1
Binge-eatingdisorder,anorexianervosa,andconstitutionalthinnessdifferin
theirassociationswithanthropometricandpsychiatricpolygenicscores
ChristopherHübel*,MohamedAbdulkadir*,MoritzHerle,RuthJ.F.Loos,GeromeBreen,
CynthiaM.Bulik,NadiaMicali
Hübel,BreenSocial,Genetic&DevelopmentalPsychiatryCentre,InstituteofPsychiatry,
Psychology&Neuroscience,King’sCollegeLondon,UKHübel,BreenUKNationalInstituteforHealthResearch(NIHR)BiomedicalResearchCentre
forMentalHealth,SouthLondonandMaudsleyHospital,London,UKHübel,BulikDepartmentofMedicalEpidemiologyandBiostatistics,KarolinskaInstitutet,
Stockholm,SwedenMicali,HerleGreatOrmondStreetInstituteofChildHealth,UniversityCollegeLondon,
London,UKHerleDepartmentofBiostatistics&HealthInformatics,InstituteofPsychiatry,Psychology
&Neuroscience,King’sCollegeLondon,UKLoosIcahnSchoolofMedicineatMountSinai,NewYork,NewYork,USAAbdulkadir,MicaliDepartmentofPediatricsGynaecologyandObstetrics,FacultyofMedicine,
UniversityofGeneva,Geneva,SwitzerlandAbdulkadir,MicaliDepartmentofPsychiatry,FacultyofMedicine,UniversityofGeneva,
Geneva,SwitzerlandBulikDepartmentofPsychiatry,UniversityofNorthCarolinaatChapelHill,ChapelHill,
NC,USABulikDepartmentofNutrition,UniversityofNorthCarolinaatChapelHill,ChapelHill,NC,
USA
*contributedequallytothiswork
Correspondenceto:
Dr.Micali,DepartmentofPsychiatry,FacultyofMedicine,UniversityofGeneva,Geneva,
Switzerland,e-mail:[email protected]
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.
2
Keywords:Eatingdisorder;AvonLongitudinalStudyofParentsandChildren(ALSPAC);
purging;constitutionalthinness,fasting,thinideal;bodydissatisfaction,UKBiobank
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
3
Abstract(232/250,unstructured)
Genome-wideassociationstudieshaveidentifiedmultiplegeneticregionsassociated
withanorexianervosa—aneatingdisordercharacterisedbyextremelylowbodyfat,
lowfat-freemass,andrelatedbehavioursandcognitions,suchasrestrictiveeatingand
bodydissatisfaction.Relativelyfewornogenome-widestudiesofothereating
disorders,suchasbulimianervosaandbinge-eatingdisorder,havebeenperformed,
despitetheirsubstantialheritability.Weaimedtoidentifytraitsthataregenetically
associatedwithbinge-typeeatingdisorders.Wecalculatedgenome-widepolygenic
scoresfor302traitanddiseaseoutcomesusingPRSicev2.2andtheirassociationwith
anorexianervosa,bulimianervosa,andbinge-eatingdisorderinupto640casesand
17,050controlsfromtheUKBiobank.Significantassociationsweretestedfor
replicationintheAvonLongitudinalStudyofParentsandChildren(upto217casesand
3018controls).Inadditiontoeatingdisorders,weassociatedpolygenicscoreswith
constitutionalthinness,anddisordered-eatingbehavioursandcognitionsacross
adolescenceatages13,14,16,and/or18years.Individualswithbinge-typeeating
disordershadhigherpolygenicscoresforotherpsychiatricdisorders,including
depression,schizophrenia,andattentiondeficithyperactivitydisorder.Bodymass-
increasingpolygenicscoreswereassociatedwithbinge-eatingbehaviour.
Anthropometricbutnotpsychiatricpolygenicscoreswereassociatedwith
constitutionalthinness,suggestingthatitisaseparateentityfromanorexianervosa.
Ourfindingsmotivateadeeperinvestigationofsharedanduniquegenomicfactorsin
bulimianervosa,binge-eatingdisorder,constitutionalthinness,anddisordered-eating
traits.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
4
Introduction
Eatingdisordersarecomplexandheritablepsychiatricconditions[1].Themost
studiedisanorexianervosa,whichischaracterisedbydangerouslylowbodyweightand
extremefearofgainingweight,whileengaginginbehavioursthatinducenegative
energybalance,suchasfastingor,insomecases,excessiveexercise.[2].Individuals
withbulimianervosaandbinge-eatingdisordershowepisodesofexcessiveovereating
(i.e.,bingeeating)accompaniedbyasenseoflossofcontrol,inwhichtheyconsume
unusuallylargeamountsoffoodinashortperiodoftime.Individualswithbulimia
nervosaalsoengageinrecurrentcompensatorybehaviors(e.g.,self-inducedvomiting,
laxative,ordiureticabuse)tocounteracttheeffectsofbingeeating[3].Incontrastto
anorexianervosa,individualswithconstitutionalthinnesspresentwiththeabsenceof
disordered-eatingbehavioursandcognitions[4],buthavepersistentlylowbodymass
indices(BMIs).Itiscurrentlyunclearhowanorexianervosadiffersfromconstitutional
thinnessonageneticlevel[5].
Twin,family,andadoptionstudiesoverthepast30yearshaveshownthateating
disordersareheritable[1].Thelargestgenome-wideassociationstudy(GWAS)todate
identifiedeightgenomicregionsassociatedwithanorexianervosaandalsofounditis
geneticallycorrelatedwithobsessive-compulsivedisorder,schizophrenia,anxiety,and
majordepressivedisorder.Thisimpliesthatanorexianervosasharesgeneticrisk
variantswiththesephenotypes.However,thestudyalsofoundthatanorexianervosa
hassignificantnegativegeneticcorrelationswithanthropometrictraits,includingBMI,
fatmass,andfat-freemass.Thismeansthatanorexianervosariskvariantsareenriched
forvariantsthatresultinlowerBMI,fatmass,andfat-freemass.[6–8].TheGWAS
additionallyhighlightedasharedgeneticbasisbetweenanorexianervosaandhigh-
densitylipoproteincholesterol,fastinginsulin,aswellasinsulinsensitivity,suggesting
thatanorexianervosamayhaveametaboliccomponent.Itisunclear,however,ifbinge-
typeeatingdisordersmayalsohaveametaboliccomponent.
Althoughtwinandfamilystudiesdemonstrateapositivegeneticcorrelation
(~0.46)betweenanorexianervosaandbulimianervosa,indicatingconsiderableshared
genetics[9,10],GWASsofbulimianervosaorbinge-eatingdisorderwithsufficientsize
haveyettobeconductedtoallowwellpoweredinvestigationsoftheirmolecular
geneticcorrelationswitheachother.Furthermore,itisnotwellunderstoodhowbinge-
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
5
typeeatingdisorderssharegeneticswithotherpsychiatric,anthropometric,and
metabolictraits.
Theextremesofthepopulationrangeofdisordered-eatingbehavioursor
cognitionsarecommoncomponentsymptomsofeatingdisorders(e.g.,extremefasting,
excessiveexercise,orbodydissatisfaction).Thegeneticunderpinningsofthese
componentsymptomsandtheirgeneticrelationshipswitheatingdisorders,psychiatric,
metabolic,andanthropometrictraitshavenotyetbeenexplored.
Tobetterunderstanddifferencesinthegeneticsofeatingdisordersandtheir
componentsymptoms,wehadthreeaimsinthisstudy:(1)Toconductpolygenicscore
analysestoidentifytraitsthataregeneticallyassociatedwitheitherofthethree
primaryeatingdisorders:anorexianervosa,bulimianervosa,orbinge-eatingdisorder
inasubsampleoftheUKBiobank[11]withattemptedreplicationintheAvon
LongitudinalStudyofParentsandChildren(ALSPAC)cohort;(2)investigatethe
associationofpolygenicscoreswithdisordered-eatingbehavioursandrelated
cognitionsatages13,14,16,and/or18yearsinALSPAC.Finally(3)wehypothesize
thatconstitutionalthinnesswouldshowadifferentgeneticarchitecturetothatofeating
disordersgivenitsassociationswithselectedpolygenicscores.
Methods
DiscoverysampleoftheUKBiobank
TheUKBiobank(ukbiobank.ac.uk)isauniqueepidemiologicalresourceto
improveprevention,diagnosis,andtreatmentofpsychiatricandsomaticillnesses.The
UKBiobankrecruitedparticipantsfromthegeneralpopulationbetween2006–2010.All
participantswerebetween40to69yearsold,wereregisteredwithageneral
practitionerthroughtheUnitedKingdom’sNationalHealthService,andlivedwithin
travelingdistanceofoneoftheassessmentcenters.TheUKBiobankisapprovedbythe
NorthWestMulti-centreResearchEthicsCommittee.Genomewidearraydataforthis
studywereavailablefor488,363individuals.Allparticipantsgavewrittenconsent.We
identifiednon-Europeanparticipantsby4-meansclusteringonthefirsttwoprincipal
componentsderivedfromthegenotypedata,andexcludedrelatedindividuals(KING
relatednessmetric>0.088,equivalenttoarelatednessvalueof0.25;N=7,765).
Weidentified1,488participantswhoeitherself-reportedanyofthethree
primaryeatingdisordersinthementalhealthquestionnaire[12]orhadan
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
6
InternationalClassificationofDiseases,version10(ICD-10)[13]hospitaldiagnosisof
F50.0orF50.1foranorexianervosa,orF50.2orF50.3forbulimianervosa.This
resultedin768(4.7%)participantswithanorexianervosa,423(2.7%)withbulimia
nervosa,and561(3.5%)withbinge-eatingdisorder.Ifparticipantsself-reportedmore
thanonediagnosis,theywereassignedtoallgroups.Werandomlysampledonesetof
controlsintheratioof1to10,resultingin15,500controlsbyusingthefollowing
exclusioncriteria:Controlsmusthaveansweredthementalhealthquestionnaire,were
notdiagnosedwithapsychiatricdisorder(i.e.,self-reportorICD-10diagnosis),or
takinganypsychotropicmedication.Thefinalanalysisincluded17,050(92%female)
Europeanparticipants(SupplementaryTable1)representing3.4%ofthegenotyped
UKBiobankparticipants(n=502,682).ThisstudywascompletedundertheUK
Biobankapprovedstudyapplication27546.
SamplefromtheAvonLongitudinalStudyofParentsandChildren(ALSPAC)
TheALSPACisapopulation-basedsampleofpregnantwomenandtheirchildrenbased
intheformercountyofAvon,UK[14–16].WomenexpectedtodeliverfromApril1,
1991,untilDecember31,1992,wereinvitedtoparticipate.Childrenfrom14,541
pregnancieswereenrolledand13,988aliveat1year.Additionally,913childrenwere
enrolledatage7years.Allwomengavewritteninformedconsent.Childrenatage14
(wave14,n=10,581),16(wave16,n=9,702),and18years(wave18,n=9,505)that
hadnotwithdrawnconsentwerefollowedup.Parentsansweredquestionnaireson
7,025adolescentsatwave14andon5,656atwave16.With6,140(58%)respondingat
wave14,5,069(52%)atwave16,and3,228(34%)atwave18whichwasusedto
augmentthevalidityoftheprobableeatingdisorderdiagnoses[17].Pleasenotethatthe
studywebsitecontainsdetailsofallthedatathatisavailablethroughafullysearchable
datadictionaryandvariablesearchtool(www.bristol.ac.uk/alspac/researchers/our-
data/).Toreducepotentialconfoundingthroughgeneticrelatednessinouranalyses,we
removedrandomlyoneindividualofeachpairthatiscloselyrelated(φ>0.2)using
PLINKv1.90[18],excluding75individualsthatwereduplicates,monozygotictwins,
first-degreerelatives(i.e.,parent-offspringandfullsiblings),orsecond-degreerelatives
(i.e.,half-siblings,uncles,aunts,grandparents,anddoublecousins,Supplementary
Table2).
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
7
Wederivedprobableeatingdisorderdiagnosesaspreviouslyreported
(SupplementaryTable3)[17],usingacombinationofadolescentself-reportaswellas
informationontheadolescentsprovidedbytheirparents,agoldstandardforchildhood
psychiatricdisorders.
EthicalapprovalfortheALSPACparticipantsofthisstudywasobtainedfromthe
ALSPACEthicsandLawCommitteeandtheLocalResearchEthicsCommittees:
www.bristol.ac.uk/alspac/researchers/research-ethics/.Consentforbiologicalsamples
hasbeencollectedinaccordancewiththeHumanTissueAct(2004)andinformed
consentfortheuseofdatacollectedviaquestionnairesandclinicswasobtainedfrom
participantsfollowingtherecommendationsoftheALSPACEthicsandLawCommittee
atthetime.
PolygenicriskscoringoneatingdisorderphenotypesinUKBiobankand
replicationinALSPAC
WeusedPRSice[19],version2.2.3.Weclumpedthesinglenucleotide
polymorphisms(SNPs)thatwerepresentbothinthesummarystatisticsofthetraitand
inthegenotypedataoftheUKBiobank(i.e.,overlappingSNPs)toobtaingenetically
independentSNPs.WeretainedtheSNPwiththesmallestpvalueineach250kilobase
windowofallthoseinlinkagedisequilibrium(r2>0.1).Wecalculated302polygenic
scoresattheiroptimalpvaluethresholdineachindividualbyscoringthenumberof
effectalleles(weightedbythealleleeffectsize)acrossthesetofremainingSNPs(fora
fulllist,seeSupplementaryTable4).Wecalculatedthepolygenicscoreusingthehigh-
resolutionscoring(i.e.,incrementallyacrossalargenumberofpvaluethresholds)
methodtoidentifythepvaluethresholdatwhichthepolygenicscoreisoptimally
associatedwiththeoutcomeandexplainsthemostvariance(i.e.,resultinginthe
highestadjustedR2forcontinuousoutcomesandNagelkerke'sR2ontheliabilityscale
forbinaryoutcomes).Weevaluatedtheassociationsbetweenpolygenicscoreand
eatingdisorderdiagnosisusinglogisticregressions,adjustedforsexandthefirstsix
principalcomponentsthatwerecalculatedontheEuropeansubsample.Toadjustfor
overfitting,wepermutedcase-controlstatusateverypvaluethreshold10,000times
and,hence,calculatedempiricalpvalues.WeconvertedtheobservedR2totheliability
scaleassumingfollowingpopulationprevalencesof3%foranorexianervosa,1.8%for
bulimianervosa,and3%forbinge-eatingdisorder[20,21].
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
8
Tocorrectformultipletesting(i.e.,302polygenicscoreregressionmodels),we
calculatedQvaluesusingthefalsediscoveryrateapproach[22,23].Wedidnotstratify
analysesbysexbecauseofthelownumberofmaleeatingdisordercases,butincluded
sexasacovariate.WeusedtheUKBiobanksampleasourdiscoverycohortandthe
SNPsthatwereincludedinthestrongestassociatedpolygenicscorewereusedtoderive
polygenicscoresintheALSPACsample,inwhichwerepeatedtheanalysisalso
adjustingforsex,andthefirstsixancestry-informativeprincipalcomponents.
Polygenicriskscoringondisordered-eatingbehavioursandcognitionsinALSPAC
Weusedthepolygenicscoresthatweresignificantlyassociatedwitheating
disordersintheUKBiobankforafollow-upanalysisintheindependentALSPACsample.
Weinvestigatedtheassociationbetweenthesepolygenicscoresanddisordered-eating
behavioursorcognitionswhicharecomponentsymptomsofeatingdisorders.Forthese
analyses,weusedtheoptimalthresholdingmethoddescribedabove.Below,we
summarisethedisordered-eatingphenotypesthatwereassessed[17].
Disordered-eatingbehaviours
Excessiveexercise.Atage14,16,and18years,participantsreportedtheir
exercisefrequencyperweekduringthepreviousyear,iftheirexercisinginterferedwith
schoolwork,iftheyexercisedtoloseoravoidgainingweight,andiftheyexercisedeven
thoughtheyweresickorinjured.
Fasting.Atage14,16,and18years,participantsreportedthefrequencyoffasting(i.e.,
noteatingforanentireday)toloseoravoidgainingweightduringthepastyear.
Bingeeating.Atage14,16,and18years,participantsreportedthefrequencyofbinge
eatingduringthepastyear—definedaseatinganunusuallylargeamountoffood,
combinedwithafeelingofbeingoutofcontrol.
Purging.Atage14,16,and18years,participantsreportedthefrequencywithwhich
theyusedlaxativesorself-inducedvomiting,orothermedicinestoloseoravoidgaining
weightduringthepastyear.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
9
Disordered-eatingcognitions
Fearofweightgain.Atage14and16years,participantsreportediftheywere
afraidofgainingweightorgettingfat,iftheyavoidedfatteningfoodsorhowtheywould
feeliftheywereaskedtoputon2kilosforthesakeoftheirhealth.
Thinidealinternalization.Atage14years,adolescentgirlsreportedhowstrongly
theyagreedwiththinwomenbeingnotattractive,womenwithlonglegsbeingmore
attractive,womenwithtoned(lean)bodiesbeingmoreattractive,tallwomenbeing
moreattractive,shapelywomenbeingmoreattractiveorshort(petite)womenbeing
moreattractive.Adolescentboysreportedhowstronglytheyagreedwiththinmen
beingmoregood-looking,menwhoareinshapebeingbetterlooking,menwithtoned
(lean)bodiesbeingmoregood-looking,menwithlargepronouncedmusclesbeing
betterlooking,tallmenbeingmoregood-looking,orbodybuilders,suchasArnold
Schwarzenegger,beingbetterlooking[24].
WeightandshapeconcernAtage14years,participantsreportedhowhappytheyhad
beeninthepreviousyearwiththewaytheirbodylooked,howmuchtheirweighthad
madeadifferencetohowtheyfeelaboutthemselves,andhowmuchtheyhadworried
aboutgainingalittleweight(e.g.,1kg).
BodydissatisfactionAtage14years,participantsreportedhowsatisfiedtheywereat
thatmomentwiththeirbodybuild/breast,stomach,waist,thighs,buttocks,hips,legs,
face,hair[25].
ConstitutionalthinnessinALSPAC
Weidentifiedparticipantswithlong-termconstitutionalthinness(i.e.,persistent
thinness)analysingobjectivelymeasuredBMI(kg/m²)at10,12,14,15,and18yearsin
theALSPACusinglatentgrowthanalysis.Additionally,participantsdidneitherendorse
weightlossbehavioursnorwereaffectedbyaneatingdisorder(forfulldetails,see
SupplementaryInfo)
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
10
Figure 1. Polygenic scores associated with eating disorders in the UK Biobank. Panel A shows psychiatric and behavioral, Panel B metabolic, and Panel C anthropometric polygenic scores that are associated with self-reported or hospital-diagnosed eating disorders in the UK Biobank sample (n = 17,050). Filled dots are statistically significant after adjustment for multiple testing through the false discovery discovery approach. Dots represent odds ratios (ORs) and error bars index 95% confidence intervals obtained via logistic regression and 10,000 permutations to obtain empirical p values.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
11
Results
PolygenicscoringoneatingdisorderphenotypesinUKBiobankandALSPAC
Aftercorrectingformultipletestingusingthefalsediscoveryrateadjustment,21
polygenicscoresweresignificantlyassociatedwiththethreeprimaryeatingdisorders
inthesubsampleoftheUKBiobank(Figure1).
Psychiatricdisordersandneurologicaldisease.Fivepolygenicscores
reflectingpolygenicliabilityforpsychiatricdisordersandoneneurologicaldisease
showedsignificantassociationswitheatingdisorders(SupplementaryTable4).The
schizophreniapolygenicscorewaspositivelyassociatedwithanorexianervosa(odds
ratio[OR]perstandarddeviationinpolygenicscore=1.18,95%confidenceinterval
[CI]:1.09,1.70;Q=0.03)andbinge-eatingdisorder(OR=1.23,95%CI:1.13,1.35;Q=
0.01).Themajordepressivedisorderpolygenicscorewaspositivelyassociatedwith
anorexianervosa(OR=1.20,95%CI:1.11,1.29;Q=0.01)andbinge-eatingdisorder
(OR=1.25,95%CI:1.14,1.36;Q=0.01).Additionally,theattentiondeficithyperactivity
disorderpolygenicscorewaspositivelyassociatedwithbinge-eatingdisorder(OR=
1.28,95%CI:1.17,1.39;Q=0.01).Thebipolardisorderpolygenicscore(OR=1.17,
95%CI:1.08,1.26;Q=0.04),theanorexianervosapolygenicscore(OR=1.32,95%CI:
1.22,1.42;Q=0.01),andthemigrainepolygenicscore(OR=1.17,95%CI:1.09,1.26;Q
=0.04)werepositivelyassociatedwithanorexianervosa.
Somatictraits.Severalpolygenicscoresindexingpolygenicloadfor
anthropometrictraits,includinghipcircumference(OR=1.39,95%CI:1.27,1.51;Q=
0.01),overweight(OR=1.35,95%CI:1.24,1.47;Q=0.01),obesityclass1(OR=1.38,
95%CI:1.27;1.51;Q=0.01),class2(OR=1.33,95%CI:1.22;1.45;Q=0.01)andclass3
(OR=1.21,95%CI:1.11,1.32;Q=0.04),BMI(OR=1.44,95%CI:1.32;1.56;Q=0.01)
andextremeBMI(OR=1.32,95%CI:1.22,1.44;Q=0.01)werepositivelyassociated
withbinge-eatingdisorder.However,itisimportanttonotethatthesepolygenicscores
werecorrelatedwitheachotherwithr=.10tor=.67(SupplementaryFigure2).
Additionally,thechildhoodobesitypolygenicscore(OR=1.22,95%CI:1.12,1.33;Q=
0.04)waspositivelyassociatedwithbinge-eatingdisorder,whiletheageatmenarche
polygenicscore(OR=0.79,95%CI:0.72,0.86;Q=0.01)wasnegativelyassociatedwith
binge-eatingdisorder.Neitherthechildhoodobesitynortheageatmenarchepolygenic
scorewereassociatedwithanorexianervosa.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
12
Theoverweightpolygenicscore(OR=1.26,95%CI:1.14,1.38;Q=0.01)andthe
waistcircumferencepolygenicscore(OR=1.25,95%CI:1.13,1.38;Q=0.02)were
positivelyassociatedwithbulimianervosa,whereastheredbloodcellcountpolygenic
scorewasnegativelyassociatedwithbulimianervosa(OR=0.80,95%CI:0.72,0.88;Q=
0.02).
Sexandagedifferences.Theadulthoodobesitypolygenicscorewasmore
stronglyassociatedwithanorexianervosathanthechildhoodobesitypolygenicscore.
Thisdifferenceinthelogoddsratioswasstatisticallysignificant(difference=-0.15,p=
0.004).Wedidnotdetectagedifferencesintheassociationsoftheobesitypolygenic
scoreswithbulimianervosaorbinge-eatingdisorder(SupplementaryTable7).We
detectednosexdifferencesintheassociationsofthebodymassindexpolygenicscore
withanorexianervosa,bulimianervosa,orbinge-eatingdisorder.
Sensitivityanalysis.Wecomparedlogisticregressionresultswhenassigning
participantstoallpossibleeatingdisorderdiagnoseswithassigningparticipants
exclusivelytooneeatingdisorderdiagnosisintheUKBiobanksample.Theeffectsizes
ofbothanalysescorrelatedwithr=0.98,indicatingthattheanalyseswerenotsensitive
toparticipantoverlap(SupplementaryFigure3).
Independentreplication.InALSPAC(SupplementaryTable8),the
overweightpolygenicscore(OR=1.38,95%CI:1.13,1.69;pempirical=0.05)andthe
obesityclass1polygenicscore(OR=1.23,95%CI:1.00;1.50;pempirical=0.001)were
positivelyassociatedwithbinge-eatingdisorder.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
13
Figure 2. Polygenic scores associated with purging, constitutional thinness, and binge eating in the Avon Longitudinal Study of Parents and Children. Sample sizes per outcome were purging (n cases= 398, n controls = 2962), constitutional thinness (n cases = 201, n controls = 6391), and binge eating (n cases = 840, n controls = 2151). Polygenic scores were calculated using PRSice v2. The optimal p value threshold to generate the polygenic score was obtained by calculating polygenic scores across multiple thresholds and permuting case-control status at each threshold 10,000 times. The polygenic score explaining the largest trait variance was used in logistic regressions including the first six ancestry principal components and sex as covariates. Dots represent odds ratios (OR) and error bars 95% confidence intervals (95% CI). Filled dots are those odds ratios that are significant after correction for multiple testing via calculation of false discovery rate-adjusted Q values.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
14
Polygenicscoresassociatedwithdisordered-eatingbehavioursandcognitions
InALSPAC,thedisordered-eatingtraitswereintercorrelatedwithr=.45-.64
(SupplementaryFigure4),explainingsomeofthesimilaritiesintheresults.
Anthropometricpolygenicscores.Severalpolygenicscoresofanthropometric
measureswerepositivelyassociatedwithdisordered-eatingbehavioursandcognitions
(Figures2,3,and4).Forinstance,theoverweightpolygenicscorewaspositively
associatedwithexcessiveexercise(OR=1.16,95%CI:1.06;1.26;Q=0.03),fasting(OR
=1.17,95%CI:1.08;1.27;Q=0.01),bingeeating(OR=1.17,95%CI:1.08;1.26;Q=
0.01),weightandshapeconcerns(β=0.13,SE=0.03;Q=6.1x10-4),body
dissatisfaction(β=0.71,SE=0.01;Q=6.1x10-4)andnegativelyassociatedwiththin
idealinternalisation(β=-0.13,SE=0.04;Q=0.03).Thesamepatternwasobserved
withadditionalpolygenicscoresforanthropometrictraitsthatareintercorrelated(r=
.06-.64,SupplementaryFigure5),includingwaistcircumference,hipcircumference,
severalcategoriesofobesity(forfullresults,seeFigures2,3,and4PanelC,and
SupplementaryTable9).Becauseofthesepropertiesweonlyreportresultsonthe
overweightpolygenicscoreinthefollowingsection.Wereportedontheassociations
betweenthebodymassindexpolygenicscoreanddisordered-eatingtraitsinALSPAC
previously,includingformalmediationanalysesandinvestigationofagedependency
[26].
Childhoodobesitypolygenicscore.Weinvestigatedtherelationshipbetween
geneticliabilityforchildhoodobesityanddisordered-eatingbehavioursandcognitions.
Thechildhoodobesitypolygenicscorewaspositivelyassociatedwithfearofweight
gain(OR=1.16,95%CI:1.07;1.25;Q=0.02),excessiveexercise(OR=1.22,95%CI:
1.13;1.32;Q=6.1x10-4),fasting(OR=1.16,95%CI:1.08;1.26;Q=0.01),bingeeating
(OR=1.19,95%CI:1.10;1.29;Q=0.002),weightshapeconcerns(β=0.17,SE=0.03;Q
=6.1x10-4),andbodydissatisfaction(β=0.80,SE=0.10;Q=6.1x10-4)inadolescence.
(SupplementaryTable10).
Sexdifferences.Overall,wedidnotdetectsexdifferences.However,onlythe
femaleBMIpolygenicscorewaspositivelyassociatedwithpurginginadolescence(OR=
1.20,95%CI:1.08;1.34;Q=0.03),butshowednodifferencesineffectsizewhen
formallytested.Nootheranthropometricorpsychiatricpolygenicscoreshowedan
associationwithpurging.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
15
Figure 3. Polygenic scores associated with fasting, excessive exercise, and fear of weight gain in the Avon Longitudinal Study of Parents and Children. Sample sizes per outcome were fasting (n cases = 777, n controls = 2678), excessive exercise (n cases = 628, n controls = 4596), and fear of gaining weight (n cases = 770, n controls = 2812). Polygenic scores were calculated using PRSice v2. The optimal p value threshold to generate the polygenic score was obtained by calculating polygenic scores across multiple thresholds and permuting case-control status at each threshold 10,000 times. The polygenic score explaining the largest trait variance was used in logistic regressions including the first six ancestry principal components and sex as covariates. Dots represent odds ratios (OR) and error bars 95% confidence intervals (95% CI). Filled dots are those odds ratios that are significant after correction for multiple testing via calculation of false discovery rate-adjusted Q values.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
16
Ageatmenarchepolygenicscore.Theageatmenarchepolygenicscorewas
negativelyassociatedwithexcessiveexercise(OR=0.86,95%CI:0.80;0.94;Q=0.02),
fasting(OR=0.87,95%CI:0.80,0.94;Q=0.02),bingeeating(OR=0.83,95%CI:0.77;
0.90;Q=0.002),weightandshapeconcerns(β=0.15,SE=0.03;Q=6.1x10-4),and
bodydissatisfaction(β=0.54,SE=0.11;Q=1.1x10-3),indicatingthatageneticliability
foralaterageatmenarcheisassociatedwithloweroddsfordisordered-eating
cognitionsorengagingindisordered-eatingbehaviour.
Psychiatricandneurologicalpolygenicscores.TheADHDpolygenicscore
waspositivelyassociatedwithfasting(OR=1.18,95%CI:1.09,1.29;Q=0.008;Figure
3,PanelB,andSupplementaryTable2,3,and4)andtheschizophreniapolygenic
scorewaspositivelyassociatedwithbingeeating(OR=1.17,95%CI:1.07;1.27;Q=
0.01).Boththemajordepressivedisorder(β=0.09,SE=0.03;Q=0.02)andthe
migrainepolygenicscores(β=0.09,SE=0.03;Q=0.03)werepositivelyassociatedwith
weightandshapeconcern.Similarly,boththedepression(β=0.40,SE=0.11;Q=0.02)
andthemigrainepolygenicscore(β=0.38,SE=0.11;Q=0.03)wereassociatedwith
bodydissatisfaction.
Polygenicscoresassociatedwithconstitutionalthinness
Theoverweightpolygenicscorewasnegativelyassociatedwithconstitutional
thinness(OR=0.63,95%CI:0.55,0.73;Q=6.1x10-4)similartootheranthropometric
polygenicscores(Figure2,PanelC)aswellasthechildhoodobesitypolygenicscore
(OR=0.54,95%CI:0.46;0.62;Q=6.1x10-4).Thechildhoodpolygenicscorewasmore
stronglyassociatedwithconstitutionalthinnessthantheadulthoodobesitypolygenic
score(pdiff=1.0x10-4),butwedidnotobservesexdifferencesforthebodymassindex
polygenicscore.Nopolygenicscoresforthepsychiatricorneurologicalillnesses
includedinouranalysiswereassociatedwithconstitutionalthinness.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
17
Figure 4. Polygenic scores associated with thin ideal internalization, body dissatisfaction, and weight and shape concerns in the Avon Longitudinal Study of Parents and Children. Sample sizes per outcome were thin ideal internalization (n = 4495, mean = 15.3, SD = 2.69), body dissatisfaction (n = 4624, mean = 21.85, SD = 7.75), weight and shape concerns (n = 4622, mean = 5.34, SD = 1.85). Polygenic scores were calculated using PRSice v2. The optimal p value threshold to generate the polygenic score was obtained by calculating polygenic scores across multiple thresholds and permuting case-control status at each threshold 10,000 times. The polygenic score explaining the largest trait variance was used in logistic regressions including the first six ancestry principal components and sex as covariates. Points represent odds ratios (OR) and error bars 95% confidence intervals (95% CI). Filled dots are those odds ratios that are significant after correction for multiple testing via calculation of false discovery rate-adjusted Q values.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
18
Discussion
Here,wefindmoleculargeneticevidencethatpsychiatricdisordersand
anthropometrictraitssharegenomicswithbinge-typeeatingdisorders(seeFigure1,
PanelC).OurUKBiobankanalysesshowedstrongpositiveassociationsbetween
bulimianervosa/binge-eatingdisorderandanthropometricpolygenicscores(e.g.,
overweightandwaistcircumference),suggestingthatbingeeatingsharesgenomic
variantswithoverweightandobesity.Bycontrast,inanorexianervosa,thedirectionof
theseassociationswasreversed.Binge-eatingdisorderalsoshowedanassociationwith
thechildhoodobesitypolygenicscorewhileanorexianervosadidnot,further
highlightingthegenomicdifferencesbetweenbinge-typeeatingdisordersandanorexia
nervosa.ThegeneticrelationshipbetweenbingeeatingandBMIhadbeenstudied
previously[27]andisreplicatedbyouranalysis,bothatthedisorderlevel(i.e.,
individualswithbinge-eatingdisorder)andsymptomlevel(i.e.bingeeating).Itis
notablethatpreviousworksuggeststhatgenomicvariantsassociatedwithBMIare
predominantlyexpressedinbraintissue[28].Ourresultsalsosuggestthatthese
genomicvariantsarerelevantforeatingdisorders,butmayactinoppositedirectionsin
binge-typeeatingdisordersandanorexianervosa.
UsingtheUKBiobankdata,wealsofoundthatpsychiatricpolygenicscores(e.g.,
schizophreniaandmajordepressivedisorder)werepositivelyassociatedwithboth
anorexianervosaandbinge-eatingdisorder(Figure1,PanelA).TheADHDpolygenic
scorewasonlysignificantlyassociatedwithbinge-eatingdisorder,whiletheanorexia
polygenicscorewasnotassociatedwithbulimianervosaorbinge-eatingdisorder.
Ourfindingsareinagreementwithpreviousstudiesthatshowedgenetic
correlationsbetweenbingeeatinganddepression[29]aswellasgeneticcorrelations
betweenanorexianervosaanddepression[7,30],migraine[31],schizophrenia[6,7],
bipolardisorder[32]andanthropometrictraits[6,7].Inlinewithourfindings,
previousGWASsshowednogenomicassociationbetweenanorexianervosaandADHD,
whereasanADHDpolygenicscorewasassociatedwithbingeeatinginaSwedishtwin
sample[33,34].Moreover,alargeevidence-basehighlightsphenotypicoverlap
betweenADHDandbingeeating[35–37].Asmostpreviousgeneticstudiesfocusedon
anorexianervosa,wepresentthefirstmoleculargeneticevidencethattheunderlying
biologybetweenbinge-typeeatingdisordersandanorexianervosadiffersandthese
differencescanbecapturedatthegenomiclevel.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
19
LargeGWASshaverecentlysuggestedtheinvolvementofametaboliccomponent
inanorexianervosathroughgeneticcorrelations[6,7];however,atthecurrentsample
size,wedidnotobserveassociationsbetweenmetabolicpolygenicscoresandthethree
primaryeatingdisordersintheUKBiobank(Figure1,PanelB).
Anthropometricpolygenicscores,includinganoverweightpolygenicscore,were
positivelyassociatedwithdisordered-eatingbehaviors(i.e.,bingeeating,fasting,
excessiveexercise)andcognitions(i.e.,weightandshapeconcerns,bodydissatisfaction,
andfearofweightgain)presentinthegeneralpopulation.Thisindicatesthatgenomic
propensityforgreaterbodymassisassociatedwithanegativeperceptionofone'sown
body,whichcanfurtherleadtocompensatorybehaviors.Previouslongitudinal
mediationanalysesonthissample[26]andonUSAmericanadolescentshavefound
thatphenotypicBMImediatestheassociationsbetweenaBMIpolygenicscoreand
weightlossbehaviorsandgoals[38].
Theassociationbetweenpolygenicliabilityforhigherbodymassandeating
disorders,relatedbehaviors,andcognitionsisfurthercomplicatedasitshowsage
dependence[8]andconfoundingthroughageatmenarche.Inbothcohorts,wefound
that:1)ageatmenarcheandBMIpolygenicscoreswerenegativelycorrelatedwitheach
otherand2)thedisordered-eatingtraitswhichwerepositivelyassociatedwiththeBMI
polygenicscorewerenegativelyassociatedwiththeageatmenarchepolygenicscore
andviceversa.Forexample,bodydissatisfactionshowedastrongpositiveassociation
withtheBMIpolygenicscorebutastrongnegativeassociationwiththeageatmenarche
polygenicscore(Figure4,PanelB&C).Phenotypicassociationsbetweenearly
menarcheandbulimianervosahavepreviouslybeenreported[39,40],butdonot
alwaysreplicate[41]andrelatedsymptomslikebodydissatisfactionshowaBMI-
mediatedassociationwithageatmenarche[42,43].Onageneticlevel,bingeeating
correlatednegativelywithageatmenarcheinatwinstudy[44]andanorexianervosa
wasnotgeneticallycorrelatedwithageatmenarcheinourpolygenicscoreanalysisnor
inpreviousGWASs[6,7].Thissuggeststhattheassociationbetweenageatmenarche
anddisorderedeatingmaybedrivenbysymptomsrelatedtobingeeatingratherthan
restriction.
Itisdifficulttodisentangletherelationamongageatmenarche,BMI,and
disorderedeating.PreviousMendelianrandomizationstudiesimplicateabidirectional
relationshipbetweenageatmenarcheandBMI[45,46],indicatingsharedcausal
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
20
mechanisms.Weobservedasimilarphenomenoninouranalysis.Toacertainextentthe
magnitudeoftheassociationbetweentheageatmenarchepolygenicscoreandthe
disordered-eatingtraitsseemstobedeterminedbythemagnitudeoftheassociation
betweenthedisordered-eatingtraitsandBMI.Ingeneral,effectswereinopposite
directions(exceptforanorexianervosa)andwarrantinvestigationusingmediationand
conditionalanalyses.
Psychiatricpolygenicscoreswereassociatedwithdisordered-eatingbehaviors
andcognitions.Inoursymptom-levelALSPACanalysis,weightandshapeconcernsas
wellasbodydissatisfactionwerepositivelyassociatedwiththemajordepressive
disorderpolygenicscore,suggestingthatindividualswhocarryagenomicpropensity
fordepressivesymptomsmayperceivetheirbodymorenegatively.Furthermore,ADHD
andBMIshowapositivegeneticcorrelation[8,47]andinouranalysistheADHD
polygenicscorewasassociatedwithbinge-eatingdisorder.Atthesymptomlevel,
however,theassociationbetweentheADHDpolygenicscoreandbingeeatingwasnot
statisticallysignificantaftercorrectionformultipletesting.Nevertheless,wefounda
significantassociationbetweentheADHDpolygenicscoreandfasting.Thissuggests
thatthesharedbiologybetweenADHDandBMImaycontributetobingeeatingand
subsequentcompensatoryfasting.Thesenovelfindingswarrantfurtherresearch;
investigatinghowgenomicvariantsassociatedwithADHD,commonlycharacterizedby
inattentionandimpulsivity,arealsoassociatedwithbingeeatingandfastingand
contributetothehighcomorbiditybetweenADHDandobesity[35–37].
Ashypothesized,constitutionalthinness(characterizedbypersistentlowBMI
andnodisordered-eatingbehaviorsorcognitions)wassignificantlynegatively
associatedwithanthropometricpolygenicscoresbutwasnotassociatedwithany
psychiatricpolygenicscores.Thisclearlydelineatesconstitutionalthinnessfrom
anorexianervosa,whichincontrastissignificantlyassociatedwithseveralpsychiatric
polygenicscores(i.e.,schizophrenia,majordepressivedisorder,andbipolardisorder)
inourUKBiobankanalysis.Ourfindingsconfirmthatconstitutionalthinnessand
anorexianervosaareseparateentitiesthatmaybecharacterizedbydifferencesinan
underlyinggenomicliabilityforpsychiatricillness.
Ourfindingsmustbeinterpretedinthelightofthefollowinglimitations:The
numberofindividualswitheatingdisordersinbothsampleswasrelativelylow,someof
theeatingdisorderdiagnosesandsymptomswereself-reported,andthesamplesonly
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
21
includedwhiteBritishparticipants.Furthermore,wewereunabletodifferentiate
betweenbinge-eating/purgingandrestrictinganorexianervosaassubtypeinformation
wasnotavailable.However,symptomsintheALSPACcohortwereobtainedbothfrom
parentsandadolescents,representingthegoldstandardinchildandadolescent
psychiatry,strengtheningthevalidityofthediagnoses.Furthermore,ouranalysis
benefittedfromtheinclusionofindividualswithconstitutionalthinnessthatwedirectly
comparedwithindividualswithanorexianervosa.Studiesoftenfocusonhighweight
andobesityandneglectthelowerendoftheweightspectrum.
Inthefuture,itwillbeimportanttocombinemeasuresofdisordered-eating
traitsandbinaryeatingdiagnoses.Usingthisapproach,researcherswillbeableto
determinewhethertheprimaryeatingdisordersmaylieonaliabilitycontinuumandif
continuousmeasuresmayfaithfullycapturetheunderlyingcontinuous"essence"of
eachdisorder,includinggenomic,biochemical,andpsychologicaltraits.Ourfindings
showforthefirsttimethatsimilaritiesexistinthegenomicpsychiatricliabilityfor
binge-typeeatingdisordersandanorexianervosa.However,wefindclear
dissimilaritiesbetweenbinge-typeeatingdisordersandanorexianervosainthe
underlyingbiologyofbodymassregulationatthegenomiclevel.Thesefindingsopenup
importantavenuesfortranslationalresearchrelevanttoEDphenotypesandoverlap
betweeneatingdisorders.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
22
Acknowledgements
Weareextremelygratefultoallthefamilieswhotookpartinthisstudy,themidwives
fortheirhelpinrecruitingthem,andthewholeALSPACteam,whichincludes
interviewers,computerandlaboratorytechnicians,clericalworkers,researchscientists,
volunteers,managers,receptionistsandnurses.Thisstudywascompletedaspartof
approvedUKBiobankstudyapplications27546toProfBreen.
Funding
ThisstudyrepresentsindependentresearchpartfundedbytheUKNationalInstitute
forHealthResearch(NIHR)BiomedicalResearchCentreatSouthLondonandMaudsley
NHSFoundationTrustandKing’sCollegeLondon.Theviewsexpressedarethoseofthe
author(s)andnotnecessarilythoseoftheUKNHS,theNIHRortheDepartmentof
Health.Highperformancecomputingfacilitieswerefundedwithcapitalequipment
grantsfromtheGSTTCharity(TR130505)andMaudsleyCharity(980).Thisworkwas
supportedbytheUKMedicalResearchCouncilandtheMedicalResearchFoundation
(ref:MR/R004803/1).TheUKMedicalResearchCouncilandWellcome(Grantref:
102215/2/13/2and217065/Z/19/Z)andtheUniversityofBristolprovidecore
supportforALSPAC.AcomprehensivelistofgrantsfundingisavailableontheALSPAC
website(http://www.bristol.ac.uk/alspac/external/documents/grant-
acknowledgements.pdf);ThisresearchwasspecificallyfundedbytheNIHR
(CS/01/2008/014),theNIH(MH087786-01).GWASdatawasgeneratedbySample
LogisticsandGenotypingFacilitiesatWellcomeSangerInstituteandLabCorp
(LaboratoryCorporationofAmerica)usingsupportfrom23andMe.NMandCB
acknowledgefundingfromtheNationalInstituteofMentalHealth(R21MH115397).CB
acknowledgesfundingfromtheSwedishResearchCouncil(VRDnr:538-2013-8864),
theNationalInstituteofMentalHealth(R21MH115397;R01MH109528;
R01MH120170;R01MH119084).Thecontentissolelytheresponsibilityoftheauthors
anddoesnotnecessarilyrepresenttheofficialviewsoftheNationalInstitutesofHealth.
Thefunderswerenotinvolvedinthedesignorconductofthestudy;collection,
management,analysis,orinterpretationofthedata;orpreparation,review,orapproval
ofthemanuscript.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
23
Conflictofinterest
Dr.BreenhasreceivedgrantfundingfromandservedasaconsultanttoEliLilly,has
receivedhonorariafromIlluminaandhasservedonadvisoryboardsforOtsuka.Dr.
BulikisagrantrecipientfromandhasservedonadvisoryboardsforShireandisa
consultantforIdorsia.ShereceivesroyaltiesfromPearson.Allotherauthorshave
indicatedtheyhavenoconflictsofinteresttodisclose.
AuthorContribution
CH,MA,MHanalysedthedata.CH,MA,MHdraftedthemanuscript.CMB,NM,RFL,and
GBsupervisedthework.Allauthorssubstantiallycontributedtotheconceptionand
interpretationofthework,revisedthemanuscriptforimportantintellectualcontent
andapprovedthefinalversion.Allauthorsagreetobeaccountableforallaspectsofthis
work.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
24
References
1. YilmazZ,HardawayJA,BulikCM.GeneticsandEpigeneticsofEatingDisorders.AdvGenomicsGenet.2015;5:131–150.
2. TreasureJ,ZipfelS,MicaliN,WadeT,SticeE,ClaudinoA,etal.Anorexianervosa.NatRevDisPrimers.2015;1:15074.
3. AmericanPsychiatricAssociation.DiagnosticandStatisticalManualofMentalDisorders.5thed.Washington,DC:AmericanPsychiatricAssociation;2013.
4. SlofR,MazzeoS,BulikCM.Characteristicsofwomenwithpersistentthinness.ObesRes.2003;11:971–977.
5. Riveros-McKayF,MistryV,BoundsR,HendricksA,KeoghJM,ThomasH,etal.Geneticarchitectureofhumanthinnesscomparedtosevereobesity.PLoSGenet.2019;15:e1007603.
6. DuncanL,YilmazZ,GasparH,WaltersR,GoldsteinJ,AnttilaV,etal.SignificantLocusandMetabolicGeneticCorrelationsRevealedinGenome-WideAssociationStudyofAnorexiaNervosa.AmJPsychiatry.2017;174:850–858.
7. WatsonHJ,YilmazZ,ThorntonLM,HübelC,ColemanJRI,GasparHA,etal.Genome-wideassociationstudyidentifieseightrisklociandimplicatesmetabo-psychiatricoriginsforanorexianervosa.NatGenet.2019;51:1207–1214.
8. HübelC,GasparHA,ColemanJRI,HanscombeKB,PurvesK,ProkopenkoI,etal.Geneticcorrelationsofpsychiatrictraitswithbodycompositionandglycemictraitsaresex-andage-dependent.NatCommun.2019;10:5765.
9. BulikCM,ThorntonLM,RootTL,PisetskyEM,LichtensteinP,PedersenNL.UnderstandingtherelationbetweenanorexianervosaandbulimianervosainaSwedishnationaltwinsample.BiolPsychiatry.2010;67:71–77.
10. YaoS,LarssonH,NorringC,BirgegårdA,LichtensteinP,DʼOnofrioBM,etal.Geneticandenvironmentalcontributionstodiagnosticfluctuationinanorexianervosaandbulimianervosa.PsycholMed.2019:1–8.
11. SudlowC,GallacherJ,AllenN,BeralV,BurtonP,DaneshJ,etal.UKbiobank:anopenaccessresourceforidentifyingthecausesofawiderangeofcomplexdiseasesofmiddleandoldage.PLoSMed.2015;12:e1001779.
12. DavisKAS,ColemanJRI,AdamsM,AllenN,BreenG,CullenB,etal.MentalhealthinUKBiobank-development,implementationandresultsfromanonlinequestionnairecompletedby157366participants:areanalysis.BJPsychOpen.2020;6:e18.
13. WorldHealthOrganization.ICD-10:InternationalStatisticalClassificationofDiseasesandRelatedHealthProblems:10threvision.Geneva:WorldHealthOrganization;1992.
14. GoldingJ,PembreyM,JonesR,ALSPACStudyTeam.ALSPAC--theAvonLongitudinalStudyofParentsandChildren.I.Studymethodology.PaediatrPerinatEpidemiol.2001;15:74–87.
15. BoydA,GoldingJ,MacleodJ,LawlorDA,FraserA,HendersonJ,etal.CohortProfile:the‘childrenofthe90s’--theindexoffspringoftheAvonLongitudinalStudyofParentsandChildren.IntJEpidemiol.2013;42:111–127.
16. FraserA,Macdonald-WallisC,TillingK,BoydA,GoldingJ,DaveySmithG,etal.CohortProfile:theAvonLongitudinalStudyofParentsandChildren:ALSPACmotherscohort.IntJEpidemiol.2013;42:97–110.
17. MicaliN,SolmiF,HortonNJ,CrosbyRD,EddyKT,CalzoJP,etal.AdolescentEatingDisordersPredictPsychiatric,High-RiskBehaviorsandWeightOutcomesinYoungAdulthood.JAmAcadChildAdolescPsychiatry.2015;54:652–659.e1.
18. ChangCC,ChowCC,TellierLC,VattikutiS,PurcellSM,LeeJJ.Second-generationPLINK:risingtothechallengeoflargerandricherdatasets.Gigascience.2015;4:7.
19. ChoiSW,O’ReillyPF.PRSice-2:PolygenicRiskScoresoftwareforbiobank-scaledata.Gigascience.2019;8.
20. MicaliN,MartiniMG,ThomasJJ,EddyKT,KothariR,RussellE,etal.Lifetimeand12-monthprevalenceofeatingdisordersamongstwomeninmid-life:apopulation-basedstudyof
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
25
diagnosesandriskfactors.BMCMed.2017;15:12.21. SminkFRE,HoekenD,OldehinkelAJ,HoekHW.PrevalenceandseverityofDSM-5eating
disordersinacommunitycohortofadolescents.IntJEatDisord.2014;47:610–619.22. BenjaminiY,HochbergY.ControllingtheFalseDiscoveryRate:APracticalandPowerful
ApproachtoMultipleTesting.JRStatSocSeriesBStatMethodol.1995;57:289–300.23. BenjaminiY,YekutieliD.Thecontrolofthefalsediscoveryrateinmultipletestingunder
dependency.AnnStat.2001;29:1165–1188.24. CalzoJP,AustinSB,MicaliN.Sexualorientationdisparitiesineatingdisordersymptoms
amongadolescentboysandgirlsintheUK.EurChildAdolescPsychiatry.2018;27:1483–1490.
25. MicaliN,DeStavolaB,PloubidisG,SimonoffE,TreasureJ,FieldAE.Adolescenteatingdisorderbehavioursandcognitions:gender-specificeffectsofchild,maternalandfamilyriskfactors.BrJPsychiatry.2015;207:320–327.
26. AbdulkadirM,HerleM,DeStavolaB,HübelC,SantosFerreiraDL,LoosRJF,etal.Polygenicscoreforbodymassindexisassociatedwithdisorderedeatinginageneralpopulationcohort.{underReview}.
27. BulikCM,SullivanPF,KendlerKS.Geneticandenvironmentalcontributionstoobesityandbingeeating.IntJEatDisord.2003;33:293–298.
28. FinucaneHK,ReshefYA,AnttilaV,SlowikowskiK,GusevA,ByrnesA,etal.Heritabilityenrichmentofspecificallyexpressedgenesidentifiesdisease-relevanttissuesandcelltypes.NatGenet.2018;50:621–629.
29. Munn-ChernoffMA,GrantJD,AgrawalA,KorenR,GlowinskiAL,BucholzKK,etal.Aretherecommonfamilialinfluencesformajordepressivedisorderandanovereating-bingeeatingdimensioninbothEuropeanAmericanandAfricanAmericanfemaletwins?IntJEatDisord.2015;48:375–382.
30. WadeTD,BulikCM,NealeM,KendlerKS.Anorexianervosaandmajordepression:sharedgeneticandenvironmentalriskfactors.AmJPsychiatry.2000;157:469–471.
31. MustelinL,RaevuoriA,KaprioJ,Keski-RahkonenA.Associationbetweeneatingdisordersandmigrainemaybeexplainedbymajordepression.IntJEatDisord.2014;47:884–887.
32. StahlEA,BreenG,ForstnerAJ,McQuillinA,RipkeS,TrubetskoyV,etal.Genome-wideassociationstudyidentifies30lociassociatedwithbipolardisorder.NatGenet.2019;51:793–803.
33. YaoS,Kuja-HalkolaR,MartinJ,LuY,LichtensteinP,NorringC,etal.AssociationsBetweenAttention-Deficit/HyperactivityDisorderandVariousEatingDisorders:ASwedishNationwidePopulationStudyUsingMultipleGeneticallyInformativeApproaches.BiolPsychiatry.2019;86:577–586.
34. CapusanAJ,YaoS,Kuja-HalkolaR,BulikCM,ThorntonLM,BendtsenP,etal.Geneticandenvironmentalaspectsintheassociationbetweenattention-deficithyperactivitydisordersymptomsandbinge-eatingbehaviorinadults:atwinstudy.PsycholMed.2017;47:2866–2878.
35. SonnevilleKR,CalzoJP,HortonNJ,FieldAE,CrosbyRD,SolmiF,etal.Childhoodhyperactivity/inattentionandeatingdisturbancespredictbingeeatinginadolescence.PsycholMed.2015;45:2511–2520.
36. CorteseS,Moreira-MaiaCR,StFleurD,Morcillo-PeñalverC,RohdeLA,FaraoneSV.AssociationBetweenADHDandObesity:ASystematicReviewandMeta-Analysis.AmJPsychiatry.2016;173:34–43.
37. LeventakouV,MicaliN,GeorgiouV,SarriK,KoutraK,KoinakiS,etal.Isthereanassociationbetweeneatingbehaviourandattention-deficit/hyperactivitydisordersymptomsinpreschoolchildren?JChildPsycholPsychiatry.2016;57:676–684.
38. NagataJM,BraudtDB,DomingueBW,Bibbins-DomingoK,GarberAK,GriffithsS,etal.Geneticrisk,bodymassindex,andweightcontrolbehaviors:Unlockingthetriad.IntJEatDisord.2019;52:825–833.
39. FairburnCG,CooperZ,DollHA,WelchSL.Riskfactorsforanorexianervosa:threeintegratedcase-controlcomparisons.ArchGenPsychiatry.1999;56:468–476.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint
26
40. Reichborn-KjennerudT,BulikCM,SullivanPF,TambsK,HarrisJR.Psychiatricandmedicalsymptomsinbingeeatingintheabsenceofcompensatorybehaviors.ObesRes.2004;12:1445–1454.
41. AlgarsM,HuangL,VonHolleAF,PeatCM,ThorntonLM,LichtensteinP,etal.Bingeeatingandmenstrualdysfunction.JPsychosomRes.2014;76:19–22.
42. Striegel-MooreRH,McMahonRP,BiroFM,SchreiberG,CrawfordPB,VoorheesC.ExploringtherelationshipbetweentimingofmenarcheandeatingdisordersymptomsinBlackandWhiteadolescentgirls.IntJEatDisord.2001;30:421–433.
43. CotrufoP,CellaS,CrematoF,LabellaAG.Eatingdisorderattitudeandabnormaleatingbehavioursinasampleof11-13-year-oldschoolchildren:theroleofpubertalbodytransformation.EatWeightDisord.2007;12:154–160.
44. BakerJH,ThorntonLM,BulikCM,KendlerKS,LichtensteinP.Sharedgeneticeffectsbetweenageatmenarcheanddisorderedeating.JAdolescHealth.2012;51:491–496.
45. BellJA,CarslakeD,WadeKH,RichmondRC,LangdonRJ,VincentEE,etal.Influenceofpubertytimingonadiposityandcardiometabolictraits:AMendelianrandomisationstudy.PLoSMed.2018;15:e1002641.
46. GillD,BrewerCF,DelGrecoMF,SivakumaranP,BowdenJ,SheehanNA,etal.Ageatmenarcheandadultbodymassindex:aMendelianrandomizationstudy.IntJObes.2018;42:1574–1581.
47. DemontisD,WaltersRK,MartinJ,MattheisenM,AlsTD,AgerboE,etal.Discoveryofthefirstgenome-widesignificantrisklociforattentiondeficit/hyperactivitydisorder.NatGenet.2019;51:63–75.
. CC-BY-NC-ND 4.0 International licenseIt is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in(which was not certified by peer review)preprint The copyright holder for thisthis version posted March 26, 2020. ; https://doi.org/10.1101/2020.03.24.20042648doi: medRxiv preprint