Distinctive Frontal and Occipitotemporal Surface Features in Neglectful Parenting
Abstract
Published: 18 March 2021
Full text
BrainSci.2021,11,387.https://doi.org/10.3390/brainsci11030387www.mdpi.com/journal/brainsci Article DistinctiveFrontalandOccipitotemporalSurfaceFeatures inNeglectfulParenting InmaculadaLeón 1,2, *,MaríaJoséRodrigo 1,2 ,IleanaQuiñones 3 ,JuanAndrésHernández‐Cabrera 1,2 andLornaGarcía‐Pentón 4 1 InstitutoUniversitariodeNeurociencia,UniversidaddeLaLaguna,38200LaLaguna,Spain; [email protected](M.J.R.);[email protected](J.A.H.‐C.) 2 FacultaddePsicología,UniversidaddeLaLaguna,38200LaLaguna,Spain 3 BasqueCenteronCognition,Brain,andLanguage;20009Donostia‐SanSebastián,Spain;i.quin[email protected] 4 MRCCognition&BrainSciencesUnit,UniversityofCambridge,CambridgeCB27EF,UK;[email protected] *Correspondence:[email protected] Abstract:Althoughthebrainsignaturesofadaptivehumanparentingarewelldocumented,the corticalfeaturesassociatedwithmaladaptivecaregivingareunderexplored. Weinvestigated whethercorticalthicknessandsurfaceareavaryinasmallgroupofmotherswhohadneglected theirchildren(24intheneglectgroup,NG)comparedtoacontrolgroupofmotherswithnon‐ne‐ glectfulcaregiving(21inthecontrolgroup,CG).Wealsotestedwhetherthecorticaldifferences wererelatedtodyadicmother‐childemotionalavailability(EA)inaplaytaskwiththeirchildren andwhetheralexithymiainvolvinglowemotionalawarenessthatcharacterizestheNGcouldplay aroleinthecortical‐EAassociations.Whole‐brainanalysisofthecorticalmantleidentifiedreduced corticalthicknessintherightrostralmiddlefrontalgyrusandanincreasedsurfaceareaintheright lingualandlateraloccipitalcorticesfortheNGwithrespecttotheCG.Follow‐uppathanalysis showeddirecteffectsoftherightrostralmiddlefrontalgyrus(RMFG)ontheemotionalavailability (EA)andonthedifficultytoidentifyfeelings(alexithymiafactor),withamarginalindirectRMFG‐ EAeffectthroughthisfactor.Thesepreliminaryfindingsextendexistingworkbyimplicatingdif‐ ferencesincorticalfeaturesassociatedwithneglectfulparentingandrelevanttomother‐childinter‐ activebonding. Keywords:neglectfulparenting;corticalthickness;surfacearea;alexithymia;mother‐childinteraction 1.Introduction Althoughmostparentscanrespondappropriatelytotheirchild’sneeds,extremecases ofinsensitivecaregivingdoexist.Infact,neglectisthemostcommonformofchildmal‐ treatment,whichinvolvesadrasticfailuretoprovidethechildwithfood,clothing,shelter, medicalcare,supervision,oremotionalsupportthatplacesthechild’ssafetyatrisk[1,2]. Beingseverelyneglectedentailsnegativebehavioral,neurobiological,andclinicalconse‐ quencesforthechild[3,4].Althoughthebrainsignaturesofadaptiveparentingarewell documented(e.g.,[5]),theneuralmechanismsassociatedwithneglectfulcaregivingare comparativelylessstudied.Nevertheless,identifyingthecorticalfeatureslinkedtoneglect‐ fulbehaviorcouldberelevanttobetterunderstandingthecycleofmaladaptiveparenting. Tothisend,thisstudyseekstoadvancework onmaladaptiveparentingbyinvesti‐ gatingwhethercorticalfeaturesvaryinmotherswhoneglecttheirchildren. Specifically, weaddressgapsinknowledgeintwodirections.First,weidentifydifferencesincortical thicknessandsurfaceareabetweenagroupofthosemotherswhohadneglectedtheir children,aneglectgroup(NG),andacontrolgroupofmothers(CG)withnon‐neglectful caregiving.Second,weexaminetowhatextentthosecorticalfeaturespotentiallyassoci‐ Citation:León,I.; Rodrigo,M.J.;Quiñones,I.; Hernández‐Cabrera,J.A.; García‐Pentón,L.DistinctiveFrontal andOccipitotemporalSurface FeaturesinNeglectfulParenting. BrainSci.2021,11,387. https://doi.org/10.3390/ b rainsci11030387 AcademicEditor:PaolaVenuti Received:4February2021 Accepted:13March2021 Published:18March2021 Publisher’sNote:MDPIstaysneu‐ tralwithregardtojurisdictional claimsinpublishedmapsandinsti‐ tutionalaffiliations. Copyright:©2021bytheauthors.Li‐ censeeMDPI,Basel,Switzerland. Thisarticleisanopenaccessarticle distributedunderthetermsandcon‐ ditionsoftheCreativeCommonsAt‐ tribution(CCBY)license(http://crea‐ tivecommons.org/licenses/by/4.0/).
BrainSci.2021,11,3872of15 atedwiththeNGwererelatedtothequalityofamother‐childinteractiveplaytask(cor‐ tical–emotionalavailabilityassociations),andwedeterminewhetheralexithymiacould playaroleintheseassociations. 1.1.NeuroanatomicalCorrelatesofNeglectfulCaregiving Previousneuroimagingresearchonmothersshowingneglectfulcaregivinghasiden‐ tifiedfunctionalandstructuralvariationsmostlyinfrontal,cingulate,andoccipitotem‐ poralareas.Inafunctionalmagneticresonanceimage(fMRI)study[6],motherswhone‐ glectedtheirchildrenversuscontrolmothersshowedreducedactivationsinresponseto infantandadultcryingfaces,bothinafrontalcorticalarea(inferiorfrontalgyrus)aswell asoccipitotemporalcorticalareas(lingualandfusiform).Inthesamestudy,motherswith neglectfulbehavioralsoshowedanattenuatedactivationtoinfantcryingfacesinleftmid‐ dlefrontalandanteriorcingulateareas,underscoringtheirspecificdifficultieswithre‐ spondingtotheinfant’semotionalcues. Inavoxel‐basedmorphometricstudy[7],grey mattervolumereductionsintherightinsula,anterior/middlecingulate,andinferior/mid‐ dlefrontalgyrus,andgreatervolumeintherightfusiformandcerebellum,wereobtained formotherswithneglectfulbehaviorascomparedtocontrolmothers. Neuroimagingresearchhasalsodocumentedthecriticalroleofregionssuchasthe medialandlateralprefrontalcortex(PFC)andanteriorcingulatecortex(ACC)inemotion regulation[8,9].Emotionregulatorycapacitiesarevitalforsupportingparents’abilityto beperceptive,responsive,andflexibleinthefaceofchallengingchildbehavior[10,11]. Likewise,occipitotemporalvisualareasspecializedintheprocessingoffacesarecrucially involvedinsensitiveparenting[5,12]. Moreover,theseareashaveshowntheirinvolve‐ mentinthealteredprocessingofachild’scryinginmotherswhoneglecttheirchildren, affectingthecaregiver’sabilitytobeattunedtotheinfant’sneeds[6,13].However,lessis knownaboutthepotentialroleofcorticalthinningandsurfaceareainthespecificneural underpinningsofbothemotionregulationandface‐responsiveareasinthecontextofne‐ glectfulcaregiving. Corticalthickness(CT)andcorticalsurfacearea(SA)aremorphologicalfeaturesof thecerebralcortexthatfollowdifferentgenetic,phylogenetic,anddevelopmentalroutes, capturingtheanatomicalandfunctionalvariabilityinthebrain[14].Studiesinnormal motheringhaveshownthatthebrainundergoeschangesincorticalthickness,withdif‐ ferencesobservedbeforeandafterpregnancy[15,16]andinprimiparousmothersversus non‐mothers[17]. Thereisalsoevidenceofcorticalvariationsinthematernalbrainina groupoffirst‐timemothers,showingthatalaterpostpartummonthwasassociatedwith increasedcorticalthicknessintheprefrontalcortex,thelateraloccipitalareas,andthefu‐ siformgyri[18],suggestingbrainstructuraladaptationstothechild’sevolvingneeds. Fromacomplementaryperspectivethatcomparesextremevariationsinmaternalcare‐ giving,thecurrentstudyexaminedpossibledifferencesincorticalfeatures(CTandSA) inthosemotherswhohadneglectedtheirchildversuscontrolmothers,andwhetherthese corticaldifferencesarerelatedtoobservedmother‐childinteractivebehavior. 1.2.RelationsbetweenBrainFeatures,PersonalityTraitsandQualityofMother‐Child Interaction Asfortheseconddirectiontakeninthisstudy,wefavoredusinganobservational mother‐childregulatorytaskoverself‐reportinstrumentsinformedbythemothers.We usedemotionalavailability(EA),measuredthroughamother‐childfreeplaytask,asa proxyforthequalityofmother‐childinteractions[19,20].Thismeasureispredictiveofthe mother’sreportedchildattachment[21].EAreferstosensitiveandresponsivedyadicex‐ changesandthecognitiveorganizationofcoordinatedmother‐childactionstowards achievingjointgoals.Mother‐childsynchronyindailyexchangesisacrucialfeaturein thistaskthatisrelatedtotheinfant’sattachmentquality[22,23].Moreover,resultsonthis taskdiscriminatedthelowperformanceintheemotionalavailabilityofneglectfulversus non‐neglectfuldyadsverywell[7].LowperformanceinEAwasalsorelatedtoreductions
BrainSci.2021,11,3873of15 inthestructuralconnectivityofwhitemattertractsinterconnectingtheface‐responsive cortexwiththelimbicandfrontalareasinmotherswithneglectfulbehavior[24]. Somedifferencesintraitalexithymiafoundinneglectfulversusnon‐neglectfulcare‐ giving[25]couldplayaroleinthecorticaldifferences–EAassociations. Alexithymiais definedasmetacognitiveimpairmentsinemotionalawarenessthatinvolvedifficultiesin identifyinganddescribingone’semotionsandminimizingemotionalexperiencebyfo‐ cusingattentionexternally[26–28],whichmayaffectthemother‐childexchanges. Brain regionshypothesizedasfeaturingNGandCGdifferenceshavebeendescribedaspartof alexithymianeuralcorrelates.Highcomparedtolowalexithymicwomenshowedre‐ ducedactivationsintheorbitofrontalcortex(OFC)andACCwhileempathizingwithan imaginedyoungchild[29].Moreover,recentevidencehasshownassociationsbetween corticalthicknessfeaturesandpersonalitytraits[30,31]. Takingthisevidenceintoaccount, wehavestartedfromamodelthatconsidersvariationsincorticalfeaturesasantecedents, presumablyrelatedtodifferencesinalexithymiaandtheperformanceinthemother‐child playtask. Finally,sincemotherswithneglectfulbehaviorhaveshownahistoryofchild‐ hoodmaltreatmentandlifestressthatusuallyentailsearlybrainalterations[1,4],itisim‐ portanttoconsideritspossibleeffectoncorticalfeatures. Insum,thisstudyaddressesthelackofevidenceinneuroimagingresearchonpossi‐ bledifferencesinthecorticalfeaturesinregionsassociatedwithneglectfulversusnon‐ neglectfulcaregiving. Thisstudyalsotriestofindevidenceoftherelationshipsofthose differentialcorticalfeatureswiththequalityofmother‐childinteractions.Followingre‐ sultsobtainedwithotherneuroimagingmeasuresinmotherswithneglectfulcaregiving reportedhere,weexpectedcorticaldifferencesinbothfrontalandoccipitotemporalareas. Wealsotentativelyexpectedthatalexithymiaasanemotionalpersonalitytraitwouldcon‐ tributetotheassociationbetweenthosecorticalfeaturesandobserveddyadicemotional availability. 2.MaterialsandMethods 2.1.Participants Forty‐fivemothers(24NGand21CG)oftheforty‐sevenscannedsamplespartici‐ patedintheexperiment.TheywereallrecruitedthroughthesamePrimaryHealthCenter. Allsubjectsgavetheirwritteninformedconsentforinclusionbeforetheyparticipatedin thestudy.ThestudywasconductedfollowingtheDeclarationofHelsinki,andtheproto‐ colwasapprovedbytheEthicsCommitteeoftheUniversityofLaLaguna(code: CEIBA2017‐0254;dateofapproval:25July2017). Generalinclusioncriteriawerebeingthe biologicalmotherofachildunderthreeyearsoldwhohadnotbeenplacedinfostercare atanypointintheirhistory,norbeenbornprematurelyorsufferedperinatalorpostnatal medicalcomplications,accordingtothepediatricians’reports.Specificinclusioncriteria foramotherintheneglectgroupwereasubstantiatedcaseofneglectregisteredinthelast 12monthsbyChildProtectiveServices(CPS)andcomplyingwiththeindicatorsofthe MaltreatmentClassificationSystem(MCS)forsevereneglect[32].Thecontrolgroup’sin‐ clusioncriteriawerehavingnegativescoresinalltheMCSneglectindicatorsandtheab‐ senceofCPSorPreventiveServicesrecordsforthefamily. Asforthemothers’sociodemographicprofile,theywereallintheirearly30s,had morechildrenandasimilarmeanageofthetargetchild,livedinruralareas,andshared asimilarlowlevelofeducation;mothersintheNGweremorelikelythanthoseintheCG toliveinsingle‐parentfamiliesandtoreceivefinancialassistance(Table1).Accordingto theneglectriskprofileratedbythesocialworkers,mostmothersintheneglectgrouphad ahistoryofchildhoodabuseorneglectandscoredpositivelyinneglectfulcaregivingin‐ dicators(seeSupplementaryS1:Participantrecruitmentstrategyandprocedure;Supple‐ mentaryS2:Riskprofilemeasures).
BrainSci.2021,11,3874of15 Table1.SociodemographicandneglectriskprofileinNeglectandControlGroups.Groupcom‐ parisonswithmeanscoreswereperformedwithtstatistic,whilethosewithpercentagevalues wereperformedwithchi‐square(χ2)statistic. NeglectGroup n=24 ControlGroup n=21 Comparisons χ2/t Sociodemographic profile Meanageofthe mother29.1(7.1)33.6(3.2)−2.82** Numberofchildren2.08(0.8)1.57(0.6)2.24* Meanageoftarget child2.7(1.5)2.3(1.9)0.82 Ruralareas(%)87.580.90.04 Levelofeducation (%): 2.53 Primary7047.6 Secondary16.628.5 >Secondary12.523.8 Single‐parentfamily50144.92* Financialhelpfrom institutions831418.7*** Neglectriskprofile Historyabuse/neglect (%)671410.5** Intimatepartnercon‐ flict(%)501.22 Chronicphysicalill‐ ness(%)1650.22 Poorhouseholdman‐ agement(%)88024.4*** Disregardhealth/edu‐ cationneeds(%)61012.4*** Disregardemo‐ tion/cognitiveneeds (%) 89520.9*** Rigid/inconsistent norms(%)67511.4*** PsychiatricDisorders factor0.67(0.99)−0.69(0.23)6.35*** *p<0.05;**p<0.01;***p<0.001. 2.2.MRIDataAcquisition High‐resolutionT1‐weightedMagnetisationPreparedRapidGradientEchoor MPRAGEanatomicalvolumeswereacquiredonaGeneralElectric3Tscannerlocatedatthe UniversityHospital’sMagneticResonanceServiceforBiomedicalResearchattheUniver‐ sityofLaLaguna.Atotalof196contiguous1mmsagittalsliceswereacquiredwiththe followingparameters:repetitiontime=8.716ms,echotime=1.736ms,fieldofview=256× 256mm2,in‐planeresolution=1mm×1mm,flipangle=12. Duringtheongoingdataacquisition,theT1‐MRIwasscreenedforclearartifactsby theMRItechnician.Participantswererescannedonthewayifanobviousartifactwas detected.Ifthetechnician(orthetrainedexperimenter)detectsanypotentialalteration
BrainSci.2021,11,3875of15 clinicallyrelevant,theimageswerepassedtotheneuroradiologistforfurtherexamina‐ tion.Alltheparticipantssignedintheconsentformtoagreetoreceiveinformationifany alterationwasdetected.Nodatawererejectedatthisstage. 2.3.CorticalThicknessandSurfaceArea AfterdataconversionintoNiftiformat,thetrainedexperimentervisuallychecked thequalityofeachT1‐MRIimageusingtheMRIcronsoftware.Theexperimenterthor‐ oughlyinspectedtheimages,takingintoaccountthesemaincriteria:motionartifacts (ghostingorblurringeffectsintheimagesduetomainlyparticipant’sgrossmovements, breathing,cardiaceffects,etc.),ringingartifacts(truncation/Gibbsartifacts)andsuscepti‐ bilityartifacts(imagedistortionsduetoinhomogeneitiesofthemagneticfield).Nodata wererejected,followedvisualinspection,andallpassedintoFreesurfer’sautomatedpre‐ processingpipeline(recon‐all).However,datafromtwocontrolparticipantshadtobe excludedduetoFreeSurfercorticalreconstructionprocess(recon‐all)failures.Freesurfer’s outputsfromallparticipantswerealsovisuallyinspected,asexplainedbelow.Nostatis‐ ticalextremeoutlierswereidentifiedbeloworabove2.5SDfortheextractedclusterbrain measuresthatmayhavedriventheresultsobtained. Corticalthickness(CT)andsurfacearea(SA)measureswereobtainedusingFree‐ Surfer(version5.1)(http://surfer.nmr.mgh.harvard.edu/,accessedon10February2020). Surfacecorticalreconstructionincluded:motioncorrection,skulltripped,automated transformationintoTalairachspace,subcorticalwhitematter,deepgraymattervolumet‐ ricstructuresegmentation,triangulartessellationofthewhitesurface(correspondingto thegrey/whitematterboundary),andpialsurface(correspondingtothepiamater),and automatedtopologycorrectionandsurfacedeformationthatoptimallyplacethe(inner) whitesurfaceandthe(outer)pialsurface[33,34].Somedeformationsincludedsurface inflationandahigh‐dimensionalnonlinearregistrationtoasphericalatlas.Thesegmen‐ tationanddeformationalgorithmsproducerepresentationsoftheCTthatistheaverage oftheclosestdistancefromthewhitetothepialsurfaceandfromthepialtothewhite surfaceateachvertexonthetessellatedsurfaces.TheSAiscomputedatthewhitesurface, whichislesssensitivetocorticalthicknessvariations,andmeasuredateachvertexasone‐ thirdoftheareaofeachtrianglethatmeetsthevertex;inotherwords,itisthesumofthe areaofallthetrianglesthatmeetthevertexdividedbythree[14].TheCTandSAmaps weresmoothedusinga15mmwithafull‐widthathalfmaximum(FWHM)Gaussian filter. ForthequalitycheckofFreeSurfer’soutputs,theindividualwhiteandpialsurfaces weredisplayedinFreeviewtoinspectvisuallyiftheyaccuratelyfollowedthegreyand whitematterboundaries.Noerrorswereobserved.FreeSurferalsoprovidedanaveraged brain(fsaverage)inMontrealNeurologicalInstitute(MNI)standardspacethatwasused tocompareandvisualizeresults. 2.4.BehavioralandPersonalityMeasures TheMini‐InternationalNeuropsychiatricInterview(M.I.N.I.6.0),intheSpanishver‐ sion[35],including15majorpsychiatricdisorders,wasadministered(seeSupplementary TableS1PsychopathologicalconditionsstratifiedbyGroup).Thescoreobtainedforeach disordercorrespondstoacumulativescoringofsymptomsandnottoacategoricalcut‐off classification.Thetwogroupsmainlydifferedinfivepsychopathologicalvariables,marked initalics,whichsurvivedtheBonferronicorrectionandweresubmittedtoaPrincipalCom‐ ponentAnalysis.Resultsgaveone‐factorsolution:“PsychiatricDisorders”(PD),withmod‐ erateinter‐correlationsamongthefivevariables,Kaiser‐Meyer‐Olkin(KMO)test=0.65,Ei‐ genvalue=2.53,withanexplainedvarianceof51%,withthecoefficientscoresinPDbeing higherintheNG(seeTable1).Noneofthemothersineithergroupwerebeingmedicated forpsychiatricdisordersatthetimeoftesting.ThecoefficientscoreinPDwasusedasa regressorintheFreeSurfermodeltocontrolasmuchaspossibleforitseffectoncortical differences.
BrainSci.2021,11,3876of15 DyadicEmotionalAvailabilitywasmeasuredinthecontextofmother‐childfreeplay usingtheEmotionalAvailabilityScale(EA):InfancytoEarlyChildhoodVersion[36].EA signifiesthequalityofemotionalexchanges,focusingonpartners’accessibilitytoeachother andtheirabilitytoreadandrespondappropriatelytoeachother’scommunications.Two externalobserversblindtothemothers’groupingsmadethevideos’ratings,andtheinter‐ raterreliabilityoftheratingsineachscale(Kappascore,0to1)wascalculated.Thescale operationalizesfouraspectsofparentalbehavior:Sensitivity(theparentshowscontingent responsivenesstochildsignalsanddemands,K=0.94);Structuring(theparentappropriately facilitatesthechild’splay,K=0.90);Non‐intrusiveness(theparentcansupportthechild’s playwithoutbeingoverdirectiveand/orinterfering,K=0.87);Non‐hostility(theparentcan behavewiththechildinawaythatisnotrejectingorantagonistic,K=0.92).Thescalealso measurestwoaspectsofchildbehavior:Responsiveness(thechild’sabilityandinterestin exploringonhisorherownandinrespondingtotheparent’sbids,K=0.92),andInvolvement (thechild’sabilityandwillingnesstoengagetheparentininteraction,K=0.86).Toobtaina simplerstructureofthesixstandardizedscales,aPrincipalComponentAnalysiswas performed(seeSupplementaryTableS2:Inter‐raterreliabilitiesandone‐factorcomponent loadingsoftheEmotionalAvailabilityScales).Theresultyieldedasingle‐factorstructure giventheexistenceofhighinter‐correlationsbetweenthevariablesinthemother‐child dyad:KMO=0.84,Eigenvalue=4.49,withanexplainedvarianceof75%.Thecoefficientscore inthissinglefactorwasusedasameasureofdyadicemotionalavailability. Finally,theTorontoAlexithymiaScale(TAS‐20)[37,38]measuresthedifficultyin identifyingandexpressingone’semotions.The20itemsarescoredona5‐pointscalefrom stronglydisagreetostronglyagree.TheTAS‐20hasathree‐factorstructure:Factor1,Dif‐ ficultyinDescribingFeelings(DDF;α=0.73);Factor2,DifficultyinIdentifyingFeelings (DIF;α=0.90);andFactor3,ExternallyOrientedThinking(EOT;α=0.50).Theoverall scoreaveragingthethree‐factorscoreswasalsocalculated. Statisticalanalysescomparingmeasuresbetweenthetwogroupsshowedsignifi‐ cantlyhigheralexithymiaandlowerdyadicemotionalavailabilityintheNGthaninthe CG(seeTable2). Table2.Pathanalysispsychologicalvariables.AlexithymiaandEmotionalAvailabilitycompari‐ sonsbetweenNeglectandControlGroups. NeglectGroup n=24 ControlGroup n=21Comparisons M(SD)M(SD)T(43)δ Alexithymia(averaged score)3.10(0.88)2.45(0.79)2.56*0.77 DifficultyinDescribing3.35(1.30)2.59(1.18)2.03*0.60 Feelings DifficultyinIdentifying2.98(1.31)2.18(1.07)2.21*0.66 Feelings ExternallyOriented Thinking3.05(0.61)2.59(0.56)2.63**0.78 EmotionalAvailability (factorScore) −0.62(0.92)0.71(0.48)−6.18***1.78 *p<0.05;**p<0.01;***p<0.001. 2.5.StatisticalAnalysesforBrainMeasures Asnopriorresearchhasexamineddifferentialcorticalfeaturesinneglectfulcaregiv‐ ing,weprobedtheentirecorticalmantlewithanexploratorywhole‐brainanalyticap‐ proachtoallowthediscoveryofafullrangeofcorticaldifferencesassociatedwithne‐ glectfulversusnon‐neglectfulcaregiving.Ananalysisofcovariance(ANCOVA)wasper‐ formedinFreeSurfertoexploreregionaldifferencesinCTandSAbetweentheNGand
BrainSci.2021,11,3877of15 CG,adjustedforage.ThemodelalsoincludedoneregressordescribedaboveasPsychi‐ atricDisorders,oncethepotentialcollinearityofthispsychiatricconditionwiththegroup wasruledout(seeSupplementaryTableS3:CollinearityindexesbetweentheGroupand thepsychopathologicalconditions).Totalintracranialvolume(TIV)wasalsoincludedas anuisancecovariateintheSAanalysis. SmoothedGaussianMonteCarlo(MCZ)simulation(10,000iterations)wasusedto performcluster‐wise(CW)correctionofmultiplecomparisons,withaninitialvertexwise cluster‐formationthreshold(CFT)≤0.001[39].However,accordingtotheseauthors,ifno clusterisfound,alessstrictCFT(≤0.005)canbeusedforcorticalthickness(CT)analyses. Atthisthreshold,CTislessaffectedbyfalse‐positiverates(FPR)thanSurfaceArea(SA)anal‐ ysesiftheFWHM>10mmisfulfilled,asisourcase(FWHM=15mm).Tofurtherminimize FPR,wecheckedwhethertheclusterssurvivingtheformercriteriaarealsohypothesis‐driven, correspondingtotheregions(e.g.,frontalandoccipitotemporal)mostlyfoundinthereported studies.Clusterformationwasperformedineachhemisphereseparatelyandthencorrected forbothhemispheresusingBonferronicorrectionatp‐value<0.05.Regionaldifferencesare describedbasedontheFreeSurferDesikan/Killianyparcellationatlas[40]. 2.6.StatisticalAnalysesforCortical–EAAssociations Pathanalyseswereplannedtotestthepotentialrelationbetweencorticalfeatures (CT/SA),alexithymia(assumingthatitdiffersbetweengroups),andemotionalavailabil‐ ity(EA)inthewholesample.Accordingtothestartingmodelexplainedabove,thecortical measure/sactedastheexogenousvariable/s,thealexithymiaoverallfactorscoreand/or specificfactorsactedasthevariable/sbetweenbrainmeasuresandEA,andEAactedas thecompletelyendogenousoutcomevariable.Thebrainmeasures’valueswerethemean thicknessandsurfaceareaextractedforthedifferentialclustersintheaboveanalyses.The alexithymiascoresmostsuitedtoenterinthepathanalysiswerethoseoftheaveraged factorand/orthethreespecificfactors,providedthattheycorrelatedwithboththecluster valuesandtheEAscores. SincemostmothersintheNG(16outof24)hadsufferedchildhoodmaltreatment (neglectorphysicalabuse),thisconditionshouldalsobeconsidered.Previousstudies showedthatahistoryofchildabusewasrelatedtocorticalvolumetricalterations[41]and poorqualityofmother‐infantinteractions[42].Basedonthisevidence,pathanalyseswere alsoplannedwherethemother’sownchildhoodmaltreatmentactedastheexogenous variable,givenitsconditionastemporalantecedent.Corticalmeasuresactedasthevari‐ ablesbetweenmaltreatmentconditionandEA,andEAactedasthecompletelyendoge‐ nousoutcomevariable.AlltheanalyseswereperformedwithRCoreteam[43]and LavaanR[44]packages. 3.Results 3.1.DifferentialCorticalFeaturesinNeglectandControlGroups TheresultsinCTwithaCFT=0.001showednoclusterformation.Then,aCFT=0.005, alsopermittedforCTwithaFWHM>10mm,wasapplied,resultinginoneclusterwitha patternofcorticalgreymatterreductioninmothersoftheNGcomparedtotheCG(CG> NG).Thecluster(p=0.014,Table3upperandFigure1)spannedseveralcontiguousana‐ tomicalregionsintherighthemisphere,involvingtherightrostralmiddlefrontalgyrus (RMFG)andthelateralandmedialorbitofrontalcortex(OFC),whichcorrespondstothe frontalregionspreviouslyhypothesized.ThestructuralMRIdataandthecovariatesscores thatsupportthefindingsofthisstudyareavailableinaG‐NodeInfrastructure(GIN)repos‐ itory(fordetails,seetheDataAvailabilitystatementattheendofthepaper).
BrainSci.2021,11,3878of15 Table3.DifferentialcorticalfeaturesinNeglectversusControlgroups.Aclustershowingreduced corticalthicknessformothersoftheneglectgroup(CG>NG)wasfoundcomprisingtherostral middlefrontalandorbitofrontalregions.Thegreatersurfaceareawasfoundinthelingualand lateraloccipitalgyrifortheneglectgroup(CG<NG).Age,totalintracranialvolumeforsurface analysesonly,andPsychiatricDisorderswereincludedasnuisancecovariates. Cluster/RegionsTotalVertexCluster‐Wise p‐ValueMaxx,y,z(mm)Max −log10(p‐Value) ControlGroup>NeglectGroup(Corticalthickness) R.Rostralmiddle frontal,lateral andmedialor‐ bitofrontal 14340.01421.1,51.5,−12.53.45 ControlGroup<NeglectGroup(Surfacearea) R.Lingualand lateraloccipital11850.00027.4,−88.3,−11.3−3.35 Figure1.ThecorticalthicknessreductioninNeglectversusControlgroups.Theclustercomprises therightrostralmiddlefrontalgyrus,aspartofthedorsolateralfrontalgyrus,andthelateraland medialorbitofrontalareas.Backgroundbrainimagesrepresenttheinferior(leftimage)andanterior (rightimage)viewsoftherightinflatedhemisphereof(FreeSurfer)fsaveragebrain.Note:p<0.05 cluster‐wisecorrected,10,000MonteCarloiterations,age,andPsychiatricDisorderswereincluded asnuisancecovariatesforthecorticalthicknessanalysis. Theresultsinthecorticalsurfacearea,applyingaCFT=0.001,revealedasignificant patternofgreatersurfaceareainmothersoftheNGwithrespecttotheCG(CG<NG)in theoccipitotemporalregionpreviouslyhypothesized.Theanalysisshowedalargecluster (p=0.0002,seeTable3bottomandFigure2)extendingovertherightlingualandlateral occipitalgyrioftherighthemisphere.
BrainSci.2021,11,3879of15 Figure2.TheincreasedcorticalsurfaceareainNeglectversusControlgroups.Theclustercom‐ priseslingualandlateraloccipitalregions.Backgroundbrainimagesrepresenttheinferior(left image)andmedial(rightimage)viewsoftherightinflatedhemisphere(FreeSurfer)fsaverage brain.Note:p<0.05cluster‐wisecorrected,10,000MonteCarloiterations,age,totalintracranial volume(TIV)andPsychiatricDisorderswereusedasnuisancecovariatesforthecorticalsurface areaanalysis. 3.2.RoleofAlexithymiainCortical‐EAAssociations Aspredicted,boththeaveragescoreofalexithymiaanditsthreefactorswerehigherfor theNGthanfortheCG(seeTable2).However,onlythefactorDifficultyinIdentifyingFeel‐ ings(DIF)correlatedbothwiththerightrostralmiddlefrontalgyrus(RMFG)corticalthick‐ ness(positively,r=0.33,p=0.05)andwithEmotionalAvailability(negatively,r=−0.30,p= 0.05).Therefore,DIFwasconsideredtobeagoodcandidatefortestingitsroleincortical‐EA associations.ApathanalysiswasperformedtotesttheroleofDIFintherelationshipbetween thecorticalthicknessscoresinthemiddlefrontalgyrus(RMFG)andEmotionalAvailability (EA)performance.Allstandarderrorsandhencethevaluesofthez‐statistichavebeenesti‐ matedfrom1000bootstrapsamples.TheresultsaredepictedinFigure3,showingthatthe alexithymiafactorDifficultyIdentifyingFeelings(DIF)wasnegativelyrelatedtoEAscores(B =−0.378,z=−2.829,p=0.005).Thepathanalysisalsorevealedsignificantdirectpositiveeffects ofRMFGthicknessonalexithymiaDIF(B=1.982,z=2.761,p=0.006),andonEAscores(B= 1.795,z=2.846,p=0.004).AmarginalindirectnegativeeffectofRMFGthicknessonEA throughalexithymiaDIFwasalsofound(B=−0.749,z=−1.850,p=0.064).However,thetotal positiveeffectofRMFGthicknessonEA,oncetheindirecteffectthroughalexithymiaDIFhad beensubtracted,remainedsignificant(B=3.400,z=3.508,p=0.000).Giventhatthenumberof estimatedparametersisequaltothenumberofdifferentinformationunitsinthecovariance matrix,nogoodnessoffitwasobtained(X 2 anddegreesoffreedomareequaltozero).