Abstract
Anorexia nervosa is compounded by severe physical morbidity, psychosocial disability and handicap comparable to schizophrenia [1], as well as the highest lethality of any psychiatric disorder [2–6]. The reported lifetime prevalence of anorexia nervosa ranges from 0.5% for narrowly defined, to 3.7% on broader diagnostic criteria [7], [8]. While there is increasing recognition in males [9], [10], young children [11] and older women [12], this doesn't necessarily equate to increasing prevalence [13], [14]. Diverse definitions and measures have been used to evaluate outcome in anorexia nervosa, with comprehensive evaluations including physical, psychological, social and temporal dimensions of change being rare [15]. Outcome studies have mostly focused on weight, menstruation and broad measures of psychological functioning in the severely ill, so that patients who have recovered based on such criteria may still be experiencing the core psychopathology of the syndrome [16], [17]. In addition, the differentiation of anorexia nervosa into subtypes is tempered by debate over what constitutes bingeing, the primacy of bulimic symptoms, and the possibility of poorer outcome, in terms of relapse, readmission to hospital and pervasively disturbed self-esteem and awareness [16]. Hence, what constitutes ‘recovery’ and ‘remission’ in anorexia nervosa [18], as opposed to diminution or absence of syndromally based diagnostic criteria, is not universally accepted. Currently, individually and flexibly tailored education, nutritional rehabilitation and psychotherapy are the mainstays of management of anorexia nervosa, in which psychotropics, including antipsychotic medications, largely play an adjuvant role [19–22].
The aim of this paper is to review the evolving utility and efficacy of atypical antipsychotic medications in the management of anorexia nervosa. This is addressed in terms of potential modulation of dysregulated serotonergic and dopaminergic systems, whether secondary to malnutrition or antecedent to the onset of clinical case definition, and beyond that of weight gain only as a possible medication side-effect [23]. The evidence for monaminergic dysregulation is addressed by discussion of clinical and biological findings, the impact of dieting, peripheral and central biological markers of monoaminergic activity in short and longer-term weight-recovered patients, and functional neuroimaging.
Antipsychotics in anorexia nervosa
Antipsychotic medications have been used in the treatment of anorexia nervosa since the 1960s. Initially, chlorpromazine, with and without insulin, had been used to facilitate weight gain, but complicated by significant medical morbidity and making no difference to outcome [24], [25]. Other typical antipsychotics such as pimozide, combined with behaviour therapy [26], and sulpiride [27] were not shown to be efficacious in short-term double-blind placebo-control and placebocrossover studies, respectively. Low dose haloperidol (1–2 mg daily) has been reported as dampening psychopathology in a small open-label study in patients with anorexia nervosa, but was confounded by concomitant use of antidepressant medications [28]. Although the core psychopathology of anorexia nervosa has generally been refractory to psychotropics [7], [29], [30], low dose atypical antipsychotic medications are currently prescribed for psychotic-like thinking, distressing anxiety and obsessionality [19], [20]. The latter may involve lessening the intensity of obsessive thoughts concerning food, and hence decrease discomfort and resistance to refeeding [31].
There are a number of issues apart from medication efficacy which impact on antipsychotic use in anorexia nervosa. Medical morbidity secondary to starvation and purging may limit the tolerability and safety of antipsychotic medication in the illness. Examples of these limitations include electrolyte and metabolic disturbance, diverse cardiac (including arrhythmias and QT [Q-wave/ T-wave] interval prolongation), gastrointestinal and neurological side-effects. The psychopathology of anorexia nervosa itself, centred on denial and control [32], [33], may affect medication adherence and in turn the utility and effectiveness of antipsychotics in this population.
Non-response to selective serotonin-reuptake medication in underweight people suffering with anorexia nervosa could be the result of inadequate nutrients to facilitate normal serotonin synthesis and function [34–37]. But, atypical antipsychotics encompass differing serotonergic and dopaminergic receptor affinities, and include: serotonin-dopamine antagonists such as olanzapine, risperidone, quetiapine, ziprasidone and clozapine; aripiprazole, a partial dopamine agonist and 5-HT2A antagonist; and amisulpiride, a dopamine D2 and D3 antagonist. Although the main target of all antipsychotic medications is the D2 receptor, the precise mechanisms of ‘atypicality’ and differentiation from ‘typical’ or first generation antipsychotics are not clear. Possible mechanisms of atypicality include: preferential cortical, rather than striatal, D2 receptor occupancy; preferential action on 5-HT2A (amisulpiride has no significant occupancy of these receptors) and D4 receptors; and looser binding to, and greater rapidity of dissociation from D2 receptors than dopamine [38–40]. In animal models, atypical antipsychotics may increase prefrontal dopaminergic release through 5-HT1A receptor stimulation, irrespective of direct 5-HT1A agonism, via 5-HT2A and D2 receptor blockade [41].
Malnutrition produces changes in neuroendocrine function, including neuropeptides and monamines, pivotal to feeding, mood and impulse control [42]. Antipsychotics modulate the neuropeptide systems of opioids, neuropeptide Y, cholecystokinin and somatostatin [43], which along with peptide YY, have specific roles in feeding structure in anorexia nervosa [44]. Cholecystokinin also modulates central dopaminergic function [45]. The correction of these neuropeptide abnormalities after recovery suggests occurrence secondary to malnutrition and weight loss in anorexia nervosa [44]. Premature normalization of the concentration of leptin (hormone signalling and regulating body fat stores) with refeeding has been postulated as a mechanism leading to the difficulty achieving and sustaining normal weight in anorexia nervosa [44]. Although increased serum leptin has been noted with atypical antipsychotic use, this is thought to be secondary to weight gain [46–48].
Let us now consider the evidence for serotonergic and dopaminergic dysregulation as a core feature of anorexia nervosa, which might be amenable to therapeutic modulation by atypical antipsychotic medication.
Serotonergic dysregulation in anorexia nervosa
The serotonergic system is involved in modulation of appetitive behaviours and postprandial satiety [49–54]. Reduced serotonergic neurotransmission and receptor activation are associated with increased food consumption and weight gain [55]. The induction of functional supersensitivity of 5-HT2C receptors by dieting, possibly in response to sparsity of central serotonin, may contribute to the development of eating disorders in predisposed individuals [56]. People suffering with anorexia nervosa show clinical and biological findings which implicate serotonergic dysregulation, including disturbed feeding, depression, impulsivity, anxiety, obsessive-compulsive features and neurobiological disturbances linked to serotonin such as thermoregulation [37],[57–61].
Although altered serotonergic activity in anorexia nervosa could be secondary to tryptophan depletion alone, disturbed activity appears to persist after both short and long-term weight-restoration in some, but not all studies. Some studies have assessed this seemingly persistent disturbance in anorexia nervosa patients by cerebrospinal fluid concentrations of 5-hydoxyindoleacetic acid (the major serotonin metabolite), which has ranged from reduced in underweight, normal in recent weightrecovery and elevated in longer-term weight recovery (6 months) [62–64], whilst in other studies there was no difference from controls [65]. Other studies have utilized serotonergic agents, such as meta-chlorophenylpiperazine [66], L-tryptophan [66], [67] and d-fenfluramine [68], to compare prolactin response, an index of central serotonergic responsiveness and activity, to controls [64], [66–70]. The findings have been inconsistent, with some studies showing variably blunted prolactin response [66–68], whilst others have not [69], [70]. Of the studies without prolactin blunting, one lacked a control group [69], but in the other [70], all subjects were of normal weight and had recommenced menstruation. The duration of weight-recovery varied in these studies, from 5 days to 3 weeks in the short term [62], [63], [66], [67], 3 months to over a year in the long term [62], [64], [69], [70], and was not noted in one study [68]. A significant proportion of patients in one study had not recommenced menstruation [64], possibly confounded by oestrogen levels impacting on central serotonergic function [70], [71]. But, the utility of amenorrhoea as a core feature of anorexia nervosa has been challenged [72]. Hence, whilst some findings suggest altered serotonergic responsiveness in postsynaptic hypothalamic and pituitary pathways in anorexia nervosa group persisting after weight-restoration, implicating changes in serotonergic function at levels higher than the hypothalamus, other studies do not.
A positron emission tomography study of central 52A receptors, utilizing [18F] altanserin, a specific 5-HT2A receptor ligand, showed altered amygdala, hippocampal and cingulate cortical binding in women, weight and symptomatically recovered from anorexia nervosa for 1 year when compared to normal controls [73]. These findings, which persisted with weight-recovery and the absence of altered grey matter volume, implicated these interconnected brain regions in the serotonergic neurocircuitry of anorexia nervosa.
Dopaminergic dysregulation in anorexia nervosa
The dopaminergic system modulates the motivation to feed [74], underpinned by restraint and emotionality underpinning linked to the dorsal striatum [75]. Dopamine is also involved in the inhibition of fat intake [76]. People suffering with restricting anorexia nervosa have clinical findings implicating the dopaminergic system, including altered reward, novelty seeking and motor activity [77], [78]. Increased central dopaminergic activity in anorexia nervosa is evidenced by impaired adrenal androgen secretion [79] and impaired prolactin response following central D2 receptor blockade [80], not present in controls. There is altered D2 receptor sensitivity in both restricting and binge-eating/purging DSM-IV anorexia nervosa subtypes compared with bulimia nervosa and normal controls, evidenced by significantly reduced growth hormone response to growth hormone releasing hormone administered with apomorphine, a selective D1 and D2 receptor agonist [81]. Although this study's diminished growth hormone response was not correlated with body mass index and the findings suggestive of reduced sensitivity of postsynaptic D2 receptors, and possibly, increased presynaptic dopaminergic secretion at the hypothalamic level, a fixed dose of apomorhine was used without correction for body weight. A fixed doseage may not have been sufficient to produce an adequate growth hormone response, which was also not corrected for bioavailability.
The findings of cerebrospinal fluid marker studies of central dopaminergic activity (homovanillic acid, a major metabolite of dopamine) in underweight anorexia nervosa patients compared with normal controls, have been contradictory, with normalization [64] or persistently increased levels after weight-restoration [77]. Additionally and similarly to the serotonergic studies, the length of weight-recovery has varied from 6 months [64] to 1 year [77] in the longer-term. In a study showing persistent alteration of dopaminergic activity in longerterm weight-recovered anorexia nervosa, this finding occurred only in the restricting and not the binge-eating/ purging subtype with significantly lower cerebrospinal homovanillic acid levels [77]. But, it was not possible to determine to what degree altered dopamine secretion, versus metabolism and clearance, contributed to these findings.
Clinical utility of atypical antipsychotics in anorexia nervosa: case reports
Recent case reports suggest that the atypical antipsychotics, olanzapine [82–88] and risperidone may benefit [89–91] the specific management of underweight young and adult anorexia nervosa patients who have not responded to other treatment modalities.
Caveats in the interpretation of these reports include very small sample sizes, open-label antipsychotic use for varying indications (psychiatric comorbidity and variably defined treatment resistance), the lack of control groups or within-subject comparison to address differing interventions and varying outcome measures limiting generalisability to other treatment settings and across the age spectrum.
Open label and single-blind trials
Olanzapine
There have been three open-label trials and one randomised comparison study of olanzapine in anorexia nervosa.
The first was a 10 week trial in 20 outpatients, predominantly female, aged 14–56 years, and 15–25 percent below ideal body weight [92]. Eighteen patients (12 with DSM-IV binge eating/purging and six with restricting subtypes) were treated with olanzapine 10 milligrams daily, and two had lorazepam concurrently. The study's baseline Positive and Negative Symptom Scale (PANSS) [93] mean score of 63 was indicative of severe global psychopathology at the outset. All patients had weekly drug monitoring and adherence sessions, whilst one continued with individual psychotherapy. Fourteen out of 20 patients completed the study, with 10 patients gaining weight over the 10 weeks, whilst three of these attained ideal body weight. The other four completers lost weight over this period. The four who had dropped out had also gained weight. There was a significant decline in the PANSS total score, thought disturbance and delusions scales and in the Hamilton Depression Rating Scale (HAM-D) [94] scores for those who gained weight, but not for those who had lost weight. The baseline HAM-D scores were not predictive of treatment outcome. Although there were significant declines in the Eating Disorder Inventory (EDI-2) [95] scores for the whole group, there were no differences in the EDI-2 drive for thinness and body dissatisfaction subscales.
The second study was a retrospective evaluation of 18 specialist eating disorders unit inpatients (gender not identified), prescribed olanzapine, ranging from 2.5 to 10 milligrams daily over 3–70 weeks, and administered a non-standardized nor validated, self-report questionnaire about symptoms before and during antipsychotic treatment [96]. Mean weight gain and other psychopathology measures were not described. This study reported reduced: anxiety before and during meals, and following stressors; difficulty in eating; and frequency of obsessive thoughts about body image and fear about being fat.
The third study looked at two groups of 23, consecutively admitted inpatients suffering with anorexia nervosa, who had been in hospital for at least 2 weeks. One group was openly prescribed olanzapine (dose ranged from 1.25 to 15 milligrams per day), and the other was not prescribed an antipsychotic [97]. Neither group differed in the rate of weight gain during hospitalization. The olanzapine group had higher eating disorder psychopathology scores on the Eating Disorders Examination Questionnaire [98], and more previous inpatient admissions and greater lengths of stay.
In a study comparing olanzapine versus chlorpromazine in 15 specialist eating disorder inpatients suffering with DSM-IV anorexia nervosa, subjects were randomly assigned to either medication and dosed according to response and tolerability [Mondraty NK, Birmingham CL, Touyz S, Sundakov V, Chapman L, Beumont P: unpublished data, 2003]. There was significantly greater reduction in ruminative thinking on the mental activities subscale of the Padua Inventory [99] in the olanzapine group in comparison to the chlorpromazine group, which was not associated with weight gain nor sedation. There was reportedly a clinically, but not statistically significant, decrease in the global and subscale scores of the EDI-2 [95] in the olanzapine group, with lower scores, including drive for thinness, bulimia, and body dissatisfaction subscale scores than the chlorpromazine group.
Risperidone
Risperidone has been studied in adolescents suffering with anorexia nervosa in two open-label studies following changeover from haloperidol.
In the first [100], 20 adolescent inpatients with DSMIV anorexia nervosa were assessed by clinical interview and the Symptom Checklist 90 [101]. Haloperidol was changed over to risperidone at 2 weeks. After 2 weeks, patients were assessed on clinical observation as more aware of their physical condition with commencement of autonomous eating. These results were reportedly stable at 6 months.
In the second study, 24 emaciated adolescent patients aged 11–22 years with a 25 percent reported reduction of maximum body weight [102] had clomipramine and haloperidol changed over to risperidone, when deemed ineffective in social withdrawal and abulia. Twenty-one of the patients completed treatment. There were reportedly significant improvements at 3 months in social life, interpersonal relationships and resumption of school activity.
Amisulpride
A single-blind comparison study of amisulpride, fluoxetine and clomipramine was conducted in 35 inpatients, diagnosed with DSM-IV anorexia nervosa, restricting subtype. These medications were administered during refeeding in a 3 month, specialist inpatient weight-restoration program, including twice weekly dietitian reviews [103]. Twelve subjects were treated with amisulpride 50 milligrams per day, 13 with clomipramine at a mean dose of 57.69 milligrams and 10 with fluoxetine at a mean dose of 28 milligrams. Baseline and 3 month assessments, consisting of clinical interview based on Long Interval Follow-up Evaluation [104] revealed a more significant increase of mean weight in the amisulpride group compared to the other groups, but no significant difference in body image disturbance, weight phobia and amenorrhoea between groups.
Discussion of trials of atypical antipsychotics in anorexia nervosa
These sparsely numbered open-label studies, one single-blinded comparison study and randomised comparison study of two antipsychotics (only one an atypical) suggest that serotonin-dopamine antagonists may be therapeutically beneficial beyond weight gain alone in severely underweight anorexia nervosa patients, in both inpatient and outpatient settings. But, there are a number of caveats precluding drawing conclusions from these studies alone. The open-label studies were limited by their inherent selection and observer biases, including the impact of the investigating team itself in the absence of study blindedness. All of the open-label studies lacked control groups or within-subject comparisons to address differing interventions, including the impact of the adherence sessions in the first olanzapine study. The retrospective olanzapine study was also limited by reliance on a non-validated, non-standardized, self-report instrument in a group of patients who were adherent to medication. The single-blind comparison of amisulpiride, fluoxetine and clomipramine also lacked a blinded placebo comparison group to assess whether any of these medications were better than the standard, 3 month inpatient weight-restoration program with regular dietetic review, on body image disturbance and weight phobia. Despite random assignment in the study comparing olanzapine and chlorpromazine, variable dosing, the comparison of an atypical antipsychotic with a typical antipsychotic (not previously demonstrated to be efficacious in the specific eating disorders psychopathology of anorexia nervosa) and the lack of a blinded placebo group were significant limitations.
All of the studies had small sample sizes in differing treatment settings and have not been replicated. The studies followed patients for a relatively short period in the course of a generally chronic illness, lasting on average 5–7 years [1]. In the 8-week olanzapine study, some of the core psychopathological phenomena of anorexia nervosa, including body image and pursuit of thinness, were not altered by the medication. This may also reflect the sensitivity of the EDI-2 to measure change over such a short period of time. The olanzapine group in one of the open-label studies [97] had greater psychopathology and utilization of inpatient care than the comparison group, suggesting that the groups were different at the outset, and that any, potential changes in weight and eating disorder psychopathology in the medication group may not have been affected. The risperidone studies did not include measures of specific eating disorders psychopathology. None of the studies measured motivation or readiness to recover from anorexia nervosa [105], [106].
Conclusions
The inconsistent findings regarding the persistence of serotonergic and dopaminergic dysregulation in anorexia nervosa highlight the difficulty in resolution of starvationinduced neuroendocrinological abnormalities, and the extent of amelioration by weight-recovery alone, from those primarily related to the onset of the illness. The diversity of these findings is also underpinned by the variety of criteria used to define outcome and recovery in anorexia nervosa, as well as what duration constitutes remission and recovery. Nonetheless, serotonergic and dopaminergic dysregulation in anorexia nervosa may represent potential therapeutic pathways for atypical antipsychotics to modulate, given their pharamacodynamics, separate to that of weight gain as a side-effect or secondary to anxiolysis or treatment of psychiatric comorbidity. Although case reports and uncontrolled (no placebo controls) trials of some atypical antipsychotic medications are encouraging, they are sparse in number, and double-blind placebo controlled studies with defined, broader and more uniform outcome measures are essential to resolve whether these medications have benefits for core eating disorders psychopathology and illness course. But, the denial and control psychopathological dimensions [32], [33] and the extent of motivation or readiness to recover [105], [106] are likely to limit the randomization of subjects into such studies.