cortex 48 (2012) 216e229 Availableonlineatwww.sciencedirect.com Journalhomepage:www.elsevier.com/locate/cortex Special issue: Research report White matter microstructural abnormalities in the frontal lobe of adults with antisocial personality disorder Frederick Sundrama,b,1,*, Quinton Deeleya,1, Sagari Sarkara, Eileen Dalya, Richard Lathama, Michael Craiga, Malgorzata Raczeka, Tom Fahya, Marco Picchionia, the UK AIMS Networkd, Gareth J. Barkerc,2 and Declan G.M. Murphya aKing’sCollegeLondon,InstituteofPsychiatry,DepartmentofForensicandNeurodevelopmentalScience,London,UK bDepartmentofPsychiatry,RoyalCollegeofSurgeonsinIreland,Dublin,Ireland cKing’sCollegeLondon,InstituteofPsychiatry,DepartmentofNeuroimaging,CentreforNeuroimagingSciences,London,UK dTheMRCAIMSConsortiumisaUKcollaborationofresearchcentresintheUKincludingtheInstituteofPsychiatry,London, TheAutismResearchCentre,UniversityofCambridge,andtheAutismResearchGroup,UniversityofOxford. TheConsortiummembersareinalphabeticalorder:BaileyAJ,Baron-CohenS,BoltonPF,BullmoreET,CarringtonS,ChakrabartiB, DalyEM,DeoniSC,EckerC,HappeF,HentyJ,JezzardP,JohnstonP,JonesDK,LaiM-C,LombardoMV,MaddenA,MullinsD, MurphyCM,MurphyDGM,PascoG,SadekS,SpainD,StewardR,SucklingJ,WheelwrightS,WilliamsSC a r t i c l e i n f o a b s t r a c t Articlehistory: Antisocialpersonalitydisorder(ASPD)andpsychopathyinvolvesignificantinterpersonal Received11February2011 and behavioural impairments. However, little is known about their underlying neurobi- Reviewed11April2011 ologyandinparticular,abnormalitiesinwhitematter(WM)microstructure.Apreliminary Revised20April2011 diffusion tensor magnetic resonance imaging (DT-MRI) study of adult psychopaths Accepted4June2011 employingtractographyrevealedabnormalitiesintherightuncinatefasciculus(UF)(Craig Publishedonline20June2011 etal.,2009),indicatingfronto-limbicdisconnectivity.However,itisnotclearwhetherWM abnormalitiesarerestrictedtothistractorareormorewidespread,includingothertracts Keywords: whichareinvolvedinconnectivitywiththefrontallobe. Antisocialpersonalitydisorder Weperformedwholebrainvoxel-basedanalysesonWMfractionalanisotropy(FA)and ASPD meandiffusivity (MD)mapsacquiredwithDT-MRItocompare15adultswithASPDand Psychopathy healthyage,handednessandIQ-matchedcontrols.Also,withinASPDsubjectswerelated Neuroimaging differencesinFAandMDtomeasuresofpsychopathy. Diffusiontensorimaging Significant WM FA reduction and MD increases were found respectively in ASPD subjects relative to controls. FA was bilaterally reduced in the genu of corpus callosum whileintherightfrontallobeFAreductionwasfoundintheUF,inferiorfronto-occipital fasciculus (IFOF), anterior corona radiata and anterior limb and genu of the internal capsule.Thesedifferencesnegativelycorrelatedwithmeasuresofpsychopathy.Alsointhe rightfrontallobe,increasedMDwasfoundintheIFOFandUF,andthecorpuscallosum andanteriorcoronaradiata.TherewasasignificantpositivecorrelationbetweenMDand psychopathyscores. * Correspondingauthor.King’sCollegeLondon,InstituteofPsychiatry,DepartmentofForensicandNeurodevelopmentalScience,POBox 50,DeCrespignyPark,SE58AFLondon,UK. E-mailaddress:[email protected](F.Sundram). 1Jointfirstauthors,theseauthorshavecontributedequallytotheproductionofthismanuscript. 2Jointlastauthor,theseniorauthorhavecontributedequallytotheproductionofthismanuscript. 0010-9452/$eseefrontmatterª2011ElsevierSrl.Allrightsreserved. doi:10.1016/j.cortex.2011.06.005 cortex 48 (2012) 216e229 217 Conclusions:ThepresentstudyconfirmsapreviousreportofreducedFAintheUF.Addi- tionally,wereportforthefirsttime,FAdeficitsintractsinvolvedininterhemisphericas well as frontal lobe connectivity in conjunction with MD increases in the frontal lobe. Hence, we provide evidence of significant WM microstructural abnormalities in frontal brainregionsinASPDandpsychopathy. ª2011ElsevierSrl.Allrightsreserved. 1. Introduction tobeaparticularlyseveresubtypeofASPD(DolanandDoyle, 2007).PsychopathyandASPDarehoweverdistinguishedfrom 1.1. Thefrontallobetheoryofantisocialpersonality behaviourssecondarytofrontallobelesionsbyhighlevelsof disorder(ASPD)andpsychopathy bothreactive(elicitedbyfrustration)andinstrumental(goal- directed)violence(Blair,2001;GlennandRaine,2009).Never- Theimportanceofthefrontallobesto socialbehaviour was theless, overlaps between traits of both psychopathy and firstrecognisedinthe19thcenturyfollowingthecaseofPhi- ASPD,andfrontallobesyndrome,haveledtothesuggestion neasGage,inwhomfrontallobedamageresultedinprofound that both PDs may result from frontal lobe abnormality personality change associated with markedly inappropriate (Damasio, 2000). Neuroimaging studies of both people with social behaviour [Harlow, 1993 (1869)]. A ‘frontal lobe’ ASPD, and of individuals with psychopathy, have provided syndrome was subsequently delineated based on clinical evidenceofabnormalitiesoffrontallobestructureandfunc- observations of the behaviour of patients with frontal lobe tionrelativetocontrolpopulations,togetherwithdeficitsin damage (Lishman, 1998) where symptoms included apathy, temporal,limbic,andotherbrainregions(seeTable1). emotionallability,alackofsocialawareness,unconcernfor For example, neuroimaging studies of adult psychopaths socialrules,impulsivity,andreactiveaggression. examining the frontal cortex have reported reduced grey It is currently recognised that there is much overlap mattervolumeinconjunctionwithreductioninthesuperior between frontal lobe syndrome and ‘functional’ or ‘non- temporalgyrus(Mulleretal.,2008),andintheprefrontalcortex organic’ personality disorders (PDs), particularly ASPD and (PFC) of ‘unsuccessful’ (caught) psychopaths, versus healthy psychopathy(Damasio, 2000).The definition ofpsychopathy controls (Yang et al., 2005). Furthermore, higher total and haschangedlittlesinceHerveyCleckleypublishedTheMaskof subfactor PCL-R scores (arrogant/deceptive, affective, and Sanity in 1941 where he described the psychopath as impulsive/unstable)wereassociatedwithreducedprefrontal acharming,callous,superficialindividual,lackingconscience greymattervolume(ibid).Similarly,prefrontalandtemporal andgenuineemotion(Cleckley,1941).ThePsychopathyChecklist corticalgreymatterthinningwasfoundinpsychopathicindi- (PCL,Hare,1980)andthelaterPsychopathyChecklisteRevised viduals,withrighthemispherereductionsrelatedtoelevated (PCL-R,Hare,1991)weredesignedtooperationaliseCleckley’s PCL-RFactor1‘Affective’facetscores(Yangetal.,2009b). conceptofpsychopathyasabasisfordiagnosingthedisorder. Other studies have identified associations between ThePCL-Rconsistsof20itemscharacterisedbroadlybytwo psychopathic traits and specific subregions of the PFC. In dimensions: Factor 1 items are primarily interpersonal or particular, the association found in brain injured patients emotionaltraitssuchasremorselessness,deception,shallow between ventromedial PFC (vmPFC) damage and reactive affectandcallousness,whereasFactor2itemsassessbehav- aggression (BlairandCipolotti,2000;Grafmanetal.,1996)is iouralsymptomssuchasviolence,criminality,anddysfunc- mirroredbyvmPFCstructuralandfunctionalimpairmentsin tionallifestyle.Foradiagnosisofpsychopathy,attributesfrom psychopaths(Tiihonenetal.,2008).Functionalneuroimaging bothofthesefactorsneedtobepresent.WhilePCL-Rscores studies of people with psychopathy have also provided (cid:1)30havetraditionallybeenusedtoclassifyanindividualas evidenceofabnormal frontallobeperfusionandabnormali- having psychopathy (Hare, 2003), more recent studies have ties of task-related activation in prefrontal and other brain arguedforascoreof(cid:1)25assufficientfordiagnosis(Edensand regionsonreversallearningparadigms(Table1).Individuals Petrila,2006;Edensetal.,2010;Rutherfordetal.,1999). with ASPD have shown similar structural and functional While the related construct of ASPD in DSM-IV-TR (Diag- prefrontal abnormalities to psychopaths e for instance nosticandStatisticalManualFourthEditioneTextRevision, reducedprefrontalgreymattervolumehasbeenfoundinboth AmericanPsychiatricAssociation,2000)includesseveraltraits antisocial adults (Raine et al., 2000) and conduct disordered present in psychopathy (e.g., lack of guilt/remorse, and children (Huebner et al., 2008), compared with healthy impulsivity),diagnosticcriteriacanbemetbasedentirelyon controls.Also,corticalthinningofthemedialfrontallobehas antisocial behaviours (e.g., violation of social norms, irre- beenfoundinASPD(Narayanetal.,2007). sponsibility, and criminality). Hence, the emotional deficits Nevertheless, structural and functional abnormalities in fundamentaltopsychopathyarenotnecessaryforadiagnosis peoplewithpsychopathyandASPDarenotrestrictedtothe ofASPD,evenifthesearepresentinsomecases.Estimatesof frontallobe.Forexample,abnormalamygdalastructureand theprevalenceofthetwodisordersalsodiffer,suggestingthat functionhaveeachbeenfoundtocorrelatewiththeemotion these are non-equivalent diagnoses. While most adult processingdeficitsobservedinASPDandpsychopathicindi- psychopathicoffendersmeetcriteriaforASPD,onlyapproxi- viduals (Gordon et al., 2004; Kiehl et al., 2001; Yang et al., matelyonethirdofthosewithASPDarepsychopathic(Hart 2009a), as well as in regions (such as fusiformeextrastriate and Hare, 1997). Psychopathy has therefore been postulated cortices) known to be modulated by the amygdala (Deeley 218 cortex 48 (2012) 216e229 Table1eSummaryofvolumetry,functionalandDT-MRIfindingsimplicatingfronto-limbicandotherbrainregion abnormalitiesinantisocialpopulations. Author Method Population Comparisongroup Finding Region (Barkataki MRI ASPD Violentandnon-violent Reducedvolume Wholebrain,bilateral etal.,2006) schizophrenia temporallobe Increasedvolume Putamen (Laaksoetal., MRI ASPDwithalcoholism Healthycontrols Reducedvolume Righthippocampus, 2000) posteriorhippocampi (Laaksoetal., MRI ASPDwithalcoholism Psychiatricpatients Volumeinversely Bilateralposterior 2001) relatedtoPCLscore hippocampus (Narayanetal., MRI ASPD Violentandnon-violent Corticalthinning MedialPFC 2007) schizophrenicsand healthycontrols (deOliveira- MRI ASPDwith Healthycontrols Reducedvolume Frontopolarcortex, Souzaetal., psychopathy Volumesinversely orbitofrontalcortex, 2008) relatedtopsychopathy anteriortemporalcortex, score superiortemporal sulcus,insula (Raineetal., MRI CommunityASPD Alcoholdependentsand Reducedvolume Prefrontalgreymatter 2000) healthycontrols (Raineetal., MRI CommunityASPD Healthycontrols Increasedlength Corpuscallosum 2003) withhighpsychopathy Increasedvolume scores Reducedthickness (Tiihonenetal., MRI ASPDwithalcohol Healthycontrols IncreasedWMvolume Bilateraloccipitallobe, 2008) dependence bilateralparietallobe, leftcerebellum Increasedgreymatter Rightcerebellum volume (Volkowetal., PET Violentoffenders Healthycontrols Reducedglucose PFC,medialtemporal 1995) metabolism cortex (Raineetal., PET Murderers Healthycontrols Reducedglucose PFC,corpuscallosum, 1997) metabolism superiorparietalgyrus, leftangulargyrus (Boccardietal., MRI Psychopathicviolent Healthycontrols Bilateraldepression Hippocampuse 2010) offenders longitudinalaxis ReducedCA1segment Hippocampuseanterior Abnormalenlargement Hippocampuse lateralborders (Craigetal., DT-MRI Psychopaths Healthycontrols ReducedFA RightUF 2009) (Glennetal., MRI Communitysamplee Communitysamplee Increasedvolumewith Striatum 2010) highpsychopathy lowpsychopathy greaterpsychopathy scorers scorers score (Mulleretal., MRI Criminalpsychopaths Healthycontrols Reducedvolume Rightsuperiortemporal 2008) gyrus (Shamay-Tsoory CT/MRI ASPDmaleswith Orbitofrontalcortex Impairedaffectiveempathy Orbitofrontalcortex etal.,2010) psychopathictraits verusnon-frontal performanceinboth lesionedmales orbitofrontalcortexlesioned andhealthycontrols andpsychopathygroup (Yangetal., MRI Unsuccessful Successfulcommunity Reducedvolume Prefrontalgreymatter 2005) community psychopathsand psychopaths healthycontrols (Yangetal., MRI Community Healthycontrols Reducedvolume Bilateralamygdala 2009a) psychopaths Surfacedeformations Amygdalanuclei: basolateral,central, cortical,lateral (Yangetal., MRI Community Healthycontrols Corticalthinning Rightfrontalcortex, 2009b) psychopaths righttemporalcortex Thinningassociatedwith Rightfrontalcortex, greaterPCLfactor2score righttemporalcortex cortex 48 (2012) 216e229 219 Table1e(continued) Author Method Population Comparisongroup Finding Region (Deeleyetal., fMRI Psychopaths Healthycontrols ReducedBOLDactivationto Fusiformgyrus, 2006) fearfulandhappyfacesin extrastriatecortex emotionprocessingtask Decreased,ratherthan Fusiformgyrus increased,BOLDactivation tofearfulfaces (Soderstrom SPECT Violentoffenders Negativecorrelationbetween Frontalandtemporal etal.,2002) withvarying interpersonalpsychopathy regions,headofcaudate, psychopathyscores factorandperfusion lefthippocampus (DeBrito MRI Communitysampleof Increasedgreymatter Medialorbitofrontal etal.,2009) boysehighversuslow concentration cortex,anteriorcingulate callous-unemotional cortex traitscorers Increasedgreymatter Bilateraltemporallobe concentrationandvolume (Kruesietal., MRI Conductdisordered Healthycontrols Reducedgreymattervolume Righttemporallobe 2004) adolescents (Huebner MRI Conductdisordered Healthycontrols Reducedgreymattervolume Bilateraltemporallobe etal.,2008) adolescentmales Lefthippocampus,left comorbidwithADHD amygdala (Sterzeretal., MRI Conductdisordered Healthycontrols Reducedgreymattervolume Bilateralanteriorinsular 2007) adolescentmales cortex Leftamygdala (Bernsetal., DT-MRI Healthyadolescentse Highscorespositively FrontalWMtracts 2009) highversuslowon correlatedwithFAand risktakingmeasure negativelywithtransverse diffusivity (Marshetal., fMRI Antisocialchildrenwith ChildrenwithADHD; ReducedBOLDactivationto Amygdala 2008) callous-unemotional healthycontrols fearfulfacesonemotional traits processingtask Reducedfunctional Betweenamygdalaand connectivity vmPFC (Fingeretal., fMRI Antisocialchildrenwith ChildrenwithADHD; AbnormalBOLDsignalto vmPFC 2008) callous-unemotional healthycontrols punishederrorsonreversal traits learningtask (Jonesetal., fMRI Antisocialboyswith Healthycontrols ReducedBOLDactivationto Rightamygdala 2009) callous-unemotional fearfulfacesonemotional traits processingtask (Stadleretal., fMRI Conductdisordered Healthycontrols ReducedBOLDactivationto Rightanteriorcingulate 2007) adolescentmales negativeaffectivepicturesin cortex emotionalprocessingtask etal.,2006).Reducedvolumeoftemporalregionsisalsoseen resonanceimaging(DT-MRI)(Basseretal.,1994b;Thiebautde inASPD(Barkatakietal.,2006).Takentogether,thesestudies Schotten et al., 2012; Catani et al., 2012) wherethe ‘connec- ofantisocialadultsandconductdisorderedchildren(withand tivity’ofneuralsystemsisassessedusingproxymeasuresof withoutpsychopathictraits)illustratereducedvolumeofthe microstructuralintegrity. temporal lobe and its constituent structures, and deficits in amygdalastructureandfunction. 1.2. Disconnectivitybetweenfrontalandotherregionsin Apartfromthesefronto-temporalstructures,otherregions psychopathy potentially relevant to psychopathy have been less exten- sivelyinvestigated.Forinstance,therehavebeenonlylimited DT-MRIisparticularlyusedintheassessmentoftissue(suchas magnetic resonance imaging (MRI) studies assessing the WM networks) where water preferentially diffuses along corpus callosum, a major white matter (WM) bundle sup- aparticularaxisalignedwiththetissue’sinternalstructureand porting interhemispheric functional integration, where is predicated on the principle that microarchitectural struc- abnormalitieshaveincludedincreasesinvolumeandlength tures,forinstance,cellmembranes,myelinsheaths,aswellas but reduction in thickness (Raine et al., 2003). Also, while intracellularmicro-organelles,actasbarrierstothediffusion reductionsinfunctionalconnectivitybetweenprefrontaland andfreemovementofwater,andthuslimitthespatialmotion limbic regions may contribute to antisocial traits (Table 1), ofthesemolecules(MalhiandLagopoulos,2008).Theassess- their underlying microstructural basis remains unknown. mentofthedirectionaldependenceofwatermoleculediffu- Overall, the wider network of abnormalities in psychopathy sion in WM is usually quantified through calculation of hasbeenrelativelyunderstudiedalthoughsuchinvestigation fractional anisotropy (FA). FA is a measure of the degree of has been made possible using diffusion tensor magnetic anisotropy or directionality where values range from 220 cortex 48 (2012) 216e229 0 (perfectly isotropic diffusion) to 1 (perfectly anisotropic Participants in both groups were medication free, spoke diffusion)(PierpaoliandBasser,1996)esoprovidingameasure English as their first language, and were right-handed as oftissueintegrity(HorsfieldandJones,2002;MoriandZhang, assessed by the Annett Handedness Questionnaire. The 2006). Mean diffusivity (MD) is another DT-MRI derived WechslerAdultIntelligenceScaleeRevised(Wechsler,1981) parameter used for reporting tissue differences which is wasusedtomeasureIQ.Allparticipants(inbothgroups)were calculated by division of the sum of the eigenvalues of the examined by formal psychiatric semi-structured interview diffusiontensor(whichcorrespondtothemagnitudeofdiffu- using ICD-10 research criteria (World Health Organisation, sion in three orthogonal directions) by three. There are 1993)inadditiontoassessmentofcasenotesforadiagnosis howeveronlylimitedstudiesexaminingneuraldisconnectivity ofASPD.Assessmentforthepresenceofcomorbidpsychiatric usingDT-MRIinpeoplewithpsychopathyand/orASPD. illness (e.g., anxiety disorders, substance misuse, schizo- A recent study by our group using DT-MRI tractography phrenia, major depression), neurological and extracerebral focusing on FA in the uncinate, inferior longitudinal and disorders that may affect brain function, and contraindica- inferior fronto-occipital fasciculi reported a significant tions to MRI scanning was also performed. Though many reductionofthismeasureinonlytheuncinatefasciculus(UF) individualswith ASPDhavea past historyofalcohol and/or of nine psychopaths compared with age- and IQ-matched substancemisuse,weattemptedtorecruit(asfaraspossible) controls (Craig et al., 2009). Additionally, a significant nega- subjectswithoutcomorbiditiesratherthantocontrolforthese tive correlation was found between measures of antisocial post hoc. None of the participants fulfilled criteria for behaviour(PCL-RFactor2scores)andtractvolumewithinthis substancemisuseordependencesyndrome6monthspriorto WM pathway, suggesting abnormal connectivity in the recruitment, with the exception of one subject who had amygdalaeorbitofrontal cortex limbic network. However, as harmfuluseofcocaine. thisstudywasconfinedtoalimitednumberofWMtracts,it 45 participants were initially recruited into the study (20 was not possible to assess WM networks on a whole brain ASPDvs25controls).However,fivewithASPDand10controls level. Consequently, the presence of deficits affecting WM respectivelywereunsuitableforfurtherassessmentfollowing connectivitywithotherbrainregionsisyettobeestablished exclusion for comorbid psychiatric disorder and contraindi- ineitherASPDorpsychopathy.Further,ourpreviousstudydid cationstoMRIprocedures.PCL-Rscoreswereobtainedfrom notexamineotherindicesofWMmicrostructure,suchasMD. casenotesderivedfromassessmentsbasedattheirspecialist Insummary,thereisincreasingevidencethatpeoplewith forensic unit by forensic psychologists fully trained in the ASPDandpsychopathymayhavesignificantdifferencesinthe administrationofthePCL-R,orbyaresearcher(QD)wherethe structure and function of frontal, limbic and other brain PCL-R had not beenadministered butwhere subjectsother- regions. However, few studies have examined the micro- wisemetinclusioncriteria.Wethereforeincluded30normal structural integrity or connectivity of their WM networks. intelligenceright-handed adultmalesubjects:15withASPD Therefore, we undertook the first DT-MRI investigation on and a mean PCL-R score of 26 (SD (cid:3) 7; range 13e34) aged awholebrainlevelofASPDandpsychopathy.Weexamined 39 (cid:3) 10 years and with full-scale IQ (FSIQ) 92 (cid:3) 13, and 15 WMnetworksandtestedthemainhypothesisthatpeoplewith healthycontrolsaged37(cid:3)11years,withFSIQ99(cid:3)12.There ASPDandpsychopathyhavesignificantdifferences,basedon werenosignificantdifferencesinageorIQbetweenpartici- DT-MRIderivedparametersofFAandMD,inmicrostructural pant groups. Those with ASPD had a history of violent integrity and connectivity as compared to healthy matched offending that encompassed manslaughter, attempted controls.Also,wetestedanadditionalhypothesisthatwithin murderandmultiplerapewithstrangulation.IntheUKitis ASPD, severity of psychopathy (as measured by PCL-R) is accepted practice to define psychopathy as a score of 25 or relatedtodifferencesintheseWMmeasures. aboveonthePCL-R(Cooke,1996;CookeandMichie,1999)and 10 of the 15in theASPD group scoredabovethis threshold. However,whilewewereabletoobtaintotalPCL-Rscoresfor 2. Methods theentirepatientcohort,itwasonlypossibletoacquireFactor 1andFactor2subscoresfor12ofthe15subjectsfromcase 2.1. Subjects notes. Study participants were recruited from three specialist 2.2. MRIacquisitionprotocol forensicinpatientunitsinsouth-eastLondon(SouthLondon andMaudsleyNationalHealthServiceFoundationTrust)and Datawereacquiredusinga1.5TGESignaLXsystem(General south-westLondon(StGeorge’sHealthcareNHSTrust)aspart Electric,Milwaukee,WI,USA),withactivelyshieldedmagnetic of our longitudinal work in assessing psychopathy, over fieldgradients(maximumamplitude40mTm(cid:4)1).Astandard aperiodofeightyears.Healthycontrolswererecruitedfrom quadrature birdcage head coil was used for both radio- the general population through the Institute of Psychiatry, frequency transmission and signal reception. Each DT-MRI King’sCollegeLondonbyadvertisement.Ethicalapprovalwas volume was acquired using a multi-slice peripherally-gated obtainedfromtheEthicsCommitteeoftheSouthLondonand echo-planar imaging (EPI) sequence, optimised for precise Maudsley Trust and Institute of Psychiatry, and St George’s measurement of the diffusion tensor in brain parenchyma, Healthcare Trust. Written informed consent was obtained from60contiguous2.5mmthicksliceswithfieldofview(FOV) fromparticipantsafterfulldescriptionofthestudy.Student’s 240 (cid:5) 240 mm and matrix size 96 (cid:5) 96, zero-filled during t tests(two-tailed) wereusedto comparethedistribution of reconstructionto128(cid:5)128,givingafinalin-planevoxelsizeof continuousdatabetweenthetwogroups. 1.875 (cid:5) 1.875 mm2 (Jones et al., 2002; Kyriakopoulos et al., cortex 48 (2012) 216e229 221 2008).Imageacquisitionwassynchronisedtothecardiaccycle givemapsoftheprobabilityofatissuebeingeitherwhiteor usingaperipheralgatingdeviceplacedonthesubject’sfore- greymatterandthresholdedtheresultingimagesatalowlevel finger.Echotimewas107msecwhiletheeffectiverepetition (10%)toprovidea(deliberatelyslightlyoverinclusive)binary timewas15ReRintervals.Durationofthediffusionencoding mask of WM. We then smoothed the original FA and MD gradients was 17.3 msec giving a maximum diffusion images,beforeapplyingthesemaskstorestrictthesubsequent weightingof1300secmm(cid:4)2.Ateachslicelocation,7images statisticaltestingtoWMonly.Asdifferentsmoothinglevels were acquired with no diffusion gradients applied, together canresultinvaryingresults(Jonesetal.,2005),intheabsence with 64 diffusion-weighted images in which gradient direc- of a specific hypothesis about the spatial extent of any tionswereuniformlydistributedinspace.Totalscantimewas abnormalities,weappliedaGaussiansmoothingfilterof5mm approximately 20 min and the relative orientations of the full-width at half maximum. Because of our overinclusive diffusion gradient vectors were based on the electrostatic mask,thesmoothedimageswillincludesomemixedgrey/WM repulsionalgorithm(Jonesetal.,1999,2002). voxels at the edges of the masked region; however, later Followingcorrectionofthediffusion-weightedimagesfor analysissteps(seebelow)furtherrestrictthestatisticaltesting image distortions introduced by the diffusion-weighting toonlycoreWMregions. gradients, in-house software was used to 1) remove non- brain tissue and 2) determine the diffusion tensor in each 2.4. Imageanalysisandcorrelations voxel (Basser et al., 1994a, 1994b). Images of 1) mean T2- weightedintensity(withnodiffusiongradientsapplied)and 2.4.1. DT-MRIgroupmapping 2)FAandMDwerecomputedforeachsubject.Fulldetailsare We examined the statistical significance of between group givenelsewhere(Jonesetal.,2002). differences in FA and MD using a non-parametric permuta- tion-based method. Locally developed software, XBAM 2.3. Pre-processingDT-MRIdata (version3.4)(InstituteofPsychiatry,http://www.brainmap.co. uk/)measuresbetweengroupdifferencesinstandardspaceat Scans were examined for image corruption or motion arte- bothvoxelandclusterlevelsbyfittingananalysisofvariance factspriortoinclusionintheimagingpipelineandnoneofthe (ANOVA)statisticalmodel.Astherewerenodifferencesinthe acquired scans demonstrated these abnormalities. The mean values in either age or IQ between groups, and the subsequentpre-processingstepsandanalyticalmethodology ranges similar, these were not entered as covariates and so have been published previously (Bloemen et al., 2010; theANOVAmodelutilisedFAorMDasthedependentvariable Kyriakopoulos et al., 2008, 2009; Sundram et al., 2010), and whilegroupclassificationwasthekeypredictorvariable. aresummarisedbelow. Given that brain changes are likely to extend over After construction of maps of FA and MD, a voxel-based a number of contiguous voxels, test statistics incorporating approach in standard space using SPM2 (Wellcome Depart- spatial information such as 3D cluster mass (the sum of ment of Imaging Neuroscience, University College London) suprathresholdvoxelstatistics)aregenerallymorepowerful within MATLAB 6.5.2 (The MathWorks, Natick, MA, USA), thanotherpossibleteststatistics,whichareinformedonlyby aligned,smoothedandsegmentedtheFAandMDimages.In data at a single voxel (Bullmore et al., 1999). The voxelwise a manner analogous to the early voxel-based morphometry statistic images were therefore thresholded at a relatively (VBM) analysis methods developed for structural T1 and/or lenientlevelofp(cid:6).05,andvoxelsthatwerespatiallycontig- T2-weightedMRimages,wefirstperformedatwo-stagenor- uous in three dimensions in the thresholded maps were malisationtostandardMontrealNeurologicalInstitute(MNI) assigned to the same cluster. The sum of voxel statistics spaceusingastudy-specific,intermediatetemplatetoreduce withineachclusterwascomputedforeachrandomisationto potentialbiasduetodifferentdegreesofwarpingthatwould formadistributionofclustermassunderthenullhypothesis. otherwise be required to match ASPD and control brains to Asnoparametricdistributionisknownforclustermass(sum astandard(controlsubject-based)template. of voxel statistics within each cluster), permutation testing The mean T2-weighted (nondiffusion-weighted, b ¼ 0) wasusedtoassessstatisticalsignificance; themassofeach images from each subject were initially registered to the clusterintheobserveddatawascomparedtothisrandomised standardEPItemplateprovidedbySPM2.Thederivedmapping distribution, and significant clusters were defined as those parametersforeachsubjectwerethenappliedtothe(inher- that had a greatercluster mass than the randomised distri- entlyco-registered)FAimages.ThenormalisedFAimagesof bution at a particular significance level. The number of all subjects were then averaged and smoothed (8 mm full- permutationsatthevoxelandclusterlevelswas1000which width at half maximum Gaussian filter) to create a new, definesthedistributionwellenoughtopermitinferenceabout study-specific,intermediatetemplatetowhicheachsubject’s changesbetweengroups(Bullmoreetal.,1999).Suchanon- FA and MD images were then re-registered. Smoothing the parametric approach also overcomes the assumption that data in order to coerce it into the appropriate statistical parametric methods adopt that the residuals of the model distributionisaprerequisiteforsomeanalyticalapproaches, tested will follow a Gaussian distribution (which has been but is not necessary for our non-parametric statistical shown to not always be true for DT-MRI data, even after approach where it serves instead to aid between-subject extensivesmoothing)(Jonesetal.,2005). anatomical matching, reduce confounds due to individual Atthisstage,weconsideredonlythosevoxelsatwhichall variation in WM anatomy and improve the signal-to-noise subjects contribute data which, along with the masking ratio. We therefore segmented the registered FA images procedureabove,restrictedtheanalysistocoreWMregions, (usingSPM’sdefaultaprioritissueprobabilityinformation)to reducing the search volume (and thus the number of 222 cortex 48 (2012) 216e229 comparisonsmade)andalsoavoidedtestingatthegrey/white 3. Results interfaces,wherethehighgrey/whitecontrastofFAimages exacerbatesanyedgeeffects.Attheclusterlevel,ratherthan 3.1. GroupcontrastsofFAusingDT-MRIgroup setasingleapriorip-valuebelowwhichweregardfindingsas mapping significant,wecalculated,forarangeofp-values,thenumber of clusters which would be expected by chance alone. 3.1.1. ASPDversuscontrolsWMFA(Fig.1,Table2) Astringentclustersignificancethresholdwasthenappliedto People with ASPD, relative to controls, had a significant renderlessthanonefalsepositiveclusterperanalysis.AsSPM reduction in WM FA; 1) bilaterally in the frontal lobe in the wasusedforimageregistration,XBAMyieldedcoordinatesof anteriorportionofthecorpuscallosum(genu);2)intheright significant clusters in MNI space and the identification of hemisphere,onlyinanteriorregionsofthebrainandinWM significant clusters and WM tracts intersecting them was tracts that included the genu of corpus callosum, anterior basedon an atlasapproachwhereseveralWMatlases were corona radiata and anterior limb and genu of the internal used in conjunction to determine the projection and local- capsule, and frontal course of the uncinate and inferior isation of WM bundles; these atlases were based on both fronto-occipitalfasciculus(IFOF);3)inthelefthemispherein single and multiple subject DT-MRI datasets in MNI space both anterior and posterior regions of the brain including (Catani and Thiebaut de Schotten, 2008; Mori et al., 2005; respectively the genu of corpus callosum and temporo- ThiebautdeSchottenetal.,2011;Wakanaetal.,2004).Given occipital course of the inferior longitudinal and IFOF, and thattheTalairachcoordinatesystemismostcommonlyused theretrolenticularpartoftheinternalcapsuleandposterior for reporting findings in the neuroimaging literature, MNI thalamicradiation. coordinatesweresubsequentlyconvertedtoTalairachspace via a non-linear transformation (Brett et al., 2002) (details 3.1.2. ASPDversuscontrolsWMMD(Fig.2,Table2) at http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach) People with ASPD, relative to controls, had a significant and the localisation of WM tracts and their respective increaseinMDonlyintherightfrontallobe.Thiswaslocal- anatomical course within cerebral lobes was additionally isedtoaclustercontainingthefrontalcourseoftheIFOFand confirmedusingtheTalairachatlas(TalairachandTournoux, UF, and the genu of corpus callosum and anterior corona 1988).WheresignificantdifferenceswerefoundoneitherFA radiata.NoregionsofincreasedMDwerefoundinthecontrol orMDmaps,thep-valueofeachsignificantclusterisreported group. in Table 2 alongside the anatomical locations of their respectiveWMtractsinTalairachspace. 3.1.3. WithinASPD,ananalysisofFactor1,Factor2and In the investigation of cluster level effects, there is an totalPCL-RscoresanddifferencesinWMFAandMD underlyingassumptionthatallregionswillequally‘smooth’ In the ASPD group, there were significant correlations and can therefore be treated equivalently (from a statistical betweenPCL-RscoresandbothWMFAandMD.MeanFAof point of view). However, while we restricted our analysis to the cluster in the frontal lobe (Cluster 2) was negatively core WM regions where signal-to-noise ratio is relatively correlatedwithFactor2(r¼(cid:4).771,p¼.003,n¼12)andtotal uniform,theeffectsofphysiologicalnoise(e.g.,motionarte- PCL-R(r¼(cid:4).685,p¼.005,n¼15)scores.Additionallyinthe facts) may still vary across the brain. We therefore concur- frontal lobe, there was a significant positive correlation rentlyinspectedthevoxel-levelmaps(whichtreateachvoxel between increased MD (Cluster 3) with Factor 2 scores independently and therefore inherently allow for such local (r¼.669,p¼.017,n¼12).Therewerenosignificantcorrela- differencesinstatistics)(Kyriakopoulosetal.,2008). tions between the cluster in the temporo-occipital cortex (Cluster1)andtotalorsubfactorPCL-Rscores. 2.4.2. PosthocanalysisofPCL-Rscores Where significant 3D clusters were found on DT-MRI group mappinganalyses,wecarriedoutposthocteststodetermine ifsignificantdifferencesinWMFAorMDwereassociatedwith 4. Discussion behaviouralvariationwithintheentiregroupwithASPD.As the clusters detected by XBAM may encompass multiple WeusedDT-MRIinconjunctionwithwholebrainvoxel-based anatomical regions (i.e., were not constrained to lie only analyses of MD and FA to compare WM microstructural within particular WM tracts), we first extracted mean FA in integrity of 15 adult males with a diagnosis of ASPD and 15 combinationwithMDforeachASPDsubject.Todothis,mask healthy controls matched for age, handedness and IQ. imageswerecreatedfromeachoftheclustersfoundbythe ReductionsofFAwerepresentinthefrontallobeoftheASPD group mapping analysis and applied to each subject’s nor- group, and these were significantly and inversely correlated malisedFAandMDimagestherebyenablingmeanFAandMD withseverityofpsychopathy(Factor2,andtotalPCL-Rscores). values to be calculated over each region for each subject. Also,increasedMDwasconcurrentlyfoundintherightfrontal These were then correlated with PCL-R scores (Factor 1: lobe of ASPD subjects which showed a significant positive ‘emotion dysfunction’; Factor 2: ‘antisocial behaviour’ and; correlationwithmeasuresofpsychopathy(Factor2scores). total PCL-R) using Pearson productemoment correlation Thefrontalcluster(Cluster2,Table2)thatshowedsignif- coefficients with Statistical Package for the Social Sciences icantFAreductionandnegativecorrelationswithseverityof (SPSS14.0forWindows,SPSSInc.,Chicago,IL,USA).Signifi- psychopathyintheASPDgroupincludedthegenuofcorpus cant correlations are reported where a Bonferroni adjusted callosum. Furthermore, it included right hemisphere struc- alphaof.025wasapplied. tures of the anterior corona radiata and anterior limb and Table2eWMFAandMDdifferencesinASPDrelativetohealthycontrols(clustersignificancethresholdforFAmapsp[.0025;clustersignificancethresholdforMDmaps p[.005). Cluster Clustersize TalairachandTournouxcoordinates Tract(s)withincluster Region ClustermeanFA ClustermeanFA p-value label (numberof ASPD(SD) Control(SD) x y z voxels) Cluster1 504 (cid:4)36 (cid:4)42 (cid:4)6 Temporo-occipitalcourseofleft Temporallobe .433(.022) .466(.029) .001378 ILFandIFOFa (cid:4)32 (cid:4)59 4 Leftposteriorthalamicradiation Occipitallobe andretrolenticularpartof internalcapsule c o r Cluster2 1027 16 47 (cid:4)8 RightIFOFandgenuofcorpus Frontallobe .383(.020) .409(.033) .000469 t e callosum x 19 30 (cid:4)8 RightUF Frontallobe 48 (cid:4)14 36 2 Leftgenuofcorpuscallosum Frontallobe (2 18 2 6 Rightgenuofinternalcapsule Sub-lobar 01 18 32 7 Rightgenuofcorpuscallosum Frontallobe 2) 23 11 9 Rightanteriorlimbofinternal Sub-lobar 2 1 capsuleandanteriorcorona 6 e radiataa 2 2 9 Clustersize TalairachandTournouxcoordinates Tract(s)withincluster Region ClusterMDASPDmm2 ClusterMDControlmm2 p-value (numberof sec(cid:4)1(cid:5)10(cid:4)3(SD) sec(cid:4)1(cid:5)10(cid:4)3(SD) voxels) x y z Cluster3 325 13 48 (cid:4)11 Rightanteriorcoronaradiata, Frontallobe .750(.026) .729(.031) .004513 genuofcorpuscallosum,and IFOFandUF a LocationofvoxelshowingmaximumFAdifference. 2 2 3 224 cortex 48 (2012) 216e229 Fig.1eReducedFAinASPDrelativetohealthycontrols[ascending2mmtransversesections]. Fig.2eIncreasedMDinASPDrelativetohealthycontrols[ascending2mmtransversesections]. cortex 48 (2012) 216e229 225 genuoftheinternalcapsuleinadditiontothefrontalcourseof andGaffan,1998).IndividualswithASPDand/orpsychopathy theUFandIFOF.Additionally,arightfrontalcluster(Cluster3, demonstrate deficits in reversal learning, a form of condi- Table2)whichencompassedthegenuofcorpuscallosumand tional associative learning characterised by a failure to anterior corona radiata, as well as the frontal course of the ‘reverse’apreviouslyrewardedresponsewhenitispunished IFOF and UF also showed a significant correlation between relative to controls (Table 1). Reversal learning deficits may increasedMDandpsychopathyscores.AlthoughFAreduction contribute to the perseveration of antisocial behaviour and wasalsofoundinposteriorregionsofthelefthemispherein high levels of recidivism characteristic of the disorder. the temporo-occipital course of the inferior longitudinal Further,fibresoftheUFconnecttheorbitofrontalcortexand fasciculus (ILF) and IFOF, in addition to the retrolenticular amygdala,whileimpairedregulationofamygdalaactivityby section and posterior thalamic radiation of the internal the orbitofrontal cortex may contribute to the behavioural capsule (Cluster 1, Table 2), these did not correlate signifi- disinhibition encountered both in ‘acquired sociopathy’ cantly with measures of psychopathy. Overall, our findings (Brower and Price,2001) andASPDandpsychopathy(Sarkar suggest that people with ASPD have WM microstructural etal.,2011).AlthoughabnormalitiesintheIFOFandanterior abnormalities involving frontal WM networks, and particu- corona radiata were also found in the frontal lobe in the larlyoftherighthemisphere. current study, functional impairments attributable to these InapreviousreportbyourgroupemployingDT-MRItrac- tractsareexaminedlaterinthediscussion. tographyexaminingFAandstreamlines(aproxymeasureof Additionally,FAdeficitswerefoundinthepresentstudyin tractvolume)oftheUF,ILFandIFOFofnineindividualswith the anterior limb of the internal capsule which contains psychopathy,FAwasfoundtobereducedonlyintherightUF; thalamo-frontal (anterior thalamic radiation), fronto- additionallythenumberofstreamlinesintheUFbilaterallywas thalamic and corticopontine fibres, and interconnects the correlated negatively with Factor 2 and total PCL-R scores dorsomedial and anterior thalamic nuclei with both the thoughFAdidnotshowanystatisticallysignificantcorrelations prefrontalandcingulatecortices.Lesionsoftheanteriorlimb (Craigetal.,2009).ThepresentfindingofreducedFAintheright of the internal capsule have been associated with ‘acquired frontal lobe encompassing the UF replicates our original sociopathy’(Molletal.,2003).Further,deficitsintheanterior finding,notingthattheseninesubjectswerealsoincludedin limb of the internal capsule have been suggested to lead to ourlargersampleof15individuals.Theincorporationofmore impairments in attention, perception and working memory subjects in our current study may have provided additional (Buchsbaum et al., 2006; Sepulcre et al., 2008). The anterior powertodetectfurthersignificantcorrelationswithFA. limboftheinternalcapsuletogetherwiththeanteriorcorona ApartfromFAdeficitsintheUF,thefrontallobeofASPD radiata (which connects the striatum with the anterior subjects in the current study also showed FA deficits in the cingulate cortex) have been postulated respectively to corpuscallosum,internalcapsule,anteriorcoronaradiataand contribute to attention impairments of alerting and conflict IFOF;thisislikelytoindicateinformationrelayimpairments processing(Niogietal.,2010).Disruptionintractssupporting between prefrontal and other brain regions e e.g., inter- thefunctionalintegrationofcorticalandsubcorticalregions hemisphericallyandwiththecingulum,pons,thalamusand involved in memory, attention, volition, learning and visual temporo-occipitalcortices(Catanietal.,2002;Schmahmann integration may contribute to the problems of people with and Pandya, 2008). Impaired functioning of the corpus cal- ASPDinadaptivelyrespondingtoalteredcontingenciesinthe losuminpsychopathshaspreviouslybeenpostulatedonthe socialandphysicalenvironment (includingsocialcuessuch basisofevidenceofprolongedinterhemispherictransfertime asfacialexpressions),expressedintraitssuchasimpulsivity relativetocontrols(HiattandNewman,2007).Lesionstudies ordifficultyinhibitingmotivatedresponses;alowthreshold of the corpus callosum have revealed its role in supporting for the discharge of aggression; and failure to learn from sensory-motor functional integration, attention, language, aversiveexperiences. interhemispheric transfer of associative learning, and WealsoreportforthefirsttimeinASPD,WMMDabnor- emotional regulation (Bellani et al., 2009; Glickstein and malitiesofthefrontallobe.MDofthediffusiontensorreflects Berlucchi, 2008; Zaidel and Iacoboni, 2003). Consequently, the magnitude of water molecule movement that is inde- thepresentfindingsofreducedFAandincreasedMDinthe pendentofdirectionandcontrastswithFAthatassessesthe corpus callosum of adults with ASPD, which also correlate directional preference of such movement (Le Bihan et al., with measures of psychopathy, may suggest that functions 2001). Increased MD has been reported for instance in mediated predominantly by the left hemisphere (e.g., vascularandneurodegenerativedisordersaffectingthebrain approach behaviour, language processing) may be relatively (Herveetal.,2005;Scolaetal.,2010),schizophrenia(Leeetal., unmodulated by functions mainly mediated by the right 2009;Narretal.,2009)andautism(Leeetal.,2007)indicating hemisphere (including behavioural inhibition and emotion lessrestrictedandthus,increasedmovementofwatermole- processing) as postulated by Hiatt and Newman (2007). This cules. Similarly in our present study, increased diffusion of proposedmechanismmayalsohelpexplaintheassociationof water molecules in the frontal lobe of those with ASPD reducedFAinthecorpuscallosumofdependentcocaineusers suggests a less coherent underlying WM microstructure. with increased impulsivity (Moeller et al., 2005) and WM Taken together with FA deficits concurrently found in the abnormalitiesinthecorpuscallosumofadolescentsengaged rightfrontallobe,suchdisorganisationofWMmicrostructure indangerousbehaviour(Bernsetal.,2009). mayhavearisenduetoabnormaldevelopmentofglialcells, Withregardtofrontallobetracts,abnormalitiesintheUF axonsorcellmembranes(Dongetal.,2004;Herveetal.,2005). are associated with impairments of conditional associative Furthermore,asfrontalabnormalitiesinthecurrentstudyare learning(GaffanandWilson,2008;Gutnikovetal.,1997;Parker lateralised and involve the right hemisphere, there may be
Description: