Pathways to improved treatment of anorexia nervosa: a need for reconceptualization

Abstract

Objective

Given anorexia nervosa’s (AN) status as one of the most difficult psychiatric disorders to treat, with the highest mortality rate, with millions of dollars spent on researching its treatment, and decades of assiduous study, it is surprising and distressing that treatment outcomes have remained essentially unimproved since the first treatments were presented over 70 years ago. Recent theorizing argues for a more personalized approach to treatment but cannot yet specify what that should look like.

Method

The current paper argues that classifying AN patients into more useful subgroups is necessary for treatment—particularly individualized treatment—to progress, rather than treating the disorder as a unitary phenomenon based on issues concerning body image. The literature on contributors and characteristics of groups of AN patients is critically reviewed including comorbidities with other psychiatric disorders, environmental stressors and life events, and pre-existing personality makeup.

Results

The examination of the data helps to identify what distinguishes individuals who respond to treatment and recover from AN from those who remain disordered. We suggest aspects of AN that contribute to treatment success or failure and suggest ways to ascertain how the disorder develops in different individuals.

Discussion

We conclude that more attention is needed to distinguish between people who develop the disorder in reaction to environmental stressors or events, people who appear to have a physiological, possibly genetic predisposition to psychiatric dysfunction, and people who have both a predisposition to disorder and an environmental trigger.

Plain language summary

• There are different etiological pathways to developing anorexia nervosa

• There are differential therapeutic outcomes for patients with different personality and etiologic backgrounds

• Subtyping anorexia nervosa based on personality and etiology could target therapy more precisely

Background

Anorexia nervosa (AN) is generally considered to be one of the most intractable psychiatric disorders, both difficult to treat successfully (e.g., [71, 81, 90, 118]); and with the highest mortality rate of any psychiatric disorder other than addictions (e.g., [3, 4, 118]). Despite millions of dollars of research money and decades of assiduous treatment studies, treatments for AN have not shown much improvement in success rates since serious efforts to treat AN began in the 1950’s and 1960’s (e.g., [81, 89, 90, 111, 118]). The number of patients who survive the disorder and who become truly symptom-free over the long term is generally on the order of only 30–40% after 9 or 10 years, though this may improve to more than 60% after as much as 20+ years (e.g., [26, 32, 36, 78, 110]). It is important to note that the prognosis for children and adolescents is generally better than the figures above suggest and has improved over time with early intervention and family-based approaches [72]. However, to date there is no universally accepted definition of clinical outcome, and these figures for both adults and adolescents could well be an over-optimistic interpretation. Interest is now turning towards a better understanding of what can best be described as “deep dive phenomenology” in AN, and as the data emerges from current studies, there may have to be a re-evaluation of clinical outcomes in AN as many of these putative characteristics of the disorder may persist longer-term [12, 74, 86, 95, 101].

One possible reason for the lack of success in treatment of AN may be how patients are identified and classified [47]. The prevailing wisdom seems to be that subgroups of patients who have different symptom clusters may respond differently to treatment and need therapies targeted more to their specific symptoms. The traditional means of classifying subtypes of AN into restricting versus binge/purging groups, with narrower subtypes based on the presence of both, does predict more drop out from treatment [122] and higher levels of eating disordered thinking and behavior [100]. Nonetheless, this distinction has not been found to identify clinically meaningful differences among these groups [97] and it is not related to degree of weight suppression or the degree of impairment in the two groups [100]. As Touyz et al. [118, p. 5] bluntly declare, “the subtyping of restrictive versus binge/purge AN, which has been integral to each and every DSM iteration, may have reached its use-by date”. Moreover, classifying AN as a body-image-based disorder along with Bulimia Nervosa, Binge Eating Disorder, and Eating Disorder Not Otherwise Specified may have forced a focus on an issue that is not actually central to AN, at the expense of the issues that are contributory and maintaining [8, 42, 118]. It has taken the published literature a very long time to reach this conclusion, but it is understandable, as the model of overvaluation of shape and weight as the central cause and maintaining factor for eating disorders has dominated the field until now [42, 83]. Moreover, the transdiagnostic assumption that individuals with AN migrate to other eating disorders [29] is more likely to reflect a reaction to food restriction and starvation than the presence of a new eating disorder. As the data from the Keys et al. [63] seminal study of human food deprivation/starvation shows, participants deprived of food in order to lose 25% of their initial body weight later gorged themselves when given unlimited access to food following the protracted period of restriction. This suggests that binge eating is more of an epiphenomenon of dysregulated eating behaviour following restriction than a sign of developing an entirely new eating disorder [8, 98].

It appears, then, that we need to identify other dimensions besides restricting versus binge eating that might help to classify different subtypes of AN, and might help us to identify how to direct treatment to different variants of AN. Le Grange [71] noted that treating all patients with the same therapies as we have for decades does not seem to be working, and we need more targeted interventions. It is, however, difficult to find reliable predictors of who will respond well to treatment for AN, or what treatment is best for which patient; in fact, it is easier to find predictors of negative outcomes of treatment [122].

In order to determine how best to treat AN, we may need to take a step back and reexamine the underlying characteristics of those with the disorder and look at the presumed contributors. This may allow us to group patients more successfully for different treatment approaches.

Factors characterizing or contributing to AN in subgroups of patients

One frequently identified characteristic of many with AN is psychiatric comorbidity. Borderline personality disorder (BPD), for example, has long been identified as a comorbidity for many AN patients [7, 11, 53]. Although such a comorbid diagnosis may not be predictive of treatment outcomes, at presentation for therapy, those AN patients with comorbid BPD exhibited greater difficulties with affect regulation, heightened distress, and were more generally disturbed than were AN patients without BPD [7], and reported more premorbid stressors and dysfunctional family environments and were more likely to need hospitalization as well as medication [53]. Over two decades ago, Westen and Harnden-Fischer [125] suggested that personality disorders should be taken into account when classifying anorexia nervosa patients and other eating disordered patients. Westen and Harnden-Fischer then used personality diagnoses of a group of eating disorder patients to identify different kinds of AN patients, who differed from each other in many respects, possibly affecting treatment outcomes. Bruce and Steiger [11] further noted that personality disorders such as OCPD, along with associated personality characteristics such as perfectionism, seem to contribute to the development of AN and may even make the disorder more resistant to treatment. A review and meta-analysis [30] found OCPD and borderline personality disorders to be frequent co-morbidities with AN. In addition, personality disorders comorbid with AN have been identified as being associated with more negative treatment outcomes (e.g., [19]) and increasing the probability of mortality in such patients [46, 55]. Attention to specific comorbid personality disorders thus seems like the sort of distinction among AN patients that we should be making in order to tailor treatment more effectively to each patient [105, 118].

Like personality disorders, depression is a diagnosis that seems to be comorbid with AN for a large proportion of patients (e.g., [99, 122]). Depression may appear after the development of AN [36, 50], but pre-morbid depression may also predict later AN symptoms, as well as being concurrently associated with eating pathology (e.g., [69, 122]). Regardless of whether depression is a precursor, a result, or a mere co-morbidity with AN, it has been found that patients with depression generally exhibit more severe symptoms of AN than do those with no comorbidity or with comorbid anxiety disorders [50]. Moreover, depression is associated with self-induced vomiting and other eating disordered behaviors in those with AN [69]. Increased depression at the start of treatment is associated with worse treatment outcomes [122], although not every study finds these differences, at least in the short term (e.g., [15, 84]). Several studies of factors associated with recovery from AN have noted that those patients with comorbid depression seem less likely to recover from AN over the longer term [28], especially if the depression is of long duration and possibly pre-morbid [61, 127]. A 22-year follow-up of AN patients identified comorbid clinical depression as a characteristic of AN patients who remained symptomatic after more than two decades, despite treatment [36]. Comorbidity of AN and any psychiatric disorder, not just depression, was associated with greater mortality from suicide or other unnatural causes [55].

Despite some mixed findings with respect to treatment outcome, the presence of depression in AN patients clearly warrants attention. This is particularly pertinent given that Zimmerman et al. [133] have identified 227 possible subtypes of depression, with many of these having a somewhat unique brain neurocircuitry. Many of these subtypes of depression show different responses to antidepressant medications and eventual clinical outcome [131]. Given this sort of variability in depressive disorder(s), we may also have been naïve in conceptualising AN as a homogeneous disorder and believing that most patients with AN will respond clinically to current evidenced-based treatment [8, 118]. This is clearly not the case for those with both AN and depression.

Similar to depression, anxiety disorders are also frequently comorbid with AN (e.g., [2, 41, 57, 113, 114, 122]), especially Obsessive–Compulsive Disorder (OCD, e.g., [37, 41, 57, 87, 93, 107]). Recent reviews report that depression and anxiety disorders such as OCD are actually the most frequent comorbid conditions with AN [49, 50]. In particular, OCD and generalized anxiety disorders appear to be associated with restrictive AN [80, 93]. It must be noted, however, that OCD, social phobia and generalized anxiety disorders may initially manifest in childhood and precede the development of AN [57, 58, 118]. Thus, unlike depression, these anxiety disorders may in fact be precursors rather than effects of AN, which may make them more resistant to change, or may even reflect an underlying genetic predisposition toward dysfunction rather than a mere effect of malnutrition [93].

There is also, however, evidence that AN may itself trigger anxiety symptoms [40], possibly from effects of the reduced weight [93, 107]. Regardless of which symptoms, anxiety or eating, appeared first, the presence of an anxiety disorder such as OCD has been shown to be associated with more negative outcomes in therapy for AN [132]. Moreover, a review of the association of anxiety with treatment outcome in AN patients found that changes in anxiety levels were not associated with changes in AN symptomatology during treatment [62]. Improving anxiety symptoms may not contribute to better eating disorder therapy outcomes and vice versa. Thus, it is not yet clear how anxiety disorders, which may precede the emergence of AN, and thus be unrelated to malnutrition, are related to AN symptoms or treatment response [50, 93, 114]. The one study that looked directly at eating disorder patients with anxiety, depression, both or neither in order to compare levels of symptomatology found that those with anxiety only were similar to patients with no comorbid disorder, while patients with depression alone were similar to those with both depression and an anxiety disorder, and were all more symptomatic than the non-comorbid group [50]. Thus, the limited data on anxiety disorders and AN suggest that anxiety is not as influential a feature of AN severity as is depression [50]. Is the anxiety the first manifestation of the disorder, or does it make a person more susceptible to developing a further disorder such as AN? Pallister and Waller [93] argue that there is evidence of a shared vulnerability for the anxiety disorders and eating disorders, which should be specifically addressed in treatment.

Manuel and Wade [79] noted that AN patients were not only more likely to suffer from depression and anxiety, but they also had more difficulties with regulating their emotions more generally. For example, patients with AN have been shown to have higher emotion avoidance, possibly as a means of escaping negative emotional states [130]. Patients with AN also often have difficulties with emotion regulation [69, 85] and use less effective strategies, including eating less in response to negative emotions in order to manage the aversive emotional states [85]. AN patients who have more difficulty in regulating their emotions and who also suffered a childhood trauma were particularly likely to do poorly in treatment, as childhood traumas appear to be associated with later difficulties in emotion regulation in this population [18]. In addition, it has been shown that patients who complete an intensive treatment program and manage to restore their weight significantly, also show improved emotion regulation, which was itself correlated with improvement in eating disorder behaviors during treatment [102]. On the other hand, difficulties in identifying positive emotional states or differentiating between emotional states was associated with increased levels of restricted eating and weight loss-related behavior, which then elevated mood in these patients. [106]. Those AN patients who are less able to identify their emotions may thus be more prone to see food restriction and weight loss as a means to make themselves feel better.

A marked fear of failure, and early indications of reluctance to engage in interpersonal relationships is at times confused by the suggestion that AN is somehow implicated in Autism Spectrum Disorder (ASD)—although this is not to say that some with AN may in fact have a dual diagnosis in this regard. While comorbidity of these disorders is possible and does occur, the symptoms of AN do not automatically indicate that ASD is also present. Alexithymia has also been used to explain the difficulties that AN patients have in differentiating emotional states [27].

Addictive behaviors are also comorbid with AN. Increased rates of substance abuse have been reported in those with AN, particularly those with binge eating and purging [64, 100]. When substance abuse is comorbid with AN it has been shown to increase problems with recovery [33] and to lead to greater mortality [35, 36, 46, 55, 94], although alcohol abuse in particular may be more an effect than a cause of AN, particularly in those already suffering from depression [34]. The use of cocaine or amphetamines may also represent attempts to moderate mood in patients unable to regulate their emotions effectively, in addition to being a weight control strategy [33].

Early traumatic experiences such as physical or sexual abuse, and vicious teasing seem to contribute to the development of anorexia nervosa, as well as other eating disorders (e.g., [5, 6, 9, 52, 88, 99]), and psychiatric disorders in general (e.g., [22, 123]). Women with AN who had experienced childhood sexual abuse were also more likely to harm or injure themselves impulsively [31]. Sjogren et al. [109] found that traumas were common in patients with moderate to severe AN, and were correlated with symptoms at the start of treatment, although this was not true later in treatment. Mares and Vroling [82], however, found that patients who experienced abuse or trauma at a young age did have longer-lasting AN.

Other external correlates and/or potential triggers for AN have been reported over the last few decades. Both past and recent studies find that environmental stressors, societal pressure to be thin, internalized belief in a thin ideal, teasing about one’s weight or shape, and even something as simple as health education classes are potential triggers of AN, possibly through the induction of negative mood states (e.g., [76, 99]). In addition, exposure to social pressures has been amplified with the advent of social media algorithms [103]. Similarly, those reacting to social pressures with lowered self-esteem and elevated body dissatisfaction have been found to be more likely to develop eating disorders and these reactions are presumed to have causal status, although the evidence tends to be correlational rather than causal [99]. All of these correlates, however, are factors present in large segments of the population, most of whom do not develop AN or any other eating disorder. These “potential triggers” resemble childhood abuse in that they are associated with AN or other eating disorders only in a small number of those exposed to the trigger or stressor. Thus, we need to look further, possibly to the psychiatric co-morbidities or other underlying susceptibilities, such as those described above for those who react to abuse by developing AN.

Individuals’ internal, cognitive behaviors or personality traits have also been implicated in the development of AN, and may help to clarify who will react to the external triggers above by developing an eating disorder. Rigid, obsessive thoughts (often but not necessarily about one’s body, weight, or appearance) and perfectionism often characterize individuals with AN (e.g., [39, 125]), and such association has also led to causal theorizing [99]. Similarly, AN patients display cognitive biases toward food words and images [99], but again, that does not necessarily indicate that such biases are etiological rather than a result of the disorder [44]. Based on clinical theorizing about what characterizes eating disorders, the Eating Disorder Inventory [38] assesses the presence of such internal traits as a lack of interoceptive awareness (i.e., an inability to identify one’s internal feelings), feelings of ineffectiveness (feeling worthless, out of control of one’s life), perfectionism, and interpersonal distrust (difficulty developing close personal relationships and general feelings of alienation) as indicators of AN. Research confirms that such personality traits as perfectionism, obsessive-compulsiveness, neuroticism, negative emotionality, harm avoidance, low self-directedness, low cooperativeness, poor interoceptive awareness, and feelings of ineffectiveness are likely predisposing and possible maintaining factors (e.g., [17, 30, 75]). These personality characteristics not only identify those with the disorder,they may indicate a susceptibility towards developing it [6, 17, 75]. Moreover, as mentioned earlier, those who chronically have difficulty expressing, identifying, and managing their own feelings or emotions also seem to be more susceptible to developing an ED in response to appropriate triggers [6].

Thus, studies on the origin or triggers of AN identify traumatic events (physical, sexual, or psychological abuse) or environmental/external influences as important precursors of the disorder. In addition, an equally large number of studies identify inborn personality and other genetic factors and psychiatric comorbidities that appear to predispose some individuals to develop a disorder such as AN. While this suggests that both types of factors may be present or even necessary for the development of AN, might these varied etiologies suggest a different way to classify AN, based on the extent of environmental triggering versus degrees of underlying physiological susceptibilities to the disorder? Could some individuals be physiologically predisposed to develop a disorder, with external factors merely directing how the disorder manifests (e.g., as AN or as depression)? And might other individuals be less physiologically pushed but still be somewhat susceptible due to personality tendencies to external triggers such as abuse, teasing, or other stressors? As we have theorized recently, might AN be similar to the old conception of depression as either reactive (based on environmental stressors overwhelming an individual’s defences) or endogenous (based on underlying brain structure or other physiological differences) [54, 74, 118]. And might these different pathways to AN call for different treatment approaches and lead to different courses and outcomes of the disorder? In addition, there may be other resilience factors possibly based on inherent personality or on environmental supports which should be taken into account when implementing treatments as well [20, 48]. We will consider this further but first must address the related question of progress of the disorder.

Who responds to treatment and recovers from AN and who does not?

Perhaps we need to reexamine the pathways to AN identified above from a different perspective, namely how the different pathways affect long-term prognosis for AN patients [105].

What interferes with recovery?

As we discussed earlier, Bruce and Steiger [11] reported that some personality disorders and personality characteristics such as perfectionism, impulse control problems, mood lability and interpersonal conflicts may be associated with difficulties in treatment and recovery. Studies on eating disorders in general have found that such personality traits do seem associated with different outcomes (e.g., [68, 73]). Moreover, as the literature shows, the presence of personality disorders in AN patients may even increase the risk of mortality [46, 55] thus indicating a serious lack of treatment response in such patients.

Similarly, comorbid depression also seems to interfere with treatment outcomes [122], if not early on (e.g., [15, 84]), certainly over the longer term [28, 36, 61, 127]. In fact, depression has been identified as the, or one of the best predictors of (negative) outcome in AN in several studies [46, 61, 127]. In fact, any difficulties with emotional regulation in AN patients seem to be associated with less positive outcomes for AN patients than an ability to express or control negative emotions [18, 102]. Adding trauma or abuse to the presence of comorbid personality or affective disorders or to emotion regulation issues also predicts worse therapeutic outcomes for AN [18, 92, 124]. Similarly, the presence of substance abuse is a negative predictor for outcome for AN patients [33, 35, 36, 46, 47, 55, 94].

What contributes to recovery?

The factors described earlier as pathways to AN also seem to predict negative outcomes for therapy. Who then develops the disorder but recovers? Looking at adolescents who had been sexually and/or physically abused, Ackard et al. [1] found that the ones who talked about their abuse with friends or parents were less likely to report eating disordered behaviors than were those who did not manage to discuss their traumatic experiences. Similarly, a study of sexual harassment in working women found that such harassment and trauma were linked to the development of eating disorder symptoms particularly in women who blamed themselves and felt most distressed [43]. A recent study of mediators between early abuse and development of AN found that impulsivity, a sense of ineffectiveness, and difficulty identifying one’s emotions were associated with the development of the disorder [6]. A 5-year prospective study found that avoidant personality disorder combined with childhood abuse predicted an elevated level of eating disorder symptomatology over the long term, even after inpatient therapy [124]. Therapeutic outcomes in abused people who develop AN seem to reflect similar personality patterns to those described above. A study of therapy outcomes in abused eating disorder patients found a pattern of positive outcomes in those who faced their traumas versus negative outcomes in those who avoided them [92], as well as worse outcomes in those who distrusted themselves, and were unable to show vulnerability with their therapists. As described above, treatment of AN is less effective in abused patients who have emotion regulation problems [18]. Female patients reporting greater childhood traumas who also suffered from emotional regulation difficulties were the ones for whom treatment was least likely to be effective [18].

Such positive outcomes for traumatized or abused AN patients who were able to use therapy to help them to face their traumatic experiences fits with the data from a study of AN (and of bulimia nervosa/BN) patients in treatment that found that those who did well and completed treatment improved not only their weight and eating behaviors, but their management of emotions, which was itself correlated with better outcomes [102]. An earlier study found that level of psychopathology and social dysfunction, along with age of onset of the disorder, presence of purging behavior, and two blood-related (physiological) factors reliably predicted which patients would improve and which would remain disordered, offering more evidence for the contribution to outcome of initial personality and social functioning levels [24]. Some people are thus more able to use what psychotherapy offers to help themselves to recover from AN.

This ability to utilize a therapeutic relationship to effect self-change may well reflect pre-existing personality differences. As mentioned earlier, Westen and Harnden-Fischer [125] found three types of AN patients: high-functioning, perfectionistic individuals, highly constricted and overcontrolled individuals, and individuals who are emotionally and behaviorally poorly controlled. While AN patients who also had bulimia nervosa (AN-BN) were present in all three personality groups, AN patients without BN did not appear in the behaviorally uncontrolled group. The authors pointed out that ignoring personality factors when classifying AN patients means that those who are functioning in their lives at a high level, but are simply self-critical are grouped together with and treated the same as other AN patients who are emotionally disturbed, have true personality disorders, and are not functioning well at all. Clearly these three groups of patients should not be treated similarly.

To test this idea, Thompson-Brenner and Westen [117] looked at data provided by knowledgeable clinicians on 145 eating disorder patients (not necessarily AN). They then categorised the patients and examined the outcome data. Those who had initially been seen as having poor emotional regulation did indeed function poorly, were more symptomatic, and improved the least over the course of therapy. A middle group of patients corresponding to their earlier characterization of being constricted and over-controlled, but not dysfunctional, did somewhat better, and the patients who were initially highest in emotional functioning improved the most. This pattern held across severity of initial eating disorder symptoms at presentation and the presence of psychiatric co-morbidities. A later study further confirmed this grouping and outcomes [115, 116]. Other researchers using different measuring instruments found a similar pattern of personality subtypes in a group of AN patients [70], with a less symptomatic subgroup of patients with relatively normal personality features, a more symptomatic group who they called overregulated who were higher in OCD and perfectionism but had only moderately greater negative affect and symptoms compared to their most symptomatic group, termed underregulated, who were most symptomatic, were called “oppositional” and most likely to engage in self-harm and other impulsive negative behaviors, and who exhibited the most negative and least positive affect. Finally, a recent study found that patients with emotionally dysregulated personality features did significantly worse in treatment, while those with healthy personality structures did well in treatment [91].

Identifying subtypes of AN that lead to differential responses to treatment

It thus seems that we should be taking personality strengths and weaknesses into account when predicting who will develop AN and when classifying AN patients for treatment [21]. In fact, work has been proceeding exploring the effects of personality on AN therapy for some time (see e.g., [81] for a review). Further studies of personality disorders and issues using Westen’s scale [126] and/or theorizing have confirmed Westen and Harnden-Fischer [125] finding of three types of personalities in AN patients (highly functioning but perfectionistic, restricted and over-controlled, and emotionally and behaviorally poorly controlled), each of which had its own affective, identity, and comorbidity constellations, which had implications for prognosis in the different groups (e.g., [39, 121]). One such study [115, 116] found the same three groups, plus two others (avoidant-insecure, and obsessional-sensitive). Again, while the more highly functioning patients did well in therapy and improved markedly, the restricted and dysregulated groups had poor outcomes (the two new groups this study identified were not associated with particularly negative outcomes). Another study of the effects of personality styles on therapy outcomes for AN patients similarly identified patients as under-controlled, over-controlled, and low in psychopathology and found that the low pathology patients did best in therapy, the overcontrolled group did worse than the first group, but the under-controlled patients fared the worst, with the poorest outcomes and the greatest likelihood of giving up on treatment altogether [129]. It should be noted, however, that these distinctions did not predict readmission rates as expected, with the low psychopathology patients being readmitted almost as much as the under-controlled group, so personality factors help predict outcomes of therapy, but are not sufficient on their own to predict long-term prognosis.

Characteristic personality traits or temperament do seem to influence not only the development of AN, but also therapy outcomes in AN patients. For the last decade or more, temperament has been taken into account in treatment in several studies (e.g., [45, 59, 81, 96, 128]). Kaye and his colleagues [59] were among the first to work on developing a specifically temperament-based treatment for AN, focusing their intervention to target traits that contribute to AN symptomatology, while keeping in mind underlying neuropsychological substrates of these traits and related symptoms. While not necessarily the main or first-line treatment, treatments based on consideration of AN patients’ temperaments have proven to be successful adjuncts and may even involve parental or family inclusion (e.g., [96]). Martinez and Craighead’s [81] review of the use of treatments for AN targeting specific personality traits suggests that such a focus may make treatment both more acceptable to patients and more effective.

Individual differences in personality can help to differentiate those who will do well in treatment from those who will struggle, and targeting treatment to these personality differences can assist in making therapy for AN more successful, but this information alone does not seem sufficient to fully direct treatment (e.g., [96, 128]). The question is then what other qualities of individuals might we also target to improve our therapeutic outcomes? We have suggested recently [118] that one aspect of AN that might be studied further for its implications for development and treatment of the disorder concerns essentially how the disorder begins, specifically whether it appears to be a reaction on the part of the patient to trauma, teasing, or other external influences (i.e., Reactive AN), or somehow more physically-based, genetic, or endogenous to the individual (i.e., Endogenous AN). As we proposed, for those AN patients for whom there appears to be an identifiable incident, event, or circumstance (such as chronic abuse, teasing, environmental pressures, etc.) that triggered or precipitated the onset of the disorder (i.e., Reactive AN), therapy targeted to acknowledging the precipitant and working through it and its implications for the individual may be most effective. For example, addressing abuse in therapy and making that, rather than eating, the focus of therapy may be most helpful for those patients whose AN seems to have resulted from the trauma of abuse (e.g., [10]). Moreover, personality variables such as willingness to use therapy in order to change (e.g., [92]), and ability to regulate and express emotional responses [18, 102] both appear to be associated with positive outcomes for AN patients, suggesting that these are the ones most able to work through their precipitating problems and recover from the disorder. These individuals seem more likely to fall into the higher functioning category identified earlier, although this may not always be the case. On the other hand, AN patients with underlying genetic/somatic contributors (such as comorbidities like depression or personality disorders) (e.g., [13, 23, 56, 65]), or those exhibiting characterological dysfunction, or pre-existing personality issues (i.e., Endogenous AN), do not have a specific issue that can be targeted in talk therapy, and may need more somatic interventions such as drug therapies, psychedelics, rTMS or even deep brain stimulation (e.g., [47, 60, 77, 104, 112, 120]). These patients are also more likely to fit into the restricted and dysregulated groups identified in the research cited earlier (e.g., [81, 126]).

The reactive-endogenous distinction raises the question of whether severe and enduring, treatment-resistant AN could be the result of endogenous genetic and/or personality factors. Duncan et al. [25] in their ground-breaking genome-wide association study in AN, have identified some key genetic markers for AN and on the basis of their results, reconceptualised AN as both a psychiatric and metabolic disorder. The importance of these findings underpinning the likely genetic contributions to the development of AN appear to have fallen under the radar and have not received the recognition that they deserve. This is a further illustration of how the field of eating disorders has continued to embrace the existing zeitgeist rather than advocating change when new data clearly warrants a reconceptualization [118]. When treatment focuses solely on either body image, eating issues, or sociocultural pressures, it does not take into account that the disorder may reflect genetic or other biological substrates that must be addressed in order for treatment to be effective. The success of new therapies such as psychedelic assisted psychotherapy (PAP) may reflect their targeting of neurobiological systems underlying the disorder such as the REBUS model proposed by Carhart-Harris and Friston [16], Koning and Brietzke [66], and Lacroix et al. [67]. The REBUS model (Relaxing Beliefs Under Psychedelics) targets the Default Mode Network to allow better integration of brain regions and to allow for more bottom-up processing than top down (brain as a prediction engine) experiences. Singleton et al. [108] have also proposed that psychedelics may in fact flatten the brain’s energy landscape which once again would allow for an increase in entropy of brain activity and thus provide the necessary flexibility for psychological growth. Such a psychedelic journey seems to encourage brain and emotional plasticity, such that individuals have reported developing new meanings of life and deepened understanding of themselves. This is consistent with the renewed interest in phenomenological psychopathology [12, 42, 51, 86, 95]. Attending clinically to the individualised journey of each AN individual undergoing PAP appears to be crucial. Phenomenology remains poorly understood in psychiatry, and iterations of DSM focus on psychopathology which is inherently different from phenomenology [51, 95]. AN is a probably the most complex of all psychiatric illnesses and incorporates many aspects of other psychiatric illnesses. Rather than being co-morbid with much of DSM-5, AN is very probably a constellation of characteristics found in several DSM disorders. At the 2025 European Council on Eating Disorders Congress in Budapest, several seminal papers made the important point as to how anorexia nervosa was comorbid with much of psychiatry. The time has now come to consider a deep dive into the phenomenology of anorexia nervosa which we believe will show it to be a highly complex psychiatric disorder as recognised by Lasegue nearly 150 years ago [119]. The individual’s own meaning of their illness may be part of what is missing from current therapies, contributing to the lack of success of current treatments. Much thought needs to be given to the reconceptualisation of AN [86].

In conclusion, we propose adding to the targeted therapies for AN a focus on a constellation of patient characteristics that include not only personality factors but underlying causal contributors to the disorder based on reactive/external triggers versus endogenous/physiological predispositions [54]. Providing therapies that address the underlying causal and maintaining issues may help us to improve our therapeutic success rates at long last. We have been focusing our therapies on a behavioral difference (binge/purge vs. restrict) but still treat the disorder as unitary (fat phobia, desire for thinness, body image distortion, etc.). We have to see that there are underlying differences in triggers, responses, and physiology/genetics that affect both outcome and which treatment will work for whom [105]. Touyz et al. [118] have alluded to the huge individual differences in aspects of anorexia nervosa patients including genetic markers for AN, in emotional responsivity or ability to recognize and respond to emotions in oneself and others, in brain plasticity and ability to integrate new experiences, in the phenomenological meaning of anorexic beliefs and behaviors to the patient [86] and in the life experiences that felt triggering or traumatic for the patient [105]. We need to be more aware of these differences in the path to AN, the experience of AN by the individual, and both the underlying genetic or physiological predispositions and the external traumas, abuse or fears that contribute to the development of AN in order to better match appropriate treatments to the individual.

Conceptualising AN as a unitary, homogeneous disorder now seems increasingly misguided, and the steadfast observance of our cherished treatment manuals may have to be reconsidered [14]. In the minds of many, this would constitute heresy, and those advocating such a departure from our current accepted norms worthy of the strongest criticism. This is not to say that some AN patients will not respond to our existing evidence-based treatments and go on to a complete recovery—they do that sometimes, but we are extremely poor at predicting which treatments work for whom, and our overall outcomes remain essentially unchanged from what they were 50 years ago [105, 118]. Adaptive and platform trials currently underway should help to clarify this. But as the world gravitates towards personalised medicine (precision psychiatry), we should also endeavour to do so for those with AN, but only once we fully understand the complex nature of this debilitating disorder [54, 74]. We are not there yet and there is an ever-increasing risk that clinicians will embrace the new world order of personalised medicine utilising their own choice of integrated therapies without fully grasping the complexities of the tasks at hand. The first rule of medicine is “primum non nocere” and in our rush to do better, we may in fact cause harm. If there is one certainty moving forward, it is that AN is not a homogeneous disorder. It will require more effort to understand the complexities of the variations of the disorder in order to develop individualised treatment strategies and then determine what works for whom. The road ahead is full of exciting possibilities but only if the potholes can be avoided.

Author contributions

JP: conceptualization, writing—original draft preparation, writing—review and editing; ST: conceptualization, writing—original draft preparation, writing—review and editing.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Availability of data and materials

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

J.P. receives royalties from PAR Associates for the EDI, and from Hogrefe and Huber for a published book. S.T. receives royalties from Taylor and Francis, Hogrefe and Huber and McGraw Hill for published book chapters. He has received honoraria from Shire/ Takeda Group of Companies for chairing the Australian Clinical Advisory Board for Binge Eating Disorder, public speaking engagements, commissioned reports as well as investigator- initiated research grants. He is an Editor in Chief of the Journal of Eating Disorders, an inaugural committee member of the National Eating Disorders Collaboration, a Member of the Technical Advisory Group (TAG) on Eating Disorders (Commonwealth of Australia) and a member of the governing council of the Australian Eating Disorders Research and Translation Centre.