeding, whereas higher doses inhibit feeding (53). Glucose administration suppresses firing in the substantia nigra dopamine neurons. There is evidence of increased hypothalamic dopamine turnover during feeding (36). This finding suggests that central dopamine mechanisms mediate rewarding effects of food as they mediate rewarding effects of intracranial self-stimulation and self-administration of psychoactive drugs. Dopamine blockers, such as pimozide, can decrease intravenous self-injection, self-stimulation, or feeding (78). Self-administration behavior is stimulated by both opiates and nonopiate peptides (neurotensin), which activate dopamine neurons in the ventral tegmental area (VTA), and these cells project in the mesocortical pathway to limbic areas including the nucleus accumbens (24). (See also Dopaminergic Neuronal Systems in The Hypothalamus and Animal models of Drug Addiction .)

The role of endogenous opioids in eating behavior is more complex. b-Endorphin, morphine, and long-acting enkephalin analogues all induce feeding when injected into the PVN (39). Opioid antagonism decreases feeding in many species, but has no effect in reducing food intake in other species. Under some physiological conditions, such as starving or insulin-induced hypoglycemia, naloxone fails to inhibit feeding. Stress-induced eating is probably driven by activation of the opioid system. Both the K-receptor agonist dynorphin and (to a lesser extent) the mu-opioid increase feeding. The major site of action for dynorphin appears to be the PVN (62). Endogenous opioids seem to influence the intake of foods with a high fat content (69). Morphine-treated rats tend to select fatty foods, whereas naloxone-treated rats tend to avoid fats (53).

Corticotropin-releasing factor (CRF) is a neuropeptide that acts within the PVN to inhibit feeding. Norepinephrine (NE) seems to inhibit the CRF inhibitory feeding affect (63). Continuous infusion of CRF produced weight loss associated with both decreased food intake and increased thermogenesis (2).

Two peptides stimulate eating behavior. The pancreatic polypeptide neuropeptide Y (NPY) increases both food and water intake when injected into the PVN. Another pancreatic polypeptide, peptide YY (PYY), is a more potent stimulator of feeding than NPY (62). Both of these peptides increase weight gain (14), and NPY specifically increases carbohydrate ingestion.

The peripheral satiety network is another important component in the regulation of eating behavior. Satiety may be produced by several gastrointestinal hormones released during the passage of food through the gut. These hormones—cholecystokinin (CCK), glucagon, somatostatin, and bombesin—all decrease food intake after pharmacologic administration. Some of these peptides inhibit feeding by activating ascending vagal fibers. CCK is the most extensively studied of these peptides. The effects of CCK, mediated by vagal fibers, have been traced to the PVN of the hypothalamus, where lesions abolish the CCK effect on feeding (15). Low doses of CCK infused into PVN attenuate feeding, and central infusion of CCK enhances feeding. There is variability of the potency of the satiety effects of CCK across various animal species, and it appears to have little satiety effect in females (71). It is of interest to note that tolerance to the appetite suppressant effects of CCK occurs with prolonged administration. Specifically, in an animal study, CCK continued to suppress meal size, but was ineffective in suppressing 24-hr intake because the animals ate more meals to maintain their food intake (77). Other peptides that appear to inhibit feeding via vagal fibers are glucagon, somatostatin, and thyroid-releasing hormone.

Bombesin is a gastric peptide that inhibits feeding independent of vagal fibers. Systemic injections of bombesin produced a potent and dose-related inhibition of normal feeding (23) and a similar effect on sham feeding (58).

A new glucostatic hypothesis postulates that a small decrease of blood glucose has informational value for the eating system that is not directly related to current metabolic need (57). This postulation is supported by the observation that a decrease in blood glucose occurs shortly before the initiation of a meal in rats with constant access to food. The decreases in glucose are small (approximately 10%), and presumably are not sufficient to decrease cellular glucose utilization in any tissue (12).


Two studies of healthy men and women placed on diets showed that serotonin function, as measured by pharmacological challenge, was altered significantly in women but not in men (27), and the availability of circulating tryptophan, a precursor necessary for serotonin synthesis, was reduced in the dieting women but not in the dieting men (1). These findings suggest that women may have a biological vulnerability for developing eating disorders. Two studies in eating disorder patients give preliminary evidence that serotonin function may be disturbed in bulimic anorectics, and this may be related to the perception of satiety in those patients. Kaye et al. (47) showed a decreased serotonin turnover in bulimic anorectics compared with restricting anorectics by measuring 5-hydroxyindoleacetic acid (5-HIAA) in cerebrospinal fluid (CSF), after probenecid. Halmi et al. (32) showed that the serotonin antagonist drug, cyproheptadine, enhanced weight gain in restricting anorectics, but decreased weight gain in bulimic anorectics, compared with amitriptyline and placebo groups. Low concentrations of CSF 5-HIAA in low-weight restricting anorectic women, when compared to themselves after weight restoration and controls, suggest a starvation-dependent alteration in serotonin functioning (48). In another study, Kaye et al. (50) found elevated CSF 5-HIAA levels in long-term weight-restored anorectic women compared to controls. This finding suggests a trait contributing to pathological feeding behavior and weight loss. A popular speculation is that the proclivity in anorectic patients to be rigid, inhibited, ritualistic, and perfectionistic might also be associated with increased CSF 5-HIAA levels. Reduced prolactin responses to m-chlorphenylpiperazine (m-CPP), a direct serotonin agonist, and L-tryptophan, a serotonin precursor, in women with anorexia nervosa in both the emaciated state and at normal weight suggest that these anorectic patients have reduced postsynaptic serotonin receptor function in the hypothalamus (11). Serotonin agonists that are effective in reducing obsessive–compulsive behaviors and that are present in relation to food and weight in anorectic patients may be useful for the prevention of relapse in this disorder. At present, no definite conclusions can be drawn from studies of serotonin functioning in anorexia nervosa. Some studies suggest that alteration in the central serotonin system could develop as a result of weight loss, persisting long after weight restoration and contributing to the resistance to weight gain seen in patients with anorexia. Other studies suggest that people who develop anorexia nervosa have a preexisting dysfunction of the homeostatic mechanisms regulating the serotonin system, which then becomes easily destabilized by food restriction and weight loss.

Because serotonin facilitates satiety, it is reasonable to suspect reduced serotonin metabolism in bulimia. Jimerson et al. (43) found that bulimic patients who binge more than twice a day had a lower CSF 5-HIAA level than did controls and those who binge less often. This suggests that highly symptomatic, bulimic patients have a decreased presynaptic release of serotonin. However, studies by Kaye et al. (45) showed that bulimic patients had normal levels of CSF 5-HIAA.

Brewerton et al. (11) showed that bulimic patients given m-CPP had a blunted prolactin response compared with controls, suggesting an abnormality of postsynaptic serotonergic neurons. Only bulimic patients with major depression had a blunted prolactin response to L-tryptophan, suggesting an involvement with both pre- and postsynaptic serotonergic neurons in the depressed bulimics. McBride et al. (59) has shown that bulimic patients have a reduced prolactin response to the serotonin agonist fenfluramine.

In summary, there is a fair amount of evidence suggesting serotonergic dysfunction in anorexia nervosa and serotonergic hypofunction in bulimia nervosa.

There are few noteworthy findings of NE dysfunction in eating disorder patients. Kaye et al. (46) reported that long-term recovered anorexia nervosa patients have low CSF NE levels. In the same study, NE levels were similar in underweight anorexia nervosa patients and in normal controls.

This study needs to be replicated with a larger number of cases. Other studies have shown that plasma NE and urinary MHPG levels are reduced during the starvation state and are increased as anorectic patients gain weight (56).

Investigations using a variety of methodologies suggest that bulimic patients have decreased activity of the sympathetic nervous system. Plasma NE concentrations in bulimia nervosa have been found to be less than those in normal controls (47), as have CSF concentrations (45). Bingeing has been shown to increase plasma NE to higher-than-normal levels in bulimic patients, and abstinence from bingeing and vomiting appears to lower these levels. These findings suggest a state-dependent effect (47). In this same study, lower CSF levels of NE were associated with amenorrhea in bulimic women both when bingeing and abstinent. In another study of actively bingeing patients, CSF MHPG concentrations were normal (43). Kaplan et al. (44), using the a2 agonist clonidine in challenge tests in depressed and nondepressed bulimics, found no evidence for adrenergic receptor abnormalities at the hypothalamic level in either group.

The changes in NE metabolism in both anorexia and bulimia nervosa are probably state-related changes associated with starvation and dieting effects.

The abnormal hedonic responses to food present in both anorexia and bulimia nervosa could be related to dopamine function. Halmi et al. (34) showed that anorexia nervosa patients have a decreased growth hormone response to L-DOPA during both the emaciated and the weight-recovered states. They also demonstrated that anorectic patients have a decreased prolactin response to chlorpromazine in both the emaciated and the weight-restored periods. This suggests that anorectic patients have an impairment at the postsynaptic dopamine receptor site. Bulimic patients without a history of anorexia nervosa appear to have lower CSF HVA levels and a less vigorous dopamine response to a clonidine challenge than do normal controls (43, 44). These findings suggest that abnormalities in the dopaminergic pathways can lead to decreased satisfaction after eating, which in turn may facilitate binge-eating behavior.

Changes associated with opioid activity in anorexia nervosa are state-related. That is, both the increased levels of opioid activity in the CSF of severely underweight anorectic patients and the decreased CSF b-endorphin levels return to a normal range with nutritional rehabilitation (49, 51). Brambilla et al. (9) found that anorectic patients had elevated plasma b-endorphin levels that did not correlate with the degree of weight loss, but rather with the depressive symptomatology.

CSF b-endorphin levels have been found to be lower in normal-weight bulimic women and correlate inversely with the degree of depression (10). In one study (75), plasma b-endorphin levels were found to be decreased in bulimics, and they correlated with the severity of eating disorder but not with the severity of depression. Conversely, another study found plasma b-endorphin levels to be increased in bulimic women when compared to controls (19). Bulimic women who binged but had not purged by vomiting for 1 month prior to this study were found to have normal plasma b-endorphin levels, whereas those actively bingeing and vomiting had elevated levels (20). At the present time, it is not clear that the changes present in the opioid system in bulimia are the result of active bingeing and vomiting, starvation, or a trait feature that might predispose a person to bulimia. The opioid antagonist naloxone decreased the size of binge-eating episodes in bulimic patients in one study (61). A double-blind study with nontoxic doses of the long-acting opioid antagonist naltrexone showed no response in diminishing the binge-eating episodes (60).

In the study of fasting and postprandial plasma CCK concentrations in anorectic women, normal levels were found both when they were underweight and following short-term weight restoration (22). These findings do not support a hypothetical role for the hypersecretion of CCK in the etiology of anorexia nervosa. Geracioti and Liddle (21) reported that, compared with normal control subjects, bulimia nervosa patients had a decreased CCK response to an experimental meal. After these patients were treated with tricyclic antidepressants and had a decrease in binge eating, an increase in the postprandial CCK response and an increase in the satiety response occurred. It remains to be answered whether CCK has a primary pathognomonic or physiological role in bulimia or whether it represents an epiphenomena reflecting chronic gastrointestinal overextension.

Underweight anorectic women have been found to have normal CSF PYY concentrations and elevated CSF NPY concentrations, which remain significantly elevated after short-term weight restoration (45). In long-term weight-restored anorectics the CSF NPY levels returned to normal in those women who regain regular menstruation and remained elevated in those who continued to be amenorrheic or oligomenorrheic. Elevations in NPY do not appear to stimulate eating in women with anorexia nervosa and may simply reflect the starvation condition.

In bulimic patients, CSF PYY concentrations were found to be normal when the bulimics were actively bingeing and purging. These levels became elevated after 30 days of abstinence from bingeing and purging. CSF NPY values remained normal before and after the bulimic patients' abstinent period (45). The elevation of PYY following abstinence from bingeing and vomiting in bulimic patients cannot be easily explained, and it needs to be replicated. It seems unlikely that the elevation of PYY is a trait in persons vulnerable to develop bulimia, and that bingeing and vomiting is an attempt to normalize some intrinsic aberration.

It was demonstrated almost 20 years ago that anorectic patients have an impaired ability to concentrate urine (74). Later studies showed that anorectic patients have abnormally high levels of plasma and CSF vasopressin that gradually return to normal with weight gain (25). Demitrack et al. (16) showed that an underweight anorectic patient also has reduced CSF oxytocin levels that return to normal with weight restoration. Demitrack et al. hypothesize that a low level of oxytocin and a high level of vasopressin could work in concert to enhance the retention of distorted thinking about food and contribute to the anorectic patient's obsessional concerns about it. Oxytocin appears to disrupt memory consolidation and retrieval, whereas vasopressin promotes the consolidation of learning. Oxytocin also inhibits vasopressin-induced adrenal corticotrophic hormone release from the anterior pituitary gland. Nishita et al. (66) showed that vasopressin secretion abnormalities with hypertonic saline infusion occur in both underweight and weight-restored anorectics, as well as in bulimic patients. Vasopressin abnormalities in the bulimic patients are probably an epiphenomena of the electrolyte changes associated with self-induced vomiting. Two studies have shown elevated levels of CRF in the CSF of patients with anorexia nervosa (41, 49). These elevated CRF levels returned to a normal range with weight gain. Central and peripheral adrenocorticotropic hormone (ACTH) and cortisol levels are elevated in underweight anorectic patients, and they returned to normal with weight restoration (29). Kaye et al. (49) did find a positive correlation between CSF corticotropin-releasing hormone (CRH) levels and depression in weight-restored but not underweight patients. CRF secretion may have a role in maintaining anorectic behaviors and initiating a relapse. There is a possibility that hypersecretion of CRH may pre-date the weight loss in some anorectic patients, particularly those where depression does exist.

Normal-weight bulimic women have been shown to have normal (76) and elevated 24-hr plasma cortisol levels (65). Nocturnal circadian patterns of cortisol production have been found to be both normal (18) and elevated (52). Also, cortisol and ACTH response to CRH administration has been shown to be normal (29) and blunted (65). Actively bingeing and vomiting bulimics have demonstrated both normal CSF and plasma ACTH and cortisol levels and decreased CSF and ACTH levels following a 30-day period of abstinence (29). These conflicting CRH findings need to be clarified and may be related to dieting behavior.

Casper et al. (13) studied menstruating, weightrecovered, anorectic patients and nonrecovered, amenorrheic, underweight patients. Recovered patients showed glucose tolerance curves and insulin responses similar to those of normal controls, whereas nonrecovered anorectic patients had persistently abnormal glucose tolerance and delayed insulin release. Glucose metabolism abnormalities appear to reverse with affective treatment in these patients.

Low fasting glucose levels have been found in bulimic women (17, 67). Goldbloom et al. (26) showed that fasting glucose, alanine, pyruvic, fat-derived metabolites, insulin, and glucagon were similar in actively bingeing and purging bulimic women and normal controls. The C peptide (secreted by the pancreas in equal amounts with insulin) level in bulimic patients was 50% that of controls, indicating a decreased clearance of insulin in the patient group. Blouin et al. (3) found that after an intravenous glucose challenge, blood glucose level and insulin/glucagon ratio remained lower in bulimics than in controls for 45 min after injection. These observed physiological differences were accompanied by an increased subjective craving for sweets, an enhanced urge to binge, and less subjective control over food intake compared to controls.


In the section of this chapter entitled "Behavioral Pharmacology of Eating," Blundell and Hill's model of eating behavior was described (4). In summary, their formulation was that the capacity to control nutrient intake to meet bodily needs requires specialized mechanisms to harmonize the physiological information in the internal milieu with nutritional information in the external environment. Two essential features are perceptual capacities to identify the characteristics of food materials in the environment and a specialized mechanism to link the biochemical consequences of the ingested food with the consumed structured form (4). One of these integrating mechanisms is the perception of hunger and satiety (37). In this section, studies of hunger and satiety perceptions of eating disorders will be reviewed along with studies of macronutrient and taste influences on eating behavior in the eating disorders.

Using an experimental meal paradigm with a liquid formula, Halmi et al. (30, 35) were able to demonstrate that eating-disorder patients have disturbances in the perception of hunger and satiety. The hunger and satiety curves before, during, and after the experimental meal demonstrated that anorexia nervosa and bulimia nervosa patients often have difficulty distinguishing between perceptions of hunger and satiety. Underweight anorectics had lower hunger levels, higher satiety levels, and less urge to eat than did normal-weight bulimics and normal-weight controls. Subjects who binged and purged were more preoccupied with thoughts of food, and normal-weight bulimics had more urge to eat after finishing a meal. Two patterns of responses emerged unaltered by treatment: (i) Those patients who were previously underweight (ANR and ANB) continued to have lower hunger ratings than did the other two groups, and (ii) ANR continued to have a higher level of satiety than did the other groups. After treatment, a difference emerged in those patients with a history of bingeing and purging. They had significantly higher palatability ratings than did ANR and controls. Urge to eat was positively correlated with monthly binge frequency for the bulimic subjects. Bulimic patients who binged the most were also the most depressed.

After establishing differences in hunger and satiety perceptions in eating-disorder patients with a standard liquid formula, Halmi et al. (31) studied the responses in anorectic and bulimic patients to meals differing in macronutrient content. In this preliminary report, there were distinct differences in the hunger and satiety responses in bulimia nervosa patients to meals that were high in carbohydrate and low in fat compared with meals that were low in carbohydrate and high in fat. In addition, the bulimic responses were different from both the unrestrained normal controls and the anorectic restrictors. It appears that the high-fat meals may indeed provoke a hunger stimulus in bulimic patients, but not in controls and not in anorectic restrictors.

The fullness or satiety ratings were consistently higher in the anorectic restrictors before both high-fat and high-carbohydrate meals; however, they remained higher compared with the other groups only after the high-fat meal. The bulimics had lower fullness or satiety ratings after the high-fat meal compared with the high-carbohydrate meal and compared with the anorectic restrictor ratings. Another interesting finding occurred with bulimics after the high-fat meal. After 8 min their ratings began to increase, and after 14 min their hunger ratings were only slightly below those they had at the beginning of the meal. This did not occur with the high-carbohydrate meal, after which the normal-weight bulimics had substantially lower hunger ratings compared with those before that meal. The hunger and satiety curve patterns before, during, and after the meal were markedly distorted in all of the eating-disorder patients.

There is ample evidence from the studies just described that the integrating processes for physiological information and perceptual capacities in the control of eating behavior is markedly disturbed in both anorexia and bulimia nervosa patients.

In a study of taste perceptions in eating-disorder patients (72), the anorectic restrictors and anorectic bulimics displayed a pronounced aversion to high-fat stimuli. The anorectic restrictors also showed a strong dislike for solutions that contained little or no sweetness. The negative responses to fat persisted following weight restoration, suggesting that this behavior may be a stable trait characteristic of anorexia nervosa. The bulimic patients did not differ in their hedonic ratings compared with control subjects. The results of the taste study match those of a cognitive preference study (73), in which all patients with a current or past history of anorexia nervosa significantly disliked high-fat, high-calorie foods compared with the control groups and the bulimic patients. After weight restoration, this intense dislike of high-fat and high-calorie foods persisted and, therefore, appears to be a stable trait characteristic in anorexia nervosa.

From the review presented here, it is obvious that there are aberrations or dysfunctions in many physiological systems in both anorexia and bulimia nervosa. Most of these physiological systems interact with each other, and most of the research conducted has studied individual isolated physiological mechanisms. Although it is a costly endeavor and difficult to accomplish, future research focusing on the interactions of these various systems would probably yield far more significant and helpful information for understanding the development of these disorders and providing clues for more affective treatment strategies.

What are some of the areas that warrant continued and more creative research? The studies of insulin secretion in anorexia and bulimia nervosa are contradictory, but there is enough evidence of changes present that warrant further investigation. CRH hypersecretion may have a role in maintaining anorectic behaviors and initiating a relapse and, therefore, deserves further study. There is considerable evidence for serotonin dysfunction in both anorexia and bulimia nervosa. Just how these aberrations interact and affect eating behavior in these disorders needs to be clarified. The persistent low CSF NE levels in long-term so-called weight-restored anorectics needs an explanation. Does this merely reflect continued aberrant eating behavior and dieting? If so, how does it affect and sustain abnormal eating behavior? Very little research has been conducted on the role of dopamine and eating behavior in the eating disorders. The abnormal hedonic food responses seen in both anorexia and bulimia nervosa could be related to dopamine function. There are only a few studies on neuropeptides in eating disorders. These studies indicate that aberrations in the neuropeptides affecting eating behavior are present in anorexia and bulimia nervosa. Creative studies in this area may show how changes in these neuropeptides and neurotransmitters affect the disturbed integrating mechanisms of hunger and satiety perceptions.

published 2000