*4.4. Perspectives*

Several perspectives suggest the need for the development of innovative treatments.

Part of the lack of evidence could be attributable to the heterogeneity of significant symptoms and the treatment responses encountered in AN. Very few studies thus far have stratified the subtypes or clinical features of the disorder. Anorexia nervosa is a heterogeneous disorder and a single "optimal" drug for all individuals is highly unlikely. Targeting more homogeneous subgroups could be helpful. Thus, growing evidence derived from various approaches supports a specific staging trajectory for anorexia nervosa, and there is preliminary evidence that interventions should be matched to the stage in the illness (for review see [196]). Interventions tailored according to the stage of the illness or the developmental trajectory could be valuable options. Certain other clinical features could be considered for more homogeneous subgroups of patients, since they could be associated with a specific form of the illness, or lower response to medication, for instance the AN subtype (AN-R, AN-BP, AN/B, AN/P), age (differentiating pre and post-pubertal adolescents and adults), gender, cultural environment, current and maximum pre-treatment BMI, associated personality traits, psychiatric comorbidities [45] or a history of childhood abuse [197,198]. A number of latent class and latent profile analyses have been performed on symptom and personality factors to stratify the endo-phenotypes spanning AN [197,199–201]. These sub-categorizations could be a starting point to homogenize samples.

Also, there is currently emphasis on using precision medicine to identify targets, which should lead to more effective treatments. Symptoms that maintain the disorder may differ across individuals and participate in the lack of evidence. A novel methodological perspective is thus to address the extreme heterogeneity within AN and to develop and adapt treatments to each individual, which could be of considerable interest in the near future [202].

Finally, beyond a categorical approach, treating one or several specific dimensions associated with the illness rather than the illness overall could be useful, since changing targets might change clinical outcomes. Some salient dimensions in AN, such as delay discounting or cognitive impulse control, could potentially be targeted by drugs or a neuromodulation approach [203].

Concerning non-psychotropic drugs used in AN, these are mainly used to treat medical conditions associated with AN. It is critical to underline that weight gain and restoration of normal weight is the first line of approach to the management of these conditions whenever possible. Therefore, refeeding modalities for weight gain or restoration, including the specific indications for enteral nutrition, oral nutritional supplementation, micronutrient supplementation and specific drugs for functional digestive disorders, seem absolutely essential, but to date no guidelines exist on nutritional therapeutic strategies for AN.

Among the drugs available, sex hormone treatments need to be discussed, as they are fairly widespread in use, but their effects are a source of considerable debate. Various arguments should be taken into account in the assessment of the risk-benefit balance of their prescription. As demonstrated by our results, oral contraceptives have failed to show any significant benefit in protecting bones among patients with AN. Although certain issues could explain this lack of effect, such as the heterogeneity of the patients included across studies, and variable treatment and follow-up durations, the suggestion is not to use oral contraception to protect bones in AN. On the other hand, transdermal estradiol

has been found to be a more physiological form of oestrogen replacement and more promising than oral administration in managing osteopenia among adolescents with AN. This explains the position of the recent British national guidelines suggesting hormone replacement therapy with 17-β-estradiol (with cyclic progesterone) rather than oral contraception, and only for young AN women (13–17 years). Furthermore, the decision to prescribe oral contraception is not without drawbacks. For example, the return of menstrual function, which indicates adequate weight restoration, is masked by the cessation of bleeding induced by contraceptives. Also, its use can provide a sense of reassurance that patients are protected against osteopenia, which can reduce efforts for weight rehabilitation. On the other hand, it is important to keep in mind that amenorrhea, occurring in 66 to 84% of women suffering from anorexia nervosa, favours the absence of contraceptive measures, explaining a particularly high rate (up to 50%) of unwanted pregnancies in this population. Finally, somato-psychic tolerance and compliance with hormonal replacement are very limited due to fat phobia, menstruation and bleeding refusal in many adolescent AN patients. Recent studies evaluated the interest of testosterone replacement therapy in AN women based on the fact that hypothalamic amenorrhea is associated with a profound androgen deficit. Transdermal testosterone replacement increases spinal BMD when administered with risedronate and can stimulate bone formation [174]. In addition, this could be an effective medication in women with treatment-resistant depression by improving depression and cognitions [175]. Otherwise, as mentioned previously, SSRIs seems to be inefficient in conditions of undernutrition, and future studies should assess optimal renutrition and adapted hormone replacement to potentiate the effectiveness of psychotropic medications [49]. To date, testosterone replacement therapy is not recommended for female nor for male AN patients. Concerning the use of medications for bone health and BMD improvement, despite the promising results of studies on sexual hormone replacement, bisphosphonates and combined interventions among adolescent and premenopausal women, the safest and most effective strategy to protect and improve bone density and prevent spontaneous fractures risk is, for now, the restoration of weight and menstrual recovery. Further studies are needed to establish standards for the treatment of osteoporosis in AN.

There are no specific treatments taking into account particular aspects of adolescent or adult male AN patients and current proposed treatments are similar to those for AN women [158]. Specific research on the male AN population is insufficient and needs to be developed, particularly with regard to testosterone and other therapies, and the benefits of specific drugs on somatic, cognitive and psychiatric functions that can influence evolution and prognosis in AN boys and men.

#### *4.5. Strengths and Limitations of This Overview*

As stated by Pollock et al. (p16) [204] "overviews are a relatively new methodological innovation, and there are currently substantial variations in the methodological approaches used within different overviews". We defined a methodology for this overview on the basis of elements from the literature, but due to the characteristic of the literature concerning medication for AN we could not totally fulfil all the quality criteria previously defined [205].

Indeed, a good quality overview should include systematic reviews with four characteristics [205]: 1. They should not substantially overlap. 2. They should focus on the precise question asked by the overview. 3. They should be high quality. 4. They should be up-to-date.

There is considerable overlap between reviews and meta-analyses selected in our overview. In order not to bias our conclusions we focused on the more recent reviews or meta-analyses; when meta-analyses and reviews were available on the same studies we focused on the meta-analysis conclusions, and when two reviews were available we focused on the more recent one including the more recent research. We selected only systematic reviews with a well-defined methodology but methodological quality was not homogeneous.

The conclusions we have drawn are limited by the methodology of the systematic reviews included, which use different criteria and objectives, over different periods, possibly leading to different conclusions and possibly biased by their selection. In order to alleviate this limitation we developed a systematic overview using the PRISMA guidelines, mainly based on meta-analyses and systematic reviews of RCTs and open comparative studies, but not on narratives reviews, retrospective studies or case reports.

Our findings are also limited by the quality of the published literature on the topic of medication for AN, which also impacts the conclusions, since most of the studies published on the subject are open studies, retrospective studies or case series or case reports, and RCTs are rare. The reason for this situation is linked to both AN patient refusal to participate in RCTs and to the clinical context, and also to the relatively low frequency of AN. Otherwise, it is also linked to ethical reasons, particularly on the refeeding topic [52] or growth hormone replacement [151]. In addition, the existing RCTs are poor quality, as they were conducted with various methodological procedures, on heterogeneous, small samples, and over different durations of follow-up. Finally, these studies were mostly conducted over short treatment durations whereas AN is a chronic disorder.
