Second Generation Antipsychotics in Adolescent Anorexia Nervosa: A New Hypothesis of Eligibility Criteria

To The Editor:

Two rigorous meta-analyses have recently shed a new light on the use of second generation antipsychotics (SGA) in anorexia nervosa (AN) (Kishi et al. 2012; Lebow et al. 2013). Both agree that SGA are not superior to placebo regarding weight recovery, and some dimensions of psychopathology (specifically, drive for thinness and body dissatisfaction). One of these even recognized a higher level of anxiety and overall eating disorder (ED) psychopathology in SGA groups than in placebo groups (Lebow et al. 2013). On the other hand, patients treated with SGA experienced more sedation/somnolence. It has to be said that the two meta-analyses described only included placebo-controlled trials. Moreover, authors agree that all included studies were limited by low sample sizes and some methodological limitations. Studies involving children and adolescents are rare and, at the moment, only one placebo-controlled trial exists for olanzapine and one (with mixed population) exists for risperidone (limiting the possibility to extract meta-analytic data from this specific population), both without encouraging results (Hagman et al. 2011; Kafantaris et al. 2011). This is at the same time intuitive, because it is undoubtedly harder to perform placebo-controlled studies in younger populations (for ethical and technical reasons), but also challenging, because it is well known that the more frequent period of onset is adolescence, and that early-onset ED could be even harder to treat than late-onset ones. Difficulties in psychopharmacology research for AN are well explained in a recent articles, which described the large amount of resources spent to plan and to start a placebo-controlled trial that is never concluded (Norris et al. 2010). Despite these findings, psychotropic drugs in AN are widely used off label, and SGA are the most used class. Rationales for using SGA could be that body misperception is believed to be a body-focused delusional thought, or that the weight gain often occurring during SGA exposure could be exploited. Moreover, SGA use fits well with some neurobiological models of AN, focusing on monoamine dysfunction (Kishi et al. 2012). It is particularly challenging to understand why SGA are being more and more used (Fazeli et al. 2012), given their side effects, such as metabolic and hormonal disturbances, extrapyramidal side effects (EPS), somnolence, and hypotension, all occurring mainly at beginning of treatment, and potentially threatening the therapeutic alliance. In this article, we will present clinical data from a set of AN adolescents, followed up naturalistically for at least 6 months, who were much improved after adding olanzapine to usual clinical management. Data will specifically focus on baseline characteristics of these patients in order to drive a hypothesis regarding which kind of adolescent AN patients could improve with psychotropic medications.

Clinical Data

We present a set of data, retrospectively collected, of five AN adolescent patients treated with olanzapine followed up for 6 months (four restrictive, one binge/purging subtype). These are five of six inpatients treated with olanzapine at Department of Mental and Physical Health and Preventive Medicine, Child and Adolescent Psychiatry Division between January 2010 and January 2012. One patient discontinued the treatment 1 week later because of marked hypotension, and then she was excluded from the study. The total number of adolescent AN patients followed by our group in that period was 45. Olanzapine was added to usual inpatient treatments, which included psychological and psychoeducational support for adolescents and parents. Mean dose of olanzapine was 5 mg, slowly titrated up, starting at 1.250 mg/day. Mean age at onset was 157 months (SD 28.8). Mean period of hospitalization before starting drug treatment was 31.6 days (SD 17.6). An extensive session with families was performed to inform them about targets of drug therapy as well as to achieve a good alliance (and they all agreed). Three patients first refused the treatment, but they all finally accepted it after dedicated sessions. All patients showed multiple comorbidities, according to American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV) criteria (American Psychiatric Association 1994), assessed with the Schedule of Affective Disorders and Schizophrenia for School-Age Children – Present and Lifetime Version (K-SADS-PL) (Kaufman et al. 1997), specifically: Four (80%) were diagnosed with bipolar disorder II (two of these would have been now diagnosed with disruptive mood dysregulation disorder, according to American Psychiatric Association, Diagnostic and Statistical Manual of Mental Disorders, 5th ed. [DSM-V]) (American Psychiatric Association 2013), and one (patient 4) was diagnosed with depressive disorder not otherwise specified (NOS), obsessive compulsive disorders, and borderline personality disorder. At the moment of hospitalization, all patients showed low overall insight (3.6±0.5), according to Scale for Unawareness of Mental Disorder (SUMD) (Amador and Strauss 1990).

After discharge, all patients started psychotherapy and 6 months later showed these results: Mean body mass index (BMI) improvement was 2.8 (p<0.001), mean kcal intake improvement was 1113.4 kcal/day (p<0.001), there was a mean reduction of 1.8 on the Clinical Global Impressions - Severity (CGI-S) (National Institute of Mental Health 1985) (p<0.001) and a mean improvement of 12.8 on the Children's Global Assessment Scale (CGAS) (Shaffer 1983) (p=0.002).

Results are summarized in Table 1.

Discussion and Hypothesis

We strongly encourage the analysis of baseline characteristics of included patients. First, all patients were nonresponders to usual inpatient treatment for ∼1 month. During this period, weight and food intake continued to fall, and psychopathology worsened, despite the psychological and psychoeducational approach. Timing in introducing a drug therapy is crucial; a good therapeutic alliance can be achieved if patients and their families perceive drug therapy as a way to improve and sustain psychological intervention, and not as main therapy for eating disturbances. Second, all patients had a comorbid diagnosis of mood disorder. This is not a rare comorbidity, as described by McElroy et al. (2011). They all exhibited euphoric states, feelings of omnipotence, temper tantrums, irritability, decrease sleep, and overactivity; these features strongly limited psychotherapy and nutritional interventions. It is not clear if overactivity is a main feature of AN or a phenomenon related to elevated mood, but it is one of most troublesome problems to deal with in the clinical management of AN. It should be always assessed and taken into account, mainly with outpatients. Moreover, an open label trial demonstrated a role for olanzapine in reducing hyperactivity in AN adolescents, mediating weight recovery (Leggero et al. 2010). Third, according to the CGAS, all our patients had a high level of functional impairment, with relevant limitations in several areas of functioning. Fourth, baseline caloric intake ranged between 465 and 880 calories/day (assessed by a professional nutritionist on the basis of a diet chart). Food intake should be given more consideration as a severity index of illness. Moreover, it is much harder to improve food intake when its baseline level is poor. Fifth, BMI, usually considered the best severity index and main outcome of placebo-controlled studies, was not particularly low in our patients. This should raise some doubts about the reliability of BMI changes in predicting a good outcome. Sixth, all patients showed low insight and awareness of illness. A debate on the role of insight in management of psychiatric patients goes beyond the aims of the present article; however, treating AN patients with low insight, sometimes with “ low insight families,” is undoubtedly harder.

A recent work by Norris et al. (2011) tried to retrospectively assess baseline characteristics in drug-treated AN patients compared with untreated ones. The authors concluded that despite their best attempt, they could not match the two groups for baseline characteristics. This conclusion is intriguing, because it shows how difficult it could be to identify the patients eligible for drug treatment. Our hypothesis is driven by Norris's work, by the discrepancies between research results (Kishi et al. 2012; Lebow et al. 2013) and clinical practice (Fazeli et al. 2012), and by our set of data (although limited by the small sample size) and other open-label studies. We hypothesize some strict criteria for considering AN adolescent patients eligible for SGA treatment. These criteria should address clinicians as to the appropriate way to treat the right patients, and should address researchers as to how to design trials with specific inclusion criteria and better-focused outcomes.

Eligibility criteria are presented below:

1. Patient is nonresponder (weight and food intake continue decreasing) for at least 1 month of psychological and psychoeducational interventions.

It is of note that we have not included low BMI as criteria, for several reasons: BMI might not be particularly low if food restriction onset was quick, and worsening; BMI might not be particularly low even with high level of psychopathology and functional impairment. We have not included core ED psychopathology (with measures such as Eating Disorder Examination [EDE] or Eating Attitudes Test [EAT-26]) as criteria, given the low impact of SGA drugs on it (Kishi et al. 2012; Lebow et al. 2013).

Marginally, we noted that, other than marked hypotension in one case, none of our patients showed clinically meaningful adverse events during the follow up period, confirming findings from a recent article by Swenne, concluding that olanzapine in adolescent AN is associated with side effects similar to those observed during olanzapine treatment in children and adolescents with diagnoses other than ED and in adults (Swenne and Rosling 2011). Our data could be relevant, albeit limited by small sample size and by the naturalistic nature of the study, because long-term safety data are rare in this population.

Conclusions

Adolescent AN is one of most troubling problems for child and adolescent psychiatrists, and this even because of the lack, or the poor quality, of evidence-based treatment strategies. Recent literature seems to raise more problems than it solves.

We propose some strict criteria for adolescent AN patients to be considered for psychopharmacological intervention with SGA. We argue that a highly selected group of AN patients could profit from SGA adjunction to usual clinical management (psychotherapy and psychoeducational support). The validity and the reliability of proposed criteria for selecting this group of patients have to be explored with future studies.