Neuroimaging in bulimia nervosa and binge eating disorder: a … · 2018. 2. 20. · REVIEW Open...

REVIEW Open Access

Neuroimaging in bulimia nervosa andbinge eating disorder: a systematic reviewBrooke Donnelly1, Stephen Touyz1, Phillipa Hay2* , Amy Burton1, Janice Russell3 and Ian Caterson4

Abstract

Objective: In recent decades there has been growing interest in the use of neuroimaging techniques to explorethe structural and functional brain changes that take place in those with eating disorders. However, to date, themajority of research has focused on patients with anorexia nervosa. This systematic review addresses a gap in theliterature by providing an examination of the published literature on the neurobiology of individuals who bingeeat; specifically, individuals with bulimia nervosa (BN) and binge eating disorder (BED).

Methods: A systematic review was conducted in accordance with PRISMA guidelines using PubMed, PsycInfo,Medline and Web of Science, and additional hand searches through reference lists. 1,003 papers were identified inthe database search. Published studies were included if they were an original research paper written in English;studied humans only; used samples of participants with a diagnosed eating disorder characterised by recurrentbinge eating; included a healthy control sample; and reported group comparisons between clinical groups andhealthy control groups.

Results: Thirty-two papers were included in the systematic review. Significant heterogeneity in the methods usedin the included papers coupled with small sample sizes impeded the interpretation of results. Twenty-one papersutilised functional Magnetic Resonance Imaging (fMRI); seven papers utilized Magnetic Resonance Imaging (MRI)with one of these using both MRI and Positron Emission Technology (PET); three studies used Single-PhotonEmission Computed Tomography (SPECT) and one study used PET only. A small number of consistent findingsemerged in individuals in the acute phase of illness with BN or BED including: volume reduction and increasesacross a range of areas; hypoactivity in the frontostriatal circuits; and aberrant responses in the insula, amygdala,middle frontal gyrus and occipital cortex to a range of different stimuli or tasks; a link between illness severity in BNand neural changes; diminished attentional capacity and early learning; and in SPECT studies, increased rCBF inrelation to disorder-related stimuli.

Conclusions: Studies included in this review are heterogenous, preventing many robust conclusions from beingdrawn. The precise neurobiology of BN and BED remains unclear and ongoing, large-scale investigations are required.One clear finding is that illness severity, exclusively defined as the frequency of binge eating or bulimic episodes, isrelated to greater neural changes. The results of this review indicate additional research is required, particularlyextending findings of reduced cortical volumes and diminished activity in regions associated with self-regulation(frontostriatal circuits) and further exploring responses to disorder-related stimuli in people with BN and BED.

Keywords: binge eating, binge episode, bulimia nervosa, binge eating disorder, eating disorders, neuroimaging,neurobiology, fMRI

* Correspondence: [email protected] Health Research Institute (THRI), School of Medicine, WesternSydney University, Sydney, New South Wales, AustraliaFull list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Donnelly et al. Journal of Eating Disorders (2018) 6:3 DOI 10.1186/s40337-018-0187-1

http://crossmark.crossref.org/dialog/?doi=10.1186/s40337-018-0187-1&domain=pdfhttp://orcid.org/0000-0003-0296-6856mailto:[email protected]://creativecommons.org/licenses/by/4.0/http://creativecommons.org/publicdomain/zero/1.0/

Plain English summaryThis paper is a systematic review of research investigatingstructural and functional differences transdiagnostically;that is, in people who have an eating disorder character-ized by binge eating, either Bulimia Nervosa (BN) or BingeEating Disorder (BED), when compared to healthy peoplewith no eating disorder or other mental illness. Using aset of fixed search terms, we completed a systematic re-view of published peer-reviewed scientific papers, identify-ing thirty-two papers that met the inclusion criteria. Themajority of papers reviewed used functional MagneticResonance Imaging (fMRI) and the rest used one or twoother neuroimaging tests. An overview and synthesis ofthe results of the papers is provided, grouped according tothe type of test completed. A small number of findingsemerged of individuals with BN or BED when they haveclinically significant symptoms, highlighting there are re-ductions in the overall size of the brain in BN and BEDand diminished activity in regions associated with selfregulation (frontostriatal circuits). Also, some studieshighlight differences in the activity within neural regionsassociated with emotional processing (amygdala), atten-tion and spatial manipulation (middle frontal gyrus) andvisual processing (occipital cortex). We discuss the impli-cations of the results and highlight recommendations forfuture neurobiological research based on our findings.

RationaleRecurrent binge eating is a debilitating symptom that isa core diagnostic criterion for bulimia nervosa (BN) andbinge eating disorder (BED); it also occurs in anorexianervosa-binge purge type (AN-BP), and is a commonfeature in other specified feeding and eating disorder(OSFED). The Diagnostic and Statistical Manual ofMental Disorders (DSM-5) [1] specifies that individualsare engaging in objective binge episodes (OBEs) at leastonce per week to reach a diagnosis of BN or BED. Thisdiffers from the proposed ICD-11 criterion where thebinge episode (BE) is not required to be objectively largeand can look like subjective binge episodes (SBEs) [2].BN involves binge episodes (BEs)1 followed by inappro-priate compensatory behaviours to avoid weight gain,such as purging, while BED involves engaging in recur-rent binge episodes with no compensatory strategies [1].BN and BED are disorders with noted social and

health consequences that typically arise in later adoles-cent and young adult years [3]. In a cross-sectionalpopulation survey of Australian adults, the three-monthprevalence of BN and BED ranged from 1.1-1.5% [4]. In2014 in the Australian population, the prevalence of re-current binge eating with or without distress was 10.1%and 13.0% in 2015 [3]. Psychiatric comorbidity, particu-larly with depression and anxiety disorders, is common[5, 7] and mortality is increased [6]. Over one in five

individuals with BN will attempt suicide during their life,with factors relating to emotion dysregulation, lifetimeanxiety and depression [7].Treatments and assessments for BN and BED have

been developed based on the current definition of BEs.Cognitive Behavioural Therapy (CBT) is the first-linetreatment for BN and BED [8]. Available psychologicaltreatments are moderately effective and medication mayoffer benefits but longer-term maintenance of effects areunclear [9–12]. Research has demonstrated that the ma-jority of individuals with BN and BED do not seek treat-ment for their eating disorder, but instead present forweight loss treatment [13, 14].In recent decades, major advances have taken place in

the field of neuroscience, which has increased knowledgeof the interrelationship between neurological processesand eating disorders [15]. However, most neuroimagingstudies have focused their attention on people with ANrather than BN or BED. Presumably, this is because theearly neurobiological literature, which focused largely onexamining structural changes associated with prolongedstarvation and malnutrition in AN with Magnetic Reson-ance Imaging (MRI), formed a foundation for ongoinginvestigation in this clinical group.The neurobiological basis of BN and BED is different

to that of AN. Specifically, BN and BED are conceptual-ized as impulsive / compulsive eating disorders with al-tered reward sensitivity and food-related attentionalbiases [16]. Alterations in the cortico-striatal circuits ofindividuals with BN and BED are similar to those re-ported in studies of people with substance abuse, withchanges in the function of the prefrontal, insular cortex,orbitofrontal cortex (OFC) and striatum [16]. In BED,individuals move from the ventral-striatal reward-basedmode of reward-related food consumption, to a dorsal-striatal impulsive / compulsive mode of reward-relatedfood consumption [16]. In BN the urge to binge eat ismediated by hyperactivity of the OFC and anterior cin-gulate cortex (ACC) and impaired inhibitory controlfrom the lateral prefrontal circuits [17]. Hyperactivity ofthe parieto-occipital regions and hypoactivation of ex-ecutive control networks in individuals with BN com-pared to HCs has also been reported [18].Increased attention has been directed to the role of

inhibitory control, or how well one can suppress in-appropriate and unwanted actions, in BN and BED e.g.[18–20]. A recent meta-analysis found impaired re-sponse inhibition in BN patients when faced with eatingdisorder-related stimuli, alongside general impairmentsin inhibitory control, when compared to HCs [19]. In arecent systematic review, no definite conclusions couldbe drawn regarding the neurocognitive profile of individ-uals with BN or BED due to the diversity in methodologyand small sample sizes within the majority of the studies

Donnelly et al. Journal of Eating Disorders (2018) 6:3 Page 2 of 24

reviewed [21]. A smaller body of research has been pub-lished over recent years examining the neurobiology of pa-tients who binge eat e.g. [22–24]. The frontostriatal area,which has a central role in controlling goal-directedthoughts and behaviours including response inhibitionand reward processing [25], has emerged as particularlyrelevant to BN [25]. Evidence suggests the diminishedfrontostriatal brain activation in BN patients contributesto the severity of symptoms [26]. Furthermore, individualswith BN display altered temporal choice behavior, thedegree of preference for immediate rewards over delayedrewards [27].Overall, it is clear that there is a rapidly expanding

body of neurobiological research in BN and BED. Com-pleting a rigorous review will provide a novel and war-ranted overview of the deficits and differences thatoccur in these clinical groups. This will increase our un-derstanding of the neurological underpinnings of BNand BED, which is critical considering the extremelyhigh rates of psychiatric comorbidities and risk of sui-cide in people with BN and BED [7, 28]. The patho-physiology of BN and BED is poorly understood and asa result, effective evidence-based treatments require fur-ther refinement [27]. Increased knowledge of these fac-tors for BN and BED will better inform treatmentsacross bulimic eating disorders.

ObjectiveThere has been a recent shift in the treatment frameworkfor eating disorders, to include knowledge of the structuraland functional changes in the brain that take place in theill state. Recent developments in neuroimaging haveallowed some clinical and psychopathological symptomsto be linked to specific neural structures and systems. Todate, the majority of this research in people with eatingdisorders has examined individuals with AN. The first aimof this systematic review is to contribute a comprehensiveunderstanding of the existing neuroimaging research inindividuals with BN or BED where BEs form the core eat-ing disorder pathology, rather than a possible clinical fea-ture. The second aim relates to the clinical utility of theDSM [1] distinction between OBEs and SBEs; specifically,whether this review identifies any neuroimaging studiesthat assist in elucidating this matter.

MethodsSearch strategyThis review was conducted according to the PreferredReporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic reviews [29].The final database searches were conducted on 14 Febru-ary 2017 and all relevant articles published up until thistime were considered against the eligibility criteria out-lined below, with no restriction on publication date to

maximise results. Key search terms included: eating dis-order, bulimi* (bulimia, bulimic), bing* (binge, bingeing),binge eating, MRI, fMRI, SPECT, PET, CT, neuro* (neuro-biology, neuronal, neurotransmitters, neuroimaging,neuroscience). The precise search strategy developed fordatabase searches is available on request from the authors.The systematic review was conducted in four stages:

1. Developing inclusion and exclusion criteria for thedatabase searches;

2. Searching selected electronic databases to identifypapers meeting inclusion criteria for the review;

3. Study selection; and,4. Appraisal and write-up of the studies that met inclu-

sion criteria.

Selection criteriaPotential peer-reviewed, published studies were identi-fied using four electronic databases: PubMed, PsycInfo,Medline and Web of Science. Additional manualsearches were conducted through reference lists. Studieswere included if they (a) were an original research paperwritten in English; (b) studied humans only; (c) usedsamples of participants with clinician-diagnosed BN orBED; (d) recruited a HC sample with no eating disorderpathology and for studies including BED participants, ahealthy, non-overweight control group was recruited;and (e) reported group comparisons between clinicaland HC groups. Studies where participants were fromadolescent or adult samples were included to broadenthe number of possible studies in the review.

Inter-rater reliabilityDue to the high number of studies screened by abstract(n=186), a random subset was screened by two co-authors(ST, AB) to establish inter-rater reliability of the selectioncriteria. Excellent inter-rater reliability was established(ST: k = 0.769, p = 0.001), (AB: k = 0.880, p = .000).

Data extraction and quality assessmentAs there is no standardised criterion for the quality as-sessment of neurobiological studies, a modified versionof the Downs and Black [30] Quality Index, adapted byFerro and Speechley [31], was used. The Ferro andSpeechley [31] index consists of 15 items of the original27-item scale and is scored dichotomously as 0 (no/un-able to answer) or 1 (yes). The index has four subscales:reporting, external validity, internal validity, and power.However within the Ferro and Speechley [31] index,some items were not applicable to the quality assess-ment of neurobiological studies. For this reason, fouritems were removed (response rate; estimates of the ran-dom variability; were staff / places / facilities where pa-tients were studied representative of the treatment the


majority of patients receive; were outcome measuresvalid and reliable) to prevent the neurobiological studiesreceiving an artificially low rating. Lastly, five additionalitems relevant to neurobiological studies of eating dis-order patients (controlled for age; controlled for BMI;controlled for handedness; satiety pre-testing; and men-strual status) were added by the authors of this system-atic review after consulting with experts in the field i.e.Psychiatrists specialising in the treatment of eating dis-orders; Staff Specialist Endocrinologists specialising ininpatient medical treatment for people with severe eat-ing disorders. This created a new modified quality as-sessment scale for this systematic review consisting of16 items. Quality ratings of the articles that were in-cluded in this systematic review ranged between sevenand 11 out of a maximum score of 16. The majority(n=18) of the papers scored 9 or more out of 16.The primary author conducted this modified quality

assessment on all studies that met inclusion criteria forthis review. To control for bias, a second author (AB)assessed 20% of the included studies using the samequality assessment. Agreement was reached over themethodology of the quality assessment and disparate rat-ings were discussed in two meetings. Excellent inter-rater reliability was established; Cohen’s kappa scoresacross the sub-set of studies ranging from (k = 0.871,p < 0.001) to perfect agreement (k = 1.0, p < .001).Following the protocol of McClelland and colleagues

[32] of a systematic review examining neuroimaging andbehavioural studies, data were extracted from the in-cluded studies and summarised in a table that was thenchecked by two other reviewers. The data related to par-ticipants (age, female/male ratio); eating disorder diag-noses; sample size; procedure, including any taskcompleted during neurobiological tests; exclusions; andbrief findings. Due to the breadth of methodologies ofthe studies that met selection criteria, a narrative synthe-sis was then conducted, grouped according to the typeof neuroimaging test and/or method used.

ResultsSearch results and selection of studiesAn initial 1,003 studies were identified using the searchstrategy described above and 811 of these were screenedby title after duplicates were removed. A total of 192 ab-stracts were screened and 149 studies were excluded atthis point. This left 43 studies screened by full-text, with11 studies excluded at this juncture. The final number ofstudies included in this review was 32. The main reasonsfor studies not meeting the criteria were: not including aBN or BED sample; not including a HC sample or onlyincluding an overweight control sample for BED studies;using a recovered clinical population where participantshad been symptom-free for over 12 months; or, using

neuropsychological measures (e.g. Stroop test) [33] butnot neurobiological measures (e.g. fMRI). The diversityin the methodologies within the studies included in thissystematic review precluded the completion of a meta-analysis. Figure 1 presents the PRISMA flowchart de-scribing the inclusion and exclusion process.

Study characteristicsA total of n= 32 studies met the inclusion criteria forthis review. Sixteen studies on participants with BN,eleven studies on participants with AN and BN, threestudies on participants with BN and BED and two stud-ies on participants with BED were selected for review.The majority of included studies (n=21) used fMRI.Seven used MRI, with one of these studies using bothMRI and Positron Emission Technology (PET). Of theremaining four studies, three utilized Single-PhotonEmission Computed Tomography (SPECT) and oneused PET. All but two of the included studies recruitedfemale eating disorder patients only; one study recruitedboth men and women [23] and one study did not reportthis [34]. Sample sizes were relatively small; see belowfor details of sample sizes grouped according to the typeof neurological test performed.

Structural differences: Summary of MRI studiesMRI produces three-dimensional, anatomical images butcannot measure metabolic rates, unlike PET or SPECT.Seven studies using MRI to investigate structural differ-ences between BN, BED and HCs met inclusion criteria.Sample sizes of MRI studies were consistently small; BNmedian and range: 10 (8-21); HC: 11 (7-21); for BEDthere was only one study, (n=17). See Table 1 for the ex-tracted data from these studies. Earlier studies tended toexamine structural differences orvolumetric differences between certain cerebral struc-

tures compared to the overall brain using a ratio calcula-tion. Hoffman et al. [35] published one of the firstneurobiological studies of BN patients; a resting-stateMRI with BN and HC participants to assess cerebral at-rophy. It was found that the BN group had a signifi-cantly lower sagittal cerebral:cranial ratio however therewas no difference between the two groups on the ventri-cle:brain ratio [35]. In a similar study, Husain et al. [36]reported the ratio of thalamus:cerebral hemisphere andmidbrain:cerebral hemisphere was significantly smallerin AN participants but not in BN or HC participants.Doraiswamy et al. [37] used MRI to investigate pituitarydifferences in AN and BN participants compared withHC subjects; results demonstrated the overall area andheight of the pituitary was significantly smaller in eatingdisorder participants compared to the control group.Several groups reported grey matter reduction. Hoffmanet al. [38] reported a significant decrease in the inferior


frontal grey matter of BN participants compared to HCs.Coutinho and colleagues [22] reported a similar finding ofreduced CN volume in BN participants, after conductinga resting-state MRI with BN and HC participants.In the only study to use MRI and PET together, Galusca

and colleagues [39] investigated cerebral serotonergic ac-tivity in severe BN-Purging (BN-P) (where severe BN-Pwas defined as at least one binge-purge episode every dayfor at least six months) and HC subjects two hours afterlunch. Widespread abnormal cerebral serotonergic activitycharacterized the BN group, however inter-individual het-erogeneity was found when individual comparisons wereconducted between each BN patient and the HC group.This ranged from isolated to widespread increases in thebinding potential of [18F]MPPF (a serotonin specificradiogland used in PET). Binding potential is the central

measure of PET [39]. Finally, Schafer and colleagues [24]conducted MRI with BN, BED and HC participants to in-vestigate grey matter volume (GMV) abnormalities. Theauthors used voxel-based morphometry to analyse specificbrain regions known to be involved in food andreinforcement processing (medial and lateral orbito-frontal cortex [OFC], insula, anterior cingulate cortex andventral / dorsal striatum). The BN and BED participantshad greater GMV in the medial OFC compared to HCparticipants. The BN group also had significantly in-creased ventral striatal volumes, part of the neural rewardsystem, when compared to the BED and HC groups [24].

Functional differences: Summary of fMRI studiesfMRI provides good coverage and excellent spatial reso-lution, relying on the interrelationship between cerebral

Records after duplicates removed N = 811

Records screened by titleN = 811

Records excluded by title

N = 619

Records identified through database search

N = 1001(PsycInfo n = 112; Medline n = 84;

Embase n = 755; Web of Science n = 50)

Records screened by abstract N = 192

Records excluded by abstract

N = 149

Full-text records screened for eligibilityN = 43

Full-text records excluded based

on criteriaN = 11

Studies to be included in synthesis

N = 32

Additional records identified through

other sourcesN = 8

IDE

NT

IFIC

AT

ION

SC

RE

EN

ING

ELI

GIB

ILIT

YIN

CLU

SIO

N

Fig. 1 PRISMA flow chart of study identification and inclusion


Table

1Characteristicsandkeyfinding

sof

includ

edstud

iesusingMRI

astheprim

arymetho

d

Autho

rs&Journal

Participants

Meanage(SD)

% Female

Proced

ure

Psychiatric

/othe

rexclusions

Find

ings

1.Cou

tinho

etal.(2015)[22]

Internationa

lJournalof

Eating

Diso

rders,48(2):206-214.

BN(n=21)HCs(n=20)

BN:31.57

(8.27)

HCs:30.9(8.79)

100%

One

resting-state

MRI

Substanceabusedisorder;

suicidalideatio

n;AxisI

disorder

otherthan

eatin

gdisorder;p

sychotropic

med

icationwith

the

exceptionof

anxiolyticsand

antid

epressants

Volumeredu

ctionin

theCNwith

inthefro

ntostriatal

circuitin

BNcomparedto

HCs

2.Doraisw

amyet

al.(1990)[37]

BiologicalPsychiatry,28:110-116.

BN(n

=10)

AN(n

=8)

HCs(n

=13)

BN:24(2.5)

AN:22.8(4.4)

HCs:27.5(5.1)

100%

One

resting-state

MRI

Major

affectivedisorder

AN&BN

vsHCs:sm

aller

pituitary

glandarea

andhe

ights

Atren

dapproaching

statisticalsign

ificance

was

foun

d:thearea

ofthepituitary

was

negatively

correlated

with

duratio

nof

illne

ss

3.Galusca

etal.(2014)[39]

TheWorldJourna

lofB

iological

Psychiatry,15:599-608.

BN-P*(n=9)

HCs(n=11)

*onlysevereBN

-Pparticipants

selected

;criterionbe

ing

atleaston

ebing

e-pu

rge

episod

e/dayforat

leastsixmon

ths

BN-P:O

nlytheage

rang

e(18-30y)was

repo

rted

.Nomeanor

SD.

HCs:Nodata,how

ever

repo

rted

tobe

age-matched

.

100%

MRI

andPET

completed

2hfollowinglunch

+ PETto

specifically

exam

ineserotone

rgic

activity

/bind

ingpo

tential

of[18 F]M

PPF(a

serotonin

specificradiog

land

used

inPETcapableof

assessingchange

inbrainserotonin)

Chron

icor

cong

enital

disease;alcoho

l,tobaccoor

drug

consum

ption;previous

orcurren

tdiagno

sisof

AN-R;m

edication

IntheBN

grou

p:oralcontraceptivepill

BNvs

HCs:increased

bind

ingpo

tentialin

four

clustersin

the

brain:

Insulaand

transverse

tempo

ral

cortex,ope

rculum

,tempo

ro-parietalcortex

Abn

ormalities

inim

pairedactivation,glucose

metabolism

orligandbind

ing

inareasinclud

inginsulaand

tempo

ralp

arietalcortex,

hipp

ocam

palreg

ion,

inter-he

misph

ericcortex,

PFCanddo

rsalraph

enu

cleus

4.Hoffm

anet

al.(1989)[35]


BN(n=8)

HCs(n=8)

BN:24.1

HCs:26.8

NoSD

repo

rted

100%

One

resting-stateMRI

Pastdiagno

sisof

AN;

curren

tdiagno

sisof

major

affectivedisorder;

alcoho

labu

se

BNvs

HCs:corticalatroph

yfoun

din

thesagittal

cerebral/cranio

ratio

(SCCR)

butno

tin

the

ventricle:brain

ratio

(VBR)

Sign

ificant

positive

correlationbe

tween

bing

efre

quen

cyandVBR

5.Hoffm

anet

al.(1990)[38]


BN(n=8)

HC(n=7)

BN24.3(3.2)

HC24.3(3.4)

100%

One

resting-stateMRI

Current

diagno

sisof

major

affectivedisorder;

alcoho

labu

seIn

theBN

&HCgrou

p:lifetim

ediagno

sisof

AN;

med

ication

BNvs

HC:Significant

decrease

ininferio

rfro

ntalgrey

matter


Table


sof

includ

edstud

iesusingMRI

astheprim

arymetho

d(Con

tinued)

Autho

rs&Journal

Participants

Meanage(SD)

% Female

Proced

ure

Psychiatric

/othe

rexclusions

Find

ings

6.Husainet

al.(1992)[36]


BN(n=12)

AN(n=12)

HCs(n=11)

BN24.5(4)

AN:25.3(7)

27.8(6)

100%

One

resting-stateMRI

IntheBN

grou

p:past

diagno

sisof

AN

ANvs.BN&HCs:

Sign

ificantlysm

aller

thalam

usandmidbrain

(mesen

ceph

alon

)area

Theratio

ofthalam

usto

cerebralhe

misph

ere

andmidbrainto

cerebral

hemisph

erewas

sign

ificantlysm

aller

inBN

&ANvs.H

Cs

however

post-hoc

testsshow

edthisresult

was

onlyrelated

toANparticipants

7.Schäferet

al.(2010)[24]

NeuroImage,50:639-643.

BN-P

(n=14)

BED(n=17)HCs(n=19)

BN-P:23.1(3.8)

BED:26.4(6.4)

HCs:22.3(2.6)

100%

One

resting-stateMRI

toexam

inestructuralbrain

abno

rmalities.G

reymatter

volumes

(GMV)

for

specificregion

sinvolved

infood

/reinforcem

ent

processing

were

analysed

viavoxel-

basedmorph

ometry:

med

ial/

lateralO

FC,

insula,A

CC,

ventral

/do

rsalstriatum

Dep

ression;

left-hande

dness;

med

ication

BNvs.BED

:greater

GMVof

med

ialand

lateralo

rbito

frontal

cortex

aswellas

ventral&

dorsalstriatum

BNvs

HCs:increased

GMVof

med

ialO

FC&ventralstriatum

BEDvs

HCs:greater

GMVof

ACC&med

ialO

FCBN

&BEDvs.H

Cs:

greatervolumes

ofthemed

ialO

FCBN

vsBED&HCs:

increasedventral

striatum

volumes;

BMIw

asne

gatively

correlated

with

striatal

grey

mattervolume

whilepu

rgingwas

positivelycorrelated

with

ventralstriatum

volume


blood flow, energy demand and neural activity [40]. Theprimary fMRI neuroimaging signal most researchersexamine is the blood oxygen level-dependent (BOLD)contrast signal, which provides a reliable measure of alocal increase in neural activity [40]. Twenty-one studiesusing fMRI as the primary neurological test met inclu-sion criteria for this review [18, 23, 25, 26, 41–57].Eleven studied participants with BN [18, 25, 26, 43, 45–47, 50, 51, 53, 58], seven examined participants with BNand AN [42, 48, 49, 54–57, 59], two examined partici-pants with BN and BED [44, 52] and one study exam-ined participants with BED [23].Overall, fMRI studies had marginally higher sample

sizes than for the MRI studies reported above; BN me-dian and range: 13 (8-32); HC: 19 (12-34); BED: onlythree studies included BED participants hence the me-dian could not be calculated, range (12-19). See Table 2for data extracted from these studies. As can be seen inTable 2, there is obvious heterogeneity across the designof the included fMRI studies. For this reason, fMRI re-sults have been grouped according to the type of stimuliused during testing (decision making and learning para-digms; food-related stimuli; body image-related stimuli).

Non-food decision making and learning paradigmsSix papers in this review utilized decision making orlearning paradigms during fMRI. Balodis et al. [23]found obese BED patients had decreased VS activity dur-ing anticipation processing while completing a monetaryincentive delay task, however during outcome processingthe group demonstrated diminished prefrontal cortex(PFC) and insula activity. Using a probabilistic learningparadigm (Weather Prediction Task), sub-threshold BNpatients demonstrated hyperactivity and overall process-ing inefficiencies in the frontostriatal system [25]. Asimilar finding was reported by Cyr et al [43], who useda reward-based virtual learning task, with abnormalfunctioning of the anterior hippocampus and frontostria-tal circuits reported in BN compared to HC participants.In the same study, BN symptom severity was signifi-cantly associated with the activation of the right anteriorhippocampus during reward processing. Two studies[45, 46] used the Simon Spatial Incompatibility Task toexamine BOLD during fMRI. In contrast to Celone et al.[25], Marsh et al. [46] found hypoactivity in the frontos-triatal circuits known to contribute to self-regulatorycontrol, including the right inferior frontal gyrus, dorso-lateral PFC and putamen, in BN compared to controls.Finally, Seitz et al. [18] reported a range of differences inthe neural networks associated with alerting, reorientingand executive attention between BN and HC partici-pants, who underwent fMRI while completing a modi-fied Attention Network Task (ANT), as well as a rangeof correlations between neural activity, BN symptoms

and scores on a measure of impulsivity associated withattention deficit hyperactivity disorder (ADHD).

Food-related stimuliSix of the fMRI studies in this review used food-specificstimuli; most of these used photographs of food items[26, 42, 52, 54, 58], however one study used actual foodstimulus of 0.5mL portions of chocolate milkshake [58],and one study asked participants to think about eatingthe food shown to them in photographs [42]. In the firststudy to examine neural reward circuitry in BN in re-sponse to actual and anticipated food intake [58], de-creased but not significantly different activation wasfound in the right precentral gyrus during both antici-pating and consuming chocolate milkshake, and de-creased activation in the left middle frontal gyrus, rightposterior insula and left thalamus, compared to the HCgroup. The precentral gyrus is the site of the primarymotor cortex while the insula has a central role in gusta-tory and hedonic taste processing and alongside themiddle frontal gyrus, it is activated in response to pleas-urable taste [58, 60, 61]. Brooks et al [42] compared BN,AN and HC participants’ neural responses to images ofhigh- and low-energy foods versus non-food items. Inresponse to food images, the BN group demonstrated in-creased activation in the visual cortex, right dorsolateralprefrontal cortex, right insular cortex and precentralgyrus. Relative to the HC group, the BN group alsoshowed deactivation in the bilateral superior temporalgyrus/insula and visual cortex, and compared to the ANgroup, had deactivation in the parietal lobe and dorsalposterior cingulate cortex and increased activation in thecaudate, superior temporal gyrus, right insula and motorarea [42].The remaining studies utilizing food images reported a

range of both decreased and increased neural activationin the binge eating groups compared to AN and HC par-ticipants. Schienle and colleagues [52] reported in-creased insula activation in women with BN in theirstudy using images of high-calorie foods, disgust-inducing and neutral images during fMRI following anovernight fast, compared to BED and HC groups. Foodimages were experienced positively across all groupswith increased activation of the orbitofrontal cortex, an-terior cingulate cortex and insula and there were nogroup differences in neural response to viewing disgust-inducing images. Women with BED demonstratedgreater medial orbitofrontal response while viewing foodimages compared to BN and HC groups, while the BNgroup demonstrated greater anterior cingulate cortexand insula activation [52]. In a study examining neuralresponse to food images used as interference during theStroop task [33] (a neuropsychological test of self-regulation utilising the frontostriatal region), women


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

1.Amiantoet

al.

(2013)

[57]

Cerebellum,12:

623-631.

AN(n=12)

BN(n=12)

HC(n=10)

AN:20(4)

BN:

23(5)H

C:24(3)

100%

One

resting-statefM

RI.

Lifetim

ehistoryof

psycho

sis,

schizoph

renia,schizoaffectivedisorder,

delusion

aldisorder,b

ipolar

I/IId

isorde

r,psycho

ticde

pression

,organicmoo

ddisorder;severemed

icalillne

ss;severe

unde

rweigh

tthat

couldno

tbe

managed

asan

outpatient;use

ofpsycho

trop

icmed

ication;ne

urolog

ical

disease

ANvs

BN&HC:g

reymatterredu

ction

AN&BN

vsHC:hyperconn

ectivity

ofthecerebe

llarne

tworkto

theparietal

cortex;increased

bilateralcon

nectivity

ofcerebe

llarICNwith

tempo

ralp

oles

BNvs

AN&HCs:GMVredu

ctionin

the

CN

2.Balodiset

al.

(2013)

[23]

Biological

Psychiatry,

73:877-886.

Obe

seBED(n=19)

Obe

seno

n-BED(n=19)

HCs(n=19)

BED43.7(12.7)

OB38.3(7.5)

HC34.8(10.7)

BED

73.7%

OB

52.6%

HC

52.6%

fMRI

completed

whilecompleting

MIDT(m

onetaryincentivede

laytask)

Intheob

eseno

n-BEDor

HCgrou

p:pasthistoryof,o

rcurren

tbing

eeatin

gor

othe

reatin

gdisorder

diagno

sis

Anticipationprocessing

:BEDvs

OB:de

creasedventrostriatal(VS)

andstriatalactivity;

OBvs

HCs:increasedVS

activity

Outcomeprocessing

:BEDvs

OB&HCs:diminishe

dactivity

inPFCandInsular

3.Bo

hon&Stice

(2011)

[58]

Internationa

lJourna

lofEating

Diso

rders,44(7):

585-595.

BNsub-threshold*

(1xBE&

comp/wk)(n=11)

BN(n=2)

HCs(n=13)

*(Sub

-BN=1xbing

eep

isod

e/week–sub-

thresholdforDSM

-IVcri-

teria,how

ever

thisfre

-qu

ency

meetsthene

wDSM

-Vcriteria

forBN

)

Not

repo

rted

per

grou

p.Forall

participants:20.3

(1.87)

100%

fMRI

exam

iningrewardcircuitrydu

ring

actual(cho

cmilkshake)

andanticipated

(tasteless

solutio

n)food

intake

Any

AxisId

isorde

r;food

allergyto

milkshake/tasteaversion

tochocolate

milkshake

IntheBN

grou

p:psycho

active

med

ications

othe

rthan

SSRIs(sertraline

&fluoxetine)

BNvs

HCs:less

activationin

right

precen

tralgyrusin

both

anticipatory

andconsum

atorycond

ition

s;less

activationin

right

anterio

rinsulawhile

anticipatingthemilkshake;andless

activationin

theleftmiddlefro

ntal

gyrus,rig

htpo

steriorinsula,left

thalam

usin

respon

seto

milkshake

4.Broo

kset

al.

(2011)

[42]

PLoS

One,

6(7):e22259.

BN(n=8)

AN-R

(n=11)

AN-BP(n=7)

HCs(n=24)

BN:25(7.1)

AN:26(6.8)

HC:26(9.5)

100%

fMRI

whileasking

participantsto

imagineeatin

gthefood

sshow

nin

photog

raph

s(72colour

photos

ofhigh

andlow

energy,sweet&savoury

food

s;72

photos

ofno

n-food

items

Lefthand

edne

ss;caffeine/alcoho

lwith

inspecified

times

preced

ingthe

fMRI;history

ofhe

adtrauma,he

aringor

visualim

pairm

ent,ne

urolog

icaldisease

IntheED

grou

ps:p

sychotropic

med

ications

othe

rthan

SSRIs

Inrespon

seto

food

vs.non

-food

images:

BNvs

HCsandAN:g

reater

activationin

visualcortex,right

dorsolateral

prefrontalcortex,right

insularcortex

andprecen

tralgyrus

BNvs.H

C:d

eactivationin

bilateral

supe

riortempo

ralg

yrus,insular

cortex,

visualcortex

BNvs

AN:red

uced

activationin

parietal

lobe

,dorsalp

osterio

rcing

ulatecortex

BNvs

AN-R:increased

activationin

bi-

lateralinferiortempo

rallob

e,leftvisual

cortex,p

osterio

rcing

ulate&leftinferio

rparietallob

eandde

activationin

right

precen

tralgyrus

BNvs

AN-BP:greateractivationin

the

leftcerebe

llum,leftparahipp

ocam

pal

gyrus,leftpo

steriorcing

ulatecortex,

right

supp

lemen

tary

motor

area,and

deactivationin

leftinferio

rtempo

ral

gyrus


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.(Con

tinued)

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

5.Celon

eet

al.

(2011)

[25]

NeuroImage,56:

1749-1757.

‘Sub

-thresho

ld’*BN

(n=18)

HCs(n=19)

*(Sub

-BN=1xbing

eep

isod

e/week–sub-

thresholdforDSM

-IVcri-

teria,how

ever

thisfre

-qu

ency

meetsthene

wDSM

-Vcriteria

forBN

)

Sub-BN

:20.67

(2.10)

HC:20.42

(1.95)

100%

fMRI

durin

gWeather

Pred

ictio

nTask

(WPT),aprob

abilisticlearning

paradigm

.

Previous

orcurren

tne

urolog

icalor

med

icaldisease;learning

disability;

substanceabuse;historyof

sign

ificantly

low

body

weigh

t(<85%

ofidealb

ody

weigh

t);p

astor

curren

tAN

Nobe

haviou

rald

ifferen

cesin

perfo

rmance

Results

demon

strate

processing

inefficienciesin

thefro

nto-striatalsys-

tem

inBN

.BNwom

ende

mon

strated

increasedoverallcateg

orylearning

-relatedactivity

intherig

htcaud

atenu

-cleusandbilaterald

orsolateralP

FCand

decreasedsupp

ressionof

thecatego

rylearning

relatedBO

LDsign

alin

thean-

terio

rcing

ulatecortex.The

directionof

theBO

LDsign

alchange

swith

inthe

fronto-striatalsystem

differsfro

mthe

initialhypo

thesis

6.Cyr

etal.

(2016)

[43]

Journa

lofthe

American

Academ

yof

Child

andAd

olescent

Psychiatry,55(11):

963-972.

BN(n=27)

HCs(n=27)

BN:16.6(1.5)

HC:16.3(2.1)

100%

fMRI

BOLD

respon

sedu

ringreward

basedspatiallearningtask

(virtual

learning

)

History

ofne

urolog

icalillne

ss;p

ast

seizures;headtraumawith

loss

ofconsciou

sness(LOC);men

tal

retardation;pe

rvasivede

velopm

ental

disorder;current

AxisId

isorde

r(other

than

depressive

/anxietydisorder

for

clinicalgrou

p)

BNvs

HCs:en

gage

dtherig

htanterio

rhipp

ocam

puswhe

nreceiving

unexpected

rewards

HCsvs

BN:eng

aged

therig

htIFCwhe

nsearchingspatially

andtherig

htanterio

rhipp

ocam

puswhe

nreceiving

expected

rewards

Overallthedata

sugg

estabno

rmal

functio

ning

oftheanterio

rhipp

ocam

pusandfro

nto-striatalcircuits

durin

greward-basedspatiallearning

Clinicalcorrelates:Severity

ofBN

was

sign

ificantlyassociated

with

activation

oftherig

htanterio

rhipp

ocam

pus

durin

grewardprocessing

7.Leeet

al.

(2017)

[44]

Neuroscience

Letters,accepted

man

uscript26/

04/17

BN(n=13)

BED(n=12)

HC(n=14)

BN:23.7(2.2)

BED:23.6(2.6)

HC:23.3(2.2)

100%

fMRI

perfo

rmed

whileparticipants

completed

theStroop

match-to-

sampletask,inwhich

participantatten-

tioniscontrolledby

aninteractionbe

-tw

eenbo

ttom

-upsensoryprocessing

andtop-do

wncogn

itive

processing

driven

mainlyby

theprefrontalcortex.

Thetask

was

mod

ified

toinclud

efood

andno

n-food

cond

ition

s.

BMI<

17.5;current

orpastpsychiatric

disorder;traum

aticbraininjury;

neurolog

icalillne

ss;current

orpastuse

ofpsychiatric

med

ications

BNvs

HC:low

eraccuracy

indicatin

gim

pairedcogn

itive

controlo

ver

interfe

rence.Highe

ractivationin

the

prem

otor

cortex

anddo

rsalstriatum

inrespon

seto

food

images

BEDvs

HC:highe

ractivationin

the

ventralstriatum

inrespon

seto

food

images

8.Marsh

etal.

(2009)

[45]

Archives

ofGeneral

Psychiatry,66(1):

51-63.

BN(n=20)

HCs(n=20)

BN:25.7(7.0)

HC:26.35(5.7)

100%

fMRI

used

toexam

ineBO

LDdu

ring

perfo

rmance

onaSimon

spatial

incompatib

ility

task

(SSIT).Twogrou

pscomparedon

patterns

ofbrain

activation.

History

ofne

urolog

icalillne

ss;p

ast


LOC;m

ental

retardation,pe

rvasivede

velopm

ental

delay

IntheBN

grou

p:curren

tAxisId

isorde

rexclud

ingmajor

depression

BNvs

HC:respo

nded

sign

ificantlymore

impu

lsivelyandmadeagreater

numbe

rof

errorson

theSSIT

BNgrou

p:Thenu

mbe

rof

objective

bing

eep

isod

escorrelated

inversely

with

thesign

ificantlyincreased

activationof

therig

htmed

ial

prefrontal,tem

poraland

inferio

rparietalcortices


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.(Con

tinued)

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

HCvs

BN:g

reater

activationin

the

anterio

rcing

ulatecortex

durin

gincorrectrespon

sesandactivated

the

striatum

morewhe

nrespon

ding

incorrectly

9.Marsh

etal.

(2011)

[46]

American

Journa

lof

Psychiatry,

168(11):1210-

1220.

BN(n=18)

HCs(n=18)

BN:18.4(2.1)

HC:17.3(2.4)

100%

fMRI

used

toexam

ineBO

LDdu

ring

perfo

rmance

onaSimon

spatial

incompatib

ility

task.Twogrou

pscomparedon

patterns

ofbrain

activation.

History

ofne

urolog

icalillne

ss;p

ast


LOC;m

ental

retardation,pe

rvasivede

velopm

ental

delay

IntheBN

grou

p:curren

tAxisId

isorde

rexclud

ingmajor

depression

BNandHCspe

rform

edcomparably

however

durin

gcorrectrespon

sesin

conflicttrialsthefro

ntostriatalcircuits

failedto

activateto

thesamede

gree

intheBN

grou

pBN

vsHC:d

emon

stratedabno

rmal

patterns

ofactivationin

the

frontostriataland

‘defaultmod

e’system

s;specifically

they

didno

thave

thesamemagnitude

ofactivity

inthe

frontostriatalcircuitsknow

nto

unde

rlie

self-regu

latory

control,includ

ingthe

right

inferio

rfro

ntalgyrus,do

rsolateral

PFC,and

putamen

10.M

arsh

etal.

(2015)

[47]

Biological

Psychiatry,77:

616-623.

BNadolescent

<19yo

(n=16)

BNadult(n=16)

HCs(n=34)

Not

repo

rted

for

either

grou

p;on

lythat

therewere

adolescentsand

adultsin

both

BN&HCs.

100%

fMRI

completed

tocompare

morph

olog

icalcharacteristicsof

their

cerebralsurface

History

ofne

urolog

icalillne

ss;p

ast


LOC;m

ental

retardation,pe

rvasivede

velopm

ental

delay

IntheBN

grou

p:curren

tAxisId

isorde

rexclud

ingmajor

depression

BNvs

HCs:Sign

ificant

redu

ctionof

localvolum

eson

brainsurface

foun

din

thefro

ntalandtempe

roparietalareas

(bilateralm

iddlefro

ntalandprecen

tral

gyri;rig

htpo

stcentralg

yrus

andlateral

supe

rior,andlateralsup

eriorand

inferio

rfro

ntalgyriof

theleft

hemisph

ere).Red

uctio

nswerealso

foun

din

tempe

roparietalreg

ions

includ

ingbilateralinferiortempo

ral

gyri,rig

htsupe

riorparietalg

yrus

and

cune

us,b

ilateralp

osterio

rcing

ulate

cortices,leftprecun

eusandfusiform

gyrus.

Enlargem

entsde

tected

inthebilateral

middle/inferio

roccipitaland

lingu

algyriandrig

htinferio

rparietallob

ulein

theBN

grou

p.Sign

ificant

inverseassociations

repo

rted

betw

eencerebralsurface

morph

olog

yandob

jectivebing

eandvomiting

episod

esin

thebilateralIFG

,PreCGand

PoCG.

11.M

iyake,

Okamoto,

Onada,Kurosaki

etal.,(2010)

[59]

Neuroimaging,

181:183-192.

BN(n=11)

AN-R

(n=11)

AN-BP(n=11)

HCs(n=11)

BN:24.5(5.8)

AN-R:22.2(4.1)

AN-BP:28.3(4.5)

HC:26.5(5.5)

100%

fMRI

with

emotionald

ecisiontask

with

distortedbo

dyim

ages

(varying

degreesof

‘thinne

ssandfatness’of

ownandhe

althyfemalebo

dyph

oto)

Presen

ceof

AxisIo

rIIdisorder

othe

rthan

ED;lefthand

edne

ssIn

theBN

andHCgrou

ps:history

ofAN

InAN-R,A

N-BPandHCs,bu

tno

tBN

,theam

ygdalawas

sign

ificantlyacti-

vatedin

respon

seto

own‘fat-im

age’

AN-BPandHCsvs

BNandAN-R:m

edial

PFCsign

ificantlyactivated


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.(Con

tinued)

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

AN-R

vsothe

rED

grou

ps:amygdala

sign

ificantlymoreactivated

inrespon

seto

‘fat’im

ageof

anothe

rwom

an

12.M

iyake,

Okamoto,

Onada,Shiraoet

al.(2010)[49]

Neuroimage,50:

1333-1339.

BN(n=12)

AN-R

(n=12)

AN-BP(n=12)

HCs(n=12)

BN:25.0(6.9)

AN-R:27.0(9.0)

AN-BP:27.2(4.8)

HC:25.4(5.8)

100%

fMRI

whilecompletingem

otionalw

ord

decision

makingtask,examining

processing

ofwords

(neg

ativebo

dyim

agewords

e.g.ob

esity;and

neutral

words).

Presen

ceof

AxisIo

rIIdisorder

othe

rthan

ED;lefthand

edne

ssNote-on

eBN

participan

tha

dahistory

ofAN

Neg

ativebo

dyim

agewords

cond

ition

:AN-R

&AN-BPvs

BN&HC:

right

amygdalasign

ificantlymore

activated

BN&AN-BPvs

HC:leftmed

ialP

FCsig-

nificantly

moreactivated

AN-R

&AN-BPvs

HCs:leftinferio

rpar-

ietallob

ulesign


Overall:results

indicatedthat

distorted

cogn

ition

ofne

gativebo

dyim

age

words

ineatin

gdisorder

patientswere

relatedto

enhanced

activationin

amygdalaandmPFC.H

owever

there

wereno

grou

pdifferences

inno

nspe

cific

negativeem

otionwords

13.M

ohret

al.

(2011)

[50]

NeuroImage,56:

1822-1831.

BN(n=15)

HCs(n=15)

BN:24.8(3.2)

HC:25.5(4.5)

100%

fMRI

whileratin

gsatisfactionandsize

estim

ationof

distortedow

nbo

dyph

otog

raph

s

History

ofsubstanceabuse;

schizoph

reniaandpsycho

ticsymptom

s;bipo

lardisorder;

neurolog

icalillne

ss;closedhe

adinjury;

lefthand

edne

ss

Theactivationpatternin

theinsula

reflected

satisfactionratin

gsof

BNand

HCs

HCsvs

BN:interm

sof

gene

ral

differences

inbo

dyim

ageprocessing

(not

specifically

durin

gsatisfaction/

percep

tioncond

ition

s),the

MFG

&rig

htpo

steriorparietalcortex

demon

stratedsign

ificantlygreater

activation(poten

tially

reflectinga

redu

cedspatialm

anipulationcapacity)

HCvs

BN:d

uringbo

dysize

estim

ation/

percep

tiontask,the

MFG

was

sign


inHCsthan

BN,and

theMFG

was

recruited

sign

ificantlymorein

thepe

rcep

tionvs

satisfactiontask

HCsvs

BN:p

osterio

rtempo

ral-o

ccipital

cortex

was

sensitive

forbo

dyim

age

distortio

n(this‘type

ofmod

ulation’was

notob

served

inBN

)HC&BN

:The

amou

ntof

bilateralinsula

activity

reflected

thepatternof

satisfactionratin

gtask.InBN

alinear

tren

doccurred

with

ade

clinein

insula

andMFG

activity

from

thinne

rto

fatter

images,alth

ough

results

weren

’tas

clearin

HCs

14.Prin

gleet

al.

(2011)

[51]

BN(n=11)

HCs(n=16)

BN:24.55

(4.97)

HC:27.38

(5.44)

100%

fMRI

toexam

ineself-referent

emotional

processing

,whe

repatientshadto

Lefthand

edne

ss;m

edication

BNvs

HCs:ratin

gof

negative

person

ality

descrip

torswas

associated


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.(Con

tinued)

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

Neuropsycho

logia

49:3272-3278.

endo

rse60

person

ality

characteristic

words

as‘me’or

‘not

me’in

rapid

even

trelatedde

sign

.

with

redu

cedactivity

intheparietal,

occipitaland

limbicareas,includ

ingthe

amygdala

15.Schienleet

al.(2009)[52]

Biological

Psychiatry,65:

654-661.

BED(n=17)

BN-P

(n=14)

HCsno

rmalweigh

t(n=19)

Con

trols-overweigh

t(C-

OW)(n=

17)

BED:26.4(6.4)

BN-P:23.1(3.8)

HC-N:22.3(2.6)

HC-O:25.0(4.7)

100%

fMRI

completed

after12-hrovernigh

tfast,w

hileparticipantsview

edthree

catego

riesof

images:highcalorie

(e.g.

icecream,frenchfries),disgust-indu

-cing

(e.g.d

irtytoilets,m

aggo

ts)and

affectivelyne

utral(e.g.

househ

old

items).

Med

ication;clinicallyrelevant

depression

;lefthand

edne

ssAllparticipantsde

mon

stratedincreased

activationin

theOFC

,ACCandinsula

high

lightingabasicappe

titive

respon

sepattern.Nogrou

pdifferences

indisgust-indu

cing

images

BEDvs.allothe

rgrou

ps:enh

anced

rewardsensitivity,stron

germed

ialO

FCactivity

whileview

ingfood

images

BNvs.allothe

rgrou

ps:g

reater

ACC

activationandinsulaactivationwhile

view

ingfood

images

16.Seitzet

al.

(2016)

[18]

European

Child

andAd

olescent

Psychiatry,1:

S185-203.

BN(n=20)

HCs(n=20)

BN:18.71

(2.53)

HC:17.90

(1.35)

100%

fMRI

whileparticipantscompleted

amod

ified

versionof

theAtten

tion

NetworkTask

(ANT),investig

ating

neuralne

tworks

associated

with

alertin

g,reorientingandexecutive

attention.

History

ofpsycho

sis;substanceabuse;

IQ<80

BNvs.H

Cs:

•Highe

rADHDscores,especially

inattention.

•Hyperactivationin

theparieto-

occipitalreg

ions

andredu

cedde

activa-

tionof

theprecun

eus,partof

the

default-mod

e-ne

tworkareasdu

ring

‘alerting’

•Po

steriorcing

ulateactivationdu

ring

alertin

gcorrelated

with

severityof

BNsymptom

s•Exploratorycorrelationanalyses

foun

dsign

ificant

associations

betw

eenne

ural

activity

inthe‘alerting’

cond

ition

inthe

bilateralm

iddlecing

ulateandglob

aleatin

gdisorder

symptom

s;sign

ificant

inversecorrelationbe

tweenactivity

inthetempe

roparietaljun

ctionand

ADHDsymptom

s;andactivity

inthe

right

parahipp

ocam

puswas

inversely

correlated

with

impu

lsivity

scores

17.Skund

eet

al.

(2016)

[26]

Journa

lof

Psychiatryan

dNeuroscience,

41(5):E69-E78.

BN(n=28)

HCs(n=29)

BN:27.54

(10.52)

HC:27.25

(6.68)

100%

fMRI

whilecompletingage

neraland

food

-spe

cific

(participantsselected

8of

theirfavourite

food

images

from

aset

of85

high

-caloriefood

spriorto

com-

pletingthetask)no

-gotask

(the

no-go

task

isasub-task

ofthego

-no-go

task

andmeasuresbe

haviou

ralinh

ibition

).

Biploardisorder;p

sychosis;history

ofhe

adinjury;neurologicdisorder;

diabetes

mellitus;n

icotine/drug

/alcoho

labu

se;lifetim

ediagno

sisof

BPD

InHCs:curren

tpsycho

trop

icmed

itatio

nIn

BN:m

edicationothe

rthan

antid

epresants

BNvs.H

Cs:redu

cedactivationin

the

right

sensorim

otor

area

(postcen

tral

gyrus,precen

tralgyrus)andrig

htdo

rsal

striatum

(caudate

nucleus,pu

tamen

)HCvs

BN(highfre

quen

cyBEson

ly):

strong

eractivationin

therig

htpo

stcentralg

yrus,right

caud

ate

nucleusandrig

htpu

tamen

18.Spang

leret

al.(2012)[53]

BN(n=12)

HCs(n=12)

BN:A

gerang

erepo

rted

only(18-

38)

100%

fMRI

whilelookingat

compu

ter-

gene

ratedim

ages

of‘thin’(BM

I=18)or

‘fat’(BMI=31)bo

dies

(and

control

InBN

:med

icationothe

rthan

antid

epressants

BN:nosign

ificant

differencefoun

din

brainactivationwhilelookingat

thin

vsfatim

ages


Table


sof

includ

edstud

iesusingfM

RIas

theprim

arymetho

d.(Con

tinued)

Autho

rs&

Journal

Participants

Meanage(SD)

% Female

Metho

dPsychiatric

/othe

rexclusions

Find

ings

Internationa

lJourna

lofEating

Diso

rders,45(1):

17-25

HC:(18-30)

cond

ition

:scram

bled

image).

Participantsinstructed

toimagine

someone

iscomparingyour

body

tothe

body

ofthewom

anin

thepicture.

BNvs.H

C:m

PFCactivationwas

sign

ificantlygreaterwhileview

ing‘fat’

images,w

ithincreasedactivity

inthe

region

sassociated

with

emotional

processing

.Nodifferences

betw

een

grou

psin

thethin

cond

ition

InthemPFC,the

peak

locatio

nof

activationforBN

patientswas

inthe

pgACCrather

than

dorsalmPFC,asit

was

forHCs

19.U

heret

al.

2004

[54]

American

Journa

lof

Psychiatry,

161(7):1238-

1246

BN(n=10)

AN(n=16)

HCs(n=19)

BN:29.80

(8.80)

AN:26.93

(12.14)

HC:26.68

(8.34)

100%

fMRI

completed

whilebe

ingpresen

ted

with

photog

raph

sof

savouryand

sweetfood

s;no

n-food

items;em

otion-

allyaversive

photog

raph

sandne

utral

stim

uli.

AxisId

isorde

rsothe

rthan

ED;

neurolog

icalor

psychiatric

illne

ssaside

from

ED;p

sychotropicmed

ication

othe

rthan

antid

epressants

BNvs

HCs:greateroccipitaland

cerebe

llaractivity

BNvs

AN&HCs:de

creasedactivation

intheanterio

randlateralP

FCin

respon

seto

food

images

(associated

with

supp

ressingun

wanted

behaviou

rs)

AN&BN

vsHC:significantly

increased

med

ialP

FCreactio

nto

food

images

20.U

heret

al.

2005

[55]

Biological

Psychiatry,

58(12):990-997

BN(n=9)

AN(n=13)

HCs(n=19)

BN:29.6(9.3)

AN:25.4(10.2)

HC:26.6(8.6)

100%

fMRI

toexam

inecerebralcorrelates

ofbo

dyim

ageactivity

whe

nparticipants

lookingat

linedraw

ings

ofun

derw

eigh

t(BMI=

<17.5),no

rmal

weigh

t(20<

BMI<25),andoverweigh

t(BMI27.5)

femalebo

dies

vs.con

trol

images

(line

draw

ings

ofho

uses)

Psycho

sis;alcoho

lordrug

depe

nden

ce;

neurolog

icalor

psychiatric

illne

ssaside

from

ED;p

sychotropicmed

ication

othe

rthan

antid

epressants

Noregion

sof

sign

ificantlyincreased

activations

ineither

eatin

gdisorder

grou

p,comparedto

thecontrol

subjects

AcrossAN,BN&HCs,thelateral

fusiform

gyrus,inferio

rparietalcortex

andlateralP

FCwereactivated

inrespon

seto

body

shapes

vs.con

trol

cond

ition

21.Vocks

etal.

2010

[56]

Journa

lof

Psychiatryan

dNeuroscience,

35(3):163-176.

AN(n=13:8

AN-R

and6

AN-BP)

BN(n=15)

HC(n=27)

AN:29.08

(9.79)

BN:28.4(7.07)

HCs:26.74(7.6)

100%

fMRI

whileparticipantslooked

at16

standardised

photog

raph

sof

theirow

nbo

dyandanothe

rwom

an’sbo

dy(BMI

19),takenwhilewearin

gabikini.

Lefthand

edne

ss;p

ersonalitydisorder

AN&BN

vs.H

Cs:whileview

ing

photog

raph

sof

theirow

nbo

dy,eating

disorder

patientsshow

edweakene

dactivity

intheleftinferio

rparietal

lobu

leANvs.BN&HCs:high

eram

ygdala

activity

whilelookingat

photog

raph

sof

anothe

rwom

an’sbo

dyANvs.BN&HCs:sign

ificantlygreater

activationin

thebilateralsup

erior

tempo

ralg

yrus


with BED had stronger activation in the VS, whilewomen with BN had greater activation in the premotorcortex and dorsal striatum [44].Skunde et al. [26] personalized visual food stimuli for

BN participants, by asking them to select eight favouritefood images to use during a measure of behavioural in-hibition, finding the BN group had reduced motor cor-tex activation including the primary motor, premotorand primary somatosensory cortices; and reduced activa-tion in the right sensorimotor area (postcentral and pre-central gyrus) and right dorsal striatum (caudatenucleus, putamen), relative to HCs. Importantly, theseresults suggest diminished frontostriatal and sensori-motor control contribute to diminished inhibitory con-trol in BN. Finally, Uher et al. [54] found women withBN had decreased activation in the anterior and lateralprefrontal cortex in response to viewing food imagescompared to AN and HC groups, and greater occipitaland cerebellar activity compared to the HC group.

Body image-related stimuliBody image distortion is a key symptom of eating disor-ders. Six studies included in the review used bodyimage-related photographs, images or tasks during fMRI[48–50, 53, 55, 56]. As in the other categories of fMRIstudies, there was heterogeneity in the stimuli used. Onestudy used negative body-related words e.g. obesity [49],reporting distorted cognition of negative body imagewords in women with BN and AN may be linked to in-creased activation in the amygdala and medial prefrontalcortex (mPFC). Two of the studies used real and dis-torted photographs of participants’ own bodies and con-trol bodies. Miyake, Okamoto, Onada, Kurosaki et al.[48] reported the amygdala and mPFC were significantlyactivated in AN and HC women in response to theirown ‘fat’ distorted image however this wasn’t found inBN women. Mohr and colleagues [50] used similar stim-uli and reported that BN participants did not recruit themiddle frontal gyrus (MFG) compared to HCs when es-timating the size of their bodies, possibly reflecting re-duced spatial manipulation. Increased activity in thelateral occipital cortex was sensitive for body size distor-tions in the control group but not in the BN group andthe authors proposed the pattern of results may underliebody size overestimation in BN [50]. Interestingly, in theBN group a linear trend was observed, wherein insulaand MFG activity declined as body photographs movedfrom thinner to actual and fatter images [50]. In a re-lated study, Vocks et al. [56] used standardized photos ofwomen’s own bodies and another woman’s body wearinga bikini, finding increased activation in the left middletemporal gyrus and middle frontal gyrus in BN and ANgroups. Two studies [53, 55] used non-naturalistic stim-uli. Spangler et al. [53] examined the activity of the

mPFC, associated with self-referencing, during fMRIwhile participants imagined their bodies being comparedto computer-generated ‘thin’ (Body Mass Index [BMI]=18) and ‘fat’ (BMI = 31) images. Women with BN expe-rienced significantly greater mPFC activation when view-ing the overweight female image compared to HCs,while no differences were found between groups whenviewing the thin female image. Lastly, Uher et al. [55]used simple line drawings of underweight (BMI =27.5) bodies and found no regions of signifi-cantly increased activation in women with BN or ANcompared to the HCs, however reported the patientgroup as a whole demonstrated weaker occipitotemporalcortex and parietal cortex activation to body shapescompared to HCs.

fMRI studies – otherThree fMRI papers that did not fit into the previous cat-egories are reviewed here. Amianto and colleagues [57]conducted one resting-state fMRI in AN and BN groups,finding grey matter reduction and decreased connectiv-ity of the cerebellar network to the parietal cortex andincreased bilateral connectivity of intrinsic connectivitynetworks (ICNs). The BN group, when compared to theAN and HC groups, demonstrated grey matter reductionin the caudate nucleus (CN), part of the dorsal striatumand frontostriatal circuit. Marsh and colleagues [47]assessed morphological measures of the cerebral surfaceand compared to HCs, the BN group had significant re-ductions of local volume on the brain surface in thefrontal and temperoparietal areas [47]. The local volumereduction in the inferior frontal regions was inverselycorrelated with age, symptom severity and pre-fMRI per-formance on the Stroop test. Pringle et al. [51] used aself-referent emotional processing task in BN, whereinparticipants had to rapidly endorse 60 personality char-acteristic words as either ‘me’ or ‘not me’. The BN groupexperienced decreased activity in the parietal, occipitaland limbic areas when processing negative personalitydescriptors compared to HCs [51].

Functional differences: Summary of SPECT and PET studies.Three studies met inclusion criteria for this systematicreview using SPECT, which provides a functional, quan-titative measure of regional cerebral blood flow (rCBF),a metric of brain function [62]. Two studies in this re-view used PET; one was described earlier that used com-bined MRI and PET [39], and the other described hereusing only PET. PET provides a measure of glucose me-tabolism rates in the brain, also a measure of brain func-tion. Sample sizes were extremely small in this categoryof papers, relative to MRI and fMRI studies; BN medianand range: 8 (5-21); HC:11.5 (9-12); BED: only one study


included participants with BED (n=8). See Table 3 forthe data extracted from these studies.Two of the SPECT studies [34, 63] measured partici-

pants’ rCBF during viewing of disorder-related stimuli orneutral stimuli. Beato-Fernandez et al. [34] examinedrCBF across AN-Restricting (AN-R), AN-Binge-Purge(AN-BP), BN-Non-Purging (BN-NP), BN-Purging (BN-P) and HCs using three SPECT imaging procedures dur-ing rest, calm visual stimulus, and after seeing their ownbody image photographs. In the BN-P and AN-R groups,increased rCBF in the right temporal area was foundwhen going from neutral to body image visual stimulus(Beato-Fernandez et al., [34]). Karhunen and colleagues[63] also found increased rCBF during exposure to foodimages compared to neutral images, in the left frontaland prefrontal cortices in obese BED participants com-pared to obese non-BED and HC participants. Nozoeand colleagues [64] completed SPECT before and afterAN, BN and HC women ate a slice of cake and foundthe BN group had the highest rCBF before eating in theleft temporal and bilateral inferior frontal regions, butthe lowest cortical activity after eating. Finally, Delvenneet al. [65] completed a resting-state PET with a chemicalmarker to assess cerebral glucose metabolism, findingwomen with BN demonstrated absolute glucose hypo-metabolism globally and regionally, most notably in theparietal and superior frontal cortices, compared to HCs.

DiscussionThis systematic review identified and reviewed 32 papersinvestigating neural differences between individuals withBN and / or BED and HC groups [18, 22–26, 34–39,42–47, 49–59, 63–65]. Included in this review were six-teen studies on participants with BN, eleven studies onparticipants with BN and AN, three studies on partici-pants with BN and BED and two studies on participantswith BED. The objectives of this review were first, toprovide a synthesis of published studies on neuroimag-ing in BN and BED. Due to the heterogeneity of the 32studies, results were reviewed according to the methodof the study and type of neurological test completed(MRI, fMRI, SPECT and/or PET). The evaluation of lit-erature indicates that it is too early to make any definiteconclusions about the neuroimaging profile of individ-uals with BN or BED. The diverse range of neuroimag-ing procedures and stimuli used during testing coupledwith small sample sizes impeded the ability to drawmany clear conclusions. Despite this, a discussion is pre-sented below which attempts to highlight the findings ina meaningful manner.In relation to the second objective, to identify neuro-

biological studies that will assist in elucidating the ap-parent decreasing clinical utility of OBEs and SBEs, nostudies were found wherein comparisons were made

between participants reporting OBEs versus SBEs. It isworth noting that a number of studies in this reviewcompared sub-groups of participants based on severity[24, 26, 35, 39, 45, 47, 52, 58], which was exclusively de-fined as the frequency of symptoms, rather than an at-tempt to quantify the actual amount of food consumedduring a BE. The frequency of BE and purging in BN isconsidered an important predictor of therapeutic out-come [26] and this review certainly highlights that moreaberrant neurobiological activity is reported in patientswith more frequent binge eating or bulimic episodes.If it is accepted that the severity of bulimic symptoms

in the context of clinical assessment and neuroimagingresearch is based on the frequency of binge and / orpurge episodes, it raises further doubt regarding the clin-ical utility and validity of continuing to attempt to quan-tify individuals’ binges as OBEs or SBEs.

MRI studiesSeven studies included in this review reported on dataobtained from MRI [22, 24, 35–39]. The findings of MRIstudies were mixed; most identified structural changesin the brains of BN and BED patients. Of the seven stud-ies, only one included both BN and BED participants[24]; two studies recruited BN and AN participants [36,37]; and four investigated BN participants only [22, 35,38, 39]. Of the two studies investigating BN and AN par-ticipants, one reported similarities in both groups interms of smaller pituitary glands, which the authors pos-tulated may have atrophied following prolonged malnu-trition [37]. The pituitary gland is central in thehypothalamic-pituitary-gonadal axis and is critical fornormal sexual and hormonal development in puberty. Arecent study [66] found age, puberty stage, testosteroneand estradiol levels predicts pituitary volume in adoles-cents, which may be relevant to this finding consideringthe onset of illness is usually during adolescence. Thesecond study reported the midbrain and thalamus dem-onstrated significant volume reduction only in the ANgroup, not in the BN and HC groups [36]. The thalamusis thought of as the ‘gate keeper’ to the cerebral cortexand is centrally involved in taste and gustatory process-ing, before communicating sensory information to thefrontal region and insula for further processing [67].One MRI study compared BN, BED and HC groups

[24], finding evidence of structural changes in thereward-learning circuit; specifically, increased grey mat-ter volumes in the medial orbitofrontal cortex (OFC) inboth BN and BED. This finding may reflect differentialreward processing as the OFC is implicated in the he-donic value of food stimuli [24, 61]. fMRI studies in thissystematic review have also highlighted alterations in themedial OFC associated with processing visual food re-ward cues in eating disorder groups compared to HCs


Table


sof

includ

edstud

iesusingSPEC

TandPETas

theprim

arymetho

d.Autho

rs&Journal

Participants

MeanAge

(SD)

%Female

Proced

ure

Psychiatric

/othe

rexclusions

Find

ings

1.Beato-Fernande

zet

al.(2011)[34]

ActasEspano

las

dePsiqiuiatria,

39(4):203-10.

AN-R

(n=11)

AN-P

(n=10)

BN-NP(n=7)

BN-P

(n=14)

HCs(n=12)

AN-R:27.1

AN-P:28.4

BN-P:30.7

BN-NP:34.7

HC:20.6

NoSD

ofmeanage

reported

Not

repo

rted

3xSPEC

Tscansto

measure

rCBF

durin

grestcond

ition

;calm

visual

stim

ulus

cond

ition

andanothe

rafter

seeing

their

ownbo

dy(film

ed).

Lefthand

edne

ss;

psychiatric

illne

ssasidefro

mED

;ne

urolog

icaldisorders

AN-R,A

N-P,BN-P,BN-NP

vsHCs:de

creasedrig

httempo

ralrCBF

whe

nmovingfro

mrestcond

ition

tone

utralvisualimage

AN-R

&BN

-Pvs

othe

rgrou

ps:increased

right

tempo

ralrCBF

goingfro

mne

utralvisualimageto

ownbo

dyvisualim

age

2.Karhun

enet

al.

(2000)

[63]

PsychiatryResearch:

Neuroimaging

Section,99:29-42.

OBBED(n=8)

OBno

n-BED(n=11)

HCs(n=12)

OBBED:

36.1(9.3)

OBno

n-BED:

45.0(10.0)

HCs:39.8(9.7)

100%

1xSPEC

Tscan

tomeasure

rCBF

whileparticipants

werelookingat

acontrol

image(land

scape)

and

1xSPEC

Tscan

while

participantswere

lookingat

apo

rtion

ofrealfood

after

anovernigh

tfast.

Lefthand

edne

ss;

‘Noothe

rdisorders

ormed

icationknow

nto

affect

thevariables

exam

ined

’(pp

31)

OBBEDvs

OBno

n-BED

&HCs:sign

ificantly

greaterincrease

inrCBF

inthelefthe

misph

ere

comparedto

therig

hthe

misph

ere,particularly

inthefro

ntaland

prefrontalcortices,

inthefood

expo

sure

cond

ition

Allgrou

psexpe

rienced

asign

ificant

increase

inhu

nger

inthefood

expo

sure

cond

ition

.In

theOB-BEDgrou

pon

ly,thiswas

associated

with

sign

ificantlyhigh

errCBF

intheleftfro

ntal

andpre-fro

ntalcortices

3.Delvenn

eet

al.(1997)[65]

Internationa

lJournal

ofEating

Diso

rders,21(4):313-320.

BN(n=11)

HCs(n=11)

BN:26.2(10.9)

HCs:25.7(2.1)

100%

RestingstatePETwith

(18-F)

fluorod

eoxyglucose

used

toevaluate

cerebral

glucosemetabolism.

History

ofelectrocon

vulsive

therapy(ECT);significant

abno

rmalities

onph

ysical

andne

urolog

ical

exam

ination;

left

hand

edne

ss;n

opsycho

activemed

ication

foraminim

umof

10days;

historyof

neurolep

ticmed

ication

BNvs

HCs:absolute

hypo

metabolism

ofglucosebo

thglob

ally

andregion

ally,n

otably

intheparietaland

supe

rior

frontalcortices.The

BNgrou

palso

show

edalower

relativeregion

alcerebralglucose

metabolism

inthe

parietalcortex

4.Nozoe

etal.

(1995)

[64]

BrainResearch

Bulletin,36(3):251-255.

BN(n=5)

AN(n=8)

HCs(n=9)

BN:21.0(2.9)

AN:24.1(7.8)

HCs:20.3(1.0)

100%

Exam

ined

rCBF

using

SPEC

Tbe

fore

andafter

food

intake

(sliceof

cake)

Lefthand

edne

ss;

abno

rmalne

urolog

ical

finding

s

BNvs

AN&HCs:

high

estrCBF

before

eatin

gin

thelefttempo

raland

bilateralinferiorfro

ntal

region

s.Also,BN

show

edless

increase

incortical

activity

post-eating

BNandANshow

edop

posite

patterns

offro

ntalreactio

nto

food

stim

uli

AN:sho

wed

nomarked

corticallateralityor

activation

inanycorticalarea

pre-eatin

gbu

tshow

edgreater

increasedcortical

activity

post-eating


[52, 54, 63]. Schafer and colleagues [24] also found in-creased volume in the VS in BN, which was significantlyand positively correlated with a lower BMI and increasedfrequency of purging. The central role of the striatum, inparticular the VS, in the processes of decision making,reward and motivation, is now well-accepted based on alarge number of human neuroimaging studies [68]. Thisfinding, in combination with increased medial OFC vol-ume, suggests a joint alteration of food reward process-ing and instrumental behaviour to reduce the chance ofweight gain as the end goal [24].The four remaining MRI studies compared BN and

HC groups only. Three studies [22, 35, 38] reportedstructural changes in terms of cortical atrophy or vol-ume reduction in the BN group, firstly in the ratio ofcerebral to cranial area [35] and in the second paper, inthe inferior frontal grey matter [38]. We believe thatHoffman and colleagues [35] made the first critical find-ing of a significant positive correlation between illnessseverity (binge frequency) and structural changes in thebrain. The third paper to identify structural changesfound volume reduction in the caudate nucleus withinthe frontostriatal circuit [22]. Lastly, Galusca et al. [39]reported widespread impairment in serotonergic activityin BN patients during PET and MRI, which is consistentwith well-known research that serotonergic activity isdisturbed in BN, and serotonin-modulating medicationsdecrease BN symptoms independently of their anti-depressant effects [69]. It is notable that Galusca andcolleagues’ [39] study recruited only severe BN-P partici-pants, where the frequency of binge-purge episodes hadto be a minimum of at least once daily for a minimumof six months, however we consider this group to bemore representative of individuals likely to seek treat-ment, due to the distress associated with symptoms atthis level.The collective MRI results reveal several key findings.

Firstly, in BN groups structural changes have been con-sistently reported, firstly in the ratio of cerebral to cra-nial area volume [35]; decreased inferior frontal greymatter [38]; reduction within the frontostriatal circuit,specifically in the caudate nucleus [22]; and volume re-duction in the pituitary gland [37]. In one study [24], in-creased medial OFC volumes characterised BN and BEDpatients while increased volumes within the VS, dorsalstriatum, medial and lateral OFC was found in BN pa-tients only, and BED patients had increased volumewithin the ACC. The only combined PET and MRI studyincluded in this review highlighted widespread serotonindysfunction in individuals with severe BN [39].

fMRI studiesTwenty-one studies in the present review used fMRI asthe primary method of investigation [18, 23, 25, 26, 41–

57]. Studies included in this review using fMRI tech-niques yielded a broad range of results. With regards tostudies of food-specific stimuli, results suggested thatpatients with bulimic pathology have less activation ingustatory and reward regions before and during eating,which may mediate the tendency to over-consume andbinge eat [58]. Schienle and colleagues [52] reportedgreater activation in the insula and the ACC in BN pa-tients, which was positively associated with frequency ofBEs. An interesting finding

Neuroimaging in bulimia nervosa and binge eating disorder: a … · 2018. 2. 20. · REVIEW Open...

Documents

Transcript of Neuroimaging in bulimia nervosa and binge eating disorder: a … · 2018. 2. 20. · REVIEW Open...