Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jan 4.
Published in final edited form as: Biol Psychiatry. 2025 Jan 17;98(3):230–236. doi: 10.1016/j.biopsych.2025.01.008

Descriptives and Genetic Correlates of Eating Disorder Diagnostic Transitions and Presumed Remission in the Danish Registry

Mohamed Abdulkadir 1, Janne Tidselbak Larsen 1, Loa Clausen 1, Christopher Hübel 1, Clara Albiñana 1, Laura M Thornton 1, Bjarni J Vilhjálmsson 1, Cynthia M Bulik 1, Zeynep Yilmaz 1, Liselotte Vogdrup Petersen 1
PMCID: PMC12759185  NIHMSID: NIHMS2123982  PMID: 39827937

Abstract

BACKGROUND:

Eating disorders (EDs) are serious psychiatric disorders with an estimated 3.3 million healthy life-years lost worldwide yearly. Understanding the course of illness, diagnostic transitions and remission, and their associated genetic correlates could inform both ED etiology and treatment. We investigated occurrences of ED transitions and presumed remission and their genetic correlates as captured by polygenic scores (PGSs) in a large Danish register-based cohort.

METHODS:

The sample comprised 10,565 individuals with a diagnosis of anorexia nervosa (AN), bulimia nervosa (BN), or eating disorder not otherwise specified (EDNOS) and with at least two registered hospital contacts between 1995 and 2018. Based on medical records, the occurrence of diagnostic transitions and periods of presumed remission were identified. Associations between 422 PGSs and diagnostic transitions and presumed remission were evaluated using Cox proportional hazard models.

RESULTS:

A minority of ED cases (14.1%–23.1%) experienced a diagnostic transition. Rates of presumed remission ranged between 86.9% and 89.8%. Higher (1 SD increase) PGSs for major depressive disorder and multisite chronic pain were positively associated with transitioning from AN to either BN or EDNOS. Higher PGSs for a measure of body fat percentage and financial difficulties were positively associated with presumed remission from AN. Having a higher PGS for mood swings was positively associated with presumed remission from EDNOS whereas higher PGS for overall health rating showed the opposite.

CONCLUSIONS:

We found that most patients with an ED did not experience diagnostic transitions but were more likely to experience a period of presumed remission. Both diagnostic transitions and presumed remission have a significant polygenic component.

Eating disorders (EDs) are serious psychiatric disorders that can have a complicated course of illness characterized by diagnostic transition from one ED to another and eventually remission for some patients (1–6). It is estimated that 3.3 million healthy life-years are lost annually worldwide due to anorexia nervosa (AN), bulimia nervosa (BN), and ED not otherwise specified (EDNOS) (7). Studies have reported varying frequencies of diagnostic transitions across EDs. For example, some case-cohort and clinical studies have found that the transition from AN to BN occurs in 13% to 54% of cases, and the transition from BN to AN occurs in 7% to 27% of cases (1–3,6). However, a large Swedish population register study (8) reported much lower transition probabilities (2%) from AN to BN and vice versa, with the transition to EDNOS occurring in 17% to 24% of AN cases and 18% to 22% of BN cases (3,8). Estimates of ED remission have also differed in the literature, ranging from 27% to 78% for AN cases and 40% to 83% for BN cases, and approximately 49% of EDNOS cases in studies with a minimum follow-up of 6 years (2,5,8). These differences could be due to variations in study design, assessment methods, or follow-up times.

It is unclear what factors determine a patient’s risk of transitioning from one ED to another or their duration of illness. Previous research suggests that greater experience of psychopathology (e.g., anxiety) may increase the likelihood of transitioning from AN to BN (1,6,9). Anthropometric factors such as body mass index (BMI) may also play a role because individuals who transition from AN to BN tend to have higher current, past minimum, and past maximum BMIs than those who do not transition (9). Similarly, higher BMI and traits such as premorbid perfectionism, state anxiety, and trait anxiety are associated with a lower occurrence of ED remission (10–12).

One interesting approach to understanding biological factors that may influence ED transition and remission is to explore genetic factors that contribute to the risk of EDs (4,13). Most genome-wide association studies (GWASs) of EDs have focused on AN (14–17) and have not examined genetic factors that influence diagnostic transition or remission. Polygenic scores (PGSs) offer a powerful method to leverage the small-to-moderate effect sizes of common single nucleotide polymorphisms (SNPs) that have been identified in a GWAS into a single continuous score that surpasses the explanatory power of individual SNPs (4). This approach has been successful in exploring the genetic relationship between traits by associating the PGS for one trait (e.g., BMI) with a phenotypic measurement of another trait (18). For example, we have previously shown that a higher BMI PGS is associated with ED symptoms in the general population (18). Novel associations that can inform us about shared etiology between 2 traits can lead to the generation of new hypotheses regarding the nature of associations and underlying shared mechanisms.

Therefore, the aims of our study were 2-fold: 1) to quantify occurrences of ED transitions and remission and 2) to investigate whether diagnostic transitions and ED remission are associated with PGSs of various complex traits in a nationwide sample of treatment-seeking individuals in Denmark (19).

METHODS AND MATERIALS

Study Population

Using data from the Danish National Patient Register (NPR) (20) and the Psychiatric Central Research Register (PCRR) (21), the study population comprised individuals 1) who were born in Denmark between May 1, 1981, and December 31, 2009; 2) who were alive and living in Denmark on their sixth birthday; 3) who received an ICD-10 primary or secondary diagnosis of AN (F50.0, F50.1), BN (F50.2, F50.3), or EDNOS (F50.8, F50.9) on or after their sixth birthday; 4) for whom genotype data were available; and 5) who had at least 2 registered ED-related hospital contacts between 1995 and 2018 after the age of 6 years. Genotype data for EDNOS cases without any other lifetime ED diagnosis were obtained from the iPSYCH2015 case-cohort sample (also including the iPSYCH2012 case cohort) (19,22), for AN cases from the Danish branch of the ANGI-DK (Anorexia Nervosa Genetics Initiative in Denmark) (23), and for BN cases from the Danish branch of the EDGI-DK (Eating Disorders Genetics Initiative in Denmark) (24). In total, we included 10,565 individuals with EDs, with their first recorded diagnosis being AN (n = 6901), BN (n = 1417), or EDNOS (n = 2247). Each individual was followed up until December 31, 2018. Sex information was obtained from the NPR and PCRR. Sex is assigned at birth; however, each individual over the age of 18 years can legally change their sex information in the registers based on their gender identity. Therefore, from here onward we report on gender. Ancestry was based on genetic information and was determined by calculating robust Mahalanobis (squared) distances using the first 2 genomic principal components (25). This approach relies on the fact that the Danish population is fairly homogeneous (22), and calculating robust distance captures individuals of European ancestry.

Outcome Data

ED Episodes.

From the inpatient and outpatient hospital contacts in the NPR and the PCRR, we defined distinct ED episodes for each individual (for detailed information, see Supplemental Methods and Figure S1). First, we combined patient records into distinct hospital contacts. We further merged contacts into ED episodes (i.e., a time period during which an individual had a diagnosis of an ED). The merging of records into contacts and episodes resulted in 3 possible ED episodes: AN, BN, or EDNOS. We also included an episode of presumed remission for individuals who were (hospital) contact free regarding their ED for at least 2 years.

Time to Diagnostic Transition.

Based on the defined ED episodes, we first grouped individuals by their first ED diagnosis. Second, we defined a diagnostic transition as an individual receiving another ED diagnosis in a subsequent ED episode. Time to diagnostic transition was calculated as the difference between the beginning of the first ED episode and the beginning of the episode in which the individual received a different ED diagnosis.

Time to Presumed Remission.

Time to presumed remission was calculated as the time difference between the date presumed remission is possible (2 years following the start of the first episode) and the date when the individual met the abovementioned criteria for presumed remission.

Genetic Data

Genotyping.

Dried blood spot samples were obtained through routine screening for congenital disorders from nearly all Danish newborns after 1981 and stored at the Danish Newborn Screening Biobank (26). Genotyping of the DNA extracted from the blood spot samples was carried out and has been described elsewhere (19,22,23,26). This study was approved by the Danish Data Protection Agency, the Danish Scientific Ethics Committee, the Danish Health Data Authority, and the Danish Newborn Screening Biobank Steering Committee. The Danish Scientific Ethics Committee, in accordance with Danish legislation, has, for this study, waived the need for informed consent in biomedical research based on existing biobanks (19).

PGS Calculations.

PGSs were calculated for ED cases with genetic data (N cases = 10,565) using LDpred2 (27) for 940 traits and with the meta-polygenic risk score (28) approach for 6 traits (i.e., attention-deficit/hyperactivity disorder, AN, autism spectrum disorder, bipolar disorder, major depression, and schizophrenia). We performed prefiltering on 940 LDpred2-calculated PGSs (e.g., duplicates and highly correlated PGSs) (for details, see Supplemental Methods and Figure S2). After filtering, a total of 422 PGSs remained for analyses and were standardized (Table S1).

Statistical Analyses

We tested the association between PGSs (one at a time) and our outcome measures using Cox proportional hazard models. We did not exclude individuals based on genetic ancestry and instead controlled for population stratification using the first 5 genomic principal components as covariates in the Cox regression models. We did not include transitions from BN to AN or EDNOS or from EDNOS to AN or BN in our PGS analysis due to the infrequent occurrence of these transitions in our sample (Table 1). We estimated hazard ratios with 95% CIs for each outcome for 1 SD increase in the PGS. We corrected for the number of PGSs tested (n = 422) by calculating the number of effective tests using the meff() function (PoolR package) on the genetic correlation matrix of the selected 422 PGSs applying the Galwey method (29). The number of effective tests was estimated to be 317; therefore, the threshold for multiple testing was set at α < 1.57 × 10−4.

Table 1.

Descriptive Statistics for the Sample by First ED Diagnosis

Total Sample, N = 10,565 AN as First Episode, n = 6901 BN as First Episode, n = 1417 EDNOS as First Episode, n = 2247
Binary Descriptives
Gender, Women 9784 (92.6%) 6425 (93.1%) 1381 (97.5%) 1978 (88.0%)
European Ancestry 9037 (85.5%) 5952 (86.2%) 1201 (84.8%) 1884 (83.8%)
Transition to
 BN – 370 (5.4%) – 153 (6.8%)
 EDNOS – 661 (9.6%) 175 (12.3%) –
 AN – – 165 (11.6%) 404 (17.8%)
 BN or EDNOS – 968 (14.1%) – –
 AN or EDNOS – – 309 (21.8%) –
 AN or BN – – – 521 (23.1%)
Presumed Remission 9273 (87.7%) 5999 (86.9%) 1273 (89.8%) 2001 (89.0%)
Relapsed From Presumed Remission 1585 (15.0%) 970 (14.0%) 290 (20.5%) 325 (14.5%)
Continuous Descriptives
Age at First Diagnosis, Years 18.1 (4.5) 17.3 (4.1) 20.9 (4.3) 18.7 (5.1)
Follow-Up Time, Years 8.9 (5.2) 9.0 (5.4) 9.1 (4.9) 8.2 (4.9)
Time to Transition, Years
 BN – 4.5 (3.3) – 3.6 (3.1)
 EDNOS – 4.7 (3.8) 4.6 (3.6) –
 AN – – 4.0 (3.4) 2.6 (2.5)
 BN or EDNOS – 4.5 (3.5) – –
 AN or EDNOS – – 4.0 (3.2) –
 AN or BN – – – 2.7 (2.7)
Time to Presumed Remission, Years 1.6 (1.8) 1.7 (1.9) 1.7 (1.76) 1.2 (1.6)
Duration of Presumed Remission, Years 5.3 (4.7) 5.4 (4.9) 4.9 (4.3) 5.0 (4.3)
Open in a new tab

Values are presented as n (%) or mean (SD).

AN, anorexia nervosa; BN, bulimia nervosa; EDNOS, eating disorder not otherwise specified.

RESULTS

Sample Description

Demographics.

We included a total of 10,565 individuals (92.6% women; 85.5% European ancestry) with genotype and ED diagnostic data in our analyses (Table 1): 6901 individuals with AN, 1417 individuals with BN, and 2247 individuals with EDNOS as their first episode. The mean age at first diagnosis for AN, BN, and EDNOS in our sample was 17.3 (SD 4.1), 20.9 (SD 4.3), and 18.7 (SD 5.1) years, respectively. The mean follow-up time was 8.9 years (SD 5.2).

Transitions.

Among individuals with AN as their first episode, 14.1% transitioned to another ED, with such a transition taking an average of 4.5 years (Table 1). The transition from AN to BN occurred in 5.4% of individuals. A higher percentage of individuals with AN (9.7%) transitioned to EDNOS, with a mean time to transition of 4.7 years. Among individuals with BN as their first episode, 21.8% individuals transitioned to AN or EDNOS: specifically, 11.6% transitioned to AN and 12.3% transitioned to EDNOS, with an average transition time of 4 and 4.6 years, respectively. Individuals who presented with EDNOS as their first episode showed the highest frequency of transitions; 23.1% of the individuals transitioned to AN or BN, with a mean transition time of 2.7 years. More EDNOS cases transitioned to AN (17.8%) than to BN (6.8%).

Presumed Remission.

Regardless of ED diagnosis at first episode, 87.7% of individuals experienced a period of presumed remission, with a mean time to presumed remission of 1.6 years (SD 1.8) (Table 1) and an average duration of the presumed remission period of 5.3 years (SD 4.7). Presumed remission occurred in 86.9% of individuals with AN as their first episode, with a mean time to presumed remission of 1.7 years (SD 1.9). Among individuals with BN as their first episode, 89.8% experienced a period of presumed remission, with a mean time to presumed remission of 1.7 years (SD 1.7). Of the individuals with EDNOS as their first episode, 89.0% experienced a period of presumed remission, with a mean time to remission of 1.2 years (SD 1.6). A minority (15%) of the cases experienced an ED diagnosis following a period of presumed remission; relapse was highest (20.5%) among those with a BN diagnosis as their first episode.

Association of PGSs With the Transition From AN to BN or EDNOS

Higher major depressive disorder (MDD) and higher multisite chronic pain PGSs were significantly associated with a 15% greater hazard of transitioning from AN to BN or EDNOS, respectively (Figure 1A; see Table S2 for full results).

Figure 1.

Figure 1.

Open in a new tab

Associations between polygenic scores (PGSs) and eating disorder diagnostic transition and presumed remission. Depicted are the associations that remained significant after correcting for multiple hypothesis testing. Associations are depicted as hazard ratios (HRs) per 1 SD increase in the PGS, with dots representing the HR point estimate and the line depicting the 95% CI. (A) Associations of PGSs with transitioning from anorexia nervosa (AN) to bulimia nervosa or eating disorder not otherwise specified (EDNOS). (B) PGSs associated with time to presumed remission for AN (in red) and EDNOS (in green).

Association of PGSs With Time to Presumed Remission

Having a higher (1 SD) PGS for a body fat percentage measure (i.e., leg fat percentage) was associated with an 8% greater hazard for presumed remission from AN (Figure 1B and Tables S3–S5). In individuals with AN, having a higher PGS for financial difficulties was also associated with a 5% greater hazard for presumed remission. Regarding EDNOS, each 1 SD increase in the PGS for overall health rating corresponded to a 12% lower hazard for presumed remission while a higher PGS for mood swings was associated with a 10% increased hazard for presumed remission.

DISCUSSION

In this Danish nationwide population study, we examined 10,565 individuals with EDs over an average follow-up time of 9 years. Overall, the occurrence of transitions (ranging between 5.4% and 23.1%) in our sample falls on the lower end of previous estimates (1–3,6). For example, in a study in which AN cases were followed weekly, about a third of all AN cases transitioned to BN during a period of 7 years (2); their transition estimate is much higher than ours (5.4%). However, our estimate of transitions is more comparable to that of another registry-based study that found estimates that ranged between 2% and 24% (8). There could be several explanations for our lower estimates of transitions. It could be that diagnostic transitions occur in between hospital visits and therefore are likely missed in register-based research. It is also possible that patients seek treatment elsewhere (e.g., private practice that is not recorded in the registries that we utilized) or decide to no longer pursue treatment. Despite differences in estimates of threshold ED transitions across studies, the general pattern that has emerged is that diagnostic stability is the norm, and if transitions do occur, they often happen within the first 5 years of illness (1–3,6,8). Furthermore, the relatively infrequent transitions between AN and BN in our study suggest, consistent with previous findings from Eddy et al. (2), that AN and BN represent distinct EDs. In contrast, the relatively more frequent transition from EDNOS to AN suggests that these transitions may be capturing a change in illness stage for individuals who have not met full diagnostic criteria for AN.

Presumed remission was very likely in our sample; depending on the diagnosis of the first episode, between 86.9% and 89.8% of individuals experienced a period of presumed remission, with only a minority (14.0%–20.5%) of individuals experiencing relapse, suggesting that this is a relatively stable state. The frequency of presumed remission in our study was comparable to partial recovery frequencies (78.4%–82.8%) reported in a longitudinal study of ED (2) and with remission (89%–95%) in a clinical study (30), suggesting that our definition likely reflects individuals who were on their way to recovery. However, our estimates for presumed remission are much higher than the only other register-based study on this topic, which reported that remission varied between 23% and 36% (8). The definition of remission in the other register-based study was defined as instances in which an ED diagnosis was no longer coded in the register at an assessment visit whereas ours was based on being contact free regarding an ED for at least 2 years. The discrepancy in results is likely due to these differing operationalizations of a remission phenotype and may suggest that our estimates for presumed remission are overestimations. However, the overall conclusion regarding remission in our study and those of the others (2,8,30) remains the same; remission is a more likely outcome than diagnostic transitions, and remission is a fairly stable state.

Regarding our PGS analysis of transitions from AN to BN or EDNOS, higher PGSs for MDD and multisite chronic pain (a phenotype defined as the sum of the number of body sites at which an individual reported chronic pain that lasted at least 3 months) corresponded to a 15% greater hazard of transitioning. Associations between AN and depression are well established, with some of the shared etiology likely due to shared genomics; MDD is genetically positively correlated (rg = 0.28, SE = 0.07, p = 8.95 × 10−25) with AN caseness (16,31,32). Our finding adds to the literature on AN and MDD and suggests that genetic variants that underlie MDD can also influence the course of AN. Similarly, multisite chronic pain has been reported to be genetically negatively correlated (rg = −0.06, SE = 0.03, p = .047) with AN (33). ED disturbances are elevated in patients with chronic pain, and findings from a longitudinal study showed that adolescents with an ED were more likely to experience frequent or persistent pain (34). Furthermore, studies suggest that pain sensitivity may be decreased in individuals with AN and BN compared with non-ED control individuals (35–37). This change in pain perception is likely not due to deficits in interoception [sense of the physiological condition of the body (38)], which are implicated in both AN and BN (39). A recent study found no evidence for a mediation effect for interoception in the association between pain experience and EDs, which suggests that pain perception in AN may be altered through a different physiological system (39). One possible shared biological process that may explain the co-occurrence of chronic pain and ED could be central sensitization, a state in which central nervous system responsiveness is increased to internal and external conditions, which is a process that is dysregulated in both conditions (34).

Regarding our PGS analysis and remission, having a higher PGS for a body fat measure (leg fat percentage) was associated with presumed remission from AN. This finding is of considerable relevance etiologically because the maintenance of low BMI and persistence of dietary restriction are unique and biologically challenging symptom profiles. How some individuals maintain a low BMI throughout the course of their illness versus conceding biologically to increased caloric intake and weight gain may rest in genetic susceptibility to both anthropometric and metabolic traits. Our findings suggest that individuals who harbor genetic variants that increase their bodyweight may counter the symptom profile (i.e., maintaining a low body weight) of AN, and therefore they contribute to presumed remission. We also report that a higher PGS for financial difficulties was associated with remission from AN. Several longitudinal studies have found a relationship between higher parental socioeconomic status (SES) and increased risk of developing AN (40,41). Furthermore, AN has been reported to be positively genetically correlated (rg = 0.20–0.27) with several measures of educational attainment, and a recent PGS study found that having a higher AN PGS was associated with higher SES in parents of AN cases (16,42). Therefore, it is likely that our findings capture the (lower) SES component of financial difficulties and may suggest that SES also plays a role in the course of AN.

Counter to our expectations, we found that having a higher PGS for an overall health rating was negatively associated with remission from EDNOS. It is important to note that about 16% of the cases with EDNOS (as their first ED diagnosis) who reached a state of presumed remission experienced at least one episode of AN. These are individuals who had not yet fulfilled criteria for AN at the time of their first assessment. Therefore, it is possible that the PGS for overall health rating is associated with restrictive eating and excessive exercise in this group, which are characteristics also seen in subthreshold AN cases and characteristics that hinder remission in the context of individuals with an ED. Lastly, we also report a positive association between a higher PGS for mood swings and remission from EDNOS. Mood swings are highly prevalent in patients with EDNOS (43), and therefore it is likely that the clinical treatment for mood swings takes precedence which might appear as presumed remission from EDNOS.

The lack of significant associations between most of the tested PGSs and our outcomes could be due to several reasons. Despite the relatively large sample size, it is possible that our study lacked power to detect associations between the PGSs and transitioning from AN to BN or EDNOS, given the infrequent occurrences of these transitions (n cases who transitioned from AN to BN or EDNOS = 976). Furthermore, it is possible that both AN transitions and ED remission are largely determined by environmental factors (e.g., treatment). Lastly, it is also possible that our transition and remission phenotypes have a unique genetic architecture that is not best captured by our selection of PGSs.

The strengths of this study include its large sample size and the long follow-up period compared with those of previous studies (1–6). However, findings from this study should be interpreted in the context of certain limitations. In Denmark, hospital-based treatment is free of charge, which lowers the barrier for individuals to seek treatment (19). However, we cannot rule out the possibility that sociodemographic factors could nevertheless have influenced treatment-seeking behavior. The diagnoses in the registers could have also been impacted by differing registration practices among hospital departments and that could have impacted our estimates of diagnostic transition and remission over time (44). For example, a clinician might assign a diagnosis of EDNOS as a first diagnosis because threshold AN or BN is still under development. Similarly, a person with AN might, on their way to recovery, go through an EDNOS stage. In the previous examples, the transition to or from EDNOS is likely an artifact of diagnostic practice. Diagnostic practices when both AN and BN criteria are met can vary between hospitals because guidelines are not consistent across diagnostic manuals. However, in our work, we have grouped hospital contacts into distinct ED episodes, and episodes that are temporally close were merged into one episode, which likely dampened the effects of such diagnostic practices. We cannot rule out the possibility that other diagnostic practices might have influenced our results. Our definition of presumed remission relied on being contact free for at least 24 months and therefore is not a direct measure of remission, which is why we refer to this phenotype as presumed remission. Therefore, it possible that some individuals, despite still having an ED, decided not to seek further treatment and therefore were erroneously considered to be in presumed remission in our study. However, it is important to note that previous clinical studies that followed patients at the symptom level found that remission typically occurs within the first 2 years (30,45). Due to a lack of symptom-level information, we were not able to explore how different operationalizations of remission (as previously investigated) could have influenced our PGS association analyses (46). Furthermore, it is important to reiterate that our study might have missed some diagnostic transitions because they occurred between hospital contacts. In addition, our study utilized data from the Danish health registers that work with the ICD-10 classification system, which does not contain information on AN subtypes (i.e., binge eating/purging and restricting); therefore, we could not investigate how transitions specifically occurred from AN subtypes to BN or EDNOS. A previous study has shown that for patients with AN, transitions are more likely to occur to BN from an AN binge eating/purging subtype; however, most individuals transitioned back into AN, suggesting that these transitions likely reflect a change in illness presentation rather than a change in diagnosis (2). Furthermore, in the same study, the authors reported high crossover frequency (~48%) between the 2 AN subtypes, suggesting that the 2 AN subtypes may not represent unique diagnostic groups (2). Genotyped BN or EDNOS cases were more likely to be included in this study if they had a lifetime diagnosis of AN or any of the other psychiatric disorders (e.g., schizophrenia, autism) ascertained for the iPSYCH case-cohort study (19,22). In other words, BN and EDNOS cases without other lifetime psychiatric diagnoses (included in the iPSYCH case-cohort study) were not captured in our sample. Therefore, our findings may not be generalizable to all BN or EDNOS cases. Another limitation is that we are unable to estimate the extent to which hospital treatment might have impacted observed ED transitions and remission. For example, treatment for AN could have led to an increased body weight, and this could have shifted the diagnosis from AN to EDNOS despite the presence of core AN symptoms. Most individuals in this study are of European ancestry, which makes the results less generalizable to non-European populations.

Conclusions

This study supports previous findings that the majority of patients with an ED do not experience diagnostic transitions and that presumed remission is likely regardless of ED diagnosis. This study highlights potential genetic influences that underlie ED transitions and remission but also points toward a possible unique genetic architecture that is not captured by our selection of PGSs. Our study focused on understanding how genetics can explain observed ED transitions and presumed remission. It will be of considerable interest for future studies to investigate environmental (and other genetic) components of ED transitions and remission. For example, one could investigate how a diagnosis of a psychiatric disorder could impact ED transitions and remission while accounting for genetics. Further research on understanding possible environmental influences on ED transitions and remission may give us a deeper understanding of the etiology of EDs and inform tertiary prevention strategies that could lead to improved outcomes.

Supplementary Material

supplementary_info_text
supplementary_tables
supplementary_key_resource_table

Supplementary material cited in this article is available online at https://doi.org/10.1016/j.biopsych.2025.01.008.

ACKNOWLEDGMENTS AND DISCLOSURES

This work was supported by the Novo Nordisk Foundation (Grant No. NNF20OC0064993). The iPSYCH data were supported by grants from the Lundbeck Foundation (Grant Nos. R102-A9118, R155-2014-1724, and R248-2017-2003) and the Universities and University Hospitals of Aarhus and Copenhagen. The Anorexia Nervosa Genetics Initiative was an initiative of the Klarman Family Foundation; additional AN genotype data were supported by a grant from the Lundbeck Foundation (Grant No. R276-2018-4581). MA acknowledges grant support from the National Institute of Mental Health (Grant No. R01MH120170). ZY acknowledges grant support from the Independent Research Fund Denmark (Grant No. 1052-00029B). CMB is supported by the National Institute of Mental Health (Grant Nos. R56MH129437, R01MH120170, R01MH124871, R01MH119084, R01MH118278, and R01MH124871), the Swedish Research Council (Award No. 538-2013-8864), and the Lundbeck Foundation (Grant No. R276-2018-4581). BJV was supported by a grant from the Lundbeck Foundation (Grant No. R335-2019-2339) and a grant from the Independent Research Fund (Grant No. 2034-00241B).

CMB reports receiving grants from Lundbeckfonden and is an author and royalty recipient at Pearson. BJV is a member of the scientific advisory board for Allelica. All other authors report no biomedical financial interests or potential conflicts of interest.

REFERENCES

  • 1.Tozzi F, Thornton LM, Klump KL, Fichter MM, Halmi KA, Kaplan AS, et al. (2005): Symptom fluctuation in eating disorders: Correlates of diagnostic crossover. Am J Psychiatry 162:732–740. [DOI] [PubMed] [Google Scholar]
  • 2.Eddy KT, Dorer DJ, Franko DL, Tahilani K, Thompson-Brenner H, Herzog DB (2008): Diagnostic crossover in anorexia nervosa and bulimia nervosa: Implications for DSM-V. Am J Psychiatry 165:245–250. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Milos G, Spindler A, Schnyder U, Fairburn CG (2005): Instability of eating disorder diagnoses: Prospective study. Br J Psychiatry 187:573–578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bulik CM, Coleman JRI, Hardaway JA, Breithaupt L, Watson HJ, Bryant CD, Breen G (2022): Genetics and neurobiology of eating disorders. Nat Neurosci 25:543–554. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Eddy KT, Tabri N, Thomas JJ, Murray HB, Keshaviah A, Hastings E, et al. (2017): Recovery from anorexia nervosa and bulimia nervosa at 22-year follow-up. J Clin Psychiatry 78:184–189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Bulik CM, Sullivan PF, Fear J, Pickering A (1997): Predictors of the development of bulimia nervosa in women with anorexia nervosa. J Nerv Ment Dis 185:704–707. [DOI] [PubMed] [Google Scholar]
  • 7.van Hoeken D, Hoek HW (2020): Review of the burden of eating disorders: Mortality, disability, costs, quality of life, and family burden. Curr Opin Psychiatry 33:521–527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Schaumberg K, Jangmo A, Thornton LM, Birgegård A, Almqvist C, Norring C, et al. (2019): Patterns of diagnostic transition in eating disorders: A longitudinal population study in Sweden. Psychol Med 49:819–827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Monteleone P, Di Genio M, Monteleone AM, Di Filippo C, Maj M (2011): Investigation of factors associated to crossover from anorexia nervosa restricting type (ANR) and anorexia nervosa binge-purging type (ANBP) to bulimia nervosa and comparison of bulimia nervosa patients with or without previous ANR or ANBP. Compr Psychiatry 52:56–62. [DOI] [PubMed] [Google Scholar]
  • 10.Dobrescu SR, Dinkler L, Gillberg C, Råstam M, Gillberg C, Wentz E (2020): Anorexia nervosa: 30-year outcome. Br J Psychiatry 216:97–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Keski-Rahkonen A, Hoek HW, Susser ES, Linna MS, Sihvola E, Raevuori A, et al. (2007): Epidemiology and course of anorexia nervosa in the community. Am J Psychiatry 164:1259–1265. [DOI] [PubMed] [Google Scholar]
  • 12.Clausen L (2008): Time to remission for eating disorder patients: A 2½-year follow-up study of outcome and predictors. Nord J Psychiatry 62:151–159. [DOI] [PubMed] [Google Scholar]
  • 13.Klump KL, Bulik CM, Kaye WH, Treasure J, Tyson E (2009): Academy for eating disorders position paper: Eating disorders are serious mental illnesses. Int J Eat Disord 42:97–103. [DOI] [PubMed] [Google Scholar]
  • 14.Boraska V, Franklin CS, Floyd JAB, Thornton LM, Huckins LM, Southam L, et al. (2014): A genome-wide association study of anorexia nervosa. Mol Psychiatry 19:1085–1094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Duncan L, Yilmaz Z, Gaspar H, Walters R, Goldstein J, Anttila V, et al. (2017): Significant locus and metabolic genetic correlations revealed in genome-wide association study of anorexia nervosa. Am J Psychiatry 174:850–858. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Watson HJ, Yilmaz Z, Thornton LM, Hübel C, Coleman JRI, Gaspar HA, et al. (2019): Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa. Nat Genet 51:1207–1214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Li D, Chang X, Connolly JJ, Tian L, Liu Y, Bhoj EJ, et al. (2017): A genome-wide association study of anorexia nervosa suggests a risk locus implicated in dysregulated leptin signaling. Sci Rep 7:3847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Abdulkadir M, Hübel C, Herle M, Loos RJF, Breen G, Bulik CM, Micali N (2022): Eating disorder symptoms and their associations with anthropometric and psychiatric polygenic scores. Eur Eat Disord Rev 30:221–236. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Pedersen CB, Bybjerg-Grauholm J, Pedersen MG, Grove J, Agerbo E, Bækvad-Hansen M, et al. (2018): The iPSYCH2012 case-cohort sample: New directions for unravelling genetic and environmental architectures of severe mental disorders. Mol Psychiatry 23:6–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Lynge E, Sandegaard JL, Rebolj M (2011): The Danish national patient register. Scand J Public Health 39(suppl):30–33. [DOI] [PubMed] [Google Scholar]
  • 21.Mors O, Perto GP, Mortensen PB (2011): The Danish psychiatric central research register. Scand J Public Health 39(suppl):54–57. [DOI] [PubMed] [Google Scholar]
  • 22.Bybjerg-Grauholm J, Pedersen CB, Bækvad-Hansen M, Pedersen G, Adamsen D, Hansen CS, et al. (2020): The iPSYCH2015 Case-Cohort sample: updated directions for unravelling genetic and environmental architectures of severe mental disorders. medRxiv 10.1101/2020.11.30.20237768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Thornton LM, Munn-Chernoff MA, Baker JH, Juréus A, Parker R, Henders AK, et al. (2018): The Anorexia Nervosa Genetics Initiative (ANGI): Overview and methods. Contemp Clin Trials 74:61–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Bulik CM, Thornton LM, Parker R, Kennedy H, Baker JH, MacDermod C, et al. (2021): The Eating Disorders Genetics Initiative (EDGI): Study protocol. BMC Psychiatry 21:234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D (2006): Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 38:904–909. [DOI] [PubMed] [Google Scholar]
  • 26.Nørgaard-Pedersen B, Hougaard DM (2007): Storage policies and use of the Danish newborn Screening biobank. J Inherit Metab Dis 30:530–536. [DOI] [PubMed] [Google Scholar]
  • 27.Privé F, Arbel J, Vilhjálmsson BJ (2021): LDpred2: Better, faster. stronger. Bioinformatics Schwartz R, editor 36:5424–5431. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Albiñana C, Grove J, McGrath JJ, Agerbo E, Wray NR, Bulik CM, et al. (2021): Leveraging both individual-level genetic data and GWAS summary statistics increases polygenic prediction. Am J Hum Genet 108:1001–1011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Galwey NW (2009): A new measure of the effective number of tests, a practical tool for comparing families of non-independent significance tests. Genet Epidemiol 33:559–568. [DOI] [PubMed] [Google Scholar]
  • 30.Stice E, Desjardins C, Shaw H, Siegel S, Gee K, Rohde P (2025): Prevalence, incidence, impairment, course, and diagnostic progression and transition of eating disorders, overweight, and obesity in a large prospective study of high-risk young women. J Psychopathol Clin Sci 134:427–437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Wade TD, Bulik CM, Neale M, Kendler KS (2000): Anorexia nervosa and major depression: Shared genetic and environmental risk factors. Am J Psychiatry 157:469–471. [DOI] [PubMed] [Google Scholar]
  • 32.Andersson P, Jamshidi E, Ekman C-J, Tedroff K, Björkander J, Sjögren M, et al. (2023): Anorexia nervosa with comorbid severe depression: A systematic scoping review of brain stimulation treatments. J ECT 39:227–234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Johnston KJA, Adams MJ, Nicholl BI, Ward J, Strawbridge RJ, Ferguson A, et al. (2019): Genome-wide association study of multisite chronic pain in UK Biobank. PLoS Genet 15:e1008164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Sim L, Harbeck Weber C, Harrison T, Peterson C (2021): Central sensitization in chronic pain and eating disorders: A potential shared pathogenesis. J Clin Psychol Med Settings 28:40–52. [DOI] [PubMed] [Google Scholar]
  • 35.de Zwaan M, Biener D, Bach M, Wiesnagrotzki S, Stacher G (1996): Pain sensitivity, alexithymia, and depression in patients with eating disorders: Are they related? J Psychosom Res 41:65–70. [DOI] [PubMed] [Google Scholar]
  • 36.Faris PL, Raymond NC, De Zwaan M, Howard LA, Eckert ED, Mitchell JE (1992): Nociceptive, but not tactile, thresholds are elevated in bulimia nervosa. Biol Psychiatry 32:462–466. [DOI] [PubMed] [Google Scholar]
  • 37.Girdler SS, Koo-Loeb J, Pedersen CA, Brown HJ, Maixner W (1998): Blood pressure-related hypoalgesia in bulimia nervosa. Psychosom Med 60:736–743. [DOI] [PubMed] [Google Scholar]
  • 38.Craig ADB (2009): How do you feel—Now? The anterior insula and human awareness. Nat Rev Neurosci 10:59–70. [DOI] [PubMed] [Google Scholar]
  • 39.Jenkinson PM, Taylor L, Laws KR (2018): Self-reported interoceptive deficits in eating disorders: A meta-analysis of studies using the eating disorder inventory. J Psychosom Res 110:38–45. [DOI] [PubMed] [Google Scholar]
  • 40.Koch SV, Larsen JT, Plessen KJ, Thornton LM, Bulik CM, Petersen LV (2022): Associations between parental socioeconomic-, family-, and sibling status and risk of eating disorders in offspring in a Danish national female cohort. Int J Eat Disord 55:1130–1142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Björkenstam E, Cheng S, Burström B, Pebley AR, Björkenstam C, Kosidou K (2017): Association between income trajectories in childhood and psychiatric disorder: A Swedish population-based study. J Epidemiol Community Health 71:648–654. [DOI] [PubMed] [Google Scholar]
  • 42.Papini NM, Presseller E, Bulik CM, Holde K, Larsen JT, Thornton LM, et al. (2024): Interplay of polygenic liability with birth-related, somatic, and psychosocial factors in anorexia nervosa risk: A nationwide study. Psychol Med 54:2073–2086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Le Grange D, Swanson SA, Crow SJ, Merikangas KR (2012): Eating disorder not otherwise specified presentation in the US population. Int J Eat Disord 45:711–718. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Schmidt M, Schmidt SAJ, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT (2015): The Danish National Patient Registry: A review of content, data quality, and research potential. Clin Epidemiol 7:449–490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Clausen L (2004): Time course of symptom remission in eating disorders. Int J Eat Disord 36:296–306. [DOI] [PubMed] [Google Scholar]
  • 46.Bardone-Cone AM, Harney MB, Maldonado CR, Lawson MA, Robinson DP, Smith R, Tosh A (2010): Defining recovery from an eating disorder: Conceptualization, validation, and examination of psychosocial functioning and psychiatric comorbidity. Behav Res Ther 48:194–202. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplementary_info_text
NIHMS2123982-supplement-supplementary_info_text.pdf (249.5KB, pdf)
supplementary_tables
NIHMS2123982-supplement-supplementary_tables.xlsx (291.1KB, xlsx)
supplementary_key_resource_table
NIHMS2123982-supplement-supplementary_key_resource_table.xlsx (20.5KB, xlsx)

RESOURCES