ebook img

Review Article The Use of Functional MRI to Study Appetite Control in the CNS PDF

14 Pages·2014·2.51 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Review Article The Use of Functional MRI to Study Appetite Control in the CNS

HindawiPublishingCorporation ExperimentalDiabetesResearch Volume2012,ArticleID764017,13pages doi:10.1155/2012/764017 Review Article The Use of Functional MRI to Study Appetite Control in the CNS AkilaDeSilva,1VictoriaSalem,1PaulM.Matthews,2,3andWaljitS.Dhillo1 1SectionofInvestigativeMedicine,DivisionofDiabetes,EndocrinologyandMetabolism,ImperialCollegeLondon, HammersmithHospitalCampus,DuCaneRoad,LondonW120NN,UK 2GlaxoSmithKlineGlobalImagingUnit,ImperialCollegeLondon,HammersmithHospitalCampus,DuCaneRoad, LondonW120NN,UK 3CentreforNeuroscience,DepartmentofMedicine,ImperialCollegeLondon,HammersmithHospitalCampus,DuCaneRoad, LondonW120NN,UK CorrespondenceshouldbeaddressedtoWaljitS.Dhillo,[email protected] Received1February2012;Accepted28April2012 AcademicEditor:AlainKtorza Copyright©2012AkilaDeSilvaetal.ThisisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense, whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited. Functionalmagneticresonanceimaging(fMRI)hasprovidedtheopportunitytosafelyinvestigatetheworkingsofthehuman brain.Thispaperfocusesonitsuseinthefieldofhumanappetitivebehaviouranditsimpactinobesityresearch.Inthepresent absenceofanysafeoreffectivecentrallyactingappetitesuppressants,abetterunderstandingofhowappetiteiscontrolledisvital forthedevelopmentofnewantiobesitypharmacotherapies.Earlyfunctionalimagingtechniquesrevealedanattenuationofbrain rewardareaactivityinresponsetovisualfoodstimuliwhenhumansarefed—inotherwords,thephysiologicalstateofhunger somehowincreasestheappealvalueoffood.Laterstudieshaveinvestigatedtheactionofappetitemodulatinghormonesonthe fMRIsignal,showinghowtheattenuationofbrainrewardregionactivitythatfollowsfeedingcanberecreatedinthefastedstateby theadministrationofanorecticguthormones.Furthermore,differencesinbrainactivitybetweenobeseandleanindividualshave providedcluesaboutthepossibleaetiologyofovereating.Thehypothalamusactsasacentralgatewaymodulatinghomeostaticand nonhomeostaticdrivestoeat.AsfMRItechniquesconstantlyimprove,functionaldataregardingtheroleofthissmallbuthugely importantstructureinappetitecontrolisemerging. 1.Introduction market.Foracomprehensiveoverviewofthepositionofgut hormonesinthetreatmentofobesity,wedirectthereaderto Whenenergyintakeexceedsenergyexpenditureinthelong areviewbyValentinoetal.,2010[2]. term, the excess accumulates as body fat. It is predicted The control of appetite and food intake can be divided that, by 2015, as many as 30% of adults in Europe will be into “homeostatic” and “nonhomeostatic” control. The obese, as defined by a body mass index (BMI) ≥30kg/m2 alteration in consumption of food that follows sensing of [1].Itiswellknownthatobesityconfersincreasedrisktoa energybalanceformsthebasisofthehomeostaticcontrolof range of diseases, including type 2 diabetes, cardiovascular appetite: following a meal, appetite is suppressed, whereas, disease, and cancer. An enhanced understanding of the following significant energy expenditure, we feel hungry. physiological mechanisms that regulate appetite is essential Postprandial satiety signals include changes in circulating forthedevelopmentofpharmacologicalstrategiestocombat concentrationsofnutrientsandguthormones.Theseinclude obesity.Inparticular,withinthefieldofneuroendocrinology, theanorectichormonesPYYandGLP-1andtheorexigenic identification of hormone-brain interactions that modulate appetite (and subsequent food intake) will aid the manip- hormoneghrelin[3],aswellasactivationofsomatosensory ulation of naturally occurring hormones to develop safe vagalafferentsconveyinginformationsuchasgastricdisten- and effective antiobesity therapies. This strategy, exploiting sion [4]. Adipokines (hormones secreted by adipose tissue, physiologicalpathwaysofappetitecontrol,holdspromisefor suchasleptin)conveyinformationtotheCNSaboutlonger- thefuturetreatmentofobesity,especiallygiventheshortage term energy stores [5]. Furthermore, it is well established of effective pharmacotherapies for obesity in the existing thatinsulinissecretedacutelyfromthepancreasinresponse 2 ExperimentalDiabetesResearch to a glucose load but is also basally elevated in states of overnutrition and obesity, thereby implying a role in con- veying both short- and longer-term information regarding energybalance.Innormalphysiology,theintegrationofthe above inputs within central neural appetite circuits results in appropriately matching food intake to metabolic need. A C C PFC The hypothalamus is widely recognised as the gatekeeper vs for this processing task, in line with other important DLPFC homeostaticfunctionscontrollingbodytemperature,energy NAc VTA expenditure, and glucose metabolism. In keeping with this, Am OFC thereismountingevidencethathypothalamicdysfunctionis H implicatedinthepathogenesisofobesity[6]. Withregardto“nonhomeostatic”control,foodintakeis driven by factors including food palatability (incorporating the key senses of sight, smell, and taste), habitual, socio- cultural, emotional, and economic influences. From an emotionalperspective,theevaluationofafoodstimulusfor Figure 1: Brain reward centres: the hypothalamus (H), as a itsmotivationalsignificanceoccursinmultiplebrainregions, homeostatic gatekeeper, has numerous connections with higher particularly the dopaminergic limbic and prefrontal reward braincentreswhichprocesssalienceandreward.Thehypothalamus areas [7]. Key areas include the amygdala, hippocampus, transmits to these higher centres information received from the insula, striatum, and orbitofrontal cortex (OFC), although periphery,suchasnutritionalstatussignalledviathepostprandial this list is by no means exhaustive. It is now widely releaseofguthormones,andinturnmodulatesmetabolicratevia accepted that, with regard to appetite control, homeostatic thesympatheticnervoussystem.Thissagittalsectionofthebrain andnonhomeostaticsystemsdonotfunctionindependently; revealstheimportantareasinvolvedinthehedoniccontrolofeating instead, there is extensive cross-modulation between them, behaviour; amygdale (Am): emotional and aversive processing; nucleusaccumbens(Nac):anticipatoryrewardprocessing;ventral along with a complex integration of inputs before a final tegmental area (VTA): numerous dopaminergic projections to decisionismaderegardingfoodconsumption(Figure1). other limbic areas; ventral striatum (VS): motivation reward; Over the past decade, functional magnetic resonance expectancyandnoveltyprocessing;anteriorcingulatecortex(ACC): imaging(fMRI)hasbecomeapopularandrapidlyadvancing decision making; orbitofrontal cortex (OFC): reward encoding; tool for investigating CNS appetite pathways in humans, prefrontal cortex (PFC): translation of external and internal cues offering the key advantage of avoiding exposure to ionising intobehaviouralresponses;dorsolateralprefrontalcortex(DLPFC): radiation. Following on from previous similar articles [8– self-control. Not shown is the insular cortex (a more lateral 10],hereweprovideanupdatedreviewofrecentadvancesin structure),whichisalsoimportantingustatoryprocessing. ourunderstandingofhumanappetitivebehaviourbasedon theresultsoffMRIstudies.Westartwithabriefoverviewof areinalow-energystate.GenerationofMRIimagesrequires thetechniqueandthensummarisefindingsfromtheearliest fMRI appetite studies, which first explored differences in application of a radiofrequency (RF) pulse at 90 degrees to B .Theprotonswillthen“tip”toalignwiththeRFpulse.In brainactivationpatternsbetweenthenormalfedandfasted 0 doingso,theygainenergy.AftertheRFpulseisswitchedoff, state and in response to glucose ingestion. This is followed the protons realign with B . Spins return to the low-energy by a discussion of a further cluster of studies, exploring 0 differencesbetween obese and lean populations. Finally, we state by emitting the absorbed energy, also in the form of a radio wave. The emitted energy can be measured by the review the results of fMRI studies investigating hormonal receivercoilandconvertedtoimages.Byslightlyalteringthe influences on appetitive brain processes and in relation to strength of the magnetic field (and therefore the frequency this and explore recent functional neuroimaging outcomes of the emitted radiation) using gradient coils across the followinginterventioninobesesubjects. volume to be imaged, spatial information can be inferred. The T relaxation time is a time constant referring to the 1 2.PrinciplesoffMRIinAppetiteStudies realignmentofspinswithB inthelongitudinalplaneafter 0 theRFpulseisswitchedoff.TheT relaxationtime,onthe 2 A full description of fMRI methodology is clearly outside otherhand,isatimeconstantreferringtothedephasingof theremitofthispaper.Itisneverthelessusefulforthenon- spins in the transverse plane after the RF pulse is switched imagingspecialisttounderstandsomeofthetechnicalitiesof off(Figure2).T andT varydependingonthetissuebeing 1 2 fMRI,inordertogainabetterappreciationofhowthefield imaged.Thestrengthofmagneticresonancesignalobtained hasprogressed,aswellassomeofitslimitations. for a particular tissue depends primarily on the proton MRI utilises the behaviour of hydrogen nuclei, which density.However,byalteringthetimebetweensuccessiveRF consist of single protons that possess angular momentum pulsesandthereforethedegreeofT andT relaxation,the 1 2 (spin).Assoonasanexternalmagneticfield(B )isapplied, image can be weighted towards one or the other of these 0 the protons in tissue tend to align with this, causing their tissue-specificproperties.Itshouldbenotedthat,inpractice, spinstoprecessaboutacircularpatharoundB .Here,they due to localised inhomogeneities in the externally applied 0 ExperimentalDiabetesResearch 3 Z(M0) x y RF pulse (a) (b) (c) (d) (e) (f) Figure2:SchematicofT andT relaxation.MRIutilisesthebehaviourofprotonswithinvaryingmagneticfieldstoproducesignalswhich 1 2 canbeconvertedintoimages.Eachhydrogennucleusinthebraincanbethoughtofasavector(inthezandx-yplanes)representingthe strengthanddirectionofitsmagneticfieldasitspinsonitsaxis(itsmagneticdipolemoment,MDM).TheMDMsoftheimagedprotons trytoalignwiththemainexternalmagneticfieldofthescanner(referredtohereasB andconventionallyshownalongthezaxisin3D 0 coordinates).Asecondmagneticfield(intheformofashortradiofrequencyRFpulse)isapplied,whichflipsalloftheMDMsfromalignment inthezdirectionintothex-yplane(a).BeforeapplicationoftheRFpulsethe,amplitudeinthez-axisismaximalwhiletheamplitudein thex-yplaneiszero.JustafterapplicationoftheRFpulsethe,amplitudeinthez-axisiszero(a)whiletheamplitudeinthex-yplaneis maximal(d).Duringrelaxation,theamplitudeinthez-axiswillslowlyincrease((b)and(d))whiletheamplitudeinthex-yplaneslowly decreases((e)and(f)).T relaxationisthetimetakenforthezvectortoregaininstrength,whereasT relaxationisthetimetakenforthe 1 2 x-yvectortodecay.ThesechangingmagneticvectorsinvoketheirownRFsignals,whicharepickedupbythereceivercoilsandinterpreted intoinformationabouttheprotondensityofthesubjectbeingscanned. magneticfield,T isshorterthanexpectedforanyparticular large numbers of images in short spaces of time. Echo- 2 tissue.ThisapparentT isreferredtoasT ∗. planar imaging (EPI), for example, is a technique whereby 2 2 fMRIutilisesadaptationsofthisclassicalMRItechnique, anentire2-dimensionalimagecanbeattainedbytherapid such that function of tissues can be analysed rather than alterationofspatialgradientsfollowingasingleRFpulse.In structurealone.Virtually,allfMRIstudiesrelyonameasure this way, the entire brain can now be functionally imaged calledblood-oxygenation-level-dependent(BOLD)contrast, within the same timeframe as the physiological changes of based on the fact that oxygenated and deoxygenated blood interest.FunctionalMRIhasthusprovidedhugelyimportant possess different magnetic properties. Increased neuronal insightsinnumerousdisciplines.However,itsuseinthefield activity in the brain elicits a local haemodynamic response, ofneuroendocrinologyhasbeenlimitedduetotheinherent whichcausesanincreaseinbloodflowgreaterthannecessary weaknessofthetechniqueatimagingthehypothalamus.This for tissue demands. This results in a locally reduced ratio small (approximately 5–8mm in diameter) brain structure of deoxyhaemoglobin to oxyhaemoglobin concentrations, liesclosetotheair-tissueboundariesofsurroundingsinuses andtheensuingdifferencesinlocalmagneticfieldinhomo- and is susceptible to signal loss due to resultant distortions geneitiescanbedetectedonaT ∗-weightedimagingproto- oftheexternalmagneticfield,whichisaparticularissuefor 2 col[11,12];thelocallyreducedratioofdeoxyhaemoglobin T2∗-weightedimagesasalreadydescribed[13]. to oxyhaemoglobin leads to a longer T ∗, producing an fMRI image reconstruction results in the selected brain 2 increased image intensity. Such protocols typically use areadividedintothousandsofvoxels,eachassignedasignal magneticgradientstogeneratetheMRsignal(gradient-echo intensity.Statisticalanalysisofeachvoxelorclusterofvoxels imaging)and,asscannerandcomputationalhardwareshave ascertains whether the signal intensity there is greater than advanced, faster sequences have been developed to acquire comparedwithallothers.Whilstthisapproachmaybeuseful 4 ExperimentalDiabetesResearch for exploratory studies, most fMRI experiments perform that oral glucose ingestion produced a profound transient ROI (region of interest) analyses whereby the functional reduction in hypothalamic BOLD fMRI signal in lean properties of a predetermined set of voxels (corresponding subjects [18]. They alluded to the hypothalamic effects to a priori anatomical areas of the brain) are analysed. being mediated by neurohumoral factors following glucose ROI analysis, with very many fewer degrees of freedom ingestion,sincethehypothalamicsignalchangeprecededany than whole-brain voxelwise analysis, has a greater chance significantriseinbloodglucose. of finding statistically significant results but entails the In2000,Liuetal.describedanewtechniquecalledtime inherent risk of missing activated areas of the brain that clustering analysis (TCA), whereby the number of voxels were not included in the original hypothesis. Resting state reachingmaximumsignalintensityasafunctionoftimewas functional connectivity analyses investigate the activational examined [19]. In this way, TCA records the time windows interrelationships between different brain regions, allowing in which maximal brain responses occur and from which for the formation of distinct neuronal networks implicated ROI analysis can be appropriately extracted. By scanning a in a coordinated brain response [14]. Resting state scans 1cmmidsagittalslice,theywereabletofurtherconcentrate are performed when the subject is not performing any resolution on the hypothalamus and found that significant purposeful task and have recently reported on blood flow hypothalamic deactivation occurred 10 minutes following tothehypothalamusasanROIwithorwithoutinfusionof glucose ingestion. They also showed that the degree to anorecticguthormones[15]. whichthehypothalamicsignalreducedpostglucoseingestion Another commonly used approach in the area of func- negativelycorrelatedwithfastingplasmainsulinlevels. tional imaging of appetite is to present the subject with Using functional MRI in a 1cm midsagittal slice, images of food (with further subdivision into appetising, Smeetsandcolleaguesreportedonrestingstateintrasubject bland, high calorie, or low calorie food) or nonfood items. hypothalamicresponsebeforeandafteringestionofglucose. A subtraction analysis is performed to see whether the dif- In their original experiment, they also found a dose- ferenceinregionalbrainactivationbetweenviewingimages of food or nonfood is altered in different physiological dependent decrease in hypothalamic BOLD fMRI signal in lean,healthysubjectsshortlyafteringestingglucosesolution, conditions (e.g., the fasted state, the fed state, or in the whichlastedfor30minutes[20].Usingthesameexperimen- presenceofexogenouslyadministeredanorectichormones). talparadigm,theywentontoshowthathypothalamicBOLD Several studies have utilised this approach; these will be signal did not decrease following ingestion of an equally discussedindepthsubsequently.Hypothalamicdataremain sweet(butcaloriedeficient)solutionofaspartame,oringes- anelusivegoalwithsuchwholebraintask-basedanalyses;in additiontotheaforementioneddifficultiesinhypothalamic tionofanequallycalorific(butnotsweet)solutionofmal- todextrose.Theytherebyinferredthathypothalamicactivity imaging,activationofsomenucleiwithinthehypothalamus was specifically glucose-sensitive and that a hypothalamic promotes food intake, whereas other nuclei are appetite- responserequiredbothsweettasteandenergycontent[21]. inhibiting[16].Therefore,usingthehungry,fastedstateasan Interestingly, from a clinical perspective, they subsequently example,theoverallhypothalamicsignalresultingfromthe demonstrated that the reduction in hypothalamic BOLD activationoforexigenicnucleiandinhibitionofanorexigenic signalfollowingglucoseingestionwasabsentinpatientswith nuclei may in fact be negligible, due to the above opposing type2diabetes,theorisingthatinappropriatehypothalamic signals cancelling each other out. In addition to developing processing of nutrient availability may be involved in the strongermagneticfieldstoenhancespatialresolution,newer perfusion imaging techniques, which are T weighted, may aetiology of the disease [22]. Using the same hypothalamic 1 provide better hypothalamic data which is not so affected BOLD imaging technique as these previous studies, this group has also alluded to the effects of gut hormones on by signal loss due to magnetic susceptibility artefacts [17]. hypothalamicactivityfollowingglucoseingestion.Inhealthy Arterialspinlabelling(ASL)usesRFpulsestomagnetically men,despitecausingathreefoldlowerriseinbloodglucose labelprotonsinbloodwatermoleculesbeforetheyreachthe concentrationthanintravenousglucoseadministration,oral tissueofinterestwhichliesintheimagingplanefromwhere glucose ingestion led to a much more robust reduction dataisacquired. hypothalamic BOLD signal [23]. As well as the effect of taste, it was postulated that hormones such as GLP-1, 3.ImagingtheHypothalamicResponseto releasedinresponsetonutrientsenteringthegut,accounted GlucoseUsingMidsagittalSliceSelection for this greater hypothalamic response to oral rather than intravenousglucose. Earlystudiesthathaveconcentratedonhypothalamicsignal However,incontrasttotheabovefindings,arecentstudy change involved conventional T ∗-weighted gradient-echo 2 by Purnell et al. in 2011 did not detect any signal change MRI pulse sequences to functionally scan a sagittal plane in the hypothalamus in response to either glucose or fruc- through the hypothalamus continuously, acquiring images tose infusion in 9 healthy, normal-weight subjects, despite before- and after- a given stimulus. By concentrating all focussing their image acquisition parameters to optimise image processing ability on this thin, central slice, greater hypothalamic signal capture by mid-sagittal T weighting 1 attention can be devoted to a ROI corresponding to the [24]. The authors postulated that differential activation of hypothalamus,althoughclearlyattheexpenseofimagingthe excitatory and inhibitory nuclei within the hypothalamus restofthebrain.Inthismanner,Matsudaetal.demonstrated may have cancelled out any overall signal change. This ExperimentalDiabetesResearch 5 illustrates the ongoing limitations when attempting to of calorific content, food images caused greater activation visualisethehumanhypothalamusinfMRIstudies. than nonfood images in the amygdala, hippocampus, and ventromedial prefrontal cortex [30]. The authors point out the importance of these areas in the expectation and 4.fMRIStudiesinNormal-Weight evaluation of reward. High-calorie foods caused particular SubjectsComparingtheResponsetoFood activation of the medial and dorsolateral prefrontal cortex, CuesintheFastedandFedStates understood to be involved in evaluating stimulus relevance within the current affective state of the individual. In Before moving into the arena of functional brain imaging contrast, low calorie foods resulted in lesser activation studies to specificallydetermine the neuroendocrinology of of these reward areas per se and greater activation in appetite control, it is useful to look at earlier studies which somatosensory areas, with a suggestion from the authors examineddifferencesinbrainactivityinhungerandsatiety. that this was due to a lesser cephalic phase response by Such differences may be attributed to a number of physi- classical conditioning to images of less appealing foods. In ological differences between the fed and fasted state, such 2005,Killgorepublishedareanalysisofthisstudydata,using as changes in concentrations of gut-derived neuropeptides an ROI approach (specifically, a subanalysis of the OFC) (PYY, GLP-1, ghrelin, CCK, and insulin to name but a and correlated BOLD activity with BMI. It was reported few) and vagal afferents conveying information about gut that,forthehighcalorieminusnonfoodcontrast,therewas distension. These early studies into appetitive behaviour a significant negative correlation between BMI and BOLD have led to our understanding of how images of food can signal in the OFC [31]. In other words, as BMI increases, trigger the brain’s reward system and how the motivational activity in the OFC becomes less food responsive. This potency of this trigger is greater in the fasted state. On highlights the more subtle role of parts of the OFC not the whole, these studies complemented earlier work which justinsimplyassigningrewardvaluetocertainstimuli,but utilised PET scanning to assess the effects of hunger and in modifying stimulus-reward associations and accordingly satietyonregionalcerebralbloodflowdifferencesandwhich redirecting feeding-related behaviour in response to new particularly implicated the OFC as a critical convergence learning [32]. Of course,in this particular study, it wasnot zone for sensory information related to rewarding stimuli possible to extend these findings to the obese population, [25–28]. sincethesubjectsstudiedwereallwithinthenormalweight Most of these fMRI studies have been task-based sub- range. traction analyses, looking at differences in regional brain In2005,St-Ongeetal.publishedtheonlyfMRIstudyto activationbetweenwhenviewingimagesoffoodornonfood datewherefastedindividualswereexposedtofourdifferent items. It is well understood that, whilst taste provides an stimuli:visualfood,visualnonfood,tactilefood,andtactile immediatereward(orpunishment)forconsumedfoods,the nonfood. In an uncorrected whole brain analysis, they visualcharacteristicsoffoodarequicklylearnedandbecome reportedthattheanteriorcingulate,superiortemporalgyrus, powerful secondary reinforcers, capable of influencing sub- hippocampus, and insula were significantly activated to a sequentfood-seekingbehaviour.Thus,showingfoodimages greaterextentduringthepresentationoffoods(whetherseen isausefulwayofexaminingtheappetitiverewardcircuitry. orfelt)overnonfooditems[33]. Inthefirststudyofitskindin2001,LaBaretal.reported In 2006, Porubska´ et al. published the results of an on9healthysubjects,whounderwentaninitialwhole-brain fMRIstudyof12normal-weightfastedsubjects.Visualfood fMRI after an 8 hour fast, followed by another postmeal stimuli(incontrasttononfoodimages)activatedtheinsular scan one hour later. A follow-up study was performed on and orbitofrontal cortices, with a positive modulation of 8 subjects who were fed a meal before the first of the two insular activity induced by subjective ratings of appetite scans to rule out the potential of habituation effects in [34]. These findings are consistent with the recognition the first group. The subjects were presented with food and of the insula as being an important region in establishing nonfood images during each scan and subtraction analysis salience [35]. Following on from this, Fu¨hrer et al. in 2008 was performed [29]. The authors’ hypothesis that variation studied12healthymalevolunteersundergoingtwoseparate in the state of hunger would modulate the response of the scanningsessions—onewhenfastedovernightandtheother amygdala and anatomically related corticolimbic structures immediately after a large meal. They performed a whole was borne out in a ROI analysis, which revealed satiety- brain, uncorrected analysis of the data and particularly induced reduction in activity in the amygdala, parahip- noted significantly enhanced activity within the OFC when pocampal gyrus, and fusiform gyrus. They suggested that hungry, again with reference to our understanding of this theamygdala,withitsextensiveneuralconnectionswiththe areaassubjectifyingtheperceivedpleasantnessoffood[36]. hypothalamus and higher brain centres, was in a pivotal We have also recently demonstrated that feeding reduces positionforintegratingresponsetovisualfoodstimuliinthe the difference in BOLD signal between viewing images of contextofnutritionalstatus.However,itwasnotedthatthis food and nonfood in several brain regions (Figure 3), but study was underpowered to detect similar satiety-induced significantly so in the insula [37]. Furthermore, in 2009, changesinmanyotherROIsstudied. Goldstone et al. reported the findings from a fMRI study In 2003, Killgore et al. used a different approach by of twenty individuals in both the fed and fasted state; scanningsatiatedrecruitswhowerepresentedwithpictures subjects viewed pictures of high calorie, low calorie, and ofhighcalorie,lowcalorie,andnonfooditems.Irrespective nonfooditemswhilstratingtheappealvalueoftheseimages. 6 ExperimentalDiabetesResearch Putamen 5.DifferingfMRIResponsesinObesity Bearing in mind the results of the above studies in normal weight individuals, it is now interesting to turn to the Insula resultsofsimilarstudiescomparingobeseandnormalweight subjects. As discussed above, in 1999, Matsuda et al. studied the hypothalamic response before and after oral glucose ingestion.Althoughtheyshowedthatoralglucoseingestion produced a profound transient reduction in hypothalamic signalin10leansubjects,thisresponsewasbothsignificantly delayed and attenuated in 10 obese individuals. More tellingly,itwasshownthat,followingglucoseingestion,the time taken to reach the maximum hypothalamic inhibitory Orbitofrontal responsewascloselycorrelatedtofastingplasmaglucoseand cortex insulin levels (which, as expected, were significantly higher in the obese group) rather than BMI per se, suggesting hypothalamicdysfunctionaseitherthecauseorconsequence ofinsulinresistance[18]. To extend the above findings to the study of reward regions,alaterstudyincorporatingvisualfoodandnonfood cuemethodologyrevealedthatBOLDfMRIactivationinthe righthippocampus(withrespecttohighcaloriefoodimages Figure3:Modulationofneuronalactivityinthefedversusfasted versusotherimages)correlatedpositivelywithfastinginsulin state. Representative whole-brain fMRI sections showing regions levels and waist circumference (but not BMI) in 12 normal wherethedifferenceinBOLDsignalbetweenviewingfoodimages weight and 12 obese adolescents who had all been fed a andnonfoodimagesisbluntedinthefedstatecomparedwiththe standard breakfast a few hours prior to scanning [42]. On fastedstate.Unpublishedimagefrom[37]. the other hand, in that study, there wassignificant negative correlation between activation of the medial right superior Theyfoundthatwhenfasted,therewassignificantlygreater frontal gyrus and the left thalamus (again, with respect to activation to high calorie over low calorie food items in the contrast between high calorie food images versus other theventralstriatum(importantinmediatinghedonicdrive images) with fasting insulin levels. The authors postulated andaction[38]),amygdala,anteriorinsula,andOFC.They thattheirresultsreflectedapermissiveroleofinsulininthe foundthathighcaloriefoodswereconsistentlyratedasmore controlofeatingbehaviourbythehippocampus.Wefurther appealing, that this was augmented with fasting and that speculatethat,inthosewithalargerwaistcircumferenceand the increase in appeal rating bias for high calorie over low high circulating levels of insulin (i.e., peripherally insulin calorie foods in the fasted state was positively correlated resistantindividuals),adegreeofcentralinsulinresistancein withactivityintheOFC[39].Inthesameyear,Schuretal. the hippocampus may explain the greater activation in this performed a study of ten normal weight, fed subjects region,despitehigherinsulinlevels. viewing images of fattening food, nonfattening food, and Studiesinvestigatingthedifferentialrewardsystemacti- nonfood items. These food images were specifically chosen vationpatternswhenlookingathighcaloriefoodsbetween based on whether the food was perceived to be compatible obese and normal weight subjects have shown significantly with an effort to lose weight. In a ROI analysis, this was greateractivationintheobesegroupinseveralbrainregions the first study to report increased hypothalamic activation implicated in food reward. In a study by Rothemund et al. when viewing pictures of fattening food compared with in 2007 of 13 obese and 13 normal weight women who nonfooditems,althoughthisfindingdidnotextendtoother had been fasted for at least 90 minutes, increasing BMI comparisons (i.e., fattening versus nonfattening or all food positivelypredictedBOLDactivationofthedorsalstriatum versus nonfood). In concordance with other studies, they (caudate/putamen), anterior insula, claustrum, posterior also found increased activation in the amygdala,insula and cingulate, postcentral cortex, and lateral OFC. This specifi- OFCwhenviewingfatteningfoodscomparedwithnonfood callypertainedtothecontrastbetweenviewinghighcalorie items[40].Alsoinagreementwithpreviousfindingswerethe food images versus nonfood images [43]. A further ROI- resultsofSiepetal.,whoin2009reportedincreasedactivity based study by Stoeckel et al. in 12 obese and 12 normal intheamygdalaandOFCinresponsetoviewinghighcalorie weight women demonstrated that pictures of high calorie versuslowcaloriefoodimages,butonlywhentheirsubjects foods (versus nonfood) produced greater activation in were fasted [41]. Their experimental protocol allowed for the obese group compared with controls in several brain the further observation that this increased activity in the regions, including the ventral and dorsal striatum, insula, amygdala and OFC was only evident when participants anterior cingulate cortex, amygdala, OFC, hippocampus, explicitlyevaluatedfoodsbutnotwhentheirconcentration ventral pallidum, and medial prefrontal cortex [44]. A wasdivertedelsewhere. furtherfunctionalconnectivityanalysisindicatedthatobese ExperimentalDiabetesResearch 7 women displayed a relative deficiency in the amygdala’s characterisation, 8 women who showed >2.5% increase in modulation of the OFC and ventral striatum, along with BMIovera6-monthperiodshowedreducedstriatalBOLD excessive modulation of the ventral striatum by the OFC fMRIresponsetochocolatemilkshakeconsumption(versus [45].Collectively,resultsfromtheabovestudiessuggestthat a tasteless control solution) compared with 12 women who a hyperreactive reward system to high calorie food cues is demonstrated stable weight [49]. These results reinforced involvedinthepathophysiologyofobesity. the authors’ hypothesis that weight gain may be associated Fascinatingly, several studies have shown that obese withreducedsensitivityofstriatalrewardcircuitry(probably individuals respond differently to satiation compared with due to the downregulation of dopamine D2 receptors), normal weight individuals, which allows us to hypothesize which may be a fundamental mechanism responsible for that, in obesity, there is a degree of dysregulation in CNS overeating. appetite regions, which may contribute to overeating and Differences between normal weight and obese indi- henceweightgain. viduals have been replicated in studies of children; in A ROI-based study by Martin et al. measured fMRI one such investigation of 10 obese and 10 normal weight BOLD activation pre- and postprandially (following a 500 children,inthefastedstate,theobesegroupshowedgreater kCalmeal)inresponsetopicturesoffood(acombinationof BOLDactivation(whenviewingfoodimagesversusnonfood highcalorieandlowcalorieimages)andnonfoodin10obese images)intheprefrontalcortexcomparedwiththenormal and10healthyweightadults.Duringthepremealcondition, weightgroup.Followingameal,thereductioninactivation obese subjects showed increased activation in the anterior oftheprefrontalcortexandnucleusaccumbenswasblunted cingulateandmedialprefrontalcortex,comparedtohealthy intheobesegroup.Furthermore,thepostmealactivationof weightcontrols[46],thusconcurringwiththeobservations theOFC(tofoodversusnonfoodimages)wasgreaterinthe of Stoeckel et al. During the postmeal condition, obese obesegroupcomparedwiththenormalweightgroup[50]. participants also showed greater BOLD activation in the medial prefrontal cortex compared with healthy weight 6.fMRIStudiesofHormonesImplicatedin controls.AfurtherrecentstudybyDimitropoulosetal.of22 AppetiteControl overweight/obesemalesand16normalweightmalesshowed thatbeforeeating,obesesubjectsshowedgreaterresponseto Having discussed the first generation of fMRI studies foodimages(versusnonfoodimages)comparedwithnormal exploringthephysiologicalresponsestofeeding,followedby weight subjects in the anterior prefrontal regions [47]. studies investigating alteration of such responses in obese Postprandially, the obese group demonstrated increased individuals, we will now direct our focus to research that response to all food images (versus nonfood) compared hasinvestigatedtheeffectsofappetitemodulatinghormones with normal weight individuals in frontal, temporal, and derivedfromthegastrointestinaltractandadiposetissueon limbic regions. Specific greater activation to high calorie thefMRIneuronalresponse.Asalludedtoearlier,suchhor- foods was seen in the obese group compared with normal mones are thought to be key mediators of both short-term weightindividualsinthelateralOFC,caudate,andanterior and longer-term energy balance; in the postprandial state, cingulatecortex[47]. anorectic hormones (including PYY, GLP-1, and insulin) Aninterestinghypothesistoexplainabnormaleatingin predominate,andinthelonger-term,high-circulatinglevels obesity was postulated by Stice et al., following an fMRI of leptin and insulin correlate with nutritional status. In studyin2008.Here,acomparisonwasmadebetweenbrain contrast,inthefasted,hungrystate,theorexigenichormone activity in 7 obese and 11 lean adolescent girls during the ghrelin is dominant amongst gut hormones. It is therefore anticipated receipt of chocolate milkshake, during actual interestingtoobservetheeffectsofexogenousadministration receipt of the milkshake, during anticipated receipt of a ofsuchhormonesonthefMRIresponseandcomparethese tasteless control solution, and during actual receipt of the results with outcomes from earlier studies comparing the tasteless control solution. In response to anticipated receipt physiological fasted and fed states. It is worth noting that, and actual receipt of the chocolate milkshake (versus the thusfar,themajorityoffMRIstudiescouplingtheexogenous tasteless solution), obese adolescent girls showed greater administration of appetite modulating hormones (with the BOLD activation bilaterally in the anterior and mid insula, exception of studies of leptin) have been conducted in frontal operculum, parietal operculum, and rolandic oper- healthy, normal weight subjects, in order to investigate the culum compared with lean adolescent girls. However, the actionofsuchhormonesinnormalphysiology. obesegirlsdisplayedlowerBOLDactivationinthecaudatein responsetoactualconsumptionofthemilkshake(versusthe tastelesssolution)comparedwiththeleangirls.Fromthese 6.1. PYY and GLP-1. PYY and GLP-1 are anorectic hor- results, the authors suggested that obese individuals show mones, released by the enteroendocrine L cells of the gut greater salience-associated responses from anticipated food following a meal; they lead to marked inhibition of food consumption in gustatory and somatosensory regions, but intakewhenadministeredtofastedhumansubjects[51–53]. weakeractivationinthecaudateduringactualconsumption In healthy human subjects, following intravenous infusion compared to lean individuals [48]. They further postulated of PYY3−36, there was not only reduced food intake, but that this reflects increased anticipatory food reward (but increasedactivityinthehypothalamusandOFC,asassessed reducedconsummatoryfoodreward)inobesity,whichmay byBOLDfMRI[15].Thisstudywasconductedwithsubjects contribute to overeating. In a further study, after baseline at rest, without engagement in a visual food-cue-based 8 ExperimentalDiabetesResearch task. Furthermore, during saline infusion visits, subjects’ 6.3.Insulin. Insulinhasacentralanorecticeffectinnormal caloricintakecorrelatedpositivelywithBOLDsignalchange physiology [56]. This must be distinguished from the in the hypothalamus, whereas this switched to a negative secondary effect of hunger resulting from hypoglycaemia correlationbetweencaloricintakeandOFCsignalonstudy that may follow the exogenous administration of insulin visits when PYY3−36 was infused. It was postulated that the (theassociatedovercompensatoryeatingmayprovidesome presenceofPYY3−36switchedregulationoffoodintakefrom explanation as to why patients with diabetes treated with a homeostatic brain region (hypothalamus) to a hedonic insulin tend to gain weight). Therefore, human studies region(OFC). investigatingthedirectphysiologicalactionofinsulininthe Using a visual food-cue-activated BOLD fMRI method, CNS have had to overcome the hurdle of maintaining eug- we recently investigated the effects of PYY3−36 and GLP- lycaemia during insulin administration, for otherwise, it is 17−36amide onneuronalactivityinbrainregionscontributing theeffectofhypoglycaemia(ratherthantheeffectofinsulin to appetitive processing and behaviour [37]. We scanned per se) that would manifest. Furthermore, to effectively 15 healthy, lean subjects on five separate occasions, during studytheactionofinsulinintheCNS,peripheraleffectsof which they each received the following interventions in the hormone should ideally be eliminated, hence even the randomised order after an overnight fast: a saline infusion; use of a euglycaemic hyperinsulinaemic clamp technique is a saline infusion following a large breakfast; an infusion unsatisfactoryinthisregard.Onegrouphasusedintranasal of PYY3−36; an infusion of GLP-1; a combined infusion of insulin administration as a method of overcoming this PreYdYu3c−e3d6 +adGlLibPi-t1u.mCoemnebrignyedinintafkuesioant oafstuhbesgeuqtuehnotrmbuonffeest technical difficulty, demonstrating that insulin resulted in reducedBOLDfMRIactivation(withrespecttofoodimages meal to a similar degree as on the day when subjects versus nonfood images) in 9 healthy, normal-weight adults received breakfast without either gut hormone. Notably, comparedwithplacebointherightandleftfusiformgyrus, the BOLD fMRI signal change (comparing exposure to food images versus nonfood images) in several a priori therighthippocampus,righttemporalsuperiorcortex,and brainROIswasreducedfollowingconsumptionofbreakfast the right frontal middle cortex [57]. Notably, peripheral (Figure 3) and also when the gut hormones were infused plasma glucose did not change following the intranasal in the fasted state. The largest reduction in BOLD signal insulin administration, thus validating their technique of occurred in the insula with combined administration of observingthecentraleffectsofinsulin. PYY3−36andGLP-1,althoughwedetectedsmallerreductions following feeding and gut hormone administration (either 6.4. Leptin. Leptin-deficient individuals are obese and singly or in combination) in the striatum and OFC. Of hyperphagic; replacement of leptin in these individuals note,ourimageacquisitionparameterswerenotdesignedto reduces food intake [58]. However, the well-recognised optimisehypothalamiccaptureinfavourofrewardregions. phenomenon of leptin resistance in obesity has precluded Furthermore,whencomparingourresultsfortheOFCwith itsdevelopmentfromatherapeuticviewpoint.fMRIstudies thoseofBatterhametal.,itmustberememberedthatadirect comparison between studies is difficult due to the method- ofleptinadministrationareneverthelessinterestingtoshed ological differences between them. Nevertheless, our study furtherlightonthephysiologicalregulationofappetite. In 2 congenitally leptin deficient human subjects, daily concluded that combined administration of the anorectic subcutaneousleptinreplacementreducedBOLDfMRIacti- gwuats ahsosromcioatneedswPiYthY3b−r3a6inanacdtivGaLtiPo-n1antodfsausbtesdequinednitveidautianlgs vation(intermsofthedifferenceinactivationbetweenview- behaviour changes similar to those after eating a full meal, ing food and nonfood images) in the nucleus accumbens- thereby reinforcing the view that PYY and GLP-1 are key caudate and putamen-globus pallidus regions compared mediatorsofpostprandialsatiety. withthecontrol,leptin-deficientstate.Furthermore,before leptin treatment, activation in the nucleus accumbens- caudate correlated positively with liking of food images in 6.2. Ghrelin. Ghrelin is a potent orexigenic gut hormone boththefedandfastedstates,whereas,afterleptintreatment, released from the stomach in response to fasting; adminis- activationinthenucleusaccumbens-caudatecorrelatedpos- tration of ghrelin to humans increases food intake [54]. A itivelywithlikingoffoodimagesonlyinthefastedstate[59]. study by Malik et al. demonstrated that in normal-weight Anotherstudythatyeardemonstratedthat,in3congenitally human subjects, intravenous ghrelin infusion increased leptin deficient human subjects, daily subcutaneous leptin hunger ratings and also increased the BOLD fMRI signal replacementreducedthedifferenceinBOLDfMRIactivation change (between food and nonfood visual cues) compared (betweenviewinghigh-calorieandlow-caloriefoodimages) with saline infusion in the amygdala, OFC, insula, visual in regions implicated in the insula, temporal and parietal areas, and striatum [55]. The fMRI methodology used in cortex compared with when subjects were not receiving this study was similar to that used in our PYY and GLP-1 leptin.Ontheotherhand,leptinreplacementincreasedthe investigation, so it is interesting to compare the results; differenceinBOLDfMRIactivation(betweenviewinghigh- whereas the anorectic hormones PYY and GLP-1 reduced calorieandlow-caloriefoodimages)intheprefrontalcortex BOLDsignalinrewardregions,theactionoftheorexigenic comparedwithwhensubjectswerenotreceivingleptin[60]. hormoneghrelindidthereverse,presumablyinkeepingwith Theauthorspostulatedthattheinsula,temporalcortex,and theopposingphysiologicalrolesofthesepeptides. parietalcortexarehungerregionsandthereforeareinhibited ExperimentalDiabetesResearch 9 Decreased activity: fed state, low-calorie food cues, anorectic gut hormones Increased activity: high- calorie food cues, obesity OFC VC H Adiposity signals Gut hormones Leptin Ghrelin Adiponectin PYY Insulin GLP-1 Figure4:Modulationoftherewardingaspectsoffood.Inthishighlysimplifiedschematic,theorbitofrontalcortex(OFC)ishighlighted asthemostimportanthubintherewardencodingnetwork.Externalfoodcues,viathevisualcortex(VC),modulatetheOFCresponse, withincreasedactivityseeninfastedandobesepatientsandinresponsetohigh-versuslow-caloriefoods.OFCactivityisalsothoughtto bemodulatedbyinputsfromthehypothalamus,whichsensesinternalinformationaboutnutritionalstatusintheformofadipositysignals (such as leptin, which gives information about longer-term energy stores) and gut hormones (which are meal dependent and therefore giveinformationaboutshorter-termnutrientavailability).Anorectic(postprandial)guthormones,suchasPYYandGLP-1,attenuateOFC activityand,infastedindividualsinduce,anOFCresponsetovisualfoodcuesmoresimilartothatmeasuredwhenfed.Conversely,the orexigenichormoneghrelinupregulatesrewardcentreactivity. byleptin,whereastheprefrontalcortexisasatietyregionand 7.ImagingbeforeandafterGastric isthusinhibitedbyweightlossandactivatedbyleptin. BypassSurgery Following this, a study by Rosenbaum et al. in 2008 showed that in 6 leptin replete obese human subjects who Themosteffectivetreatmentforobesityavailableatpresent achieved an initial 10% weight loss (between 36 and 62 is Roux-en-Y gastric bypass surgery, leading to sustained days) through dietary means, there was increased BOLD weight loss of approximately 30% and often a very prompt fMRIsignaldifference(betweencomparingfoodimagesand resolutionoftype2diabetes[62]. nonfoodimages)inseveralregionsincludingthebrainstem Although the mechanisms behind the weight loss seen and parahippocampal gyrus compared with the baseline following Roux-en-Y bypass surgery remain to be fully state before weight loss. These changes were reversed by 5 elucidated, they are unlikely to be solely due to restrictive weeks of subcutaneous leptin administration. Interestingly, ormalabsorptivemeansalone.Roux-en-Yresultsingreater other regions, including the hypothalamus and cingulate weight loss than adjustable gastric banding, and this is gyrus responded in the opposite direction following weight thoughttobeduetoanumberofadditionalneurohormonal loss, but these changes remained sensitive to reversal by 5 causes [63]. These include elevated postoperative levels of weeksofsubcutaneousleptinadministration[61].Thisstudy endogenousanorecticguthormonessuchasPYYandGLP-1 reinforcedtheideathatweightlossleadstoastateofrelative leptindeficiency,whichmayberesponsiblefortherebound [64–66],increasesinenergyexpenditureandmetabolicrate hyperphagiaandsubsequentweightgaininmanydieters. [67], changes in taste preference away from calorie dense Figure4showsasimplifiedschematicofthemodulation foods[68,69],increasedbileaciddeliverytotheileum[70], of appetite centres by peripheral signals via the hypothala- andchangesingutmicrobiota[71].Recentstudiesimaging musandinobesity. appetiteinpatientsbeforeandafterweightlossareimportant 10 ExperimentalDiabetesResearch in helping us to understand some of the CNS-mediated regions implicated in reward processing when food cues effects. (and in particular high calorie food cues) are presented to In 2011, Van De Sande-Lee performed fMRI studies on subjects. Obese subjects seem to show heightened response 8 lean control subjects and 13 obese patients before- and to such food cues, which may fundamentally contribute to after- Roux-en-Y gastric bypass, by which an average 30% increased wanting of food (and thus increased subsequent reduction in body weight had been achieved [72]. They foodintake)inobeseindividuals.ThefMRIdataalsoseems replicated the TCA experiment described earlier [19]. In to be in agreement that the consumption of a meal (or addition,theyalsoperformedafunctionalconnectivityanal- othersurrogatemarkersoffeeding,suchasariseinglucose) ysis. Their TCA findings mirrored that found previously— tends to deactivate key appetite regions in normal weight namely that the reduction in hypothalamic signal seen 5– individualsand,furthermore,thatthissatietyresponsemay 10 minutes following glucose ingestion in normal controls also be abnormal in those with obesity. Further studies wasdiminishedintheobese(preoperative)group.However, shouldleadtoconsolidationofexistingconceptsontherole a recovery of this impaired hypothalamic response was of individual brain regions and their interconnections in seen postoperatively when the obese group had lost weight physiologicalandpathologicalsatietyresponses. following gastric bypass surgery. Interestingly, they also Building on this fundamental understanding of the reported,inleansubjects,ahighleveloffunctionalconnec- change in BOLD signal between the fasted and fed states, tivity between the hypothalamus and the orbitofrontal and recentstudieshavesuggestedthat,ingeneral,orexigenicsig- somatosensorycortices.Thiswasnotseeninobesesubjects, nals,whichareinabundanceinthefastedstate,increasethe but postoperative weight loss was reported to re-establish BOLD response in brain reward regions, whereas anorectic theseconnectivitypatterns[72]. signals,whichpredominateinthefedstate,correspondingly In the same year, a further ROI-based study of 10 deactivate the same regions. Subtle differences in brain obese female patients 1 month before- and after- Roux- responses between these studies may reflect the slightly en-Y gastric bypass, there was attenuation of fMRI BOLD different roles of the different hormonal signals in short- signal (when viewing high-calorie versus low-calorie food term versus longer-tem energy homeostasis and also may images) in the ventral tegmental area, ventral striatum, reflect differential action on various appetite pathways. putamen, lentiform nucleus, posterior cingulated, and dor- Nevertheless, studies in obese patients, assessing the effect somedial prefrontal cortex following the bypass procedure ofgastricbypasssurgeryonthefMRIBOLDresponse,have [73]. In this study, hunger was also reduced after surgery. furtherconcurredwiththeresultsofstudiesinvestigatingthe Althoughcirculatingguthormonelevelswerenotmeasured, BOLD response following the exogenous administration of the authors speculated that postsurgical reductions in the orexigenic and anorectic hormones, thus consolidating the orexigenic hormone ghrelin could have explained their overallbodyofworkinthearea. imaging observations. This would certainly be consistent AlthoughfMRIisatpresentusedonlyasaresearchtool withthepreviousfMRIfindingsbyMaliketal.[55]following to study appetite, the overall goal in the future will be to exogenous ghrelin administration. Equally, it is possible translate this research into the clinical arena. An obvious that elevated levels of PYY and GLP-1 after-bypass would role would be to help develop pharmacological treatments explain the imaging findings, consistent with our recent for obesity, by means of assessing the fMRI response to fMRI findings [37] following exogenous administration of administration of novel agents in obese individuals. Alter- theseanorectichormones.Indeed,theelevationinPYYand natively, we may envisage the use of fMRI as a tool to help GLP-1 levels achieved in our study following exogenous predict response to gastric bypass surgery. To aid effective administration of the hormones wassimilar to the elevated comparison of results, it would be enormously useful to endogenous levels of PYY and GLP-1 observed in previous strive towards greater consistency between studies, particu- studiesfollowinggastricbypasssurgery[65].Thesefindings larly with regard to experimental design, image acquisition arestriking,giventheneedtodevelopnonsurgicalsolutions and image analysis. The studies outlined in this paper have forthetreatmentofobesity.Todate,allcentrallyactingdrugs establishedkeyconcepts,anditisexpectedthatthecoming developed to treat obesity have failed due to cardiovascular decade should see consolidation of these findings. We are and mood altering side effects, as a result of actions fortunatethatfunctionalMRIstudieshaveallowedscientists outside pathways solely governing appetite. Therefore, it is to investigate for the first time in humans, the intricacies foreseeablethat,inthefuture,elevatingcirculatinglevelsof of appetitive processing. As this technology continues to PYY and GLP-1 (by means of exogenous administration) evolve,itwillremainanimportanttoolforinvestigatingbasic to similar levels as found naturally after a gastric bypass physiologyaswellasevaluatingnewtherapies. proceduremayrepresentanovelpharmacologicalmethodof treatingobesity. ConflictofInterests 8.Conclusion Theauthorsdeclarethatthereisnoconflictofinterests. The use of fMRI to study appetite control in humans is a rapidly evolving field. There appears to be a degree of Authors’Contribution consistency between studies, suggesting that, in the fasted state, greater BOLD activation is seen in several brain A.DeSilvaandV.Salemcontributedequallytothiswork.

Description:
Later studies have investigated the action of appetite modulating hormones on the. fMRI signal, showing appetite (and subsequent food intake) will aid the manip- ulation of 3, Article ID dmp051, pp. 276–292 . 400–409, 2008.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.