Melita L. Daley, Carolyn Nguyen and Robert T. Rubin

“A young woman thus afflicted, her clothes scarcely hanging together on her anatomy, her pulse slow and slack, her temperature two degrees below the normal mean, her bowels closed, her hair like that of a corpse—dry and lustreless, her face and limbs ashy and cold, her hollow eyes the only vivid thing about her—this wan creature whose daily food might lie on a crown piece, will be busy with mother’s meetings, with little sister’s frocks, with university extension and with what you please else of unselfish effort, yet on what funds God only knows.”¹

History and Epidemiology

AN and BN are diseases primarily of adolescent girls; approximately 95% of AN cases are female.¹⁰ Although there is documentary evidence of AN occurring in the Middle Ages, the first medical accounts appeared in the 17th century as A Discourse upon Prodigious Abstinence by John Reynolds in 1669 and Phthisiologia; or, a Treatise of Consumptions by Richard Morton in 1689.^3,11,12 AN as a modern clinical entity derives from publications in 1873 by Charles Lasègue and William Gull, who referred to the illness as hysterical anorexia,^13,14 and Gull’s publication in 1874 entitled Anorexia Nervosa.¹⁵ Bliss and Branch³ commented, “It is revealing to note the extraordinary differences in the description of the same condition by the two men. While Gull’s comments were as direct and precise as a pathologic report, Lasègue conveyed a sense of the spirit and feeling of these people, the nuances of their disturbed relationships, and the subtleties of their intrapsychic turmoil” (p. 14). The early 20th-century history of AN primarily involves its distinction from physical illnesses such as Simmonds’ cachexia.¹⁶

Although binge eating (bulimia) has long been recognized as part of the symptomatology of AN, BN as a distinct syndrome was first proposed by Russell in 1979.¹⁷ He elaborated two criteria for BN: an irresistible urge to overeat followed by self-induced vomiting or purging, and a morbid fear of becoming fat. In common with AN, such patients kept their body weight somewhat below normal but not to the same extent as AN patients. They also tended not to develop amenorrhea and were more active sexually. Russell considered BN to be an “ominous variant” of AN, in that comorbid depressive symptoms were often severe and distressing, leading to a high risk of suicide.

In the 1990s, there were opposing views about whether to treat binge eating disorder (BED) as a separate entity. Some argued that too little was known about it and other recurrent forms of overeating, and inclusion would be a source of diagnostic confusion. Others argued that there was a relatively pure group of individuals with Eating Disorder Not Otherwise Specified (EDNOS) who met criteria for BED and not BN, and that recognizing BED as a new disorder would stimulate research and clinical programs for these patients. Since 1994, BED became a provisional eating-disorder diagnosis and was included in the appendix of the fourth edition text revision of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR).¹⁸

In the past 30 years, careful theoretical formulations of the psychodynamics of these eating disorders have been augmented by elucidation of their physiology, especially their neuroendocrinology. Psychopharmacologic and psychotherapeutic treatment methods have been refined, and there now is a reasonably effective armamentarium with which to manage these disorders.

The reported incidence of AN has varied between about 0.5 and 15 cases per 100,000 population per year.19–21 The large variation is related to the diagnostic criteria used, the methods of case ascertainment, and the population studied (e.g., clinic versus community). The prevalence of AN ranges from 0.5% to 1.0% among females, with males being affected about one tenth as frequently, although the prevalence has been considerably higher in some studies.^20,21 A meta-analysis of 42 studies revealed the crude mortality rate in AN (due to all causes of death) to be 5.9%, which is substantially greater than that for female inpatients and for the general population.²²

The prevalence of BN varies between 2% and 4%.20,21,23,24 Both AN and BN, as well as subsyndromal anorexia and bulimia, are far more common in certain groups of young women, notably athletes and ballet dancers, where occupational demands place a premium on thinness. Primary and secondary amenorrheas in these individuals are extraordinarily common. Overweight and obesity are common comorbidities of BED, which has an estimated prevalence rate between 0.7% and 3%.²⁵

Since the 1980s, large-scale twin and family studies have suggested that eating disorders are familial.26,27 There is cross-transmission between AN and BN, and they appear to share a common vulnerability, but the transmissible elements remain elusive. Twin and family studies suggest that major depressive disorder and substance dependence most likely do not share a common etiology with the eating disorders. The genetic contribution is likely predisposing rather than determining, in that sociocultural circumstances, as well as personal psychological stressors, are risk factors.^28,29 The risk of developing AN is higher in female twins than in male twins, but male members of opposite-sex pairs have almost the same risk as in female twins.³⁰ This suggests that an intrauterine factor, such as sex steroid hormones, also influences the development of AN.

Diagnosis

The DSM-IV-TR¹⁸ provides the current clinical criteria for AN (Table 4-1) and BN (Table 4-2) and EDNOS, which includes the provisional diagnostic category of BED. Since the DSM became a criterion-based system with DSM-III in 1980, it has undergone two revisions, and several diagnostic categories have undergone considerable change across these revisions. In contrast, the criteria for AN have remained relatively stable, although future revision might be useful, such as the elimination of amenorrhea as a criterion.³¹ The criteria for bulimia are in greater flux. For example, BED, characterized as recurrent episodes of binge eating and associated distress without inappropriate compensatory purging or other behaviors, has been included in the DSM-IV-TR Appendix as a set of criteria for further study.¹⁸ There have been recommendations to include BED as a distinct eating disorder in the upcoming DSM-V.^32,33 Although there are far fewer men with eating disorders than women, the clinical features are similar.³⁴

The DSM-IV-TR criteria for AN, BN, and BED highlight the many identifiable behavioral and psychological aspects of these illnesses.³⁵ If the patient and her family are queried about these aspects, considerable information pointing toward the diagnosis can be gleaned, psychiatric consultation can be considered early in the diagnostic process, and specialized and expensive laboratory testing often can be avoided.

For AN, a key feature of the weight loss is the patient’s refusal to maintain body weight, which often manifests as an active resistance to increasing caloric intake. As illustrated in Fig. 4-1, the cachexia can be severe, and as mentioned, self-starvation can lead to death. If the onset is in childhood or early adolescence, there may be a failure to make the expected weight gain during the active growth phase. This refusal to gain or maintain body weight is rooted psychologically in the second criterion, an intense fear of gaining weight or of becoming fat, even though the patient is underweight. There is a subjective distortion of body image such that the emaciated individual appears to herself to be either at an acceptable weight or even fat. Indeed, initiation of treatment by insisting on the patient’s eating can lead to purging behavior that did not occur prior to treatment. How inflexible a given patient may be in refusing to gain weight, however, can vary considerably.³⁶

A corollary of this is the third criterion, disturbance of the experience of one’s body weight or shape (e.g., undue influence of body weight or shape on self-evaluation/self-esteem and/or denial of the seriousness of the weight loss), even though there may be clear adverse physical sequelae in addition to the current fourth criterion, secondary amenorrhea. For the endocrinologist, questions that should be asked of the patient and her family to address these psychological and behavioral aspects of AN include the following (adapted from the Structured Clinical Interview for DSM-IV³⁷):

Rather than rigidly adhering to the DSM-IV-TR algorithm for diagnosing AN—that is, the reduction of body weight to less than 85% of expected and complete amenorrhea for at least 3 months—the clinician should consider both the patient’s core symptoms and current medical impairment; such impairment may require treatment even though body weight may be greater than 85% of expected and/or menses still may occur sporadically.³¹

For BN, the distinguishing features are uncontrollable binge eating of a definitely larger-than-normal amount of food (first criterion) that is clearly repetitive (third criterion). Here too, one’s sense of self is unduly influenced by weight and body shape (fourth criterion). Compensatory behaviors such as self-induced vomiting, purging, fasting between binges, and exercise are invoked to prevent weight gain (second criterion). The frequency and chronicity of the bingeing, and especially the compensatory behaviors, help distinguish BN from overeating in general. And because binge eating and purging can occur as part of AN as a subtype, a fifth criterion for BN is that it does not occur exclusively during episodes of AN. The validity of the DSM-IV-TR eating disorders classification has been questioned, owing to the high rates of diagnostic crossover among the AN subtypes and BN. Recent findings support the longitudinal distinction between AN and BN, but they may not support the AN subtyping schema.38,39 BED also is being distinguished from BN^18,40 because obesity rather than inanition often results, and the endocrinologic and other metabolic changes are different from those of AN and BN.

For the endocrinologist, questions that should be asked of the patient and her family to address the criteria for BN include the following (adapted from the Structured Clinical Interview for DSM-IV³⁷):

These questions and those regarding AN should be phrased in a way that is comfortable to both the health care professional and the patient. Depending on the patient, questions about both bulimic and anorexic behaviors may need to be asked; the patient’s history and presenting symptoms should dictate the emphasis of the interview.⁴¹ The sample questions are given to indicate that one can (and should) ask about eating behavior forthrightly. If the topics of the questions are followed, information about all the diagnostic criteria for both AN and BN will be gleaned. The patient should also be asked if this is the first time these behaviors have occurred or if there have been past episodes; in the case of the latter, a careful history taking about each episode, its severity and duration, and treatments and their success should be done. This information may yield important clues as to how to approach the patient therapeutically during the current episode.

Comorbid symptoms of depression and anxiety occur with some frequency in both AN and BN and should be assessed, because their persistence during treatment may portend poorer outcomes.42–48 Symptoms of depression and anxiety are exaggerated by malnutrition and improved with nutrition. AN patients with obsessive-compulsive disorder tend to have an obsessional need for symmetry/exactness and compulsions toward ordering and arranging.⁴⁹ Mild to moderate negative mood states and obsessive symptoms can persist after recovery in both anorexic and bulimic individuals,^50,51 suggesting that these traits may contribute to the pathogenesis of eating disorders.⁵² As well, comorbid impulse control disorders (ICD), the most common being compulsive buying disorder and kleptomania, occur more frequently in individuals with binge eating subtypes and are associated with greater use of laxatives, diuretics, appetite suppressants, and fasting.⁵³

There also is a high incidence of comorbid obsessive-compulsive disorder (OCD) in anorexic and bulimic women and their families, as well as increased rates of AN and BN in persons with primary OCD. It may be that the core eating disorder symptoms (e.g., fear of fatness, pursuit of thinness) are a specific type of obsession. Symmetry, ordering, and perfectionism are the most common target symptoms in women with AN and BN and often persist after recovery. Leckman and others⁵⁴ delineated four symptom dimensions of OCD, one of which was symmetry and ordering; these occurred most commonly in men. OCD in men and eating disorders in women may be gender-specific expressions of a common psychobiology related to obsessions with symmetry and ordering compulsions.

There are primary central nervous system (CNS) conditions for which anorexia or bulimia may be an associated symptom; these include both organic pathology⁵⁵ and “functional” illnesses. As an example of the latter, one of the authors was asked to consult on a hospitalized adolescent girl with electrolyte disturbances suggestive of Bartter’s syndrome. With no evidence of somatic causes of the patient’s cachexia, and with nurses’ observations of bizarre eating habits, a diagnosis of AN was entertained, and a psychiatric interview of the patient during endocrine rounds was requested. After introductions as to who the psychiatrist was and why he was there, questioning began with what the patient ate (lettuce and carrots only) and why she ate only those two items (the computer in the hospital basement was giving her instructions to do so). Elucidation of additional symptoms of psychosis led to a presumptive diagnosis of schizophrenia with secondary anorexia and referral for inpatient psychiatric care.

Genetics

Eating disorders in both women and men are familial, as shown by both family and twin studies.56–61 The heritability of both AN and BN has been estimated between 33% and 84%.⁶² Recent findings corroborate previous research implicating the transition from early to mid-adolescence as a critical time for the emergence of a genetic diathesis for disordered eating.⁶³ Studies of the genetics of AN and BN have been primarily by linkage analysis and association studies. In linkage analysis, correlations are determined between the occurrence of a disease in families and the inheritance of specific chromosomal regions in those families. In association studies, nucleotide polymorphisms are searched for in specific candidate genes suggested by chromosomal linkage studies and/or the known functions of the gene product.

Linkage studies in AN and BN families have suggested susceptibility genes on several chromosomes, but the strength of the associations has depended on the diagnostic stringency of the cases. For example, in 192 families with at least one affected relative pair with AN, BN, and related eating disorders, modest linkage was found with a marker on chromosome 4.⁶⁴ However, when the analysis was restricted to 37 families in which at least two relatives had a diagnosis of the restricting subtype of AN (without bingeing and purging), there was a much stronger linkage to a marker on chromosome 1p. With reference to specific behavioral traits, two variables were identified in a sample of 196 families with an AN proband: drive for thinness and obsessionality.⁶⁵ When these variables were incorporated into a linkage analysis, there again was highly significant linkage to a region on chromosome 1, as well as linkages of lesser significance to regions on chromosomes 2 and 13. In contrast, linkage analysis of 308 families with a BN proband yielded a high linkage score to chromosome 10.⁶⁶ In a subset of 133 families in which two or more BN members reported self-induced vomiting, an even higher linkage was found to a region on chromosome 10p. Another region on chromosome 14q was suggestive of linkage. These studies can serve to suggest specific chromosomal regions for association studies.

In association studies, researchers have targeted serotonin, dopamine, and norepinephrine-related genes. The most promising results have been with the serotonergic system, because serotonin (5-HT) neurotransmission inhibits feeding behavior.56,67 Several studies have suggested an association between polymorphisms in the promoter region of the 5-HT2A receptor gene and both AN and BN, but the results have been inconsistent, the association has been rather weak, and the functionality of the 5-HT2A polymorphism is unknown.^68,69 Studies reporting an association between a polymorphism in the promoter region of the 5-HT transporter gene and eating disorders have been inconsistent as well. In one study, a relationship was suggested between a polymorphism in the 5-HT1B receptor gene and BN; in another study, a relationship was suggested between a polymorphism in the 5-HT2C receptor gene and vulnerability to rapid weight loss following reduced food intake in AN. On the other hand, studies of the 5-HT receptor genes 2C, 7, and 1Dβ, as well as the tryptophan hydroxylase gene (the rate-limiting enzyme in the biosynthesis of 5-HT), have been negative. Phenotypically, reduced serotonergic activity, as measured by platelet ³H-paroxetine binding, has been found in both AN and BN and was unrelated to diagnostic subtype or ancillary dimensions such as impulsive behavior or depression.⁷⁰ The use of serotonin uptake inhibiting antidepressants in the treatment of AN and BN will be discussed later.

The dopaminergic system also has received attention because dopamine (DA) neurotransmission has been implicated in feeding behavior and in motor activity, amenorrhea, and distortion of body image.56,57 One study each of D3 and D4 receptor polymorphisms and food restriction in AN have been negative. One of three studies of polymorphisms in the catechol-O-methyl transferase gene (an enzyme in DA metabolism) in AN was positive, but the other two were negative.

Other candidate genes controlling hormones and proteins putatively related to food intake that have been studied in eating disorders include the estrogen receptor, the uncoupling proteins UCP-2 and UCP-3, pro-opiomelanocortin, the melanocortin-4 receptor, leptin, agouti-related protein, neuropeptide Y, cholecystokinin (CCK), the β₃-adrenergic receptor, and tumor necrosis factor. All but one have been negative.56,71 Thus, although the familial clustering of both AN and BN is clear, there is no solid evidence that single nucleotide polymorphisms in any of the aforementioned genes are other than mildly related to either of these eating disorders.

General Physical and Laboratory Findings

The general physical and laboratory findings in AN and BN are presented in Tables 4-3 and 4-4.^72,73 Metabolic changes and medical complications occur secondary to chronic starvation and malnutrition and to bingeing and purging.^7,73–75 Malnutrition-associated disturbances as severe as pulmonary bronchiectasis and emphysema have been reported.⁷⁶ Cardiac complications are also well recognized in patients with eating disorders, including the more common abnormalities of bradycardia, orthostatic hypotension, and (although less common) the risk for more serious abnormalities, including prolongation of the QT interval and silent pericardial effusion.⁷⁷

Gastrointestinal (GI) disturbances occur in both AN and BN.⁷⁸ Delayed gastric emptying and delayed colonic transit time resulting in constipation are common in AN patients and are associated with complaints of early satiety, bloating, and abdominal distension, leading the patient to feel fat and avoid eating. GI disturbances in BN include increased gastric capacity, decreased gastric relaxation, delayed gastric emptying, and abnormal function of the enteric autonomic nervous system. Abnormalities of GI function in AN and BN are reversible to various degrees with improvement in the underlying disorders. Treatments targeted at these abnormalities, such as agents to improve gastric emptying in AN and neurotransmitter modulators to reduce bingeing in BN, have met with varying success and require further study.

Increasing recognition is being given to osteopenia as an early and serious complication of AN.79–82 Adolescence is a time when optimal bone mineralization supporting physical growth is critical, and adolescent girls with AN have relatively poor bone mineral accrual. Several factors contribute to osteopenia in AN, including estrogen, androgen, and insulin-like growth factor deficiency; elevated cortisol, ghrelin, and peptide YY; excessive exercise; and nutritional deficiencies such as calcium and vitamin D. Multifaceted treatment of AN is required to restore bone mineral density to the normal range.

The electrolyte disturbances of AN and BN depend on whether purging is primarily by vomiting or by abuse of laxatives or diuretics (or both) and can be life threatening. The medical management of these cases can be complex and must be individualized. Refeeding syndrome may occur in severely malnourished patients as a result of severe shifts in fluids and electrolytes, in particular phosphate, from extracellular to intracellular spaces. These electrolyte shifts can lead to cardiovascular, neurologic, and hematologic complications and can be associated with significant morbidity and mortality.⁸³ Comerci⁷² presented detailed case histories of an anorexic and a bulimic patient, including critiques of their treatment, which highlight the intricacies of successful medical management of these disorders.

Neuroimaging

This is a growing area of eating-disorders research supported by major improvements in imaging technology.84,85 In AN, enlargement of cortical sulci and subarachnoid spaces suggestive of cerebral, and on occasion cerebellar, atrophy have been reported, the changes being positively correlated with poor neuropsychological test performance and being reversible to varying degrees with weight gain.^86–88 Reduced gray matter volume in the anterior cingulate cortex has been reported in AN compared to healthy controls, the amount of the decrease being correlated with illness severity.⁸⁹ Studies in recovered AN and BN patients have differed, showing either persistent alterations or normalization of gray- and white-matter volumes.^90,91

With regard to CNS metabolites, proton magnetic resonance spectroscopy (¹H-MRS) indicates higher ratios of choline-containing compounds to creatine and N-acetylaspartate, as well as reduced myoinositol, in the frontal white matter and occipital gray matter in association with decreased body mass index (BMI), suggesting starvation-associated increased cell membrane turnover.⁹² Consistent with these findings, ³¹P-MRS has shown altered phosphodiester and phosphomonoester peaks in malnourished anorexics, suggesting that reduced body mass alters CNS cellular membrane phospholipid metabolism.⁹³ Reduced blood flow in frontal, parietal, and frontotemporal cortex, as shown by single photon emission computed tomography (SPECT), has been reported in AN, which reverted to normal perfusion with clinical remission.⁹⁴ Another SPECT study suggested reduced blood flow in the anterior cingulate in patients with AN compared to controls, both before and after weight restoration.⁹⁵

Functional neuroimaging (fMRI) also is delineating activation of limbic and paralimbic areas that may be involved with calorie fear, including exaggerated activation of the caudate and reduced activation of the insula to taste stimuli in women recovered from AN compared to normal women.96,97 These areas also have been implicated as neural substrates of obsessive-compulsive and depressive symptoms.⁹⁸ Decreased food-stimulated activation of several cortical areas has also been shown by fMRI in chronically ill anorexics, compared to those exhibiting long-term recovery.⁹⁹ Positron emission tomography (PET) imaging has demonstrated reduced 5-HT1A and 5-HT2A receptor binding in frontal, parietal, and occipital cortex in both AN and BN patients, which may persist after recovery in both conditions.^8,100,101 This may tie in with the candidate gene studies of these receptors discussed earlier; correlative studies have yet to be undertaken.

Neuroendocrinology

Abnormal hormone profiles and responses to challenge are closely related to the “starvation” status of AN and BN patients.102–106 Hormone abnormalities also may be present (but to a lesser extent) in normal-weight women with BN. The presence of starvation in AN is evident from the weight loss, but it may not be recognized in normal-weight BN: Although bulimic women often maintain a normal weight, they do so by restricting food intake when not bingeing and purging, and they often have poorly balanced meals. Starvation-induced depletion of hepatic glycogen stores results in free fatty acids and ketone bodies replacing glucose as the primary energy source. This shift from glycogenolysis to lipolysis and ketogenesis is associated with an increase in free fatty acids and their metabolites. β-Hydroxybutyric acid levels are elevated in both AN and BN,¹⁰⁷ indicating that bulimic patients are nutritionally depleted despite their normal body weight.¹⁰⁸

The relationship of starvation and eating disorders to neuroendocrine function is most clearly seen for the pituitary-gonadal axis. As mentioned, secondary amenorrhea is one of the criteria for AN in postmenarcheal women,⁷⁴ and oligomenorrhea occurs in about 50% of bulimics. Table 4-5 lists the major endocrine disturbances that occur in AN and BN.^{103,109–112} The secondary amenorrhea is a direct result of altered gonadotropin secretion. Serum sex hormone–binding globulin may be increased, and both estrogen and testosterone are decreased.¹¹³ As indicated in Table 4-3, there are physical signs of severe estrogen deficiency. The luteinizing hormone (LH) response to LH-releasing hormone stimulation is blunted, but the follicle-stimulating hormone response is usually normal.

Fig. 4-2 illustrates circadian serum LH profiles of girls at different pubertal stages and in a 21-year-old woman with AN of 14 months’ duration whose body weight was 60% of ideal weight.¹⁰⁹ This patient’s menarche was at age 14, and during the period of weight loss, her LH profile was that of a prepubertal girl. BN patients also may have decreased gonadotropin secretion if they have lost 15% or more of their body weight, but they usually have normal circulating gonadotropins and continue their menses.

With reference to the hypothalamic-pituitary–adrenal cortical (HPA) axis (see Table 4-5), the abnormalities in AN and in reduced-weight BN^111,114,115 are strikingly similar to those occurring in 30% to 50% of patients with major depression.^111,116,117