Anorexia Nervosa: Neurobiological Basis, Medical Complications, Family-Based Treatment (FBT), CBT-E, and Refeeding Protocols

Anorexia nervosa (AN) is a severe psychiatric disorder characterized by persistent restriction of energy intake, intense fear of gaining weight, and disturbance in the experience of body weight or shape. Among all psychiatric illnesses, AN carries the highest standardized mortality ratio (SMR), estimated at 5.86 in a comprehensive meta-analysis by Arcelus et al. (2011), with approximately 5–6% of affected individuals dying per decade of illness. This mortality reflects both the direct medical consequences of starvation and a significantly elevated suicide rate — roughly 18% of deaths in AN are attributable to suicide.

The DSM-5-TR distinguishes two subtypes: the restricting type (AN-R), in which weight loss is accomplished primarily through dieting, fasting, or excessive exercise, and the binge-eating/purging type (AN-BP), in which the individual engages in recurrent episodes of binge eating or purging behaviors such as self-induced vomiting or laxative misuse. This subtyping carries clinical significance: AN-BP is associated with greater psychiatric comorbidity, higher impulsivity, and worse medical prognoses compared to AN-R.

Epidemiological data from the DSM-5-TR estimate the 12-month prevalence of AN among young females at approximately 0.4%. Lifetime prevalence estimates range from 0.9% to 4.0% in women and 0.1% to 0.3% in men, depending on the diagnostic threshold applied. Incidence rates peak sharply between ages 15 and 19, with studies from Western European registries (e.g., the UK General Practice Research Database) suggesting an incidence of approximately 13.7 per 100,000 person-years in females aged 15–19. Importantly, AN prevalence appears relatively stable across recent decades despite increased detection, though there is growing recognition of cases in males, older adults, and non-Western populations previously thought to be minimally affected.

The gender ratio is approximately 10:1 female-to-male, although emerging data suggest this ratio narrows when broader diagnostic criteria or subclinical presentations are considered. Males with AN are underdiagnosed, present later, and may exhibit atypical features such as a focus on muscularity rather than thinness, complicating recognition in clinical settings.

The DSM-5-TR diagnostic criteria for anorexia nervosa (307.1 / ICD-10: F50.0) require three core features:

• Criterion A: Restriction of energy intake relative to requirements, leading to a significantly low body weight in the context of age, sex, developmental trajectory, and physical health.
• Criterion B: Intense fear of gaining weight or of becoming fat, or persistent behavior that interferes with weight gain, even though at a significantly low weight.
• Criterion C: Disturbance in the way in which one's body weight or shape is experienced, undue influence of body weight or shape on self-evaluation, or persistent lack of recognition of the seriousness of current low body weight.

The DSM-5-TR specifies severity based on BMI: mild (BMI ≥ 17 kg/m²), moderate (BMI 16–16.99), severe (BMI 15–15.99), and extreme (BMI < 15). However, clinicians must exercise caution: BMI thresholds are arbitrary and fail to capture medical acuity in many cases. A patient who has lost 30% of body weight over three months but remains at a BMI of 17.5 may be far more medically unstable than a patient with chronic AN and a BMI of 14 who has been weight-stable for years. Rate and trajectory of weight loss are often more clinically informative than absolute BMI.

The ICD-11 (6B80) offers a somewhat more flexible framework, replacing rigid BMI cutoffs with the concept of "significantly low body weight" that is not attributable to another medical condition or substance, while maintaining the core psychopathological features.

Differential diagnosis requires careful exclusion of:

• Medical causes of weight loss: Hyperthyroidism, Addison's disease, celiac disease, inflammatory bowel disease, malignancy, and superior mesenteric artery (SMA) syndrome. A basic laboratory panel (CBC, CMP, TSH, ESR/CRP) and clinical history typically distinguish these conditions, though AN and medical illness can co-occur.
• Avoidant/Restrictive Food Intake Disorder (ARFID): Distinguished from AN by the absence of weight/shape-related psychopathology. Patients with ARFID restrict intake due to sensory sensitivity, fear of aversive consequences (e.g., choking, vomiting), or apparent lack of interest in eating — not because of body image disturbance.
• Major Depressive Disorder: Appetite suppression and weight loss occur commonly in depression, but the driven quality of food restriction, body image distortion, and fear of weight gain are absent.
• Somatic Symptom Disorder and Illness Anxiety Disorder: Patients may restrict eating due to somatic preoccupation, but the core AN psychopathology is absent.
• Bulimia Nervosa: If the individual is at a significantly low body weight and engages in binge-purge behaviors, the diagnosis is AN binge-eating/purging type, not bulimia nervosa. BN is diagnosed only when weight is not significantly low.

A common diagnostic pitfall is atypical anorexia nervosa (coded under Other Specified Feeding or Eating Disorder, OSFED), in which all criteria for AN are met except the individual's weight falls within or above the normal range despite significant weight loss. Research — including studies from the REAL (Registry of Eating and Allied Disorders) — demonstrates that patients with atypical AN can present with equivalent or even worse medical instability (bradycardia, orthostatic hypotension, electrolyte disturbances) compared to those meeting full weight criteria. Clinical severity should never be dismissed based on BMI alone.

The neurobiology of AN involves complex interactions among serotonergic, dopaminergic, and opioidergic neurotransmitter systems, with dysregulation of cortico-striatal and insular circuits underlying the disorder's characteristic aberrations in reward processing, interoception, and cognitive flexibility.

Serotonin (5-HT) System

The serotonin system is the most extensively studied neurochemical system in AN. Research by Walter Kaye and colleagues has demonstrated that individuals with AN — both acutely ill and weight-restored — exhibit alterations in 5-HT_1A and 5-HT_2A receptor binding. PET imaging studies using [¹¹C]WAY-100635 have shown increased 5-HT_1A receptor binding in cortical and limbic regions in recovered AN patients, suggesting a trait-related alteration rather than a state effect of malnutrition. The Kaye model posits that individuals with AN have a baseline state of elevated serotonergic tone, which produces chronic dysphoria and anxiety. Dietary restriction paradoxically reduces tryptophan availability (the precursor to serotonin), thereby lowering 5-HT activity and temporarily alleviating aversive mood states. This creates a powerful negative reinforcement cycle: starvation becomes anxiolytic, reinforcing further restriction.

Dopamine and Reward Processing

Dopaminergic dysfunction in AN centers on altered reward processing within the mesolimbic and mesocortical pathways. Functional MRI studies consistently demonstrate that patients with AN show reduced ventral striatal activation (particularly the nucleus accumbens) in response to food cues and palatable tastes, but paradoxically increased activation in the dorsal caudate — a region involved in habit-based decision making. This pattern suggests a shift from flexible, reward-driven eating toward rigid, habitual dietary restriction. PET studies have shown elevated D2/D3 receptor binding in the dorsal striatum of recovered AN patients, which may reduce the subjective experience of reward from food and contribute to the trait-level anhedonia frequently observed.

Interoception and the Insular Cortex

The insula — critical for processing internal bodily signals including hunger, satiety, and visceral sensations — shows consistent functional abnormalities in AN. Patients demonstrate reduced interoceptive accuracy and atypical insular activation during tasks requiring detection of bodily states. Neuroimaging studies reveal hypoactivation of the anterior insula during taste processing and altered connectivity between the insula and prefrontal regions. This interoceptive deficit may explain the well-documented clinical observation that patients with AN report inability to detect hunger signals, even at dangerously low weights.

Prefrontal–Striatal Circuitry and Cognitive Control

AN is associated with a pronounced imbalance between frontal cognitive control systems (dorsolateral prefrontal cortex, DLPFC; dorsal anterior cingulate cortex, dACC) and limbic/ventral striatal reward systems. Hyperactivation of frontal control regions during food-related decisions, combined with hypoactivation of reward circuits, produces the phenotype of excessive self-control and denial of appetitive signals. Set-shifting deficits — poor cognitive flexibility — are among the most replicated neuropsychological findings in AN, persisting even after weight restoration, suggesting they represent an endophenotype rather than a consequence of malnutrition.

Genetics and Heritability

Twin studies consistently estimate the heritability of AN at 50–60%, comparable to bipolar disorder and schizophrenia. The largest genome-wide association study (GWAS) to date, conducted by the Psychiatric Genomics Consortium Eating Disorders Working Group (Watson et al., 2019; published in Nature Genetics), identified eight significant loci associated with AN. Crucially, this study revealed that AN has significant positive genetic correlations with obsessive-compulsive disorder (OCD), major depression, anxiety disorders, and schizophrenia — but also negative genetic correlations with BMI, fat mass, and metabolic traits (including type 2 diabetes and fasting insulin). This finding has prompted reconceptualization of AN as a metabo-psychiatric disorder, in which metabolic and psychiatric genetic risk factors converge. Specific candidate genes implicated include those involved in serotonergic function (HTR1D, HTR2A), brain-derived neurotrophic factor (BDNF), and opioid receptor systems (OPRD1).

Neuroendocrine and Neuropeptide Dysregulation

Starvation in AN produces widespread neuroendocrine disruption: elevated cortisol (hypothalamic–pituitary–adrenal axis dysregulation), suppressed gonadal hormones (functional hypothalamic amenorrhea), reduced leptin, elevated ghrelin, and alterations in peptide YY and cholecystokinin. The suppression of leptin — an adipokine signaling energy sufficiency — contributes to hyperactivity, amenorrhea, and bone loss. Some of these endocrine abnormalities normalize with weight restoration, but others (e.g., cortisol dysregulation, bone mineral density loss) may persist for years.

Anorexia nervosa produces medical complications affecting virtually every organ system, many of which are potentially life-threatening. Understanding these complications is essential for acute medical management and long-term monitoring.

Cardiovascular

Cardiac complications are the leading cause of medical death in AN. Sinus bradycardia (heart rate < 60 bpm, often < 40 bpm in severe cases) is nearly universal. QTc prolongation — driven by electrolyte disturbances (hypokalemia, hypomagnesemia, hypophosphatemia) and direct myocardial effects of starvation — increases the risk of torsades de pointes and sudden cardiac death. Structural changes include reduced left ventricular mass, pericardial effusion (present in up to 35% of severely ill patients), and mitral valve prolapse. The ipecac cardiomyopathy, though rarer with declining ipecac availability, remains a known complication in purging subtype.

Gastrointestinal

Gastroparesis (delayed gastric emptying) occurs in approximately 50% of patients with AN and contributes to early satiety, bloating, and nausea that perpetuate food avoidance. Superior mesenteric artery (SMA) syndrome — compression of the duodenum between the aorta and SMA due to loss of the intervening fat pad — can cause complete gastric outlet obstruction. Constipation is almost universal. Elevated hepatic transaminases occur in both the starvation state (autophagic hepatitis) and during refeeding. Acute pancreatitis can occur during refeeding.

Endocrine and Metabolic

Hypothalamic amenorrhea results from suppression of pulsatile GnRH secretion and is a hallmark of the disorder, though the DSM-5-TR removed amenorrhea as a required criterion to accommodate males and pre-menarchal/post-menopausal females. Bone mineral density loss is severe: up to 85% of women with AN have osteopenia and up to 40% meet criteria for osteoporosis. The mechanism involves estrogen deficiency, hypercortisolism, IGF-1 suppression, and elevated sclerostin. Bone loss may be only partially reversible with weight restoration, and fracture risk remains elevated long after recovery. Low T3 syndrome (euthyroid sick syndrome) is common and represents an adaptive metabolic response; it does not warrant thyroid hormone replacement.

Hematologic

Gelatinous marrow transformation (serous fat atrophy) occurs in severe AN, producing pancytopenia — leukopenia, anemia, and thrombocytopenia. Leukopenia with absolute neutrophil counts below 1,500/µL is present in approximately 30–40% of hospitalized AN patients. Despite low white cell counts, infections are relatively uncommon until extremely low weights, likely due to preserved innate immunity.

Neurological

Structural neuroimaging consistently demonstrates cortical gray matter volume reduction and ventricular enlargement (pseudoatrophy) in the malnourished state. While gray matter volume partially recovers with weight restoration, white matter recovery may lag, and some deficits may persist. Peripheral neuropathy can occur due to vitamin deficiencies (B12, thiamine). Seizures may result from severe hypoglycemia or electrolyte disturbances.

Dermatologic

Lanugo hair (fine, downy hair on trunk and extremities) reflects the body's thermoregulatory response to fat depletion. Russell's sign — calluses or scarring on the dorsum of the hand from self-induced vomiting — is pathognomonic for purging behavior. Acrocyanosis, dry skin, hair loss, and brittle nails are common.

Family-Based Treatment (FBT), often referred to as the Maudsley Approach, is the most empirically supported intervention for adolescents with AN and is recommended as the first-line treatment by NICE, APA, and most international eating disorder guidelines. Developed at the Maudsley Hospital in London, FBT is a manualized outpatient treatment (Lock & Le Grange, 2013) that proceeds through three phases over approximately 9–12 months (typically 15–20 sessions).

Structure and Phases

• Phase 1 (Weight Restoration): Parents are empowered to take temporary control of their child's eating. The therapist externalizes the illness, supports parental authority, and orchestrates a family meal session in which the therapist coaches the family in encouraging the patient to eat. The emphasis is on action — parents must feed their child — and on reducing parental guilt and blame.
• Phase 2 (Returning Control to the Adolescent): Once the patient is gaining weight steadily and eating with reduced resistance, control over food is gradually transferred back to the adolescent in an age-appropriate manner.
• Phase 3 (Adolescent Issues and Termination): Focus shifts to general adolescent developmental issues (identity, autonomy, peer relationships) and relapse prevention.

Empirical Evidence

The landmark randomized controlled trial by Lock et al. (2010) — a multisite study comparing FBT to Adolescent-Focused Individual Therapy (AFT) in 121 adolescents with AN — demonstrated that FBT produced significantly higher full remission rates at end of treatment (42% vs. 23%) and at 12-month follow-up (49% vs. 23%). Full remission was defined as achieving ≥ 95% of expected body weight (EBW) plus an Eating Disorder Examination (EDE) score within 1 SD of community norms.

A Cochrane systematic review (Fisher et al., 2019) concluded that family therapy approaches are superior to individual therapy for adolescents with AN at 6–12 months post-treatment, with a relative risk of remission favoring family-based approaches (RR approximately 1.5–2.0 depending on the comparison). However, the evidence base remains relatively small, with most trials involving fewer than 150 participants, and there is significant heterogeneity in outcome definitions across studies.

Moderators and Limitations

FBT is most effective when initiated early in the illness course (duration < 3 years) and when obsessive-compulsive comorbidity is absent or well-managed. Predictors of poor FBT response include: higher baseline eating disorder psychopathology, the binge-purge subtype, co-occurring psychiatric disorders (particularly OCD and depression), single-parent households, and marked parental criticism/hostility (high expressed emotion). Approximately 30–50% of adolescents do not achieve full remission with FBT alone, highlighting the need for alternative or augmentative approaches.

FBT has limited direct evidence for adults with AN, though adaptations for young adults (18–25 years) are being tested. The model's reliance on parental involvement makes it inapplicable for many adult patients.

Enhanced Cognitive Behavioral Therapy (CBT-E), developed by Christopher Fairburn, is a transdiagnostic treatment originally designed for all eating disorders. The "enhanced" framework addresses not only the core eating disorder psychopathology (dietary restraint, overconcern with weight/shape/eating) but also four maintenance mechanisms that operate across eating disorder diagnoses: clinical perfectionism, core low self-esteem, interpersonal difficulties, and mood intolerance. CBT-E for AN (the "broad" version) is delivered over 40 sessions across approximately 40 weeks and includes a focused module on weight regain.

Evidence for CBT-E in AN

Fairburn et al. (2013) published a cohort study of 99 patients with AN treated with CBT-E, demonstrating that approximately 60% of treatment completers achieved a BMI ≥ 18.5 by end of treatment, with gains maintained at 60-week follow-up. However, the intent-to-treat completion rate was lower — approximately 64% completed treatment — meaning the overall remission rate on an intent-to-treat basis was approximately 38–40%.

The most important head-to-head comparison for adults is the ANTOP trial (Zipfel et al., 2014), a large multicenter RCT conducted in Germany that randomized 242 adult outpatients with AN to focal psychodynamic therapy, CBT-E, or optimized treatment as usual (TAU). At end of treatment, BMI gains did not differ significantly among the three groups (approximately 0.9–1.4 kg/m² gain). At 12-month follow-up, focal psychodynamic therapy showed a slight advantage in global outcome (recovery rate of 35% vs. 29% for CBT-E and 13% for TAU), though the clinical significance of differences between the active treatments was debatable. The ANTOP trial's key finding was that no single psychotherapy demonstrated clear superiority for adult AN, underscoring the treatment-resistant nature of the disorder.

Other Therapeutic Approaches

Specialist Supportive Clinical Management (SSCM) — a combination of clinical management (nutritional counseling, weight monitoring, psychoeducation) with supportive therapeutic techniques — has performed comparably to structured therapies in several trials (McIntosh et al., 2005; the landmark Christchurch study). SSCM's strong performance has been interpreted as evidence that the nonspecific therapeutic factors (therapeutic alliance, consistent monitoring, expert guidance) may be as important as specific psychotherapeutic techniques in AN.

Maudsley Model of Anorexia Nervosa Treatment for Adults (MANTRA), developed by Ulrike Schmidt and colleagues, addresses the maintaining mechanisms of AN through a cognitive-interpersonal framework targeting rigid thinking styles, avoidance of emotions, pro-anorexia beliefs, and unhelpful responses from close others. The MOSAIC trial (Schmidt et al., 2015) compared MANTRA to SSCM in 142 adult outpatients with AN and found no significant difference in BMI outcomes at 12 months, though MANTRA showed some advantages in eating disorder psychopathology and patient satisfaction.

The overall picture for adult AN treatment is sobering: no single psychotherapy produces full remission in more than approximately 30–40% of patients on an intent-to-treat basis, dropout rates range from 20–40%, and head-to-head trials consistently fail to show clear superiority of one approach over another.

In contrast to bulimia nervosa and binge-eating disorder, pharmacotherapy has not demonstrated consistent efficacy for the core symptoms of AN. This represents one of the most significant unmet needs in eating disorder treatment.

Antidepressants

SSRIs are frequently prescribed in clinical practice but lack compelling evidence for weight restoration or relapse prevention in AN. The landmark study by Walsh et al. (2006) — a double-blind RCT of fluoxetine versus placebo following weight restoration — found no difference in time to relapse between groups. The prevailing neurobiological explanation is that serotonergic medications require adequate tryptophan availability (derived from dietary protein) to function, and in the malnourished state, tryptophan depletion renders SSRIs ineffective. Tricyclic antidepressants carry unacceptable cardiac risk in malnourished patients (QTc prolongation, arrhythmias).

Atypical Antipsychotics

Olanzapine is the most studied antipsychotic in AN, with its appetite-stimulating and anxiolytic properties providing theoretical rationale. A meta-analysis by Dold et al. (2015) found that olanzapine produced small but statistically significant BMI gains compared to placebo (weighted mean difference approximately 0.7 kg/m²), but effects on eating disorder psychopathology were inconsistent. The largest RCT (Attia et al., 2019) randomized 152 patients with AN to olanzapine (mean dose 7.8 mg) versus placebo and confirmed a modest but significant weight advantage for olanzapine (approximately 0.7 kg/m² greater BMI increase) with no significant improvement in psychological symptoms. Olanzapine's role is thus best characterized as a modest adjunct — NNT estimates for clinically meaningful weight gain are in the range of 5–8 — and must be weighed against metabolic side effects.

Other Agents

Zinc supplementation (14 mg elemental zinc daily) has shown modest benefits in accelerating weight gain in several small trials. D-cycloserine, a partial NMDA receptor agonist used as a cognitive enhancer in exposure therapy, showed preliminary promise for AN-related food anxiety but has not been replicated in larger studies. Dronabinol (synthetic THC) has produced mixed results. Intranasal oxytocin, targeting social-emotional processing, remains investigational.

Refeeding syndrome is a potentially fatal constellation of metabolic disturbances occurring when nutrition is reintroduced after a period of starvation. The hallmark is hypophosphatemia, driven by insulin-mediated intracellular phosphate shifts when carbohydrate intake resumes, but the syndrome also encompasses hypokalemia, hypomagnesemia, thiamine deficiency, fluid retention, and cardiac decompensation.

Pathophysiology

During starvation, the body shifts from glucose-based to fat-based metabolism. Intracellular stores of phosphate, potassium, and magnesium become depleted despite normal (or even elevated) serum levels. When carbohydrate is reintroduced, insulin secretion resumes, driving glucose — and with it, phosphate, potassium, and magnesium — into cells. Serum phosphate drops precipitously. Phosphate is essential for ATP synthesis, 2,3-DPG production (oxygen delivery to tissues), and cellular membrane integrity. Severe hypophosphatemia (< 1.0 mg/dL) can produce rhabdomyolysis, respiratory failure, cardiac arrhythmias, seizures, and death.

Clinical Risk Factors

The NICE guidelines identify high-risk criteria for refeeding syndrome: BMI < 16, unintentional weight loss > 15% within 3–6 months, little or no nutritional intake for > 10 days, and low baseline electrolytes. Patients meeting these criteria require close monitoring including twice-daily electrolytes for the first 3–5 days of refeeding, continuous cardiac telemetry, and daily weights.

"Start Low, Go Slow" vs. Higher-Calorie Refeeding

Historically, refeeding protocols for AN adopted a highly conservative approach — "start low, go slow" — initiating feeds at 10–20 kcal/kg/day (often 800–1,200 kcal/day) and advancing gradually. However, emerging evidence has challenged this dogma. The landmark REAL (Refeeding in Anorexia in a Lenient manner) study and related work by the O'Connor group (Garber et al., 2016; Whitelaw et al., 2010) from the Royal Children's Hospital Melbourne demonstrated that higher initial caloric prescriptions (1,500–1,900 kcal/day, advancing to 2,500–3,500 kcal/day) achieved significantly faster weight restoration without increased incidence of refeeding syndrome, provided that prophylactic phosphate supplementation and electrolyte monitoring were rigorously implemented.

A 2020 RCT by Garber et al. (published in JAMA Pediatrics) randomized 120 adolescent inpatients with AN to lower-calorie refeeding (approximately 1,400 kcal/day initial) versus higher-calorie refeeding (approximately 2,000 kcal/day initial). The higher-calorie group achieved clinical stability and weight restoration goals significantly faster, with shorter hospital stays (approximately 3 days shorter), and without increased rates of refeeding hypophosphatemia — in part because both groups received prophylactic phosphate supplementation.

Current Best Practice

Contemporary protocols at major eating disorder centers typically initiate refeeding at 1,500–2,000 kcal/day for medically stable patients (or lower for extremely malnourished patients with BMI < 12 or frank medical instability), with:

• Prophylactic phosphate supplementation (Neutra-Phos or sodium/potassium phosphate, 500–1,000 mg/day)
• Thiamine supplementation (100–300 mg IV or PO daily for the first 5–7 days, given before or concurrent with refeeding to prevent Wernicke encephalopathy)
• Electrolyte monitoring (phosphate, magnesium, potassium, sodium) twice daily for the first 3–5 days, then daily
• Cardiac telemetry for high-risk patients (QTc > 500 ms, heart rate < 40 bpm)
• Caloric advancement of 200–400 kcal every 1–2 days as tolerated, targeting 3,000–4,000 kcal/day for weight restoration in many adolescent and young adult patients

The shift toward higher initial calories represents one of the most important practice changes in AN medical management over the past decade, reducing length of stay and improving outcomes while maintaining safety when proper monitoring protocols are followed.

Psychiatric comorbidity is the rule rather than the exception in AN, with over 70% of individuals meeting criteria for at least one co-occurring psychiatric disorder during their lifetime. These comorbidities significantly complicate treatment, impair functioning, and worsen prognosis.

Anxiety Disorders

Lifetime prevalence of any anxiety disorder in AN ranges from 55–65%. Critically, anxiety disorders — particularly social anxiety disorder and generalized anxiety disorder — typically predate the onset of AN by several years, supporting the model that anxious temperament is a premorbid vulnerability factor. OCD co-occurs in approximately 25–40% of AN patients, a rate far exceeding population prevalence. OCD symptoms may intensify during malnutrition (partly due to serotonergic dysfunction) and may partially improve with weight restoration, though many patients require targeted OCD treatment.

Major Depressive Disorder

Lifetime MDD prevalence in AN ranges from 40–60%. Diagnostic disentanglement is challenging because starvation itself produces depressive symptoms (low mood, concentration difficulties, social withdrawal, insomnia). Clinicians are advised to reassess depressive symptoms after weight restoration before initiating antidepressant treatment, as many depressive symptoms resolve with nutritional rehabilitation alone.

Obsessive-Compulsive and Related Disorders

Beyond formal OCD, obsessive-compulsive personality disorder (OCPD) traits are present in approximately 20–35% of AN patients. Rigidity, perfectionism, and need for order are among the most treatment-resistant features and often persist after weight restoration.

Substance Use Disorders

Substance use disorders co-occur in approximately 12–27% of AN patients, with significantly higher rates in the binge-eating/purging subtype (up to 35%) compared to the restricting subtype. Alcohol and stimulant misuse are most common.

Personality Disorders

Personality disorder comorbidity is substantial, though estimates vary widely by assessment method. Avoidant personality disorder is most common in AN-R (approximately 15–25%), while borderline personality disorder is more frequent in AN-BP (approximately 15–25%). Personality disorder comorbidity predicts chronicity and poorer treatment response.

Autism Spectrum Disorder

Emerging research — particularly from the work of Janet Treasure, Kate Tchanturia, and colleagues — highlights significant overlap between AN and autism spectrum disorder (ASD). Estimates suggest that 20–35% of women with AN meet screening criteria for ASD, though the validity of autism assessment during acute malnutrition is debated. Shared features include rigid thinking, restricted interests, difficulties with social communication, and sensory sensitivity. Patients with AN and co-occurring autistic traits show poorer response to standard treatments (particularly those requiring cognitive flexibility or social processing, such as group therapy).

Long-term outcome studies paint a mixed picture. The most frequently cited longitudinal data come from the Steinhausen (2002) meta-analysis of 119 studies spanning five decades, which found that across all studies, approximately 46% of patients with AN eventually achieve full recovery, 33% show improvement but with residual symptoms, and 20% develop a chronic course. Mortality was approximately 5%.

Favorable Prognostic Factors

• Younger age at onset (adolescent onset carries better prognosis than adult onset)
• Shorter duration of illness before treatment initiation (especially < 3 years)
• Higher BMI at presentation (less severe weight loss)
• Restricting subtype (vs. binge-purge, which carries worse prognosis)
• Absence of psychiatric comorbidity (particularly no comorbid OCD, personality disorder, or substance use)
• Intact family support and functioning
• Early weight gain in treatment — this is one of the most robust predictors. In FBT, weight gain of ≥ 2.3 kg by session 4 (approximately week 4) strongly predicts end-of-treatment remission (Le Grange et al., 2014).

Poor Prognostic Factors

• Duration of illness > 7–10 years (severe enduring AN, or SE-AN)
• Binge-purge subtype
• Very low BMI at presentation (< 13)
• Comorbid personality disorders (especially borderline PD)
• Previous treatment failures
• High expressed emotion in the family
• Autistic traits

Severe and Enduring AN (SE-AN)

Approximately 20–25% of individuals with AN develop a chronic, treatment-refractory course, typically defined as illness duration exceeding 7 years with multiple failed treatment attempts. For this population, treatment goals shift from full recovery toward improving quality of life, medical stabilization, harm reduction, and functional rehabilitation. The concept of SE-AN remains controversial — some clinicians argue that it may lead to therapeutic nihilism and premature withdrawal of recovery-oriented care.

Long-Term Medical Consequences

Even after recovery, individuals with a history of AN face persistent medical risks. Bone mineral density may never fully normalize, with fracture risk remaining elevated for decades. Fertility is generally restored with weight recovery, but may be impaired for those with prolonged amenorrhea. Studies suggest increased cardiovascular mortality risk even in recovered patients, potentially related to lasting cardiac structural changes or persistent metabolic dysregulation.

AN treatment occurs across a continuum of care, and appropriate level-of-care determination is critical for outcomes. The APA Practice Guidelines and AED (Academy for Eating Disorders) Medical Care Standards provide frameworks for these decisions.

Outpatient Treatment

Appropriate for medically stable patients with BMI typically > 17 (though not an absolute threshold), intact vital signs, no acute electrolyte disturbances, and motivation for treatment. FBT (for adolescents) or CBT-E/MANTRA/SSCM (for adults) are delivered in this setting.

Intensive Outpatient (IOP) and Partial Hospitalization (PHP)

IOP typically involves 3–4 supervised meals/snacks per week with group and individual therapy (9–12 hours/week). PHP provides daily programming (5–6 days/week, 6–10 hours/day) including all meals and snacks with therapeutic support. These levels are appropriate for patients failing outpatient treatment, those requiring meal support but not 24-hour medical supervision, or as step-down from higher levels of care.

Residential Treatment

24-hour non-hospital care in a structured environment with supervised eating, therapy, and medical monitoring. Indicated for patients who are medically stable but unable to maintain adequate nutrition in less intensive settings. Length of stay typically ranges from 30–90 days. Evidence for residential treatment effectiveness is largely observational, with reported weight restoration success rates of approximately 50–75% during the admission, though relapse rates after discharge are substantial (30–50% within 1 year).

Inpatient Medical Hospitalization

Criteria for inpatient medical admission include: heart rate < 50 bpm (< 40 in some guidelines), systolic blood pressure < 90 mmHg, temperature < 36°C (96.8°F), symptomatic orthostasis, severe electrolyte disturbances, QTc > 500 ms, BMI < 15 (some guidelines < 13), acute medical complications (syncope, seizures, organ failure), or acute suicidality. Inpatient stays focus on medical stabilization and initiation of refeeding protocols, with transition to lower levels of care once medically stable.

Despite decades of research, AN remains among the most difficult-to-treat psychiatric disorders, and significant gaps in the evidence base persist.

Neuroimaging and Biomarkers

Advances in functional connectivity MRI, diffusion tensor imaging, and PET neuroreceptor mapping are refining our understanding of circuit-level dysfunction. The ENIGMA Eating Disorders Working Group is conducting the largest-ever mega-analysis of brain structural data in AN, aiming to identify replicable neuroanatomical biomarkers. However, the field has not yet identified biomarkers with sufficient sensitivity or specificity for clinical use in diagnosis, treatment selection, or prognosis.

Novel Pharmacological Targets

The reconceptualization of AN as a metabo-psychiatric disorder has opened new pharmacological avenues. Research is exploring psilocybin-assisted therapy (with early pilot data suggesting potential for disrupting rigid cognitive patterns), ketamine and esketamine (targeting glutamatergic dysfunction and rapid mood effects), GLP-1 receptor modulation, and agents targeting the endocannabinoid system. However, all of these remain in early-phase investigation, and no novel pharmacotherapy has demonstrated efficacy in an adequately powered RCT for AN to date.

Neuromodulation

Deep brain stimulation (DBS) targeting the subcallosal cingulate (Brodmann area 25) and nucleus accumbens has been explored in small case series for severe enduring AN, with preliminary data showing BMI improvement in some patients. Repetitive transcranial magnetic stimulation (rTMS) targeting the left DLPFC has shown mixed results in small pilot trials, with some evidence of reduced food-related anxiety and improved decision-making. These approaches remain experimental.

Microbiome Research

The gut-brain axis in AN is an area of growing interest. Patients with AN show markedly altered gut microbiota composition (reduced diversity, altered Firmicutes-to-Bacteroidetes ratios), which does not fully normalize even after weight restoration. Whether microbiome interventions could serve as adjunctive treatments remains speculative.

Limitations of the Evidence Base

Major limitations include: small sample sizes in most RCTs (the largest AN trials involve 150–250 participants); heterogeneous outcome definitions (no consensus on the definition of "recovery" or "remission"); high dropout rates (20–40% in most trials); underrepresentation of males, racial/ethnic minorities, and individuals with higher body weights (atypical AN); and limited data on long-term outcomes beyond 12–24 months post-treatment. The absence of a clearly effective pharmacological treatment and the modest, undifferentiated effects of psychotherapies in adults highlight the urgency of new treatment development.