The DSM-IV-TR criteria for establishing the diagnosis of anorexia or bulimia nervosa appear in Tables 2 and 3, respectively.

Although DSM-IV-TR criteria allow clinicians to diagnose patients with a specific eating disorder, eating disorder symptoms frequently occur along a continuum between those of anorexia nervosa and those of bulimia nervosa. Weight preoccupation and excessive self-evaluation of weight and shape are primary symptoms in both disorders, and many patients demonstrate a mixture of both anorexic and bulimic behaviors. For example, 50%–64% of patients with anorexia nervosa develop bulimic symptoms, and some patients who are initially bulimic develop anorexic symptoms (476, 477). Patients with atypical features who deny a fear of weight gain, accurately appraise their bodies as malnourished, and deny distorted perceptions of their body constituted about 20% of the patients admitted to a specialty eating disorder program (478). Denial of a fear of weight gain was found in 28% of anorexia nervosa patients assessed via a structured interview (479).

Anorexia nervosa appears in two subtypes: restricting and binge eating/purging; this classification into subtypes is based on the presence or absence of binge eating or purging symptoms. Patients with anorexia nervosa can alternate between the bulimic and restricting subtypes at different periods of their illness (480–483). Among patients with the binge eating/purging subtype of anorexia nervosa, further distinctions can be made between those who both binge and purge and those who purge but do not objectively binge. Patients with bulimia nervosa can be subclassified into the purging or nonpurging subtype. Patients with the nonpurging subtype use inappropriate methods to compensate for binge eating, including fasting and excessive exercising, as opposed to patients with binge eating disorder, who do not use inappropriate compensatory strategies. Many patients, particularly younger patients, have combinations of eating disorder symptoms that cannot be strictly categorized as anorexia or bulimia nervosa and are technically diagnosed as EDNOS (484). The value of requiring persistent amenorrhea as a criterion for diagnosing anorexia nervosa has been questioned (24).

Patients with anorexia and bulimia nervosa often experience associated psychiatric symptoms and behaviors. Social isolation is common in patients with anorexia nervosa. Depressive, anxious, and obsessional symptoms; perfectionistic traits; rigid cognitive styles; and a lack of interest in sex are often present among patients with the restricting type of anorexia nervosa (363). Early in the course of their illness, patients with anorexia nervosa often have limited recognition of their disorder and experience their symptoms as intrusive repetitive thoughts; sometimes there is a corresponding limited recognition of the disorder by patients' families. Depressive, anxious, and impulsive symptoms, as well as sexual conflicts and disturbances with intimacy, are often associated with bulimia nervosa. Although patients with bulimia nervosa are likely to recognize their disorder, shame or guilt frequently prevents them from seeking treatment for it at an early stage (485). In one subgroup of patients with bulimia nervosa (the "multi-impulsive" bulimic patients), significant degrees of impulsivity have been observed and are manifested as stealing, self-harm behaviors, suicidality, substance use, and sexual promiscuity (486, 487). Patients with anorexia nervosa of the binge eating/purging subtype may also be suicidal and engage in self-harming behaviors.

In the psychodynamic literature, patients with anorexia nervosa have been described as having difficulties with separation and autonomy (often manifested as enmeshed relationships with parents), affect regulation (including the direct expression of anger and aggression), and negotiation of psychosexual development. These deficits may make women who are predisposed to anorexia nervosa more vulnerable to cultural pressures for achieving a stereotypic body image (142, 169, 488, 489).

Psychodynamic issues in bulimic patients have been understood in a number of ways, ranging from viewing bulimic symptoms as manifestations of impulsivity or problems with emotion regulation and dissociative states to viewing them along a spectrum of self-harming behaviors commonly seen in patients with borderline personality organization (363, 490–492).

Some of the clinical features associated with eating disorders may result from malnutrition or semistarvation (493, 494). Studies of volunteers who have submitted to semistarvation experiments and semistarved prisoners of war report the development of food preoccupation, food hoarding, abnormal taste preferences, binge eating, and other disturbances of appetite regulation as well as symptoms of depression, obsessionality, apathy and irritability, and other personality changes (279). In patients with anorexia nervosa, some of these starvation-related phenomena, such as abnormal taste preference, may completely reverse with refeeding, although it may take considerable time after weight restoration for them to abate completely. However, some of these symptoms may reflect both preexisting and enduring traits, such as obsessive-compulsiveness, which are then further exacerbated by semistarvation. Such symptoms, therefore, may be only partially reversed with nutritional rehabilitation (82, 495). Complete psychological assessments may not be possible until some degree of weight restoration is achieved. Although patients with bulimia nervosa may appear to be physically within the standards of healthy weight, they may also show psychological and biological correlates of semistarvation, such as depression, irritability, and obsessionality, and may be below their personally optimum weight range, even at a weight considered to be "normal" according to population norms (496, 497). Furthermore, even at normal weight, body composition may be abnormal.

Common physical complications of anorexia nervosa are listed in Table 5. Amenorrhea of even a few months may be associated with osteopenia, which may progress to potentially irreversible osteoporosis and a correspondingly higher rate of pathological fractures (498, 499). If fracture risk is substantial, patients should be cautioned to avoid high-impact exercises. Pain in the extremities may signal stress fractures that may not be evident on X-rays but may be detected in abnormal bone scan results. Patients with anorexia nervosa who develop hypoestrogenemic amenorrhea in their teenage years that persists into young adulthood are at greatest risk for osteoporosis because they not only lose bone mass but also fail to form bone at a critical development phase (207). Osteopenia may be present in women who have been recovered from anorexia nervosa for up to 21 years (500). In addition, prepubertal and early pubertal patients are also at risk for permanent growth stunting (501, 502).

Acute complications of anorexia nervosa include dehydration, electrolyte disturbances (with purging), cardiac compromise with various arrhythmias (including conduction defects and ventricular arrhythmias), gastrointestinal motility disturbances, renal problems, infertility, premature births, other perinatal complications, hypothermia, and other evidence of hypometabolism (43). Death from anorexia nervosa is often proximally due to cardiac arrest secondary to arrhythmias (503).

Common physical complications of bulimia nervosa are listed in Table 6. The most serious physical complications occur in patients with chronic, severe patterns of binge eating and purging and are of most concern in very-low-weight patients (504).

Laboratory abnormalities in anorexia nervosa may include leukopenia with relative lymphocytosis, abnormal liver function, hypoglycemia, hypercortisolemia, hypercholesterolemia, hypercarotenemia (the latter two findings attributed to reduced catabolism), low serum zinc levels, electrolyte disturbances, and widespread disturbances in endocrine function. Low potassium levels may result from purging by any of several methods and can lead to potentially fatal cardiac arrhythmias. Sometimes abnormalities in serum chloride or bicarbonate levels precede low potassium levels. Electrolyte abnormalities can occur quickly and require ongoing monitoring in patients with extensive vomiting or laxative and/or diuretic abuse alone or in combination with low weight. In such patients, electrolyte levels should be repeated periodically to assess for abnormalities. Thyroid abnormalities may include low T₄ levels, even though thyroid-stimulating hormone levels are in the normal range; the low T₄ levels reverse with weight restoration and generally should not be treated with hormone replacement therapy (200, 505–507). Normal serum phosphorus values may be misleading because they do not reflect total body phosphorus depletion (which is usually reflected in serum phosphorus only after refeeding has begun). In early malnutrition, when many other laboratory measures may still be within normal limits, serum complement component 3 and 4 and serum transferrin may be abnormally low and serve as indicators of nutritional status (71).

Abnormal findings on magnetic resonance images reflect changes in the brain (508). White matter and cerebrospinal fluid volumes appear to return to the normal range after weight restoration. However, gray matter volume deficits, which correlate with the patient's lowest recorded BMI, may persist even after weight restoration (99, 509, 510). Some patients show persistent deficits in their neuropsychological testing results that have been shown to be associated with poorer outcomes (511).

It is important to consider that laboratory findings in anorexia nervosa may be normal in spite of a patient's profound malnutrition. For example, patients may have low total body potassium levels even when serum electrolytes are normal and thus may be prone to unpredictable cardiac arrhythmias (512).

Laboratory abnormalities in bulimia nervosa may include electrolyte imbalances such as hypokalemia, hypochloremic alkalosis, mild elevations of serum amylase (most often salivary in origin), and hypomagnesemia and hypophosphatemia, especially in patients who abuse laxatives (513–515).

Estimates of the incidence or prevalence of eating disorders vary depending on the sampling and assessment methods, and many gaps exist in our current knowledge base. The reported lifetime prevalence of anorexia nervosa among women has ranged from 0.3% for narrowly defined to 3.7% for more broadly defined anorexia nervosa (25, 516, 517). With regard to bulimia nervosa, estimates of the lifetime prevalence among women have ranged from 1% to 4.2% (516, 518, 519). Some studies suggest that the prevalence of bulimia nervosa in the United States may have decreased slightly in recent years (520), whereas the prevalence of anorexia nervosa may have increased slightly (421, 521). Eating disorders are more commonly seen among girls and women, with estimates of the male-female prevalence ratio ranging from 1:6 to 1:10 (516). The prevalence of anorexia nervosa and bulimia nervosa in American children and younger adolescents is not well documented.

In the United States, eating disorders appear to be about as common in young Hispanic and Native American women as in Caucasian women and less common among African American and Asian women (522–524). Although studies have shown that preadolescent African American girls report a higher drive for thinness than Caucasian girls (525, 526), the drive for thinness increases significantly in Caucasian girls during puberty and remains unchanged in African American girls (527). Disordered eating is prevalent in many other countries. In a Scandinavian study of girls and boys ages 14–15 years, 0.7% of the girls and 0.2% of the boys reported a lifetime prevalence of anorexia nervosa, and 1.2% of the girls and 0.4% of the boys reported a lifetime prevalence of bulimia nervosa (528). Studies in Japan suggest that the prevalence of eating disorders is on the rise there. Recent data indicate that >50% of female college students report a history of significant and persistent dieting, 40% use diet pills or drinks to lose weight, and 18% report a BMI <18.5 kg/m² (529). However, the latter finding requires cautious interpretation, because appropriate BMI ranges might vary by ethnic grouping; for example, the normal range of BMI might actually be lower in Asian populations than in North American and European populations (530). Eating disorder concerns and symptoms do appear to be increasing among Chinese women exposed to Western culture and modernization in cities such as Hong Kong (531–533). The prevalence of disturbed eating disorders attitudes, as assessed by surveys, also appears to be high in other non-Western countries such as Iran, nonwhite South Africa, and Fiji (429, 534, 535).

First-degree female relatives of patients with anorexia and bulimia nervosa have higher rates of eating disorders compared with relatives of control subjects (536–539). In addition, relatives of individuals with anorexia and bulimia nervosa have increased rates of eating disorders that do not meet full diagnostic criteria compared with relatives of control subjects (538, 539). Identical twin siblings of patients with anorexia or bulimia nervosa also have higher rates of these disorders, with monozygotic twins having higher concordance than dizygotic twins. Families of patients with bulimia nervosa have been found to have higher rates of substance abuse (particularly alcohol use disorders), affective disorders, and certain personality traits, including elevated levels of perfectionism and an increased sense of ineffectiveness (540, 541). In Fiji, the prevalence of binge eating disorder is comparable to that in the United States (542).

High rates of co-occurring psychiatric illness are found in patients seeking treatment at tertiary-level psychiatric treatment centers. Lifetime co-occurring major depression or dysthymia has been reported in 50%–75% of patients with anorexia (323, 324) and bulimia (324, 331) nervosa. Estimates of the prevalence of bipolar disorder among patients with anorexia or bulimia nervosa are usually around 4%–6% but have been reported to be as high as 13% (325). The lifetime prevalence of OCD among anorexia nervosa patients has been as high as 25% (82, 323, 543), with OCD frequently predating the onset of anorexia nervosa (341, 461). Obsessive-compulsive symptomatology has been found in a large majority of weight-restored patients with anorexia nervosa treated in tertiary-level care centers (544). OCD is also common among patients with bulimia nervosa (82, 331, 543). Co-occurring anxiety disorders, particularly social phobias, are common among patients with anorexia and bulimia nervosa (82, 310, 323, 331, 543). Substance abuse has been found in as many as 23%–40% of patients with bulimia nervosa. Among patients with anorexia nervosa, estimates of those with substance abuse have ranged from 12% to 18%, with this problem occurring primarily among those with the binge eating/purging subtype (308, 310, 323, 545).

Co-occurring personality disorders are frequently found among patients with eating disorders, with estimates ranging from 42% to 75%. The associations between bulimia nervosa and Cluster B and C disorders (particularly borderline personality disorder and avoidant personality disorder) and between anorexia nervosa and Cluster C disorders (particularly avoidant personality disorder and obsessive-compulsive personality disorder) have been reported (546, 547). Eating disorder patients with personality disorders are more likely than those without personality disorders to also have concurrent mood or substance use disorders (308, 331). Co-occurring personality disorders are significantly more common among patients with the binge eating/purging subtype of anorexia nervosa than among patients with the restricting subtype or in normal-weight patients with bulimia nervosa (349).

Sexual abuse has been reported in 20%–50% of patients with bulimia (346) and anorexia (221, 548) nervosa, although sexual abuse may be more common in patients with bulimia nervosa than in those with the restricting subtype of anorexia nervosa (346, 549). Childhood sexual abuse histories are reported more often in women with all psychiatric disorders, including eating disorders, than in women from the general population (549). Women who have eating disorders in the context of sexual abuse appear to have higher rates of comorbid psychiatric conditions than other women with eating disorders (314, 346). Furthermore, individuals with bulimia nervosa are reported to have experienced higher rates of other types of trauma besides childhood sexual abuse, including adult rape and molestation, aggravated assault, and physical neglect (332, 550, 551).

Although the overall percentage of individuals who fully recover from anorexia nervosa is modest, it is well established that younger patients who receive prompt and appropriate intervention have a much better full recovery rate. For example, in the study by Strober et al. (19), >70% of adolescents had a full and lasting recovery 5 years after the onset of comprehensive treatment. Although some patients improve symptomatically over time, a substantial proportion continue to have body image disturbances, disordered eating, and other psychiatric difficulties (163, 324, 552). In one 10-year follow-up study, a relapse rate of 42% was seen during the first posthospitalization year for patients with anorexia nervosa (553). A review of a large number of studies of patients who were hospitalized or who received tertiary-level care and were followed up at least 4 years after the onset of illness indicates that "good" outcomes occurred in 44% of the patients (i.e., weight restored to within 15% of recommended weight for height and regular menstruation established), although these criteria are clearly insufficient to consider a patient as recovered or even as having restored weight to an adequate level. Poor outcomes occurred in about 24% (weight never reached within 15% of recommended weight for height; menstruation absent or at best sporadic), and intermediate outcomes occurred in about 28% (163). Approximately 5% of the patients died. Overall, about two-thirds of anorexia nervosa patients continue to have enduring morbid food and weight preoccupation, and up to 40% have bulimic symptoms. Even among those who have good outcomes as defined by restoration of weight and menses, many have other persistent psychiatric symptoms, including dysthymia, social phobia, obsessive-compulsive symptoms, and substance abuse (323, 554).

Among adolescents with anorexia nervosa, approximately 50%–70% recover, 20% are improved but continue to have residual symptoms, and 10%–20% develop chronic anorexia nervosa (163). In a 10- to 15-year follow-up study of adolescent patients hospitalized for anorexia nervosa—76% of whom met criteria for full recovery—time to recovery was quite protracted, ranging from 57 to 79 months depending on the definition of recovery (19, 478). Anorexia nervosa patients with atypical features, such as denying a fear of gaining weight or denying distorted perceptions of their bodies, had a somewhat better course (478). Although good outcomes were observed in only 35% of 80 patients in Eisler et al.'s 5-year follow-up study (155), outcomes were good in 62% of the 21 patients who had been ill for <3 years and whose illness began before age 19.

Diagnostic migration occurs in patients with anorexia nervosa, reflecting the development of binge eating and/or purging behavior. The most frequent change among diagnostic categories is from anorexia nervosa, restricting type, to anorexia nervosa, binge eating/purging type; most changes occur by the fifth year after the onset of illness (477, 553). In one study, >50% of patients with anorexia nervosa, restricting type (both adolescents and adults) developed bulimic symptomatology over the course of follow-up, and only a small fraction of patients with anorexia nervosa, restricting type remained in that diagnostic subtype (555). Factors leading to the development of bulimic symptoms among patients with anorexia nervosa, restricting type are not well understood, nor is the precise time course of this development.

Mortality rates in eating disorders, specifically anorexia nervosa, are among the highest in the mental disorders. The prognosis of anorexia nervosa does not appear to have improved during the 20th century (163, 556, 557). Harris and Barraclough (558) calculated the standardized mortality ratios (SMRs) for all causes of death in 152 English language reports from a MEDLINE search on the mortality of mental disorder. The highest risk of premature death from natural and unnatural causes was related to eating disorders and substance abuse. Another study analyzing 10 large samples of individuals with eating disorders found strong evidence for an elevated SMR in eight of these samples, with a definitely elevated SMR for anorexia nervosa and no conclusion for bulimia nervosa. Lower weight at presentation was associated with a higher SMR. Mortality also varied with age at presentation, with an SMR of 3.6 for those presenting under age 20 years; 9.9, for ages 20–29 years; and 5.7, for age 30 years or older. Among female patients, the risk of death was 0.59% per year (559).

Deaths among male patients from anorexia nervosa have also been studied. In a recent report, two national registers, the National Patient Register (NPR) and the Causes of Death Register (CODR), were examined in Norway for deaths related to anorexia nervosa that occurred during a 9-year period (1992–2000). The medical record or death certificate listed anorexia nervosa as a diagnosis or cause of death for 66 individuals. Rates of death related to anorexia nervosa were 6.46 and 9.93 per 100,000 deaths for the NPR and the CODR, respectively. A substantial percentage of deaths (43.9%) in both registers occurred at or above age 65 years. For the NPR, the mean age at the time of death was 61 years, and 31% of deaths occurred among men. For the CODR, the mean age at the time of death was 49 years, and 18% of deaths occurred among men (560).

In other analyses, approximately 5.6% of patients diagnosed with anorexia nervosa die per decade of illness (561), and female anorexia nervosa patients are reportedly 12 times more likely to die than women of a similar age in the general population (321). The most common causes of death are suicide and starvation-related effects. The suicide rate among women with anorexia nervosa is up to 57 times higher than that for women of a similar age in the general population (321). Lower weight at presentation, longer duration of illness, and severe alcohol use appear to be associated with higher risk of mortality (321, 562).

Nielsen (563) conducted a literature review of mortality studies in eating disorders and concluded that methodological problems created biases to the eating disorder mortality data. The major problems with these studies were small sample sizes and loss of patients to follow-up. Mortality and morbidity for anorexia nervosa, bulimia nervosa, and related disorders are likely to be underreported because they go unrecognized by clinicians. Patients' denial of illness may result in their avoidance of treatment at an early phase and the later development of multiple chronic physical problems, with associated morbidity and mortality (321, 563).

A shorter duration of illness and younger age at onset have been associated with a better outcome; lower initial minimum weights, vomiting, binge eating, purgative abuse, chronicity of illness, and obsessive-compulsive personality symptoms are reported to be unfavorable prognostic features (163). However, many of these prognostic indicators have not been consistently replicated and may be more reliable predictors of short-term but not long-term outcomes. In general, adolescents have better outcomes than adults and younger adolescents have better outcomes than older adolescents.

Although the literature on the long-term course and prognosis of bulimia nervosa remains limited, studies over the last decade have begun to clarify these issues. First, studies have shown that in untreated community samples, there are modest degrees of spontaneous improvement over a 1- to 2-year period, with roughly a 25%–30% reduction in binge eating, purging, and laxative abuse (564, 565). The overall short-term success rate for patients receiving psychosocial treatment or medication has been reported to be 50%–70% (324). Relapse rates of 30%–85% have been reported for successfully treated patients at 6 months to 6 years of follow-up (329, 566).

In a 5-year period, most individuals with bulimia nervosa in the community continue to have some form of an eating disorder of clinical severity, with about 33% remitting each year and another 33% relapsing to full diagnostic criteria, which suggests a relatively poor prognosis for this untreated group (567). In a naturalistic longitudinal study of 110 treatment-seeking women with bulimia nervosa, 73% achieved full recovery (no bingeing or purging for at least 8 consecutive weeks) at some point during a median of 7 years of follow-up, and 36% of those relapsed (568). A 6-year follow-up of patients treated for bulimia nervosa found that 60% of the patients were rated as having a good outcome, 29% as having an intermediate outcome, and 10% as having a poor outcome; 1% were reported as having died (569). A review of the treatment literature by this same group (570) found that, over time, social adjustment tended to normalize in some patients but that a fairly large group experienced chronic symptomatology and impairment; there was little crossover to anorexia nervosa or binge eating disorder. The longest follow-up study to date (562), with a mean follow-up of 11.5 years, found that the number of women who continued to meet full diagnostic criteria for bulimia nervosa declined over time. At long-term follow-up, 30% continued to engage in recurrent binge eating and purging behaviors. Subsequent analysis of this data set concluded that although menstrual irregularities were common at follow-up, the baseline presence of illness appeared to have little impact on these patients' later ability to achieve pregnancy (571). The results of this follow-up were interpreted to indicate that treatments with demonstrated efficacy for short-term outcome appeared to improve psychosocial functioning at long-term outcome among women with bulimia nervosa (572). A review of other literature in this area concluded that no consistent evidence exists to support the idea that early intervention implies a better long-term outcome (573).

A variety of factors have been examined as possible predictors of outcome. The available literature suggests that outcomes for patients with illness onset in adolescence are better than for those with later onsets (556). Although the data are highly variable, evidence suggests that comorbidity with OCD may be associated with a longer duration of illness (574) and that comorbidity with personality disorders may alter the natural course of illness (575). Overevaluation of shape and weight and a history of childhood obesity may be negative predictor factors (576), whereas a history of substance use disorders at intake or misuse of laxatives during the follow-up period may predict suicide attempts (577). The overall conclusion is that considerable variability occurs in the natural course of this illness, with persistence of symptoms at long-term follow-up in a significant subgroup of patients.

EDNOS is a commonly used diagnosis, being given to >50% of patients with eating disorders who present to outpatient treatment settings (263). EDNOS variants consisting of mixtures of anorexia and bulimia nervosa symptoms appear to be particularly common among adolescents. This heterogeneous group of patients consists largely of subsyndromal cases of anorexia or bulimia nervosa (e.g., those who fail to meet one criterion, such as not having 3 months of amenorrhea or having fewer binge eating episodes per week than required for a strictly defined diagnosis) as well as the substantial group of patients with binge eating disorder.

Because the diagnosis of EDNOS includes individuals with diverse eating disorder presentations, it is predictable that the course of EDNOS will be highly variable. Indeed, an early study of an unselected EDNOS population found a varied course of illness and low rate of recovery over 30 months (267). In addition, in patients with a variety of eating disorders who were followed over time, it appears that considerable movement occurred from one eating disorder diagnostic category to another, including EDNOS (263).

Binge eating disorder occurs in about 2% of community cohorts and is common among patients seeking treatment for obesity at hospital-affiliated weight programs (1.3%–30.1% prevalence), with studies using more rigorous interview-based measures typically reporting lower rates (578, 579). About 33% of these patients are male. Binge eating disorder typically begins in adolescence (at least by retrospective recall) or early adulthood and occurs more frequently in adults than in adolescents, but patients generally do not present for treatment until adulthood. (580). A well-established concomitant feature of binge eating disorder is that obese individuals who binge eat are more likely than those who do not binge eat to display comorbid axis I psychopathology, particularly major depressive disorder, with lifetime rates of 46%–58% (313, 334, 335, 337, 581).

Important observations have been made regarding the course of binge eating disorder. A 5-year community study of young women with binge eating disorder reported that a majority of the women had recovered spontaneously by 5-year follow-up. However, the age of participants in this study was considerably younger than that of most patients presenting for binge eating disorder treatment, making the generalizability of these findings uncertain (567). Another community study that followed patients over a 6-month period reported that about half of patients remaining in the study continued to meet binge eating disorder criteria, whereas symptoms of the other half partially remitted (567, 582). A 6-year study (583) that followed intensively treated binge eating disorder patients found that approximately 57% had a good outcome, 35% an intermediate outcome, and 6% a poor outcome; 1% of the patients had died. Although shorter-term remission is not necessarily maintained on a longer-term basis, clinical samples and shorter-term studies of binge eating disorder treatment have often reported high rates of response to minimal interventions (e.g., placebo) (584). Taken together, these lines of evidence suggest that the course of binge eating disorder is rather unstable over time. Treatment appears to be associated with a fairly positive long-term response, but it is difficult to know how many patients might have recovered without specific treatment.

The presence of binge eating may be predictive of weight gain over time. The aforementioned study of Fairburn et al. (567) reported that the prevalence of obesity in that group of patients had nearly doubled by the end of the follow-up period. Follow-up data from several treatment studies (271, 272, 585, 586) suggest that the persistence of binge eating may be associated with weight gain over time.

Family and twin studies suggest a strong genetic component in the development of anorexia and bulimia nervosa (587–589), but the specifics of exactly what vulnerabilities are transferred and the mechanisms whereby they contribute to the pathogenesis of eating disorders need to be identified. The evidence also suggests that anorexia and bulimia nervosa may share genetic transmission with anxiety disorders and major depression (590, 591).

Further investigation of genetic contributions to vulnerability for eating disorders has occurred with two types of analyses: linkage studies and association studies for polymorphisms of specific genes. Evidence from a large international, multisite study suggests the presence of an anorexia nervosa susceptibility locus on chromosome 1p (592) and a susceptibility locus for bulimia nervosa on chromosome 10p (593). In affected sibling pairs who ranked high for "drive for thinness" and "obsessionality" traits, suggestive linkages were found on chromosomes 1, 2, and 13 (594). Association studies for polymorphisms of specific genes with specific behavioral covariates have produced many contradictory findings. For example, four studies were positive for a polymorphism of –1438 G/A in the promoter 5HT2A gene, and three studies were negative for this polymorphism (reviewed by Hinney et al. [595]). A meta-analysis of all the association studies of the 5HT2A gene in anorexia nervosa showed a persistent significant effect of the –1438 allele (596).

Preliminary evidence suggests that the norepinephrine transporter gene (NET) and monoamine oxidase A gene (MAOA) contribute to the increased risk for anorexia nervosa, restricting type. A serotonin transporter gene (SERT), known to be associated with anxiety, is preferentially transmitted to children with anorexia nervosa when the more active MAOA variant is also transmitted (597). The findings regarding these three genes (MAOA, SERT, and NET) in relation to susceptibility to anorexia nervosa require replication. Other studies suggest significant associations between anorexia nervosa and the serotonin gene HTR1D and the opioid gene OPRD1 (598).