The Role of Circadian Rhythms and Sleep in Anorexia Nervosa

Hannah Wilcox; Valentina Paz; Richa Saxena; John W Winkelman; Victoria Garfield; Hassan S Dashti

doi:10.1001/jamanetworkopen.2023.50358

. 2024 Jan 4;7(1):e2350358. doi: 10.1001/jamanetworkopen.2023.50358

The Role of Circadian Rhythms and Sleep in Anorexia Nervosa

Hannah Wilcox ^1,², Valentina Paz ^2,^3,⁴, Richa Saxena ^1,^2,^5,⁶, John W Winkelman ^5,⁷, Victoria Garfield ⁴, Hassan S Dashti ^1,^2,^5,^6,^8,^✉

¹Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital and Harvard Medical School, Boston

²Center for Genomic Medicine, Massachusetts General Hospital and Harvard Medical School, Boston

³Instituto de Psicología Clínica, Facultad de Psicología, Universidad de la República, Montevideo, Uruguay

⁴MRC Unit for Lifelong Health & Ageing, Institute of Cardiovascular Science, University College London, London, United Kingdom

⁵Division of Sleep Medicine, Harvard Medical School, Boston, Massachusetts

⁶Broad Institute, Cambridge, Massachusetts

⁷Sleep Disorders Clinical Research Program, Massachusetts General Hospital and Harvard Medical School, Boston

⁸Division of Nutrition, Harvard Medical School, Boston, Massachusetts

Accepted for Publication: November 16, 2023.

Published: January 4, 2024. doi:10.1001/jamanetworkopen.2023.50358

^✉

Corresponding Author: Hassan S. Dashti, PhD, RD, Massachusetts General Hospital, 55 Fruit St, Edwards 4-410C, Boston, MA 02114 (hassan.dashti@mgh.harvard.edu).

Author Contributions: Dr Dashti had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Saxena, Dashti.

Acquisition, analysis, or interpretation of data: Wilcox, Paz, Winkelman, Garfield, Dashti.

Drafting of the manuscript: Wilcox, Dashti.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Wilcox, Dashti.

Administrative, technical, or material support: Paz, Dashti.

Supervision: Saxena, Winkelman, Garfield, Dashti.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by grant R00HL153795 from the National Institutes of Health to Dr Dashti.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 3.

^✉

Corresponding author.

PMCID: PMC10767597 PMID: 38175645

Key Points

Question

What is the association between anorexia nervosa, circadian rhythms, and sleep traits?

Findings

In this genetic association study, a bidirectional association between anorexia nervosa and morning chronotype was found using mendelian randomization. In addition, an association between anorexia nervosa and insomnia was found using mendelian randomization and a polygenic risk score for anorexia nervosa in a clinical biobank.

Meaning

These findings suggest that anorexia nervosa is a morningness eating disorder, in contrast to most other evening-based psychiatric diseases, and corroborate the association between anorexia nervosa and insomnia, concordant with earlier observational studies.

This genetic association study investigates the association between anorexia nervosa and circadian and sleep traits through mendelian randomization and tests the associations between a polygenic risk score for anorexia nervosa and sleep disorders in a clinical biobank.

Abstract

Importance

Observational studies have associated anorexia nervosa with circadian rhythms and sleep traits. However, the direction of causality and the extent of confounding by psychosocial comorbidities in these associations are unknown.

Objectives

To investigate the association between anorexia nervosa and circadian and sleep traits through mendelian randomization and to test the associations between a polygenic risk score (PRS) for anorexia nervosa and sleep disorders in a clinical biobank.

Design, Setting, and Participants

This genetic association study used bidirectional 2-sample mendelian randomization with summary-level genetic associations between anorexia nervosa (from the Psychiatric Genomics Consortium) and chronotype and sleep traits (primarily from the UK Biobank). The inverse-variance weighted method, in addition to other sensitivity approaches, was used. From the clinical Mass General Brigham (MGB) Biobank (n = 47 082), a PRS for anorexia nervosa was calculated for each patient and associations were tested with prevalent sleep disorders derived from electronic health records. Patients were of European ancestry. All analyses were performed between February and August 2023.

Exposures

Genetic instruments for anorexia nervosa, chronotype, daytime napping, daytime sleepiness, insomnia, and sleep duration.

Main Outcomes and Measures

Chronotype, sleep traits, risk of anorexia nervosa, and sleep disorders derived from a clinical biobank.

Results

The anorexia nervosa genome-wide association study included 16 992 cases (87.7%-97.4% female) and 55 525 controls (49.6%-63.4% female). Genetic liability for anorexia nervosa was associated with a more morning chronotype (β = 0.039; 95% CI, 0.006-0.072), and conversely, genetic liability for morning chronotype was associated with increased risk of anorexia nervosa (β = 0.178; 95% CI, 0.042-0.315). Associations were robust in sensitivity and secondary analyses. Genetic liability for insomnia was associated with increased risk of anorexia nervosa (β = 0.369; 95% CI, 0.073-0.666); however, sensitivity analyses indicated bias due to horizontal pleiotropy. The MGB Biobank analysis included 47 082 participants with a mean (SD) age of 60.4 (17.0) years and 25 318 (53.8%) were female. A PRS for anorexia nervosa was associated with organic or persistent insomnia in the MGB Biobank (odds ratio, 1.10; 95% CI, 1.03-1.17). No associations were evident for anorexia nervosa with other sleep traits.

Conclusions and Relevance

The results of this study suggest that in contrast to other metabo-psychiatric diseases, anorexia nervosa is a morningness eating disorder and further corroborate findings implicating insomnia in anorexia nervosa. Future studies in diverse populations and with subtypes of anorexia nervosa are warranted.

Introduction

The circadian clock is the intrinsic biological system that produces 24-hour cyclical processes called circadian rhythms.^1,2 Circadian rhythms affect a wide spectrum of human physiology, including cycles in body temperature and sleep-wake patterns. Chronotype, the preference for morningness or eveningness, is a correlate of circadian rhythms and an individual’s propensity for earlier or later timing for sleep and other activities and is associated with psychological, neurological, gastrointestinal, and respiratory disorders.^3,4

There is a bidirectional association between circadian rhythms and psychiatric disorders. For example, tendency for an evening chronotype is associated with depression and schizophrenia.^5,6 Conversely, bipolar disorder, anxiety, and schizophrenia onset result in alterations in the rhythmicity of core clock genes, and interventions that target the circadian system, such as bright light therapy, are considered valuable treatments.^7,8,9 There is also some evidence implicating the circadian system in eating disorders, primarily centered on the link between binge eating disorder and evening chronotype in alignment with other psychiatric disorders.^10,11,12 Characteristics of generally healthy adults with an evening chronotype, such as emotional eating, later food timing, and impulsivity, further support the role of eveningness tendency in binge eating–like behaviors.^13,14 However, the relevance of the circadian system in anorexia nervosa remains largely unknown.

Anorexia nervosa is a major eating disorder characterized by severe food intake restriction, weight loss, and intense fear of weight gain¹⁵ and has among the highest mortality rates of all psychiatric diseases.¹⁶ Some evidence suggests the role of the circadian clock: for example, morningness tendency, with lack of appetite, of which the most prominent is the well-recognized morning anorexia and evening hyperphagia dichotomy that characterizes patients with night eating syndrome,¹⁷ but also with evening preference in small observational studies.¹⁸ Additional evidence includes early morning awakening insomnia among people with anorexia nervosa, indicative of perturbations in sleep-wake cycles.¹⁹ Directionality of the association and whether any effects are causal or mediated by mood and anxiety pathophysiology that often cluster with anorexia nervosa remain unknown.²⁰

Genetic analyses can further interrogate the link between anorexia nervosa and circadian and sleep traits. Genome-wide association studies (GWAS) for anorexia nervosa revealed several loci.^21,22 Mendelian randomization (MR) uses genetic variants that are robustly associated with a trait to examine potential causal effects on outcomes of interest.²³ Due to the properties of genetic variants, MR is less likely to be affected by unmeasured confounding factors or reverse causation than observational studies. The application of MR has provided additional evidence supporting observational data on hypoleptinemia among patients with anorexia nervosa.^24,25 It is also speculated that there is a bidirectional association between sleep and eating processes, such that eating pathology disrupts sleep, and conversely, dysregulation in sleep influences eating behaviors.²⁶ The aim of this study was to investigate the bidirectional association between anorexia nervosa and chronotype and other sleep traits through MR and a polygenic risk score (PRS) for anorexia nervosa.

Methods

GWAS for Anorexia Nervosa

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization (STROBE-MR) reporting guidelines.³¹ The largest GWAS meta-analysis for anorexia nervosa from the Eating Disorders Working Group of the Psychiatric Genomics Consortium aggregated data from participants of European ancestry across 33 data sets and included 16 992 cases and 55 525 controls, of which 87.7% to 97.4% of cases were female, 49.6% to 63.4% of controls were female, and 5.3% of the samples were from the UK Biobank.²² The case definition of anorexia nervosa was based on a lifetime diagnosis of anorexia nervosa from hospital or register records, structured clinical interviews, online questionnaires based on standardized criteria (ie, Diagnostic and Statistical Manual of Mental Disorders [Third Edition Revised]; Diagnostic and Statistical Manual of Mental Disorders [Fourth Edition]; International Classification of Diseases, Eighth Revision; International Classification of Diseases, Ninth Revision [ICD-9]; or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10]), or a self-reported diagnosis. The GWAS identified 8 independent genome-wide significant (P < 5 × 10⁻⁸) single-nucleotide variants (SNVs). SNV-based heritability was estimated at 11% to 17%, reflecting the polygenicity of anorexia nervosa.

GWAS for Self-Reported Chronotype and Sleep Traits

The primary outcome was chronotype. Summary statistics for the largest GWAS of European ancestry predominantly from the UK Biobank for chronotype (morning preference) was retrieved from the Sleep Disorder Knowledge Portal²⁷ (eMethods in Supplement 1). Four sleep traits were also considered secondary outcomes, including daytime napping, daytime sleepiness, and sleep duration from the Sleep Disorder Knowledge Portal²⁷ and insomnia from the Center for Neurogenomics and Cognitive Research in Amsterdam.²⁸ Traits were determined using primarily self-reported data (eMethods in Supplement 1).

Statistical Analysis

MR Primary Analyses

Bidirectional 2-sample MR between anorexia nervosa and chronotype and sleep traits was conducted using the largest publicly accessible GWAS studies in participants of European ancestry (eTable 1 in Supplement 2) using the “TwoSampleMR” R package. All data sources obtained participant informed consent and relevant ethical approval; as we used publicly available data which does not constitute human participant research, per the US code of federal regulations (45 CFR 46.102), no additional institutional review board approvals were necessary. The primary analysis considered chronotype (morning preference) and 4 secondary sleep traits: daytime napping, daytime sleepiness, insomnia, and sleep duration. For each trait, a genetic instrument was generated using lead GWAS variants (eTables 2-15 in Supplement 2), and standard data harmonization was conducted (eMethods in Supplement 1). The inverse variance weighted (IVW) regression method was used as the primary analysis. The IVW method yields an unbiased estimate in the absence of horizontal pleiotropy.^23,29 A range of sensitivity analyses robust to the presence of horizontal pleiotropy were tested, including MR-Egger and weighted median estimator (WME) methods. Consistency in effect sizes and overlap of 95% CIs were examined.³⁰ A threshold of .05 was used for statistical significance in the MR for our primary analysis with chronotype and secondary analyses with sleep traits. All analyses were performed between February and August 2023.

MR Sensitivity and Secondary Analyses

For any significant finding from the IVW analyses, additional sensitivity analyses were conducted including MR pleiotropy residual sum and outlier (PRESSO),³² leave-one-out analysis,²⁹ MR-Egger intercept,²⁹ I² quantification,³³ and a Steiger test³⁴ (eMethods in Supplement 1). Secondary analyses were conducted to examine the robustness of the findings including MR analyses for a binary self-report chronotype variable and 3 objective measures of chronotype,³⁵ and MR using female-only summary statistics from the insomnia GWAS.²⁸

PRS for Anorexia Nervosa in the Mass General Brigham Biobank

The Mass General Brigham (MGB) Biobank is the health care enterprise clinical biobank from the MGB health care network in Boston, Massachusetts, and links electronic health records to genetics and lifestyle data. The present protocol was approved by the MGB institutional review board. Patients provided blood samples for genotyping with written consent (eMethods in Supplement 1). All genetic analyses were restricted to participants of European ancestry. A PRS for anorexia nervosa based on effect estimates from the recent meta-analysis GWAS for anorexia nervosa of European ancestry was calculated for each patient.²² The PRS was generated using PRS–continuous shrinkage (PRS-CS) based on bayesian regression that places a continuous shrinkage prior on SNV effect sizes.³⁶ The UKBiobank EUR-ancestry linkage disequilibrium (LD) panel was set. A total of 1 734 712 SNVs were retained in the PRS following clumping. The PRS was standardized with a mean of 0 and an SD of 1.

Disease designation for anorexia nervosa and sleep disorders was based on PheCode classifications, which aggregate clinically relevant ICD-9 and ICD-10 billing codes into disease groups as described in the PheCode catalog³⁷ (eMethods in Supplement 1). Anorexia nervosa cases were based on the anorexia nervosa-specific code (305.21), which uses ICD-9 billing code 307.1 and ICD-10 billing code F50.0. Association analyses for the anorexia nervosa PRS were conducted for anorexia nervosa and sleep disorders with at least 1% prevalent cases, using the PheWAS R package.³⁸ Associations were tested using logistic regression models with adjustments for age, sex, genotyping array, batch, and 10 principal components of ancestry to account for genetic substructure (minimal model), then further adjusted for body mass index, employment status, smoking status, exercise, alcohol intake, and education attainment (fully adjusted model). Significance was considered at the Bonferroni threshold for 9 tested sleep disease outcomes (P < .05 / 9 tests).

Enrolled patients were invited to complete an optional Health Information Questionnaire, which included questions on sleep habits (eMethods in Supplement 1). The following circadian and sleep variables from self-report were examined for associations with the PRS: time in bed, sleep debt, sleep midpoint, and social jetlag. Associations were tested using adjusted linear regression models as described previously, and in sensitivity analyses, further adjusted for PheCodes for depression (296.2) and anxiety (300.1), 2 prevalent comorbidities.³⁹ Significance was considered at the Bonferroni threshold for 4 circadian and sleep outcome traits (P < .05 / 4 tests). All analyses were conducted using R software version 4.2.3 (R Project for Statistical Computing).

Results

MR Analyses for Anorexia Nervosa and Chronotype

The anorexia nervosa GWAS included 16 992 cases (87.7%-97.4% female) and 55 525 controls (49.6%-63.4% female). We first conducted bidirectional 2-sample MR between anorexia nervosa and chronotype. Using IVW MR, we found a bidirectional association between anorexia nervosa and morning chronotype (Table 1 and Figure 1). Genetic liability for anorexia nervosa was associated with a more morning chronotype (β = 0.039; 95% CI, 0.006-0.072), and conversely, genetic liability for morning chronotype was associated with increased risk of anorexia nervosa (β = 0.178; 95% CI, 0.042-0.315). Bidirectional effect sizes were consistent using MR-Egger and WME (Table 1 and Figure 1).

Table 1. MR Primary Analyses Results for the Associations Between Anorexia Nervosa and Chronotype and Sleep Traits.

Exposure	SNVs, No.	IVW		WME		MR-Egger
Exposure	SNVs, No.	β (SE)^a	P value	β (SE)^a	P value	β (SE)^a	P value
Anorexia nervosa with circadian and sleep traits
Chronotype (morning)	8	0.039 (0.017)	.02	0.043 (0.017)	.01	0.157 (0.052)	.02
Daytime napping	8	−0.003 (0.015)	.83	0.011 (0.009)	.23	0.040 (0.061)	.54
Daytime sleepiness	8	−0.008 (0.006)	.16	−0.005 (0.007)	.41	0.014 (0.023)	.57
Insomnia	8	−0.005 (0.037)	.89	−0.004 (0.035)	.91	0.311 (0.088)	.01
Sleep duration	8	−0.020 (0.011)	.08	−0.021 (0.014)	.13	−0.094 (0.039)	.05
Circadian and sleep traits with anorexia nervosa
Chronotype (morning)	315	0.178 (0.070)	.01	0.175 (0.086)	.04	0.177 (0.211)	.40
Daytime napping	112	0.351 (0.244)	.15	0.117 (0.262)	.66	0.433 (0.869)	.62
Daytime sleepiness	40	0.225 (0.469)	.63	0.752 (0.518)	.15	1.422 (1.767)	.43
Insomnia	467	0.339 (0.147)	.02	0.306 (0.172)	.08	−0.490 (0.568)	.39
Sleep duration	73	−0.084 (0.170)	.62	−0.164 (0.187)	.38	−0.681 (0.642)	.29

Open in a new tab

Abbreviations: IVW, inverse variance weighted; MR, mendelian randomization; SNV, single-nucleotide variant; WME, weighted median estimator.

^{^a}

β Values reflect the difference in the outcome for each 1-unit increase in the exposure.

Figure 1. — Forest plot of MR primary, sensitivity, and secondary results for the associations between anorexia nervosa and chronotype (morning), including accelerometer-derived objective chronotype measures of midpoint of least active 5 hours, midpoint of most active 10 hours, and sleep midpoint. β Values reflect the difference in the outcome for each 1-unit increase in the exposure, with blue boxes denoting statistically significant results. Error bars represent 95% CIs. Primary analyses included inverse variance weighted (IVW) MR, weighted median estimator (WME), and MR-Egger, and sensitivity analyses included MR pleiotropy residual sum and outlier (PRESSO).

We performed a series of sensitivity analyses to further assess the presence of horizontal pleiotropy. Results from the MR-Egger intercept were not significant for horizontal pleiotropy with anorexia nervosa as the exposure or with chronotype as the exposure. Findings from MR PRESSO suggested possible pleiotropy with chronotype as the exposure (P < .001), however this was mitigated by removing 3 outlier variants, resulting in estimates consistent with the main analysis (β = 0.197; 95% CI, 0.067-0.327) (Figure 1; eTable 16 in Supplement 2). Assessment of horizontal pleiotropy using MR PRESSO was not significant for pleiotropy with anorexia nervosa as the exposure. Leave-one-out analysis did not suggest undue influence by single outliers or systemic bias (eTables 17-18 in Supplement 2). The calculated I² quantity was 49.6% with anorexia nervosa as the exposure and 48.2% with chronotype as the exposure. The Steiger test held true in both directions.

The observed direction of the association between chronotype and anorexia nervosa was consistent in secondary analyses using summary statistics for binary chronotype (morning vs evening preference) as the outcome (β = 0.014; 95% CI, 0.002-0.026) (Figure 1 and Table 2). In sensitivity analyses, there was no evidence of horizontal pleiotropy using the MR-Egger intercept test or MR PRESSO. Leave-one-out analysis did not suggest undue influence by single outliers or systemic bias (eTable 19 in Supplement 2). The I² quantity was calculated as 41.0%, and the Steiger test held true.

Table 2. MR Secondary Analyses Results for the Associations Between Anorexia Nervosa and Chronotype and Sleep Traits.

Exposure	SNVs, No.	IVW		WME		MR-Egger
Exposure	SNVs, No.	β (SE)^a	P value	β (SE)^a	P value	β (SE)^a	P value
Anorexia nervosa with chronotype
Binary chronotype (morning)	8	0.014 (0.006)	.02	0.012 (0.006)	.07	0.056 (0.019)	.02
Midpoint of least active 5 h, accelerometer	8	−0.031 (0.021)	.14	−0.027 (0.027)	.33	−0.104 (0.083)	.26
Midpoint of most active 10 h, accelerometer	8	−0.032 (0.021)	.12	−0.038 (0.028)	.18	−0.050 (0.082)	.57
Sleep midpoint, accelerometer	8	−0.026 (0.021)	.22	−0.031 (0.028)	.26	−0.083 (0.083)	.36
Chronotype and sleep traits with anorexia nervosa
Midpoint of least active 5 h, accelerometer	5	−0.244 (0.301)	.42	−0.239 (0.216)	.27	−0.850 (0.458)	.16
Female-only insomnia	272	0.324 (0.157)	.04	0.289 (0.171)	.09	−0.063 (0.740)	.93

Open in a new tab

Abbreviations: IVW, inverse variance weighted; MR, mendelian randomization; SNV, single-nucleotide variant; WME, weighted median estimator.

^{^a}

β Values reflect the difference in the outcome for each 1-unit increase in the exposure.

We found no evidence for the association between anorexia nervosa and chronotype using accelerometer-derived sleep timing (Figure 1 and Table 2). However, MR direction of associations for risk of anorexia nervosa were consistent.

MR Analyses for Anorexia Nervosa and Sleep Traits

In MR, we found an association between insomnia and increased risk of anorexia nervosa, and no associations between anorexia nervosa and daytime napping, daytime sleepiness, or sleep duration (Table 1). Genetic liability for insomnia was associated with increased risk of anorexia nervosa using IVW MR (β = 0.339; 95% CI, 0.052-0.627); however, no association was observed for the genetic liability of anorexia nervosa with increased risk of insomnia (Table 1 and Figure 2). In sensitivity analyses, effect sizes were consistent using WME but not MR-Egger models. Results from the MR-Egger intercept were not significant for horizontal pleiotropy. However, results from MR PRESSO were significant (P < .001), and subsequently mitigated through removing 6 outlier variants resulting in estimates consistent with the main analysis (β = 0.449; 95% CI, 0.195-0.703) (Figure 2; eTable 16 in Supplement 2). Leave-one-out analysis did not suggest influence by single outliers or systemic bias (eTable 20 in Supplement 2). The I² quantity was calculated as 51.5%, and the Steiger test proved false, suggesting the direction of causality of insomnia on anorexia nervosa cannot be ascertained.

Figure 2. — β Values reflect the difference in the outcome for each 1-unit increase in the exposure, with blue boxes denoting statistically significant results. Error bars represent 95% CIs. Primary analyses included inverse variance weighted (IVW) MR, weighted median estimator (WME), and MR-Egger, and sensitivity analyses included MR pleiotropy residual sum and outlier (PRESSO).

When using effect sizes from the female-only insomnia GWAS, we also found that genetic liability for insomnia was associated with increased risk of anorexia nervosa using IVW MR (β = 0.324; 95% CI, 0.015-0.632) (Table 2 and Figure 2). Results from the MR-Egger intercept were not significant for horizontal pleiotropy. However, results from MR PRESSO were significant (P < .001) and subsequently mitigated through removing 4 outlier variants, resulting in estimates consistent with the main analyses (β = 0.312; 95% CI, 0.021-0.603) (Figure 2; eTable 16 in Supplement 2). Leave-one-out analysis did not suggest influence by single outliers or systemic bias (eTable 21 in Supplement 2). The I² quantity was calculated as 54.0%, and the Steiger test was also false.

Anorexia Nervosa in the MGB Biobank

To further examine associations between anorexia nervosa and sleep disorders in a disease-enriched and independent health enterprise clinical biobank, we conducted a series of analyses in the MGB Biobank. A total of 47 082 adult patients of European ancestry with genetic data were included in the MGB Biobank analyses (mean [SD] age, 60.4 [17.0] years; 25 318 [53.8%] female). We first examined associations between the anorexia nervosa PRS and anorexia nervosa diagnosis based on PheCodes. Each SD increase in the PRS was associated with 36% higher odds of anorexia nervosa (odds ratio [OR], 1.36; 95% CI, 1.14-1.63) (Table 3). In association analyses with 9 prevalent sleep disorders, we found associations with higher organic or persistent insomnia. Each additional SD in the anorexia nervosa PRS was associated with 10% higher odds for organic or persistent insomnia (OR, 1.10; 95% CI, 1.03-1.17), which remained unchanged when accounting for lifestyle factors. Associations with other sleep disorders including sleep apnea and restless legs syndrome were not significant. In addition, among patients with self-reported sleep habits (n = 16 109), there were no associations evident between the anorexia nervosa PRS and time in bed, sleep midpoint, sleep debt, or social jetlag (eTable 22 in Supplement 2).

Table 3. Associations Between Polygenic Risk Score for Anorexia Nervosa and Anorexia Nervosa and Sleep Disorders in the Mass General Brigham Biobank (n = 47 082)^a.

Disease description	Phecode	Model 1			Model 2
Disease description	Phecode	No. of cases/No. of controls	OR (95% CI)	P value	No. of cases/No. of controls	OR (95% CI)	P value
Anorexia nervosa	305.21	182/17 106	1.36 (1.14-1.63)	<.001	57/6417	NA	NA
Organic or persistent insomnia	327.41	1343/31 569	1.10 (1.03-1.17)	.005	403/11 266	1.18 (1.05-1.33)	.006
Insomnia	327.4	5748/35 974	1.04 (1.00-1.07)	.04	1637/11 266	1.01 (0.94-1.07)	.87
Restless legs syndrome	327.71	1281/31 507	1.07 (1.00-1.14)	.06	397/11 266	0.99 (0.88-1.11)	.84
Central/nonobstructive sleep apnea	327.31	555/30 781	0.95 (0.86-1.05)	.32	172/11 266	0.83 (0.70-0.99)	.04
Sleep apnea	327.3	4161/34 387	0.99 (0.95-1.03)	.66	1190/11 266	1.02 (0.95-1.10)	.64
Parasomnia	327.5	497/30 723	1.02 (0.92-1.13)	.70	154/11 266	1.01 (0.83-1.21)	.95
Obstructive sleep apnea	327.32	6722/36 948	1.01 (0.97-1.04)	.71	1952/11 266	1.01 (0.96-1.08)	.63
Hypersomnia	327.1	989/31 215	1.01 (0.94-1.09)	.72	314/11 266	1.03 (0.90-1.18)	.65
Sleep disorders	327	3048/33 274	1.01 (0.96-1.05)	.73	898/11 266	1.00 (0.92-1.08)	.99

Open in a new tab

Abbreviations: NA, not applicable; OR, odds ratio.

^{^a}

Effect sizes are per SD increase in the polygenic risk score for anorexia nervosa. Associations were tested using adjusted logistic regression models. Model 1 was adjusted for age, sex, genotyping array, batch, and 10 principal components of ancestry (minimal model). Model 2 was further adjusted for body mass index, employment status, smoking status, exercise, alcohol intake, and education attainment (fully adjusted model). Findings for anorexia nervosa are missing for model 2 due to insufficient number of cases. Significance was considered at the Bonferroni threshold for 9 tested sleep disease outcomes (P < .05 / 9 tests).

Discussion

Using MR, we provide evidence for bidirectional associations between anorexia nervosa and morning chronotype, which were consistent in a series of sensitivity and secondary analyses. We also found associations between insomnia and higher risk for anorexia nervosa concordant with earlier observational studies; however, these effect sizes were possibly confounded by psychosocial comorbidities. We did not find associations between anorexia nervosa and other sleep traits, including daytime napping, daytime sleepiness, and sleep duration.

The bidirectional association between anorexia nervosa and morning chronotype proved robust to sensitivity analyses using alternative approaches to ascertain chronotype and to major bias due to pleiotropy. Results using accelerometer-derived measures of chronotype were consistent with findings from self-report implicating morningness with anorexia nervosa but were ultimately not significant likely due to limited power (85 670 participants in the accelerometer GWAS in comparison with 449 734 participants in the self-report GWAS). In a clinical biobank, we did not observe associations between the anorexia nervosa PRS and self-reported sleep midpoint, a measure of chronotype, possibly due to limited power or confounding comorbidities. As many as 62% and 54% of people diagnosed with an eating disorder also have an anxiety disorder and a mood disorder such as depression, respectively.³⁹ Anxiety and depression, and most other psychiatric diseases, have consistently been associated with eveningness, potentially confounding previous findings using observational data.^40,41 Furthermore, anorexia nervosa is most common at a younger age, peaking in onset around ages 15 to 16 years, when people experience a delay in chronotype toward eveningness.^42,43 MR mitigates these confounding effects through the use of random genetic variation, allowing for robust analysis of the connection between anorexia nervosa and chronotype independent of psychosocial comorbidities and resolution of seemingly conflicting evidence from observational studies.

Our findings suggest that anorexia nervosa is another eating disorder with likely a circadian feature. First, in observational data, anorexia nervosa was associated with breakfast and lunch skipping, reinforcing the connection between anorexia and food temporality.⁴⁴ Second, morning bright light therapy was deemed less effective in anorexia nervosa treatment compared with binge eating disorders and bulimia nervosa, 2 eating disorders presumed to be evening based.^7,9 Third, compared with healthy controls, malnourished patients with anorexia nervosa have an enhanced and advanced cortisol awakening response often exhibited with earlier chronotype.^45,46 Lastly, patients with anorexia nervosa often experience early morning awakening insomnia indicative of perturbations in sleep-wake cycles.¹⁹ Contradictory evidence from observational data linking anorexia nervosa with evening chronotype¹⁸ or showing no associations with sleep timing⁴⁷ are small and may be confounded by psychosocial comorbidities. The precise role of the circadian system in the etiology of anorexia nervosa remains unclear.

In addition, we found a potential association between insomnia and increased risk of anorexia nervosa. Although findings from the IVW method were inconsistent in sensitivity analyses using MR-Egger, a low I² test statistic⁴⁸ indicated that the MR-Egger result was likely biased.⁴⁹ Our findings provide evidence for a robust association between insomnia and anorexia nervosa without bias due to horizontal pleiotropy; however, the direction of causality cannot be proven. The results from MR were further corroborated in a clinical biobank using a PRS for anorexia nervosa, in which we observed positive associations between anorexia nervosa and organic or persistent insomnia. Directionality cannot be inferred from this cross-sectional analysis.

The clinical implications of our novel bidirectional chronotype findings are unclear; however, our study provides several opportunities for future research into the prevention and treatment of anorexia nervosa. Our findings suggest the role of morningness as an underrecognized risk factor for anorexia nervosa whose importance should be further investigated in the context of other established risk factors. If deemed a relevant risk factor, interventions promoting later sleep schedules may be considered to ameliorate risk conferred by a morning chronotype. Our findings also direct research into circadian-based mechanisms for future anorexia nervosa treatment. Unlike morning bright light therapy, which has resulted in mixed findings in preliminary studies, bright light therapy in the evening, as is recommended for patients with early morning awakening insomnia, may be efficacious in the prevention and treatment of anorexia nervosa.⁵⁰ Whether this would impact transdiagnostic treatment of more evening-based psychiatric comorbidities, eg, depression, remains unknown. Replication analyses and mechanistic studies are crucial to warrant any circadian-based therapies suggested by our findings.

This study has several strengths. We leveraged the largest GWAS for anorexia nervosa, chronotype, and sleep traits available at the time of analysis. In addition, we provided evidence of robust replication of the anorexia nervosa genetic variants in an independent clinical biobank facilitated by a higher prevalence of anorexia nervosa (0.3% of samples) in our clinical MGB Biobank, far exceeding other biobanks in previous studies including the BioMe Biobank (0.02% of samples).⁵¹ However, this prevalence is likely not sufficient to interrogate associations between anorexia nervosa diagnosis and other genetic variants, such as a PRS for chronotype or insomnia.

Limitations

There are some important limitations to note. As the anorexia nervosa GWAS focused on broadly defined anorexia nervosa based on billing codes and/or self-report, we were unable to interrogate anorexia nervosa subtypes, such as restrictive, binge-eating/purging, or atypical anorexia nervosa.²² Our analyses only considered participants of European ancestry, as the genetic discoveries were from predominantly European cohorts and the transferability of variants into non-European populations remains unknown.⁵² It is worth noting that anorexia nervosa prevalence remains currently highest among European and Western cultures, is rapidly rising among young adolescent girls,⁵³ and is likely underdiagnosed in men and individuals from minoritized racial and ethnic groups,⁵² necessitating future large investigations on early or late onset anorexia nervosa⁵⁴ and anorexia in non-European populations. We presented MR P values unadjusted for multiple testing because our study is exploratory in nature. Therefore, we acknowledge that individual findings should be interpreted with caution, and that additional studies are necessary to verify these findings. Additionally, as the anorexia nervosa GWAS included 5.3% samples from the UK Biobank, there was some sample overlap in the 2-sample MR analysis, which may have led to the overestimation of effect sizes.⁵⁵

Conclusions

Our study found bidirectional associations between anorexia nervosa and morning chronotype, suggesting that in contrast to other metabo-psychiatric diseases, anorexia nervosa is distinctly a morningness eating disorder. In addition, the associations between anorexia nervosa and insomnia corroborate earlier findings.

Supplement 1.

eMethods.

Click here for additional data file.^{(559.4KB, pdf)}

Supplement 2.

eTable 1. Genome-wide association studies for anorexia nervosa and circadian and sleep traits

eTable 2. SNVs and proxies included in genetic instrument for anorexia nervosa with chronotype

eTable 3. SNVs and proxies included in genetic instrument for anorexia nervosa with binary chronotype

eTable 4. SNVs and proxies included in genetic instrument for anorexia nervosa with daytime napping

eTable 5. SNVs and proxies included in genetic instrument for anorexia nervosa with daytime sleepiness

eTable 6. SNVs and proxies included in genetic instrument for anorexia nervosa with insomnia

eTable 7. SNVs and proxies included in genetic instrument for anorexia nervosa with sleep duration

eTable 8. SNVs and proxies included in genetic instrument for anorexia nervosa with accelerometer derived midpoint of least active 5 hours, midpoint of most active 10 hours, and sleep midpoint

eTable 9. SNVs and proxies included in genetic instrument for chronotype with anorexia nervosa

eTable 10. SNVs and proxies included in genetic instrument for daytime napping with anorexia nervosa

eTable 11. SNVs and proxies included in genetic instrument for daytime sleepiness with anorexia nervosa

eTable 12. SNVs and proxies included in genetic instrument for insomnia with anorexia nervosa

eTable 13. SNVs and proxies included in genetic instrument for female only insomnia with anorexia nervosa

eTable 14. SNVs and proxies included in genetic instrument for sleep duration with anorexia nervosa

eTable 15. SNVs and proxies included in genetic instrument for accelerometer derived midpoint of least active 5 hours with anorexia nervosa

eTable 16. Variants identified as outliers by MR-PRESSO

eTable 17. Leave-one-out analysis for results of anorexia nervosa with chronotype. IVW MR conducted with listed SNV removed in each respective analysis

eTable 18. Leave-one-out analysis for results of chronotype with anorexia nervosa. IVW MR conducted with listed SNV removed in each respective analysis

eTable 19. Leave-one-out analysis for results of anorexia nervosa with binary chronotype. IVW MR conducted with listed SNV removed in each respective analysis

eTable 20. Leave-one-out analysis for results of insomnia with anorexia nervosa. IVW MR conducted with listed SNV removed in each respective analysis

eTable 21. Leave-one-out analysis for results of female only insomnia with anorexia nervosa. IVW MR conducted with listed SNV removed in each respective analysis

eTable 22. Association between polygenic risk score for anorexia nervosa and self-reported sleep traits in the Mass General Brigham Biobank (n =16,109)

Click here for additional data file.^{(206.9KB, xlsx)}

Supplement 3.

Data Sharing Statement

Click here for additional data file.^{(14.3KB, pdf)}

References

1.Zee PC, Attarian H, Videnovic A. Circadian rhythm abnormalities. Continuum (Minneap Minn). 2013;19(1 Sleep Disorders):132-147. doi: 10.1212/01.CON.0000427209.21177.aa [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Takahashi JS, Hong HK, Ko CH, McDearmon EL. The genetics of mammalian circadian order and disorder: implications for physiology and disease. Nat Rev Genet. 2008;9(10):764-775. doi: 10.1038/nrg2430 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Burgess HJ, Fogg LF. Individual differences in the amount and timing of salivary melatonin secretion. PLoS One. 2008;3(8):e3055. doi: 10.1371/journal.pone.0003055 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Knutson KL, von Schantz M. Associations between chronotype, morbidity and mortality in the UK Biobank cohort. Chronobiol Int. 2018;35(8):1045-1053. doi: 10.1080/07420528.2018.1454458 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Jones SE, Lane JM, Wood AR, et al. Genome-wide association analyses of chronotype in 697,828 individuals provides insights into circadian rhythms. Nat Commun. 2019;10(1):343. doi: 10.1038/s41467-018-08259-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Daghlas I, Lane JM, Saxena R, Vetter C. Genetically proxied diurnal preference, sleep timing, and risk of major depressive disorder. JAMA Psychiatry. 2021;78(8):903-910. doi: 10.1001/jamapsychiatry.2021.0959 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Oldham MA, Ciraulo DA. Bright light therapy for depression: a review of its effects on chronobiology and the autonomic nervous system. Chronobiol Int. 2014;31(3):305-319. doi: 10.3109/07420528.2013.833935 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Walker WH II, Walton JC, DeVries AC, Nelson RJ. Circadian rhythm disruption and mental health. Transl Psychiatry. 2020;10(1):28. doi: 10.1038/s41398-020-0694-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Beauchamp MT, Lundgren JD. A systematic review of bright light therapy for eating disorders. Prim Care Companion CNS Disord. 2016;18(5). doi: 10.4088/PCC.16r02008 [DOI] [PubMed] [Google Scholar]
10.Menculini G, Brufani F, Del Bello V, Moretti P, Tortorella A. Circadian rhythms disruptions and eating disorders: clinical impact and possible psychopathological correlates. Psychiatr Danub. 2019;31(suppl 3):497-502. [PubMed] [Google Scholar]
11.Romo-Nava F, Guerdjikova AI, Mori NN, et al. A matter of time: A systematic scoping review on a potential role of the circadian system in binge eating behavior. Front Nutr. 2022;9:978412. doi: 10.3389/fnut.2022.978412 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Kandeğer A, Eğilmez Ü, Selvi Y. Feeding and eating disorders in the context of circadian rhythms. Alpha Psychiatry. 2021;22(6):278-284. doi: 10.5152/alphapsychiatry.2021.21151 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Garaulet M, Vizmanos B, Muela T, et al. Evening types as determined by subjective and objective measures are more emotional eaters. Obesity (Silver Spring). 2023;31(5):1192-1203. doi: 10.1002/oby.23749 [DOI] [PubMed] [Google Scholar]
14.Dashti HS, Chen A, Daghlas I, Saxena R. Morning diurnal preference and food intake: a mendelian randomization study. Am J Clin Nutr. 2020;112(5):1348-1357. doi: 10.1093/ajcn/nqaa219 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Moskowitz L, Weiselberg E. Anorexia nervosa/atypical anorexia nervosa. Curr Probl Pediatr Adolesc Health Care. 2017;47(4):70-84. doi: 10.1016/j.cppeds.2017.02.003 [DOI] [PubMed] [Google Scholar]
16.Chesney E, Goodwin GM, Fazel S. Risks of all-cause and suicide mortality in mental disorders: a meta-review. World Psychiatry. 2014;13(2):153-160. doi: 10.1002/wps.20128 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Allison KC, Lundgren JD, O’Reardon JP, et al. Proposed diagnostic criteria for night eating syndrome. Int J Eat Disord. 2010;43(3):241-247. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Bat-Pitault F, Da Silva C, Charvin I, Da Fonseca D. Sleep disturbances in anorexia nervosa subtypes in adolescence. Eat Weight Disord. 2021;26(6):1845-1852. doi: 10.1007/s40519-020-01033-z [DOI] [PubMed] [Google Scholar]
19.Lacey JH, Crisp AH, Kalucy RS, Hartmann MK, Chien CN. Weight gain and the sleeping electroencephalogram: study of 10 patients with anorexia nervosa. BMJ. 1975;4(5996):556-558. doi: 10.1136/bmj.4.5996.556 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Goel NJ, Sadeh-Sharvit S, Trockel M, et al. Depression and anxiety mediate the relationship between insomnia and eating disorders in college women. J Am Coll Health. 2021;69(8):976-981. doi: 10.1080/07448481.2019.1710152 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Duncan L, Yilmaz Z, Gaspar H, et al. ; Eating Disorders Working Group of the Psychiatric Genomics Consortium . Significant locus and metabolic genetic correlations revealed in genome-wide association study of anorexia nervosa. Am J Psychiatry. 2017;174(9):850-858. doi: 10.1176/appi.ajp.2017.16121402 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Watson HJ, Yilmaz Z, Thornton LM, et al. ; Anorexia Nervosa Genetics Initiative; Eating Disorders Working Group of the Psychiatric Genomics Consortium . Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa. Nat Genet. 2019;51(8):1207-1214. doi: 10.1038/s41588-019-0439-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Davies NM, Holmes MV, Davey Smith G. Reading mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601. doi: 10.1136/bmj.k601 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Hebebrand J, Muller TD, Holtkamp K, Herpertz-Dahlmann B. The role of leptin in anorexia nervosa: clinical implications. Mol Psychiatry. 2007;12(1):23-35. doi: 10.1038/sj.mp.4001909 [DOI] [PubMed] [Google Scholar]
25.Peters T, Antel J, Naaresh R, et al. Suggestive evidence for causal effect of leptin levels on risk for anorexia nervosa: results of a mendelian randomization study. Front Genet. 2021;12:733606. doi: 10.3389/fgene.2021.733606 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Christensen KA, Short NA. The case for investigating a bidirectional association between insomnia symptoms and eating disorder pathology. Int J Eat Disord. 2021;54(5):701-707. doi: 10.1002/eat.23498 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Costanzo MC, von Grotthuss M, Massung J, et al. ; AMP-T2D Consortium . The type 2 diabetes knowledge portal: an open access genetic resource dedicated to type 2 diabetes and related traits. Cell Metab. 2023;35(4):695-710.e6. doi: 10.1016/j.cmet.2023.03.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Watanabe K, Jansen PR, Savage JE, et al. ; 23andMe Research Team . Genome-wide meta-analysis of insomnia prioritizes genes associated with metabolic and psychiatric pathways. Nat Genet. 2022;54(8):1125-1132. doi: 10.1038/s41588-022-01124-w [DOI] [PubMed] [Google Scholar]
29.Burgess S, Thompson SG. Interpreting findings from mendelian randomization using the MR-Egger method. Eur J Epidemiol. 2017;32(5):377-389. doi: 10.1007/s10654-017-0255-x [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512-525. doi: 10.1093/ije/dyv080 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Skrivankova VW, Richmond RC, Woolf BAR, et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: the STROBE-MR Statement. JAMA. 2021;326(16):1614-1621. doi: 10.1001/jama.2021.18236 [DOI] [PubMed] [Google Scholar]
32.Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693-698. doi: 10.1038/s41588-018-0099-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557-560. doi: 10.1136/bmj.327.7414.557 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13(11):e1007081. doi: 10.1371/journal.pgen.1007081 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Jones SE, van Hees VT, Mazzotti DR, et al. Genetic studies of accelerometer-based sleep measures yield new insights into human sleep behaviour. Nat Commun. 2019;10(1):1585. doi: 10.1038/s41467-019-09576-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Ge T, Chen CY, Ni Y, Feng YA, Smoller JW. Polygenic prediction via bayesian regression and continuous shrinkage priors. Nat Commun. 2019;10(1):1776. doi: 10.1038/s41467-019-09718-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Wei WQ, Bastarache LA, Carroll RJ, et al. Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record. PLoS One. 2017;12(7):e0175508. doi: 10.1371/journal.pone.0175508 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Denny JC, Ritchie MD, Basford MA, et al. PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations. Bioinformatics. 2010;26(9):1205-1210. doi: 10.1093/bioinformatics/btq126 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Hambleton A, Pepin G, Le A, Maloney D, Touyz S, Maguire S; National Eating Disorder Research Consortium . Psychiatric and medical comorbidities of eating disorders: findings from a rapid review of the literature. J Eat Disord. 2022;10(1):132. doi: 10.1186/s40337-022-00654-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Merikanto I, Partonen T. Eveningness increases risks for depressive and anxiety symptoms and hospital treatments mediated by insufficient sleep in a population-based study of 18,039 adults. Depress Anxiety. 2021;38(10):1066-1077. doi: 10.1002/da.23189 [DOI] [PubMed] [Google Scholar]
41.Kivelä L, Papadopoulos MR, Antypa N. Chronotype and psychiatric disorders. Curr Sleep Med Rep. 2018;4(2):94-103. doi: 10.1007/s40675-018-0113-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.PLOS . Sleep-wake rhythms vary widely with age as well as amongst individuals of a given age. ScienceDaily. Published online June 21, 2017. Accessed June 12, 2023. https://www.sciencedaily.com/releases/2017/06/170621140458.htm
43.van Eeden AE, van Hoeken D, Hoek HW. Incidence, prevalence and mortality of anorexia nervosa and bulimia nervosa. Curr Opin Psychiatry. 2021;34(6):515-524. doi: 10.1097/YCO.0000000000000739 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Lavender JM, Utzinger LM, Crosby RD, et al. A naturalistic examination of the temporal patterns of affect and eating disorder behaviors in anorexia nervosa. Int J Eat Disord. 2016;49(1):77-83. doi: 10.1002/eat.22447 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Monteleone P, Scognamiglio P, Monteleone AM, et al. Abnormal diurnal patterns of salivary α-amylase and cortisol secretion in acute patients with anorexia nervosa. World J Biol Psychiatry. 2011;12(6):455-461. doi: 10.3109/15622975.2011.590226 [DOI] [PubMed] [Google Scholar]
46.Randler C, Schaal S. Morningness-eveningness, habitual sleep-wake variables and cortisol level. Biol Psychol. 2010;85(1):14-18. doi: 10.1016/j.biopsycho.2010.04.006 [DOI] [PubMed] [Google Scholar]
47.Kleppe MM, Kessler U, Rekkedal GÅ, Skjåkødegård HF, Danielsen YS. Differences in sleep patterns between patients with anorexia nervosa and healthy controls: a cross-sectional study. J Eat Disord. 2023;11(1):76. doi: 10.1186/s40337-023-00799-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Sanderson E, Spiller W, Bowden J. Testing and correcting for weak and pleiotropic instruments in two-sample multivariable mendelian randomization. Stat Med. 2021;40(25):5434-5452. doi: 10.1002/sim.9133 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Bowden J, Del Greco M F, Minelli C, Davey Smith G, Sheehan NA, Thompson JR. Assessing the suitability of summary data for two-sample mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int J Epidemiol. 2016;45(6):1961-1974. doi: 10.1093/ije/dyw220 [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Lack L, Wright H, Kemp K, Gibbon S. The treatment of early-morning awakening insomnia with 2 evenings of bright light. Sleep. 2005;28(5):616-623. doi: 10.1093/sleep/28.5.616 [DOI] [PubMed] [Google Scholar]
51.Xu J, Johnson JS, Signer R, et al. ; Eating Disorders Working Group of the Psychiatric Genomics Consortium . Exploring the clinical and genetic associations of adult weight trajectories using electronic health records in a racially diverse biobank: a phenome-wide and polygenic risk study. Lancet Digit Health. 2022;4(8):e604-e614. doi: 10.1016/S2589-7500(22)00099-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Huckins LM, Signer R, Johnson J, Wu YK, Mitchell KS, Bulik CM. What next for eating disorder genetics? replacing myths with facts to sharpen our understanding. Mol Psychiatry. 2022;27(10):3929-3938. doi: 10.1038/s41380-022-01601-y [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Hoek HW, van Harten PN, Hermans KME, Katzman MA, Matroos GE, Susser ES. The incidence of anorexia nervosa on Curaçao. Am J Psychiatry. 2005;162(4):748-752. doi: 10.1176/appi.ajp.162.4.748 [DOI] [PubMed] [Google Scholar]
54.Watson HJ, Thornton LM, Yilmaz Z, et al. Common genetic variation and age of onset of anorexia nervosa. Biol Psychiatry Glob Open Sci. 2021;2(4):368-378. doi: 10.1016/j.bpsgos.2021.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Burgess S, Davies NM, Thompson SG. Bias due to participant overlap in two-sample Mendelian randomization. Genet Epidemiol. 2016;40(7):597-608. doi: 10.1002/gepi.21998 [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