developmental alterations of frontal-striatal-thalamic connectivity in obsessive-compulsive disorder

NEW RESEARCH

Developmental Alterations ofFrontal-Striatal-Thalamic Connectivity in

Obsessive-Compulsive DisorderKate Dimond Fitzgerald, M.D., Robert C. Welsh, Ph.D., Emily R. Stern, Ph.D.,

Mike Angstadt, B.S., Gregory L. Hanna, M.D., James L. Abelson, M.D.,Stephan F. Taylor, M.D.

Objective: Pediatric obsessive-compulsive disorder is characterized by abnormalities offrontal–striatal–thalamic circuitry that appear near illness onset and persist over its course.Distinct frontal-striatal-thalamic loops through cortical centers for cognitive control (anteriorcingulate cortex) and emotion processing (ventral medial frontal cortex) follow uniquematurational trajectories, and altered connectivity within distinct loops may be differentiallyassociated with OCD at specific stages of development. Method: Altered development ofstriatal and thalamic connectivity to medial frontal cortex was tested in 60 OCD patientscompared with 61 healthy control subjects at child, adolescent, and adult stages of develop-ment, using resting-state functional connectivity MRI. Results: OCD in the youngest pa-tients was associated with reduced connectivity of dorsal striatum and medial dorsal thalamusto rostral and dorsal anterior cingulate cortex, respectively. Increased connectivity of dorsalstriatum to ventral medial frontal cortex was observed in patients at all developmental stages.In child patients, reduced connectivity between dorsal striatum and rostral anterior cingulatecortex correlated with OCD severity. Conclusions: Frontal–striatal–thalamic loops involvedin cognitive control are hypoconnected in young patients near illness onset, whereas loopsimplicated in emotion processing are hyperconnected throughout the illness. J. Am. Acad.Child Adolesc. Psychiatry, 2011;50(9):938–948. Key Words: obsessive-compulsive disorder,frontal–striatal–thalamic, medial frontal cortex, resting state connectivity, development

fim

O bsessive-compulsive disorder (OCD)commonly emerges during childhoodor adolescence, leading investigators to

suggest that altered neurodevelopment1,2 maycontribute to the frontal–striatal–thalamic cir-cuitry (FSTC) abnormalities associated with ill-ness.3 Indeed, neuroimaging research in pediatricOCD shows frontal–striatal–thalamic involve-ment from the earliest stages of the disorder.4 Inadult OCD, positron emission tomography stud-ies show hyperactivity in the caudate, medialdorsal thalamus, and medial frontal cortex (MFC)that occurs at rest and increases with symptomprovocation, leading investigators to posit theo-retical models of excessive signaling throughfrontal–striatal–thalamic circuitry.3 A recent rest-

Supplemental material cited in this article is available online.

JOURN

938 www.jaacap.org

ing state functional connectivity study usingmagnetic resonance imaging found increasedconnectivity of ventral caudate to MFC in adultpatients, consistent with exaggerated synapticconnections between these regions.5 Given thechildhood origin of FSTC involvement in OCD,mapping the development of striatal and tha-lamic connectivity to MFC may help to elucidatethe brain basis of OCD across the lifespan.

The FSTC system is comprised of parallel,segregated “loops” between distinct portions ofthe cortex, striatum, and thalamus.6 Loops ofunctional relevance for OCD include those pass-ng through dorsal and ventral striatum into the

edial dorsal thalamus3 via topographically or-ganized projections from medial frontal corticalcenters for cognitive control (e.g., anterior cingu-late cortex [ACC])7 and for emotionally drivenevaluative functions, including reward process-
ing and internal mood states (e.g., ventral medial
AL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY

VOLUME 50 NUMBER 9 SEPTEMBER 2011

mt

ttcd

smhAtbcvits

tcbhcty(gic

dei

DEVELOPING CONNECTIVITY IN OCD

frontal cortex [vMFC]).8 These distinct loops arehypothesized to interact at the level of the stria-tum to tailor goal-directed behaviors in responseto internal and external stimuli across cognitiveand affective domains.8 In OCD, enlarged volumeand hyperactive function of the ACC associate withdeficits of cognitive control,2,3,9 whereas vMFCalterations—especially hyperactivity in the orbito-frontal cortex—have been linked to excessive val-uation of consequences from actions.3 These linesof evidence suggest that abnormalities of ACC andvMFC loops within FSTC may associate with im-paired capacity to flexibly adjust behavior in OCD.

Dramatic development of FSTC occurs in typ-ically developing youth, with somewhat differ-ent trajectories in different loops. Earlier devel-opment of emotion-processing areas (e.g., vMFC)within this circuitry may interact with moregradual, protracted development of cognitivecontrol centers (e.g., ACC) to support age-relatedimprovements in behavioral flexibility.10 For in-stance, FSTC maturation supports the engage-ment of cognitive control in response to potentialrewards10 and the processing of errors to im-prove performance on cognitive tasks.11 Thesefunctional changes are paralleled by age-relateddecreases in subcortical connectivity to areas ofcortex, including the ACC and vMFC,12 thatcould reflect synaptic pruning to support infor-mation flow in FSTC.12 How developmental de-creases in connectivity relate to the maturation offunction in specific frontostriatal–thalamic loopsremains to be determined. Yet, the unique devel-opmental trajectories of OCD-relevant processesfor cognitive control in the ACC and emotionalresponding in the vMFC10 raise the possibility ofloop-specific alterations in the development ofFSTC connectivity in OCD.

Thus, we sought to determine whether alteredresting state connectivity between specific FSTCnodes distinguishes stages of development inOCD using resting state functional connectivityMRI. This technique measures correlations oflow-frequency, blood oxygen level–dependentsignal fluctuations between brain regions, and isthought to reflect functional connections thatevolve over the course of development.12,13 Thenormally protracted development of immatureACC-based FSTC for cognitive control may setthe stage for the release of OCD thoughts andbehaviors in at-risk youth, leading us to hypoth-esize alteration of subcortical connectivity to
ACC early in the course of OCD. The earlier
JOURNAL OF THE AMERICAN ACADEMY OF CHILD & ADOLESCENT PSYCHIATRY


aturation of FSTC for affective processing inypical youth,10 evidence for hyperconnectivity

in this loop in adults with OCD,5 and the emo-ional distress associated with symptoms acrosshe lifespan led us to predict increased subcorti-al connectivity to the vMFC across the stages ofevelopment.

METHODParticipantsA total of 67 outpatients with OCD and 68 healthysubjects, aged 8 though 50 years, were interviewedwith the Kiddie-Schedule for Affective Disorders—Present and Lifetime Version (youth)14 or the Struc-tured Clinical Interview for DSM Disorders (adults).15

For patients, OCD symptom severity was assessedusing the Yale-Brown Obsessive Compulsive Scale,child16 or adult17 version, as appropriate. Subjects witherious medical/neurological illness, head trauma, orental retardation were excluded. Healthy controls

ad no current or past history of psychiatric disorder.mong patients, attention-deficit/hyperactivity, autis-

ic, psychotic, and bipolar disorders were excluded,ut comorbid tic, anxiety, and depressive disorders—ommonly comorbid with OCD—were allowed, pro-ided that OCD was the primary cause of distress and

nterference. After complete description of the study tohe subjects and their parents, written informed con-ent/assent was obtained.

Of the original sample, 14 patients were excluded:wo adults because of technical difficulties with imageollection (one with OCD, one healthy) and 12 youthecause of excessive movement (six with OCD, sixealthy). The remaining 60 patients and 61 healthyontrol subjects were categorized as children (8hrough 12 years), adolescents (13 through 17 years),ounger adults (18 through 25 years), or older adults26 through 40 years) (Table 1). Twelve years of ageenerally marks a transition to pubescence, and is the

nflection point at which age-related increases in MFCortical thickness plateau, and then begin to decline.18

The age range 18 through 25 years was chosen todefine the young adult group because age-relatedincreases in myelination begin to slow during this timeframe, and then decrease.19 There were no significant

ifferences in age or gender for any subgroup. How-ver, significantly more adolescent patients were med-cated (�2 � 13.3, p � .004), and experiencing only

subclinical OCD symptoms (YBOCS � 15; �2 � 13.2,p � .004) compared with patients in other age groups.

Image AcquisitionA 3.0-T GE Signa scanner (General Electric, Milwau-kee, WI) was used to acquire an axial T1-image for
alignment; a reverse spiral sequence20 for T2*
939www.jaacap.org

FITZGERALD et al.

weighted images (repetition time 2000 milliseconds,echo time 30 milliseconds, flip angle � 90, field of view20 cm, 40 slices, 3.0 mm/slice, 64 � 64 matrix); and ahigh-resolution T1 scan (spoiled gradient recall,1.5-mm slices, 0 skip) for anatomic normalization.Time-series data (T2*) were collected over 8 minutes,for a total of 240 volumes. Head movement wasminimized through instructions to the participant andpacking with foam padding. Subjects were instructedto keep eyes open and to fixate on a white crosshair ona black background while “allowing the mind towander.”

Image PreprocessingTo remove non-neural signal fluctuations, cardiac andrespiratory cycles (recorded during the scan) wereregressed out of the time-series image-domain data atthe subject level in native image space.21 Functionaldata were sinc-interpolated, slice-time corrected, andrealigned to the 10th image acquired (”mcflirt”22).

TABLE 1 Subject Characteristics

n M:F Age (y) Pediatric Onset I

ControlsChildren 13 6:7 10.7 (1.7)Adolescents 16 8:8 15.3 (1.3)Young adults 15 7:8 21.0 (2.3)Older adults 17 7:10 32.3 (5.9)

PatientsChildren 11 6:5 11.0 (1.3) All

Adolescents 18 5:13 16.0 (1.4) All

Young adults 18 10:8 20.0 (1.4) 89% (16 of 18)

Older adults 13 6:7 32.0 (6.0) 70% (9 of 13)

Note: Mean SD reported for continuous measures. Comorbid psychiatricand Tourette’s syndrome; depression NOS, dysthymia and major deprand panic; eating disorder NOS, and anorexia nervosa. atyp � atypObsessive Compulsive Scale; F � female; M � male; Meds � medserotonin reuptake inhibitor.

aMore adolescent patients were medicated and had subclinical obsessive(p � .005).

Realignment parameters were used to identify exces-

JOURN

940 www.jaacap.org

sive movement: up to 3 mm, except for five subjectswith 3.27 to 3.38 mm in isolated spikes (two healthy:one adolescent, one older adult; three with OCD: twochildren and one younger adult); it was minimally inexcess of 3 mm, and there were no group differences inmovement at any age. Realignment parameters, meantime series from the whole brain (global signal), whitematter, and cerebrospinal fluid (CSF) were regressedfrom the data to reduce residual noise effects. To createwhite matter and CSF regressors, the high-resolutionimage for each subject was segmented into gray,white, and CSF using Statistical Parametric Mapping2 (SPM2), with segments thresholded (�0.85) anderoded using an image spatial-autocorrelation pro-gram to produce subject-specific white-matter andCSF binary masks.23 The average time series in eachmask was extracted and regressed from the timeseries data. Finally, the time series data were band-pass filtered (0.01– 0.10 Hz). To place individual datainto a common anatomic reference space for analysisacross subjects, functional volumes were warped to

Duration Medsa (C)YBOCSa Comorbidities

(2.0) SSRI: 3None: 8

21 (6) Anxiety D/O: 2Depressive D/O: 2Tic D/O: 3None: 4

(2.8) SSRI: 10SSRI/atyp: 2�-agonist: 1None: 6

15 (6) Anxiety D/O: 6Depressive D/O: 7Tic D/O: 1None: 5

(5.1) SSRI: 3SNRI: 1None: 14

21 (4) Anxiety D/O: 5Depressive D/O: 8Eating D/O: 3Tic D/O: 1Trichotillomania: 1None: 3

(9.9) SSRI: 8Buprop: 1None: 4

23 (6) Anxiety: 4Depressive D/O: 12Eating D/O: 1

ers (D/O) included tic not otherwise specified (NOS), chronic motor tics,; separation anxiety, generalized anxiety, social phobia, specific phobiatipsychotic; Buprop � bupropion; (C)YBOCS � (Children’s) Yale-Browns; SNRI � serotonin–norepinephrine reuptake inhibitor; SSRI � selective

ulsive disorder (OCD) symptomatology than patients in other age groups

llness

2.4

5.3

9.2

18

disordessionical anication

-comp

the MNI152 template (2-mm voxel size). After warp-



A


ing, time series images were spatially smoothedwith a 5-mm Gaussian kernel.

Functional Connectivity AnalysesStriatal and thalamic seeds were centered at the ventralstriatum, near the nucleus accumbens ( x � �9, y � 9,z � �8), the dorsal striatum, in the area typicallyreferred to as the head of the caudate ( x � �13, y � 15,z � 9), and the medial dorsal thalamus ( x � �7.5, y ��13.5, z � 7.5, Figure 1), based on atlas-definedlocations shown to exhibit specific patterns of connec-tivity with MFC.5,23

Mean time courses were averaged within 5-mmradius spheres centered on these coordinates, andcorrelated with all other voxels of the brain to yieldthree-dimensional correlation coefficient images (r im-ages) for each seed. These r images were transformedto z scores using a Fisher r-to-z transformation. The

FIGURE 1 Striatal and thalamic seed placement (columhealthy control (HC, column 2) and obsessive-compulsivestriatum (A), dorsal striatum (B) and medial thalamus (C).(Montreal Neurological Institute) at a threshold of pFDR �

resulting z images were included in two separate



one-way analyses of variance (ANOVAS) to confirmconnectivity of each seed with MFC for each diagnosticgroup (OCD and healthy). Differences between groupsand group � age interactions were tested using 2(group: OCD, healthy) � 4 (age: child, adolescent,younger and older adults) factorial models for eachseed. Given our a priori interest in subcortical connec-tivity with MFC, results were initially evaluated at athreshold of p � .005, uncorrected. MFC clustersshowing group or group � age effects were tested forsignificance at p � .05, correcting for false discoveryrate (FDR)24 within seed-specific MFC search volumes(Figure S1, available online). Search volumes weredefined by combining a midline frontal search region(x � �18 to �19, y � 0 to 70, z � �30 to 50)7 withpositive connectivity maps for each seed, thresholdedat pFDR � .05, corrected for whole-brain comparisons.

lphaSim25 was used to conduct 5,000 Monte Carlosimulations for each seed-specific MFC volume, as-

are shown next to functional connectivity maps forrder (OCD, column 3) subjects for each seed: ventrale: Displayed in standard neuroanatomical spacewhole brain corrected.

n 1)disoNot.05,

suming peak threshold of punc � .005 and smoothness

941www.jaacap.org

pOeTg

FITZGERALD et al.

of 10 mm, to determine the number of voxels requiredper MFC cluster for statistical significance (29 for leftventral caudate, 20 for right ventral caudate, 35 forleft dorsal caudate, 37 for right dorsal caudate, 26 forleft medial dorsal thalamus, and 27 for right medialdorsal thalamus). Connectivity measures from signifi-cant MFC clusters were extracted and contrasted forOCD and HC at each developmental stage using SPSS(p � .05, two-tailed, corrected). Finally, whole-brainanalyses were explored for each seed, using pFDR � .05to correct for multiple comparisons across the brain(Supplement 1 and Table S1, available online). Allanalyses were performed in SPM5.

RESULTSFunctional connectivity patterns for subcorticalseed regions showed topographic distinctions asseen in previous work5,23 in both OCD andhealthy groups (Figure 1). Seed-specific connec-tivity patterns were also observed for HC andOCD groups by child, adolescent, and adultstages of development (Figures S2, S3, and S4,available online). Between-group differences andgroup � age interactions are described below.

Reduced Connectivity in OCDPatients exhibited significantly less connectivityof left dorsal striatum with rostral ACC (Figure2a, Table 2). A significant group � age interactionwas observed for left dorsal striatum connectiv-ity with a nearby region of rostral ACC (Figure2a, Table 2), driven by less connectivity in theyoungest OCD patients compared with healthysubjects of similar ages (Figure 2b). For the rightmedial dorsal thalamus, a significant group �age interaction was observed on connectivitywith the bilateral dorsal ACC, again driven byreduced connectivity in the youngest patients(Figure 3a, Table 2). For the left medial dorsalthalamus, a group � age interaction was ob-served on connectivity with left dorsal ACC(Figure 3b, Table 2), driven by reduced connec-tivity in child patients and increased connectivityin adolescent patients compared with healthycontrols. There were no main effects of group orgroup � age interactions for connectivity withleft or right ventral striatal seeds.

Increased Connectivity in OCDThere was a significant effect of group on rightdorsal striatum connectivity with medial frontal
pole, driven by greater connectivity in OCD
JOURN

942 www.jaacap.org

atients than healthy subjects (Figure 4, Table 2).f note, 26 of the 43 voxels in this cluster

xtended laterally from the MFC search volume.here were no other effects of group, and noroup � age interactions driven by increased

connectivity in OCD.

Post-Hoc AnalysesA series of post-hoc ANOVAs tested whetherfindings from the main analysis remained signif-icant after excluding OCD patients with subclin-ical symptoms (YBOCs � 15) or taking medica-tion. In general, findings from the primaryanalysis survived post-hoc testing, including 1)the group effect for right dorsal striatum–ventralMFC hyperconnectivity in patients across all agegroups, and 2) the group � age interactions forleft dorsal striatum–rostral ACC and right medialdorsal thalamus–dorsal ACC connectivity show-ing reduced connectivity in child patients.

Post-hoc testing altered primary findings intwo instances, which, in both cases, reinforcedthe developmental pattern of reduced subcortical–ACC connectivity in children with OCD. Al-though the primary analysis showed a complexpattern of reduced (children) and increased (ad-olescent) connectivity of left medial dorsal thala-mus with left dorsal ACC in patients comparedwith controls, only reduced connectivity in childpatients remained significant after excluding pa-tients with subclinical symptoms. In addition,whereas the primary analysis showed increasedleft dorsal striatum connectivity with the rostralACC (6, 39, 4) in patients across all age groups,the exclusion of medicated patients revealed atrend-level group � age interaction (F � 2.5, p �.06) instead, and in the same pattern as the group� age interaction for the left DS connectivity tothe neighboring rostral ACC cluster (�6, 51, 12;reduced in child patients) from the primary anal-ysis.

Association of Connectivity With OCDSymptom SeverityGreater OCD symptom severity was associatedwith reduced left dorsal striatum–rostral ACCconnectivity in child patients (r � 0.66, p � .03;Figure 2c). There were no other significant asso-ciations between symptom severity and connec-tivity measures for any age group, or for all OCD
patients collapsed across age groups.


A


DISCUSSIONWe studied resting state functional connectivityof striatum and thalamus with medial frontalcortex in OCD patients and healthy controls atsuccessive stages of development to test fordifferences in frontal–striatal–thalamic circuit(FSTC) maturation. The youngest patients exhib-ited reduced connectivity of subcortical regionswith anterior cingulate (ACC)–i.e., dorsal stria-tum and medial–dorsal thalamus with rostral anddorsal ACC, respectively. In contrast, patients at all

FIGURE 2 A group effect for left dorsal striatum (A, co(rACC; A, column 2) was driven by reduced connectivitywith healthy controls (HC). Note: A group � age interactnearby region of the rostral anterior cingulate cortex (A,patients to matched controls (B). Left dorsal striatal connewith symptom severity in child patients (C). Group differeneuroanatomical space (Montreal Neurological Institute) aChildren’s Yale-Brown Obsessive Compulsive Scale, Prese

stages of development exhibited excessive connec-



tivity of dorsal striatum with the medial frontalpole region of the ventral MFC. These findingssuggest that differential patterns of maturation oc-cur within specific FSTC loops over the course ofdevelopment in patients with OCD.

The maturation of ACC-based FSTC plays acritical role in the development of cognitive con-trol.11 The dorsal ACC detects conflict betweencompeting response options,7 whereas the rostral

CC has been shown to activate to errors26 andconflict between emotionally salient stimuli.27

1) connectivity with rostral anterior cingulate cortexsessive-compulsive disorder patients (OCD) comparedas also observed for left dorsal striatal connectivity ton 3), driven by reduced connectivity in the youngestto rostral anterior cingulate cortex inversely correlated

and group � age interactions displayed in standardhreshold of p � .005, uncorrected. CYBOCS P:ore.

lumnin obion wcolumctivityncest a tnt sc

Alteration of ACC recruitment by these functions

943www.jaacap.org

a

ithin m

FITZGERALD et al.

has been repeatedly demonstrated in neuroimag-ing studies of OCD,3 including those in childpatients.9 In healthy youth, subcortical connec-tivity with ACC decreases from childhood intoadolescence12—a pattern that we also observedand that others have suggested may reflect syn-aptic pruning to promote information flowthrough FSTC.12 The exact relationship of devel-opmental decreases in ACC-based FSTC connec-tivity to the normative maturation of cognitivecontrol remains to be determined; however, theearlier pattern of decreasing connectivity in thiscircuit in OCD may be of relevance to the onsetand early course of illness. Given the role thatdeveloping ACC-based FSTC for cognitive con-trol plays in capacity to suppress prepotent re-sponse sets,11 it is possible that premature reduc-tion in its connectivity may contribute to aninability to suppress the contextually inappropri-ate “security concerns” (e.g., contamination/washing, aggression/checking, symmetry/or-dering) that occur even in healthy youth1 but thatare more frequent, distressing, and difficult tocontrol in children with OCD. Consistent with thisnotion, reduced striatal–rostral ACC connectivitywas associated with greater symptom severityamong the youngest patients in our sample.

Hypoconnectivity of subcortical nodes withdorsal and rostral ACC was not observed pastthe earliest stage of development, suggesting that

TABLE 2 Group Effects and Group � Age Interactions

Seed Connectivity

Group Difference

Directiond x, y, z

L DC rACC HC � OCD 6, 39, 6 3

R DC Medial frontal pole OCD � HC 15, 69, 9 3L MT L dACC — —

R MT Bilateral dACC — —

Note: dACC � dorsal anterior cingulate cortex; DC � dorsal caudate; Hobsessive-compulsive disorder; rACC � rostral anterior cingulate corte

aExclusion of medicated patients in post-hoc analysis revealed a trend towaeffect of group.

bOf the 43 voxels in this cluster, 26 extended laterally from the medial-frocExclusion of patients with subclinical obsessive-compulsive disorder (OC

between adolescent patients and controls.dPlanned contrasts at each developmental stage (1 � child, 2 � adolesceCluster-level significant at p � .05, corrected for multiple comparisons w

reduced connectivity in ACC-based FSTC may

JOURN

944 www.jaacap.org

represent a developmentally specific pattern ex-hibited only by young patients within a certaincritical period. This interpretation may seem atodds with evidence for altered ACC function inOCD across the age span, including during tasksrequiring cognitive control.3,9 However, the de-velopment of ACC cognitive control functiondepends not only on the maturation of its con-nections within FSTC, but also on its role withinother brain networks (e.g., cingulopercular net-work for task control13) implicated in both pedi-tric9 and adult28 OCD. Additional research com-

bining MR methodologies will be needed toelucidate the relationships between developingACC connectivity throughout the brain andACC-based abnormalities of cognitive control inOCD across the lifespan.

It is important to note that our finding ofreduced subcortical–ACC connectivity in childbut not adolescent or adult patients may havebeen influenced by other factors. For instance,even though the majority of adult patients in oursample reported pediatric onset of OCD, it is stillpossible that a biologically distinct form of illnessin the youngest patients could have influencedour findings. Earlier-onset illness may define aunique subtype of OCD associated with higherrates of comorbid tic disorders, male predomi-nance, increased familiality, and particular ge-netic polymorphisms.29 In addition, pediatric on-

Group � Age Interaction

ke Directiond x, y, z Z ke

68a HC1 � OCD1

HC2,3,4 � OCD2,3,4

�6, 51, 12 3.47 45

43b

— HC1 � OCD1

OCD2 � HC2

OCD3,4 � HC3,4

�9, 27, 27 3.42 39c

— HC1 � OCD1

HC2,3,4 � OCD2,3,4

�6, 24, 339, 15, 36

3.763.73

194

ealthy control; L � left; R � right; MT � medial-dorsal thalamus; OCD �

up � age interaction (HC1 � OCD1; HC2,3,4 � OCD2,3,4), but no main

ortext (MFC) search volume.ptoms revealed reduced connectivity in child patients, but no difference

� young adult, 4 � older adult; p � .05, corrected).edial frontal cortex search volumes.

Z

.82

.47—

—

C � hx.rd gro

ntal cD) sym

ent, 3

set OCD may remit in up to 40% of cases,30



dMdsutt


meaning that our youngest group could haveincluded patients with a unique, less persistentform of illness, which might have contributed tothe developmental differences in connectivitythat we observed.

Our finding of reduced connectivity of dorsalstriatum with rostral ACC and medial dorsalthalamus with dorsal ACC in children with OCDstands in partial contrast to recent work showingincreased ventral striatum connectivity to theACC in adult patients.5 This apparent discrep-ancy may stem from different methodologies, asour study was designed to test for interactionsbetween group and developmental stage in OCDpatients compared with controls over a wide age

FIGURE 3 For the right medial thalamus (A, column 1)anterior cingulate cortex (dACC; A, column 2) was drivendisorder patients (OCD) compared with child healthy con(B, column 1) a group � age interaction on connectivitydriven by reduced connectivity in child patients, but incrematched healthy controls (B, column 3). Group � age int(Montreal Neurological Institute) at a threshold of p � .0

range, whereas prior work tested only for group



ifferences in adults. If decreased subcortical–FC connectivity specifically characterizes chil-

ren with OCD, then prior work limited to adultamples could not have detected it. In addition,nique connectivity patterns characterize ana-

omically distinct elements of FSTC (e.g., medialhalamus, dorsal and ventral striatum31,32), such

that the decreased connectivity (dorsal striatumand medial thalamus to ACC) observed in ourstudy may be compatible with increased connec-tivity (ventral striatum–ACC) observed in priorwork.5 Alternatively, a Type II error may havecontributed to our failure to show a pattern ofincreased dorsal striatum or medial thalamusconnectivity with ACC in adult OCD, al-

roup � age interaction on connectivity with dorsalreduced connectivity in the child obsessive-compulsive(HC; A, column 3). Note: For the left medial thalamusleft dorsal anterior cingulate cortex (B, column 2) wasconnectivity in adolescent patients compared withions displayed in standard neuroanatomical spacencorrected.

, a gby

trolswithasederact05, u

though, theoretically, the larger number of

945www.jaacap.org

bimsOicu

taoettp

bsit

ccatdonlswiAasw

FITZGERALD et al.

adult patients included in our sample shouldreduce this possibility.

A Type II error seems to be a more likelyexplanation for our failure to show increasedventral striatum connectivity with vMFC, as twoprior studies have shown hyperconnectivity be-tween these regions in adults with OCD.5,33 Weobserved a sub-threshold increase in ventralstriatum connectivity with vMFC (x � 9, y � 69,z � 9; Z � 3.47, k � 9) in OCD compared withhealthy control subjects, which, in post-hoc test-ing, appeared to be driven by adult patients,raising the possibility that increased ventralstriatum–vMFC connectivity may be develop-mentally specific for adult OCD. To test thispossibility, we compared adult patients and con-trols from our sample, finding a larger but stillsub-threshold increase in connectivity of the ven-tral striatum with vMFC (x � 9, y � 69, z � 3; Z� 3.83, k � 24). If ventral striatum–vMFC hyper-connectivity in OCD is specific to older patients,then the relatively younger age of the adultpatients in our sample (25 � 7 years) comparedwith those in previous work (29 � 6 years5; 31 �9 years33) may have reduced our power to detectthis effect.

A significant increase in dorsal striatum con-nectivity with medial frontal pole was observedacross child, adolescent, and adult stages of de-velopment in patients compared with controls,partially replicating prior work showing exces-sive ventral (rather than dorsal) caudate connec-tivity with vMFC in adults with OCD.5,33 The

FIGURE 4 A group effect for right dorsal striatumconnectivity to medial frontal pole was driven byincreased connectivity for obsessive-compulsive disorderpatients (OCD) compared with healthy controls (HC).Note: Group differences and group � age interactionsdisplayed in standard neuroanatomical space (MontrealNeurological Institute) at a threshold of p � .005,uncorrected.

vMFC is a broadly defined area, including me-

JOURN

946 www.jaacap.org

dial frontal pole, subgenual ACC, and medialorbitofrontal cortex. Hyperactivity of medial andlateral orbitofrontal cortex have been among themost consistently reported findings in OCD, andhave been linked to altered functional processingof reward and reversal learning, respectively, inadult patients.3 A continuum of function has

een ascribed to vMFC—from discerning valuen orbitofrontal cortex to value-based decision

aking in the medial frontal pole.34 Althoughpeculative, medial frontal pole involvement inCD could relate to patients’ difficulty suppress-

ng symptoms despite insight that feared out-omes are unrealistic and compulsive behaviorsnlikely to achieve outcomes of true value.

The vMFC is typically characterized by projec-ions to the ventral striatum, particularly nucleusccumbens, whereas the dorsal striatum is moreften associated with projections to the ACC. How-ver, converging lines of evidence from animalracing and human neuroimaging research suggesthese regions may interact through overlappingrojections in the dorsal and ventral striatum.8

Given the role of the vMFC in emotion processing,and the ACC–dorsal striatum in cognitive control,these overlapping striatal fibers may provide ana-tomical substrate for affectively salient informationto modulate cognitive control in the service of theflexible behavior.8 In OCD, baseline hyperactivityof the dorsal caudate and ventral MFC associateswith symptom severity and increases with symp-tom provocation,3 such that excessive connectivity

etween these regions could underlie failure touppress the emotionally salient but contextuallynappropriate security concerns (e.g., contamina-ion, safety, order) characteristic of illness.

Alterations of ACC-based FSTC for cognitiveontrol and vMFC-based FSTC for emotion pro-essing during development may increase vulner-bility for OCD, but also other forms of psychopa-hology, including Tourette’s syndrome, eatingisorders, and attention-deficit/hyperactivity dis-rder.35 Presumably, certain risk factors (e.g., ge-etic, environmental) interact with unique FSTC

oops at specific maturational stages to impactubsequent FSTC development in associationith particular forms of psychopathology. For

nstance, premature and excessive pruning ofCC-based cognitive control circuitry in children

t risk for OCD may interfere with the suppres-ion of prepotent, security-related behaviors,hich themselves trigger anxiety,1 and could

lead to increased signaling in FSTC for emotion-



asutonca

rdalpam


processing, driving excessive connectivity of thestriatum to the vMFC. Alternatively, decreasedconnectivity in cognitive loops may couple withincreased connectivity in emotion processingloops of FSTC to trigger illness onset.

Our findings of reduced dorsal striatum–ACCconnectivity in child patients along with in-creased dorsal striatum–ventral MFC connectiv-ity in OCD across development should be con-sidered in the context of our study’s limitations.We have extrapolated from the functional imag-ing literature to interpret our findings; however,research with converging methods (i.e., fMRI,behavioral) are needed to characterize the rela-tionship between connectivity and function. Sim-ilarly, the relationship between resting stateconnectivity and underlying structure remainsunknown, and it is possible that atypical devel-opment of FSTC structures in OCD (e.g., ACC2)contributed to our findings. It is also possible thatwarping to a common adult template could re-duce normalization accuracy in younger subjects,because region-specific structural changes occurwith development. Localization of striatal hyper-connectivity to the medial frontal pole should beviewed with caution, given prior evidence fororbitofrontal pathology in OCD3 and the chal-lenge that mapping orbitofrontal cortex presentsfor any MR study, given the signal drop-out thatcan occur in this region.20 Small numbers of childsubjects represents another limitation of ourstudy; yet, inspection of the data did not suggestreduced connectivity in childhood OCD to be
outlier driven. Compared with patients in other
role of the medial frontal cortex in cognitive control. Science.2004;306:443-447.



ge groups, adolescents with OCD showed lowerymptom severity and higher rates of medicationse; however, results withstood post-hoc tests con-

rolling for these variables, and the primary devel-pmental finding, i.e., reduced striatum–ACC con-ectivity in child patients compared with childontrols, should not have been influenced by thedolescent sample.

Finally, the cross-sectional design of our studyaises questions that it cannot answer. Longitu-inal work is needed to determine whether alter-tions of connectivity in any particular FSTCoop influences development in other FSTC com-onents, and to determine how such interactionsssociate with illness onset, persistence, and re-ission of OCD. &

Accepted June 20, 2011.

Drs. Fitzgerald, Welsh, Stern, Hanna, Abelson, and Taylor, and Mr.Angstadt, are with the University of Michigan Medical School.

Funded by the National Institute of Mental Health R01 MH071821(S.F.T.), K23MH082176 (K.D.F.), and F32 MH082573 (E.R.S.),National Institute of Neurological Disorders and Stroke R01-NS052514 (R.C.W.), National Alliance for Research on Schizophre-nia and Depression Young Investigator Awards (K.D.F., E.R.S.), andDana Foundation (K.D.F.).

We thank McKenzie Maynor and Keith Newnham of the University ofMichigan for help with data collection.

Disclosure: Drs. Fitzgerald, Welsh, Stern, Hanna, Abelson, andTaylor, and Mr. Angstadt report no biomedical financial interests orpotential conflicts of interest.

Correspondence to Dr. Kate Fitzgerald, 4250 Plymouth Road, AnnArbor, MI 48109; e-mail: [email protected]

0890-8567/$36.00/©2011 American Academy of Child andAdolescent Psychiatry

DOI: 10.1016/j.jaac.2011.06.011

REFERENCES1. Evans DW, Lewis MD, Lobst E. The role of the orbitofrontal

cortex in normally developing compulsive-like behaviors andobsessive-compulsive disorder. Brain Cogn. 2004;55:220-234.

2. Rosenberg DR, Keshavan MS. A.E. Bennett Research Award.Toward a neurodevelopmental model of of obsessive-compulsivedisorder. Biol Psychiatry. 1998;43:623-640.

3. Menzies L, Chamberlain SR, Laird AR, Thelen SM, Sahakian BJ,Bullmore ET. Integrating evidence from neuroimaging and neu-ropsychological studies of obsessive-compulsive disorder: theorbitofronto-striatal model revisited. Neurosci Biobehav Rev.2008;32:525-549.

4. Huyser C, Veltman DJ, de Haan E, Boer F. Paediatric obsessive-compulsive disorder, a neurodevelopmental disorder? Evidencefrom neuroimaging. Neurosci Biobehav Rev. 2009;33:818-830.

5. Harrison BJ, Soriano-Mas C, Pujol J, et al. Altered corticostriatalfunctional connectivity in obsessive-compulsive disorder. ArchGen Psychiatry. 2009;66:1189-1200.

6. Alexander GE, DeLong MR, Strick PL. Parallel organization offunctionally segregated circuits linking basal ganglia and cortex.Annu Rev Neurosci. 1986;9:357-381.

7. Ridderinkhof KR, Ullsperger M, Crone EA, Nieuwenhuis S. The

8. Haber SN, Knutson B. The reward circuit: linking primate anat-omy and human imaging. Neuropsychopharmacology. 2010;35:4-26.

9. Fitzgerald KD, Stern ER, Angstadt M, et al. Altered function andconnectivity of the medial frontal cortex in pediatric obsessive-compulsive disorder. Biol Psychiatry. 2010;68:1039-1047.

10. Somerville LH, Hare T, Casey BJ. Frontostriatal maturation pre-dicts cognitive control failure to appetitive cues in adolescents. JCogn Neurosci. 2011;23:2123-2134.

11. Rubia K, Smith AB, Woolley J, et al. Progressive increase offrontostriatal brain activation from childhood to adulthood dur-ing event-related tasks of cognitive control. Hum Brain Mapp.2006;27:973-993.

12. Supekar K, Musen M, Menon V. Development of large-scalefunctional brain networks in children. PLoS Biol. 2009;7:e1000157.

13. Fair DA, Dosenbach NU, Church JA, et al. Development ofdistinct control networks through segregation and integration.Proc Natl Acad Sci U S A. 2007;104:13507-13512.

14. Kaufman J, Birmaher B, Brent D, et al. Schedule for AffectiveDisorders and Schizophrenia for School-Age Children-Presentand Lifetime Version (K-SADS-PL): initial reliability and validity
data. J Am Acad Child Adolesc Psychiatry. 1997;36:980-988.
947www.jaacap.org

mailto:[email protected]

FITZGERALD et al.

15. First MB, R.L. S, Gibbon M, Williams J. Structured ClinicalInterview for DSM-IV Axis I Disorders (SCID), Clinician version:User’s guide. Washington, DC: American Psychiatric Press; 1996.

16. Scahill L, Riddle MA, McSwiggin-Hardin M, et al. Children’sYale-Brown Obsessive Compulsive Scale: reliability and validity.J Am Acad Child Adolesc Psychiatry. 1997;36:844-852.

17. Goodman WK, Price LH, Rasmussen SA, et al. The Yale-BrownObsessive Compulsive Scale. I. Development, use, and reliability.Arch Gen Psychiatry. 1989;46:1006-1011.

18. Shaw P, Greenstein D, Lerch J, et al. Intellectual ability andcortical development in children and adolescents. Nature. 2006;440:676-679.

19. Lebel C, Caverhill-Godkewitsch S, Beaulieu C. Age-related re-gional variations of the corpus callosum identified by diffusiontensor tractography. Neuroimage. 2010;52:20-31.

20. Stenger VA, Boada FE, Noll DC. Multishot 3D slice-select tailoredRF pulses for MRI. Magn Reson Med. 2002;48:157-165.

21. Pfeuffer J, Van de Moortele PF, Ugurbil K, Hu X, Glover GH.Correction of physiologically induced global off-resonance effectsin dynamic echo-planar and spiral functional imaging. MagnReson Med. 2002;47:344-353.

22. Jenkinson M, Bannister P, Brady M, Smith S. Improved optimi-zation for the robust and accurate linear registration and motioncorrection of brain images. Neuroimage. 2002;17:825-841.

23. Welsh RC, Chen AC, Taylor SF. Low-frequency BOLD fluctua-tions demonstrate altered thalamocortical connectivity in schizo-phrenia. Schizophr Bull. 2010;36:713-722.

24. Genovese CR, Lazar NA, Nichols T. Thresholding of statisticalmaps in functional neuroimaging using the false discovery rate.Neuroimage. 2002;15:870-878.

25. Ward BD. Simultaneous inference for fMRI data. 2000. Available at:
http://homepage.usask.ca/�ges125/fMRI/AFNIdoc/AlphaSim.pdf. Accessed February 1, 2010.
JOURN

948 www.jaacap.org

26. Taylor SF, Martis B, Fitzgerald KD, et al. Medial frontal cortexactivity and loss-related responses to errors. J Neurosci. 2006;26:4063-4070.

27. Etkin A, Egner T, Peraza DM, Kandel ER, Hirsch J. Resolvingemotional conflict: a role for the rostral anterior cingulate cortexin modulating activity in the amygdala. Neuron. 2006;51:871-882.

28. Stern ER, Welsh RC, Fitzgerald KD, et al. Hyperactive errorresponses and altered connectivity in ventromedial and frontoin-sular cortices in obsessive-compulsive disorder. Biol Psychiatry.2011;69:583-591.

29. Dickel DE, Veenstra-VanderWeele J, Cox NJ, et al. Associationtesting of the positional and functional candidate gene SLC1A1/EAAC1 in early-onset obsessive-compulsive disorder. Arch GenPsychiatry. 2006;63:778-785.

30. Bloch MH, Craiglow BG, Landeros-Weisenberger A, et al. Predictorsof early adult outcomes in pediatric-onset obsessive-compulsivedisorder. Pediatrics. 2009;124:1085-1093.

31. Di Martino A, Scheres A, Margulies DS, et al. Functional connec-tivity of human striatum: a resting state FMRI study. CerebCortex. 2008;18:2735-2747.

32. Fair DA, Bathula D, Mills KL, et al. Maturing thalamocorticalfunctional connectivity across development. Front Syst Neurosci.2010;4:10.

33. Sakai Y, Narumoto J, Nishida S, et al. Corticostriatal functionalconnectivity in non-medicated patients with obsessive-compulsivedisorder [published online ahead of print November 8, 2010]. EurPsychiatry. doi:10.1016/j.eurpsy.2010.09.005.

34. Grabenhorst F, Rolls ET. Value, pleasure and choice in the ventralprefrontal cortex. Trends Cogn Sci. 2011;15:56-67.

35. Marsh R, Maia TV, Peterson BS. Functional disturbances within
frontostriatal circuits across multiple childhood psychopatholo-gies. Am J Psychiatry. 2009;166:664-674.


http://homepage.usask.ca/~ges125/fMRI/AFNIdoc/AlphaSim.pdf

http://homepage.usask.ca/~ges125/fMRI/AFNIdoc/AlphaSim.pdf

F


SUPPLEMENT 1Outside of the medial frontal cortex (MFC), thewhole brain search revealed significant group �age interactions of right medial thalamus connec-tivity to right hippocampus (x � 39, y � �24, z ��6; Z � 4.55, k � 162, pFDR � .05), and tomidcingulate extending into the precentralgyrus (x � 15, y � �6, z � 36; Z � 4.98; k �111, pFDR � .02). Planned contrasts revealed theinteraction for right mdT– hippocampus to bedriven by greater connectivity for child pa-tients compared with child controls. The group �age interaction for mdT–midcingulate connectiv-ity derived from a combination of reduced con-nectivity for patients compared with controls atthe youngest developmental stage, and increasedconnectivity for patients compared with controlsat young adulthood.

No other significant group or group � ageinteractions were observed in the whole-brainsearch for any other seed.



IGURE S1 Seed-specific masks of medial frontalcortex. Note: The medial frontal cortex search volumesfor the left and right dorsal caudate (DC), medial-dorsalthalamus (MT) and ventral caudate (VC). Volumes arenoted in mm3.

948.e1www.jaacap.org

FITZGERALD et al.

TABLE S1 Whole-Brain Analysis: Effects of Group and Group � Age Interactions

Seed

Group Difference Group � Age Interaction

Connected Region Direction x, y, z Z kb Directiona x, y, z Z k*

R MT Midcingulate HC1� � OCD1�

HC20 � OCD2

0

OCD3� � HC3�

OCD40 � HC4

0

15, �6, 36 4.98 111

R MT R. hippocampus OCD1� � HC1�

OCD2� � HC2�

HC3� � OCD3�

HC4� � OCD4�

39, �24, �6 4.55 162

Note: MT � medial–dorsal thalamus.aPlanned contrasts of obsessive-compulsive disorder (OCD) subjects and healthy control (HC) subjects at each developmental stage (1 � child, 2 �

adolescent, 3 � young Adult, 4 � older adult; p � .05, corrected).
bCluster-level significant at pFDR � .05, corrected for multiple comparisons across whole brain.
hctna

FIGURE S2 Left ventral striatum connectivity maps forhealthy control (HC, column 2) and obsessivecompulsive disorder (OCD, column 3) subjects acrossthe stages of development. Note: Displayed in standardneuroanatomical space (Montreal Neurological Institute)at a threshold of p � .005, uncorrected. Right ventralstriatum connectivity maps were qualitatively similar.

JOURN

948.e2 www.jaacap.org

FIGURE S3 Left dorsal striatum connectivity maps forealthy control (HC, column 2) and obsessiveompulsive disorder (OCD, column 3) subjects acrosshe stages of development. Note: Displayed in standardeuroanatomical space (Montreal Neurological Institute)t a threshold of p � .005, uncorrected. Right dorsal

striatum connectivity maps were qualitatively similar.




FIGURE S4 Left medial thalamus connectivity mapsfor healthy control (HC, column 2) and obsessivecompulsive disorder (OCD, column 3) subjects acrossthe stages of development. Note: Displayed in standardneuroanatomical space (Montreal Neurological Institute)at a threshold of p � .005, uncorrected. Right medialthalamus connectivity maps were qualitatively similar.


VOLUME 50 NUMBER 9 SEPTEMBER 2011 948.e3www.jaacap.org

developmental alterations of frontal-striatal-thalamic connectivity in obsessive-compulsive disorder

Documents