Differential prefrontal gray matter correlates of treatment response to fluoxetine or cognitive-behavioral therapy in obsessive–compulsive disorder

European Neuropsychopharmacology (2013) 23, 569–580

0924-977X/$ - see frohttp://dx.doi.org/1

nCorresponding au785—31 andar Ala N

E-mail address: m

www.elsevier.com/locate/euroneuro

Differential prefrontal gray matter correlatesof treatment response to fluoxetine orcognitive-behavioral therapy inobsessive–compulsive disorder

Marcelo Q. Hoextera,b,c,n, Darin D. Doughertyc, Roseli G. Shavitta,Carina C. D’Alcantea, Fabio L.S. Durana, Antonio C. Lopesa,Juliana B. Diniza, Marcelo C. Batistuzzoa, Karleyton C. Evansc,Rodrigo A. Bressanb, Geraldo F. Busattoa, Euripedes C. Miguela

aDepartment and Institute of Psychiatry, University of S~ao Paulo Medical School, S~ao Paulo, BrazilbInterdisciplinary Laboratory for Clinical Neuroscience (LiNC), Department of Psychiatry, FederalUniversity of S~ao Paulo (UNIFESP), S~ao Paulo, BrazilcDepartment of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA

Received 14 December 2011; received in revised form 28 June 2012; accepted 30 June 2012

KEYWORDSOCD;Neuroimaging;Gray matter;Serotonergic reuptakeinhibitor;Cognitive-behavioraltherapy

nt matter & 20120.1016/j.euroneur

thor at: Departmorte-sala 9 (PROTO

qhoexter@gmail.

AbstractNearly one-third of patients with obsessive–compulsive disorder (OCD) fail to respond toadequate therapeutic approaches such as serotonin reuptake inhibitors and/or cognitive-behavioral therapy (CBT). This study investigated structural magnetic resonance imaging (MRI)correlates as potential pre-treatment brain markers to predict treatment response intreatment-naıve OCD patients randomized between trials of fluoxetine or CBT. Treatment-naıve OCD patients underwent structural MRI scans before randomization to a 12-week clinicaltrial of either fluoxetine or group-based CBT. Voxel-based morphometry was used to identifycorrelations between pretreatment regional gray matter volume and changes in symptomseverity on the Yale–Brown Obsessive–Compulsive Scale (Y–BOCS). Brain regional correlations oftreatment response differed between treatment groups. Notably, symptom improvement in thefluoxetine treatment group (n=14) was significantly correlated with smaller pretreatment graymatter volume within the right middle lateral orbitofrontal cortex (OFC), whereas symptomimprovement in the CBT treatment group (n=15) was significantly correlated with largerpretreatment gray matter volume within the right medial prefrontal cortex (mPFC). Nosignificant a priori regional correlations of treatment response were identified as commonbetween the two treatment groups when considering the entire sample (n=29). These findingssuggest that pretreatment gray matter volumes of distinct brain regions within the lateral OFC

Elsevier B.V. and ECNP. All rights reserved.o.2012.06.014

ent and Institute of Psychiatry, University of S~ao Paulo Medical School, Rua Ovıdio Pires de CamposC), CEP 05403-010, S~ao Paulo, SP, Brazil. Tel.: +55 11 3069 7896.

com (M.Q. Hoexter).

M.Q. Hoexter et al.570

and mPFC were differentially correlated to treatment response to fluoxetine versus CBT in OCDpatients. This study further implicates the mPFC in the fear/anxiety extinction process andstresses the importance of lateral portions of the OFC in mediating fluoxetine’s effectiveness inOCD. Clinical registration information: http://clinicaltrials.gov-NCT00680602.& 2012 Elsevier B.V. and ECNP. All rights reserved.

1. Introduction

Serotonin reuptake inhibitors (SRIs), cognitive-behavioraltherapy (CBT) or a combination of both are first linetreatments for obsessive–compulsive disorder (OCD) (Koranet al., 2007). However, at least one-third of patients withOCD fails to respond to adequate therapeutic approaches(Mancebo et al., 2006). Therefore, reliable measures topredict which treatment would be more effective in OCDwould be of great clinical value. These measures could beused to guide clinical decisions toward the best treatmentchoice for individuals with OCD (e.g., pharmacotherapyversus CBT).

In this regard, neuroimaging methods have demonstratedsome utility for identifying brain patterns predictive oftreatment outcome in psychiatric disorders (Evans et al.,2006). There are some neuroimaging treatment studies ofmood and anxiety disorders suggesting different mechan-isms of action for SRIs and CBT (Benazon et al., 2003; Brodyet al., 1998; Gilbert et al., 2000; Hoexter et al., 2012;Martin et al., 2001; Rosenberg et al., 2000). However, giventhat there are also studies showing similar effects (Baxteret al., 1992; Schwartz et al., 1996), the exact neurobiolo-gical underpinnings that predict response to each specificintervention are still not well understood. Brain structuresencompassing cortico-subcortical circuits, known to bemodulated by serotonergic transmission (el Mansari et al.,1995; Soumier et al., 2009) and implicated in OCD patho-physiology (e.g., orbitofrontal, anterior cingulate and tem-porolimbic cortices, striatum and thalamus) (Saxena andRauch, 2000; Valente et al., 2005), are fine candidates.Animal studies have suggested that SRIs were able tomodulate serotoninergic neurotransmission earlier in sub-cortical regions and later in the cerebral cortex (el Mansariet al., 1995). In addition, regions within the medialprefrontal cortex (mPFC), due to their role in fear extinc-tion (Milad et al., 2005; Phelps et al., 2004), should beinvestigated. More specifically, the mPFC is a key regulatorybrain region that participates in the inhibition of condi-tioned fear responses (thought to underlie some obsessive–compulsive symptoms) and in the mediation of the extinc-tion process of such fear responses (Milad et al., 2005;Phelps et al., 2004). Therefore, this seems a plausibleregion through which CBT exerts its therapeutic action.The specific mPFC sub-regions that predict fear extinctioninclude the medial orbitofrontal, rostral anterior cingulateand subcallosal cortices (Milad et al., 2005). Moreover,there is evidence showing that successful pharmacologicaltreatments for OCD seems to preferentially exert asubcortical–cortical ‘bottom-up’ regulation in the brain,whereas CBT seems to elicit cortical–subcortical ‘top-down’effects (Derryberry and Tucker, 1992; Goldapple et al.,2004; Tucker et al., 1995).

The majority of the published neuroimaging outcomestudies in OCD, using positron emission tomography (PET),have associated pretreatment resting regional metabolismwith treatment response (Brody et al., 1998; Evans et al.,2006; Saxena et al., 1999, 2003). The most consistent findingacross these studies has been an inverse correlation ofpretreatment orbitofrontal cortex metabolism with symptomimprovement following SRI exposure (Brody et al., 1998;Saxena et al., 1999). Notably, only one metabolic PET studycompared pharmacotherapy and CBTwith regard to responseprediction. Brody and colleagues showed that lower pre-treatment orbitofrontal cortex metabolism correlated withbetter response to fluoxetine, whereas higher metabolicactivity within this region correlated with a better responseto CBT (Brody et al., 1998). However, the treatment arms inthis study were not randomized, limiting the interpretationof the reported findings. Furthermore, most of the afore-mentioned studies included patients that had been previouslytreated with SRIs, an important confounder, which likelycontributed to the variability observed across studies.Though metabolic neuroimaging studies have been conductedto evaluate predictors of response in OCD patients withrelative success (Brody et al., 1998; Saxena et al., 1999,2003), morphometrical studies are lacking. This is unfortu-nate, given that morphometric magnetic resonance imaging(MRI) is widely available, is cheaper, requires no exposure toradiation, and involves a simpler acquisition protocol com-pared to other neuroimaging methods described above. Theheuristic value of using regional brain volumes to predictclinical response has been explored in several studies thatinvestigated other psychiatric disorders such as depressionand schizophrenia (Garner et al., 2009; MacQueen, 2009).Thus, the primary aim of the present study is to validate anoptimized voxel-based morphometry (VBM) approach toinvestigate pretreatment structural MRI correlates as poten-tial markers to predict changes in symptom severity intreatment-naıve OCD patients randomized to a clinical trialof either fluoxetine or group-based CBT.

We hypothesized that pretreatment volumetric measureswithin the principal brain structures implicated in OCD (i.e.,orbitofrontal, anterior cingulate and temporolimbic cortices,striatum and thalamus) (Saxena and Rauch, 2000; Valenteet al., 2005) would be correlated with subsequent treatmentoutcome. Specifically, given the effects of serotonin inmodulating frontal–subcortical circuits (el Mansari et al.,1995; Soumier et al., 2009) we hypothesized that widespreadmeasurements of gray matter (GM) volume within theseterritories would be correlated with treatment response tofluoxetine. Conversely, we hypothesized that treatmentresponse to CBT would be correlated with measurements ofGM volume within the mPFC, reflecting the primacy ofcognitive restructuration and extinction processes attributedto this region (Milad et al., 2005; Phelps et al., 2004).

GM correlates of treatment response in OCD 571

2. Experimental procedures

2.1. Participants

Forty-one treatment-naıve OCD patients participated in this study.Participants were recruited from a sample of 623 patients who werereferred to our OCD clinic from 2006 to 2008 (Hoexter et al., 2009,2012) and came from a larger clinical trial designed to compare theeffectiveness of fluoxetine and group CBT in a setting closer toclinical practice (trial registration: NCT00680602) (Belotto-Silvaet al., 2012). Initially, potential subjects were screened by tele-phone by a trained clinical psychologist who provided informationabout the protocol and checked for the presence of obsessive–compulsive symptoms as well as for major inclusion/exclusioncriteria. After the telephone screening, potential OCD subjects(n=369) were personally interviewed by OCD expert psychiatrists.From those, 291 patients were excluded from the present study fornot being treatment-naıve. Additional reasons for exclusion were asfollows: OCD was not the primary diagnosis (n=5); initial Y–BOCSscore o16 (n=6); refused to participate (n=13), MRI administrativeproblems (n=5), use of dental braces (n=4) and other reasons(n=4). As shown in Figure 1, three subjects were also excluded dueto the presence of silent gross brain alterations on their MRI images(one patient had a pineal cyst, another one had a nodule suggestiveof a cavernoma located in the right posterior parietal region, andthe third had a bifrontal lesion adjacent to the cerebral falx).Therefore, 38 patients were allocated to receive fluoxetine (n=19)or group CBT (n=19).

Subjects were recruited from primary psychiatric services, orwere selected through the local media (radio, television, news-papers, and internet). To take part in this investigation, patientswere required to have: (1) age between 18–65 yr; (2) Diagnostic andStatistical Manual of Mental Disorders (DSM-IV) (4th edition) criteria

Figure 1 Randomized clinical trial enrollment flowchart andfollow-up.

for OCD as the primary diagnosis and (3) pretreatment Yale–BrownObsessive–Compulsive Scale (Y–BOCS) score Z16 for obsessions andcompulsions or Z10 for only obsessions or compulsions. Individualswere excluded if they presented (1) previous exposure to any kindof psychotropic medication (benzodiazepines, antipsychotics, anti-depressants, stimulants, mood stabilizers); (2) previous exposure toat least 12 sessions of CBT; (3) history of head injury with loss ofconsciousness; (4) past or current substance abuse or dependence;(5) lifetime history of psychosis; (6) suicide risk; (7) any organicdisorders that could affect the central nervous system; (8) contra-indications for MRI scanning; and (9) pregnancy. Patients with otherpsychiatric comorbid disorders, as determined by the StructuredClinical Interview for DSM-IV (SCID-I) (First et al., 1997) andadditional modules for tic, impulse control and attention-deficit/hyperactivity disorders, were not excluded. The main reason forexcluding some patients who participated in the larger clinical trialwas that the great majority of them were not treatment-naıve.

Of the 29 patients who completed the full treatment protocol,10 met DSM-IV criteria for current major depression and 10additional patients had a past history of depression and/or dysthy-mia. Other SCID-I DSM-IV lifetime diagnoses included panic disorderwith agoraphobia (n=1), panic disorder without agoraphobia (n=2),agoraphobia (n=1), social phobia (n=16), specific phobia (n=8),post-traumatic stress disorder (n=6), generalized anxiety disorder(n=11), bipolar II disorder (n=2), somatization (n=1), body dys-morphic disorder (n=4), eating disorders (n=4), trichotillomania(n=2), skin picking (n=6), intermittent explosive disorder (n=6),chronic tics (n=3), Tourette syndrome (n=1), and attention-deficit/hyperactivity disorder (n=2). Of note, the most frequent comorbiddisorders were similarly distributed between groups (p40.05; chi-squared tests) as follows: major depression (43% for fluoxetine-treated patients and 48% for CBT-treated patients), bipolar (43% forfluoxetine-treated patients and 48% for CBT-treated patients),generalized anxiety disorder (43% for fluoxetine-treated patientsand 48% for CBT-treated patients), social phobia (43% for fluoxetine-treated patients and 48% for CBT-treated patients), specific phobia(43% for fluoxetine-treated patients and 48% for CBT-treatedpatients), post-traumatic stress disorder (43% for fluoxetine-treatedpatients and 48% for CBT-treated patients), panic disorder (43% forfluoxetine-treated patients and 48% for CBT-treated patients),eating disorder (43% for fluoxetine-treated patients and 48% forCBT-treated patients), chronic tics/Tourette (43% for fluoxetine-treated patients and 48% for CBT-treated patients) and attention-deficit/hyperactivity disorder (43% for fluoxetine-treated patientsand 48% for CBT-treated patients).

Several psychopathological assessment instruments, including theY–BOCS (Goodman et al., 1989), Dimensional Yale–Brown Obsessive–Compulsive Scale (DY–BOCS) (Rosario-Campos et al., 2006), BeckDepression Inventory (BDI) (Beck et al., 1961) and Beck AnxietyInventory (BAI) (Beck et al., 1988) were administered before andafter 12 weeks of treatment (Hoexter et al., 2009) by interviewerswho were blind to the type of treatment. In addition, to ensurereliability of diagnosis and symptom severity evaluations, all inter-viewers who have participated in this study were trained for thepurposes of this study. The training of the interviewers consisted ofwatching at least 5 videotaped interviews and 5 live interviews with anexpert researcher. The inter-rater reliability of the researchers whoparticipated in this study was 96%, as recently reported by our group(Miguel et al., 2008).

All subjects provided written informed consent, which had beenapproved by the Institutional Review Board of the University of S~aoPaulo Medical School.

2.2. Neuropsychological assessment

A concise neuropsychological investigation that involved measuresof global intelligence, motor coordination and attention was

M.Q. Hoexter et al.572

obtained at baseline to examine possible between group effectsthat might account for the observed findings. The WechslerAbbreviated Scale of Intelligence (WASI) was used to estimateglobal intelligence coefficient (IQ) (Wechsler, 1999), the FingerTapping Test assesses motor speed and motor control and was usedto measure motor coordination (Reitan and Wolfson, 1985) and theTrail Making Test was used to investigate attention given it assessesvisual attention (Reitan and Wolfson, 1985).

2.3. Allocation

A specific method of randomization, known as sequential allocation(Fossaluza et al., 2009), was used in this study with the aim tominimize potential differences between groups. It was performedby a computer program that sequentially allocated patients to eachtreatment group, in which prognostic factors such as initial Y–BOCSscore, gender, and current age were inserted in the model tobalance possible confounders.

2.4. Treatment

Subjects allocated to the medication group received fluoxetine for12 weeks, starting at 20 mg/day in the first week, with weeklyincrements of 20 mg/day until reaching the maximum dosage of80 mg/day (Diniz et al., 2009). Medical appointments to monitorclinical improvement, treatment compliance, side effects andadverse events were scheduled every four weeks. No other psy-choactive medication or psychotherapy was offered during thisperiod. All patients who completed fluoxetine treatment reachedthe maximum dose of 80 mg/day, except one patient who could nottolerate doses higher than 20 mg/day.

The CBT protocol was conducted in groups coordinated by apsychologist with several years of experience and emphasizedexercises of exposure and response prevention, as well as cognitivetechniques, such as correction of thoughts and beliefs, and relapseprevention strategies. A detailed explanation of the group CBTprotocol can be found elsewhere (Belotto-Silva et al., 2012).Subjects were divided into subgroups of 6–8 patients and attendedweekly, 2-h sessions, for 12 weeks. Briefly, the first CBT sessionelucidated the rationale for exposure and response prevention(E/RP) therapy, followed by practical E/RP exercises. In thesubsequent sessions, commonly dysfunctional beliefs about OCDwere explained and cognitive techniques for their corrections weregiven, followed by practical exercises of E/RP and cognitiverestructuring. At the end of each session, homework exercises wereassigned to each patient. In the final sessions, the focus was onstrategies for relapse prevention and in reviewing E/RP andcognitive techniques. The rationale for using a group instead ofan individual CBT protocol was based on the low cost-effectivenessof this approach, which allows the treatment of greater samples ofpatients.

2.5. Imaging acquisition and processing

Imaging data were obtained using a 1.5 T GE Signa scanner (GeneralElectric, Milwaukee WI, USA) using a T1-3D SPGR sequence.Contiguous axial images (1.6 mm) across the entire brain wereacquired with the following parameters: TE=4.20 ms, TR=10.5 ms,flip angle=15, acquisition matrix=256� 192; and were interpolatedusing ZIP2 to a final voxel size of 0.94 mm� 0.94 mm� 0.80 mm(248 slices). All images were reviewed by a neuroradiologist withthe purpose of identifying artifacts during image acquisition and thepresence of silent gross brain alterations. VBM processing wasperformed with the SPM5 package (Wellcome Department ofImaging Neuroscience, London, United Kingdom), executed inMatlab (Mathworks, Sherborn, Massachusetts), using the default

parameters implemented in the VBM5 Toolbox (http://dbm.neuro.uni-jena.de/vbm/). This protocol uses the unified segmentationapproach implemented in SPM (Ashburner and Friston, 2005), whichintegrates the processes of: tissue classification; MRI inhomogeneitybias correction; and spatial normalization to the standard SPMT1-MRI template, using linear (12-parameter affine) and non-lineartransformations, based on 152 healthy subjects from the MontrealNeurological Institute (MNI). To increase the quality of imagesegmentation, the extension of Hidden Markov Random Fieldapproach available in the VBM5 tool Box was used (Cuadra et al.,2005). Subsequently, the final tissue maps of gray matter, whitematter and cerebro-spinal fluid were modulated by the Jacobiandeterminants derived from the spatial normalization to the MNIstandard space, thus allowing brain structures to have their volumeinformation restored. The analysis of modulated images permits theinvestigation of differences in regional brain volume instead ofdensity (Good et al., 2001). Voxel sizes of segmented and spatiallynormalized images are 1mm x 1mm x 1mm. Finally, in order toreduce variations generated by inter-individual differences in theanatomy of the gray and white matter brain compartments, imageswere smoothed using a 12 mm Gaussian kernel.

2.6. Statistical analysis

2.6.1. Clinical, demographical and behavioral dataChi-squared test was used to evaluate categorical variables.Comparisons of means between groups were carried out using theMann–Whitney U test, whereas Wilcoxon test was used for within-group comparisons. Statistical analysis was performed using theStatistical Package for the Social Sciences (SPSS Inc). A P-value of0.05 (two-tailed) was set as the significance threshold for thestatistical analyses.

2.6.2. Neuroimaging dataFirstly, we investigated the presence of significant linear correla-tions between the percent reduction of baseline Y–BOCS scores andpretreatment regional GM volumes with VBM using the generallinear model, based on random Gaussian field theory (Friston et al.,1994). Both direct and inverse relationships were assessed. Onlyvoxels with values above an absolute threshold of p=0.05 fordifferentiating GM from other tissues entered the analyses. Result-ing statistics at each voxel were transformed to z-scores anddisplayed as SPMs into standardized space, at an initial thresholdof z43.09 (po0.001, uncorrected for multiple comparisons). Suchvoxelwise investigation was performed across each subgroup ofpatients—those receiving fluoxetine and those receiving CBT, andacross the entire cohort of subjects, with covariance for the globalamount of GM in the brain and age. Then, each SPM was inspectedon a hypothesis-driven fashion, searching for voxel clusters inspecific brain regions where findings were predicted a priori:orbitofrontal, ventral and dorsal anterior cingulate, temporolimbiccortices (amygdala, hippocampus and parahippocampal gyrus),caudate, putamen and thalamus. Such voxel-wise search of eachmap was performed using the small volume correction (SVC)method, with the purpose of constraining the total number ofvoxels included in the analyses. Each region was circumscribed bymerging the spatially normalized ROI masks that are availablewithin the Anatomical Automatic Labeling (AAL) SPM toolbox. SinceAAL does not contain separate ROIs for ventral and dorsal ACC, wehave parcellated the ACC in ventral (rostral, subcallosal andsubgenual) and dorsal portions in accordance to McCormick et al.(2006), by drawing a vertical plane in the coronal plane at therostral edge of the genu of the corpus callosum, i.e. where theportions of the corpus callosum in each hemisphere are no longerconnected (McCormick et al., 2006). Anatomical masks were usedseparately in each hemisphere (left and right, respectively),resulting in search volumes of: 7704 and 7976 voxels for the

GM correlates of treatment response in OCD 573

superior lateral orbitofrontal cortex; 7104 and 8120 voxels for themiddle lateral orbitofrontal cortex; 13,520 and 13,631 voxels forthe inferior lateral orbitofrontal cortex; 5752 and 6848 voxels forthe medial orbitofrontal cortex; 7032 and 6420 voxels for theventral anterior cingulate gyrus; 4168 and 4084 voxels for the dorsalanterior cingulate gyrus; 1760 and 1984 voxels for the amygdala;7456 and 7568 voxels for the hippocampus; 7824 and 9056 voxels forthe parahippocampal gyrus; 7696 and 7952 voxels for the caudatenucleus; 8039 and 8438 voxels for the putamen and 8435 and 8174voxels for the thalamus. Findings of these hypothesis-driven, SVC-analyses were reported as significant if surviving family-wise error(FWE) correction for multiple comparisons (po0.05), with furthercriteria for voxel clusters to comprise at least 20 voxels over therespective ROI. Maps were also inspected for linear correlationsbetween the percent reduction of baseline Y–BOCS scores andpretreatment regional GM volumes in other territories not pre-viously assessed in ROI-based morphometric studies. Such unpre-dicted findings were considered as significant only if they survivedcorrection for multiple comparisons over the whole brain (FWE-corrected po0.05). In all analyses, we converted MNI coordinatesof voxels of maximal statistical significance to the Talairach andTournoux (1988) system (Brett et al., 2002). Finally, in order toinvestigate the potential confounding effects of percentage ofimprovement in BDI, Y–BOCS at pretreatment, and BDI at pretreat-ment within these territories, three sets of control maps wereperformed as follows: (1) linear correlation between GM volumesand percentage of improvement in BDI; (2) linear correlationbetween GM volumes and Y–BOCS at pretreatment, and (3) linearcorrelation between GM volumes and BDI at pretreatment. Bothdirect and inverse correlations were tested across the entire cohortand for each subgroup of subjects.

3. Results

3.1. Clinical and demographic data

Patients allocated to either group were not significantlydifferent in terms of major clinical and demographicalcharacteristics (Table 1). Additionally, groups did not differin measures of global IQ, motor coordination, or attention.

3.2. Treatment response

Twenty-nine subjects (14 fluoxetine-treated and 15 CBT-treated patients) completed the full 12-week treatmentprotocol (Figure 1). As a group, patients exhibited asignificant reduction in OCD symptom severity, as measuredby the Y–BOCS (reduction of 35.3%, z-score: �4.34,po0.001). Individually, the fluoxetine and CBT subgroupsdisplayed significant OCD symptom improvement (reductionof 38.7%, z-score=�2.98, p=0.003 and 32.6%, z-score=�3.20, po0.01 respectively). In addition, as a group,patients experienced a significant reduction in depressivesymptoms as measured by the BDI (reduction of 34.6%,z-score=�2.95, po0.01). Of note, this BDI score reductionwas statistically significant only for the CBT subgroup alone(reduction of 33.1%, z-score=�2.42, po0.05), whereasthere was just a tendency for significance in the fluoxetinesubgroup (reduction of 36.6%, z-score=�1.88, p=0.06).Anxiety symptoms, measured with the BAI, decreased inboth groups, but did not reach statistical significance forfluoxetine-treated patients (Table 2).

3.3. Gray matter volume as a potential marker oftreatment response

The relationship between regional GM volume and percentreduction of baseline Y–BOCS scores was assessed firstconsidering each treatment modality separately (fluoxetineand CBT).

In fluoxetine-treated patients (n=14), a significant corre-lation was found in a locus within the right middle lateralorbitofrontal cortex (Figure 2, left). Specifically, the smallerthe right middle lateral orbitofrontal cortex pretreatmentvolume, the greater the reduction of Y–BOCS scorespost-treatment. In CBT-treated patients (n=15), a locus ofsignificant correlation was observed within the right mPFC,namely in the subgenual anterior cingulate cortex (Figure 2,right). That is, the larger this region within the right mPFCpretreatment volume, the greater the reduction of Y–BOCSscores post-treatment. Further analysis of the entire brainrevealed no unpredicted loci of significant correlation foreither of the treatment groups (p=0.05, corrected formultiple comparisons). For exploratory purposes, given thatour study is most probably underpowered to detect unpre-dicted findings at a strict threshold (p=0.05, corrected formultiple comparisons) and that the traditional ‘‘OCD neu-robiology model’’ may have overlooked other regions thatnow appear to be involved in OCD, such as the posteriorcingulate and parietal cortices (Menzies et al., 2008), dataof unpredicted GM volumes that were correlated withtreatment response using a less stringent threshold ofpo0.001 for the whole brain (uncorrected for multiplecomparisons) are listed in Supplementary Table S1.

Control maps were used to assess whether these loci ofsignificant correlations between GM volumes and percentageof Y–BOCS reduction might be confounded by other clinicalvariables, such as pretreatment severity of OCD symptoms,pretreatment severity of depressive symptoms, or improve-ment in severity of depressive symptoms. None of thesecontrol analyses yielded significant correlations at the lociidentified in the initial analysis for either treatment group.Subsequently, all analyses were then repeated consideringthe group as a whole (n=29). When both types of treatmentwere analyzed together, no significant correlations werefound in the loci within the principal territories hypothesizedto be involved in the pathophysiology of OCD. Moreover, wehave performed between- group VBM analyses (fluoxetineversus CBT-treated groups) in order to investigate potentialdifferences in GM volumes before treatment. The only apriori hypothesized brain region that showed a statisticallysignificant difference was the right hippocampus, which wassmaller in fluoxetine-treated patients compared to the CBT-treated patients (20 voxels; peak coordinates: 40, �28, �15;z=3.62; peak voxel p-FWEo0.05). There were also nobetween-group clusters of regional changes of greater orsmaller GM volumes in other unpredicted brain regionsconsidering a P valueo0.05, corrected for multiple compar-isons over the whole brain.

Finally, GM volume values were extracted from thespatially normalized images of each patient using a sphereof 3 mm of radius surrounding the voxel of peak statisticalsignificance and exported to the SPSS software to graphi-cally illustrate the correlations between regional GMvolume and the percent reduction of Y–BOCS scores.

Table 1 Clinical and demographic characteristics of OCD patients.

Variable All (n=29) Fluoxetine (n=14) CBT (n=15) Analysisf

Mean (SD) Mean (SD) Mean (SD) U z P value

Age (yr) 33.2 (10.6) 33.1 (11.6) 33.3 (10.0) 99.5 �0.24 0.81Education (yr) 14.5 (2.6) 13.6 (3.3) 68.5 �0.52 0.6Age of onset (yr) 13.1 (8.0) 11.4 (6.9) 14.7 (8.9) 93.5 �0.51 0.61Illness duration (yr) 20.1 (10.3) 21.6 (9.0) 18.7 (11.5) 82.5 �0.98 0.33Y–BOCS score 25.5 (5.3) 23.5 (4.9) 27.3 (5.2) 62.5 �1.86 0.06

DY–BOCS scoreAggressiona 6.0 (4.8) 5.9 (5.1) 6.1 (4.7) 103.5 �0.07 0.95Sexual/religiousb 3.8 (4.9) 4.9 (5.5) 2.8 (4.2) 83.5 �1.05 0.35Symmetryc 7.8 (3.9) 8.2 (3.2) 7.3 (4.5) 88.5 �0.72 0.47Contaminationd 6.1 (5.2) 5.1 (5.5) 7.1 (4.8) 88.0 �0.76 0.44Hoardinge 3.1 (3.4) 3.6 (3.0) 2.7 (3.9) 81.0 �1.10 0.27

BDI 16.9 (9.2) 16.1 (8.0) 17.7 (10.3) 97.5 �0.33 0.74BAI 16.0 (12.4) 15.1 (11.0) 16.9 (13.9) 101.0 �0.17 0.86Global IQ 93.9 (11.7) 95.4 (10.6) 92.6 (12.9) 93.0 �0.52 0.6

N (%) N (%) N (%) v2 Df P valueGender, female 18 (62) 8 (57) 10 (66) 0.28 1 0.6Caucasian 25 (86) 13 (93) 12 (80) 0.22g 1 0.64

Socioeconomic statush

Classes a/b (higher) 17 (59) 8 (57) 9 (60) 0.24 1 0.88Classes c/d/e (lower) 12 (41) 6 (43) 6 (40)

Right-handed 28 (96) 14 (100) 14 (93) 0g 1 1Current MDD 10 (34.5) 6 (42.9) 4 (26) 0.28 1 0.6Lifetime MDD 16 (55.2) 8 (57.1) 8 (53.3) 0.04 1 0.84

Abbreviations: BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; CBT, cognitive-behavioral therapy; Df, degrees offreedom; DY–BOCS, Dimensional Yale–Brown Obsessive–Compulsive Scale; IQ: intelligence coefficient; NA, not applicable; OCD,obsessive–compulsive disorder; SD, standard deviation; U, Mann–Whitney U test; w2, Pearson’s chi-squared test; yr, years; Y–BOCS,Yale–Brown Obsessive–Compulsive Scale; z, z-score.

aObsessions about harm due to aggression/injury/violence/natural disasters and related compulsions.bObsessions concerning sexual/moral/religious obsessions and related compulsions.cObsessions about symmetry/‘just-right’ perceptions, and compulsions to count or order/arrange.dContamination obsessions and cleaning compulsions.eObsessions and compulsions related to hoarding.fFluoxetine-treated versus CBT-treated patients.gThe chi-square test with Yates’ continuity correction was used when less than 5 patients were analyzed for a given variable.hThe socio-economic breakdown was determined according to the criteria of the Brazilian Association of Market Research

Institutes—ABIPEME, which is more representative of the Brazilian population. It is based on the on total household assets like TV sets,stereosystems, and motor vehicles, as well as the head-of-family’s schooling. This classification divides individuals into classes A, B, C,D, and E, based on the composite scores.

M.Q. Hoexter et al.574

The Pearson product-moment correlation analysis for theright middle lateral orbitofrontal cortex in the fluoxetinegroup yielded an r of 0.83 (po0.001) (Figure 3, left),whereas that for the right mPFC in the CBT group yieldedan r of 0.81 (po0.001) (Figure 3, right).

4. Discussion

This study investigated structural brain correlates as poten-tial markers of response in treatment-naıve OCD patientsenrolled in a randomized clinical trial comparing fluoxetineand CBT. In this sample, our findings demonstrate that

particular morphometric MRI measurements of brain struc-tures were correlated with treatment response in OCDpatients. Specifically, we found that smaller pretreatmentGM volume within the right middle lateral orbitofrontalcortex for fluoxetine-treated patients and larger pretreat-ment GM volume within the right mPFC for CBT-treatedpatients were associated with a greater reduction inobsessive–compulsive symptom (OCS) severity. Interestingly,we did not find correlations between improvement in OCSand pretreatment GM volumes within any brain regionspredicted a priori when the treatment groups were ana-lyzed together. Thus, our results suggest that independentpretreatment GM volume markers may be involved in

Table 2 Clinical variables of OCD patients before and after treatment.

Rating scale Group Pretreatment Posttreatment Difference z score P valueb

Mean (SD) Mean (SD) (%)

Y–BOCS All (n=29) 25.5 (5.3) 16.5 (8.1) 35.3 �4.34 o0.001Fluoxetine (n=14) 23.5 (4.9) 14.4 (6.3) 38.7 �2.98 o0.01CBT (n=15) 27.3 (5.2) 18.4 (9.4) 32.6 �3.20 o0.01

BDI All 16.2a (8.6) 10.6a (8.3) 34.6 �2.95 o0.01Fluoxetine 16.1 (8.0) 10.2 (6.2) 36.6 �1.88 0.06CBT 16.4a (9.4) 10.9a (10.1) 33.1 �2.42 0.016

BAI All 16.0 (12.4) 11.2 (8.7) 30 �2.55 o0.05Fluoxetine 15.1 (11.1) 11.4 (8.6) 24.5 �1.26 0.21CBT 16.9 (13.9) 11.1 (9.1) 34.3 �2.48 o0.01

Abbreviations: BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; CBT, cognitive-behavioral therapy; SD, standarddeviation; Y–BOCS, Yale–Brown Obsessive–Compulsive Scale.

aBDI data were missing for one patient.bWilcoxon test for paired comparison.

Figure 2 Loci of significant correlations between pretreatment gray matter volume and subsequent response to Fluoxetine (left)and CBT (right). Left: negative statistically significant correlation between pretreatment gray matter volume within the rightmiddle lateral orbitofrontal cortex (39 voxels; peak z score]=3.70; Talairach coordinates=45, 48, �3; Brodmann area 10; peak voxelp-FWE=0.029) and Y–BOCS improvement following treatment with fluoxetine. Right: positive statistically significant correlationbetween pretreatment gray matter volume within the right medial prefrontal cortex (mPFC), namely the subgenual anteriorcingulate cortex (36 voxels; peak z score]=3.45; Talairach coordinates=7, 27, �6; Brodmann area 25/32; peak voxel p-FWE=0.046)and Y–BOCS improvement following treatment with CBT. ]The peak z score refers to the voxel of peak statistical significance withineach volume of interest. The volume of interest for the right middle lateral orbitofrontal cortex comprised 8120 voxels and thevolume of interest for the right medial prefrontal cortex comprised 6420 voxels.

GM correlates of treatment response in OCD 575

treatment response to specific interventions rather than beassociated with a nonspecific general treatment response.These findings converge with previous longitudinal morpho-metric studies performed in OCD patients suggesting thatpharmacotherapy and psychotherapy may have differentmechanisms of action (Benazon et al., 2003; Brody et al.,1998; Gilbert et al., 2000; Goldapple et al., 2004; Hoexteret al., 2012; Martin et al., 2001; Rosenberg et al., 2000).

Despite some evidences that OCD and major depressivedisorder are mediated by distinct but partially overlapping

neural systems (Cardoner et al., 2007) and that the twosyndromes may have different neurobiological substratesinvolved in treatment response (Saxena et al., 2003), it isimportant to note that the association between regional GMvolumes and percent reduction in baseline Y–BOCS scoresfor each treatment modality did not appear to be attribu-table to confounding factors such as age, severity of OCD,depression at the time of MRI data acquisition, and didnot appear to be influenced by reduction in severity ofdepression symptoms following treatment. Additionally,

Figure 3 Correlation between pretreatment gray matter volume and change in Y–BOCS score after treatment with fluoxetine (left)and CBT (right). Left: graphically correlation between pretreatment gray matter volume within the right middle lateralorbitofrontal cortex and the percentage of improvement in Y–BOCS score after treatment with fluoxetine. Specifically, for thisregion, the Pearson product-moment correlation yielded r=0.83; po0.001. Right: graphically correlation between pretreatmentgray matter volume within the right medial prefrontal cortex (mPFC), namely the subgenual anterior cingulate cortex and thepercentage of improvement in Y–BOCS score after treatment with CBT. Specifically, for this region, the Pearson product-momentcorrelation yielded r=0.81; po0.001.

M.Q. Hoexter et al.576

these findings did not seem to be confounded by between-group differences, given that OCD patients in each groupwere quite similar in terms of clinical and sociodemographiccharacteristics.

4.1. Morphometric correlates as a potentialmarker of treatment response to fluoxetine

Our findings of an inverse correlation between GM volume inthe middle lateral orbitofrontal cortex and OCD symptomimprovement in fluoxetine medicated subjects are intri-guing. Given the neurotrophic effects of SRIs and otherserotonergic modulators in the prefrontal cortex (Czehet al., 2007; Soumier et al., 2009), it is possible that theorbitofrontal cortex constitutes the site of neural substratewhere SRIs have the greatest anti-obsessional effects (elMansari et al., 1995). Evidences from animal models suggestthat fluoxetine treatment counteracts the inhibitory effectof chronic stress on prefrontal cortex cytogenesis (newbornglial cells) (Czeh et al., 2007) and that serotonergic regula-tion induced by serotonergic agonists increases gliogenesisin this region (Soumier et al., 2009). It may be that thesmaller the size of the orbitofrontal cortex, the greater thepotential for modulatory effect of the SRIs. In PET studies,greater activity within the lateral orbitofrontal cortex hasbeen directly associated with OCD severity (Rauch et al.,1994) and activity within this subregion decreases inindividuals who show a positive response to SRIs (Saxenaet al., 1999). This indicates that in OCD patients, the lateralsubregions of the orbitofrontal cortex may be an importantregion for pharmacological response. Interestingly, recentfindings of reduced activation of the lateral orbitofrontalcortex observed in OCD patients and in their unaffectedrelatives during a reversal learning task (Chamberlain et al.,2008), suggest that abnormalities within this territory are amarker of vulnerability to OCD, and thus, represents a keyregion to early pharmacological intervention. However,given that the orbitofrontal cortex participates in a wide

range of normal and psychopathological processes (Menzieset al., 2008), more complex models will likely be moreaccurate.

4.2. Morphometric correlates as a potentialmarker of treatment response to CBT

Results from the current study also demonstrated thatpatients with larger pretreatment GM volumes within theright mPFC were more likely to respond to CBT. It isnoteworthy that the CBT protocol used in this studyemphasized exercises of ERP. After a systematic, repeated,and prolonged exposure to certain anxiogenic situations andrestraint from completing compulsions, the anxietyresponse to a feared stimulus is inhibited, resulting in theextinction of the feared response. Convergent data fromanimal and human models implicate the mPFC in theinhibition of conditioned fear responses and in the media-tion of the extinction (Milad et al., 2005; Phelps et al.,2004). Moreover, it is known that the mPFC has denseprojections to other brain regions, such as the amygdala,that mediate conditioned fear responses (Quirk et al.,2006). Remarkably, the maximal peak focus within themPFC reported herein (Talairach coordinates: 7, 27, �6) islocated in close proximity to those foci within the mPFCshown to be functionally (Talairach coordinates: 6, 25, �11and 2, 35, �8) (Milad et al., 2007) and morphometrically(Talairach coordinates: 6, 28, �15 and 4, 30, �12) (Miladet al., 2005) correlated to fear extinction retention inhealthy individuals. Our findings of greater GM volumewithin this region of the mPFC are in accordance with theresults of Milad and colleagues (Milad et al., 2005) whofound that thicker GM within this region predicted betterextinction memory in healthy volunteers. Thus, our findingof a positive correlation between the subgenual anteriorcingulum volume and treatment response to CBT supportsthe concept that the mPFC is a key region involved in themediation of fear extinction learning (Milad et al., 2005;

GM correlates of treatment response in OCD 577

Phelps et al., 2004). Accordingly, one might speculate thatOCD patients with greater mPFC volume would demonstratea better response to CBT because they are able to retainmore effectively the extinction learning after this interven-tion. Interestingly, data from a recent neuroimaging studydemonstrated that lower activity in the subgenual anteriorcingulate (Talairach coordinates: 8, 30, �8) predictedtreatment response to anterior capsular stimulation inrefractory OCD patients (Van Laere et al., 2006). It isbecoming evident, with the increasing number of deepbrain stimulation (DBS) and ablative studies performed intreatment refractory OCD patients, that invasive proce-dures are an important treatment option for these severelyaffected patients (Greenberg et al., 2010). Although itremains unclear how neurosurgery mediates OCD symptomimprovement, one possible hypothesis is that DBS andablative procedures facilitate behavior therapy (Greenberget al., 2006). Thus, the mPFC could constitute a keymodulatory region by which surgery facilitates fear extinc-tion in refractory OCD patients. On the other hand, giventhat pretreatment metabolic activity within the mPFC,specifically in the subgenual anterior cingulate cortex, hasalso been associated with treatment response to severalmodalities of interventions (pharmacotherapy, stereotacticablation and deep brain stimulation) in mood (Doughertyet al., 2003; Mayberg et al., 1997; Saxena et al., 2003) andother anxiety (Evans et al., 2009) disorders, our findingsshould be interpreted with caution. Nonetheless, theobserved association between mPFC volume and CBTresponse was seen even after controlling for severity andimprovement of depressive symptoms.

4.3. Limitations

The present study has certain limitations: (1) the modestsample size may have limited our power to detect moresubtle correlations between GM volumes and Y–BOCSchanges and to explore gray matter volume correlations ofspecific OCD dimensions, (2) the potential confoundinginfluences of comorbid diagnoses, (3) the absence of aplacebo-treated group, and (4) the lack of monitoringfluoxetine serum levels and the CBT engagement. Weacknowledge that group CBT dedicates a lesser amount oftime to deal with each patient individually. Nevertheless,the ‘‘group factor’’ may provide social support and strengthmotivation to handle the E/RP exercises. Likewise, giventhat the ultimate goal of neuroimaging prediction studies isto provide useful information to guide clinical decision-making, the inclusion of a real-world heterogeneous OCDsample can maximize the generalizability of our findingstowards a clinical application. Additionally, this study hadother strengths that support the interpretation of itsfindings. First, patients were treatment-naıve, which mini-mizes the confounding effects of previous therapeuticinterventions. Second, patients participated in a rando-mized controlled clinical trial, enabling comparisons fromtwo different interventions considered first line treatmentsfor OCD. Third, it should be acknowledged that recruitingadult treatment-naıve patients with OCD is quite difficultand this is the largest longitudinal imaging study performedso far with such a population in the OCD literature.

Our results must be also interpreted with caution due tolimitations in the VBM approach, such as systematic align-ment and registration errors during spatial normalization,inaccuracy in the segmentation process of GM, whitematter, and cerebrospinal fluid into distinct compartments,and error introduced by spatial smoothing (Jones et al.,2005). Finally, given the number of statistical tests, we arevulnerable to making Type I error. Therefore, our findingswarrant independent replication with a larger sample sizebefore generalization.

While the present findings are intriguing, it remains anopen question as to whether morphometric MRI will be auseful tool in identifying pre-treatment volumetric brainprofiles predictive of treatment response in OCD. Futurestudies combining different morphometric (VBM, manualROI, cortical thickness) and functional methods in largesamples of OCD patients are warranted to deal with thelimitations presented above. Additionally, the future inves-tigation of OCD samples with fewer comorbidities andspecific subtypes may serve to validate that the presentfindings are not partially due to changes in comorbidsymptoms. In this context, the inclusion of groups withother primary psychiatric disorders would also be valuableto establish whether the morphometric prediction proper-ties of the lateral orbitofrontal cortex and mPFC in mediat-ing fear extinction are specific to OCD or can be generalizedto other mood and anxiety disorders that also benefit fromSRIs and behavioral approaches.

5. Conclusions

This study has shown that pretreatment GM volumes ofdistinct brain regions within the prefrontal cortex wereinvolved with response to specific treatment modalities forOCD patients. Measurements of GM volume within thelateral orbitofrontal cortex were correlated with treatmentresponse to fluoxetine, whereas measurements of GMvolume within the mPFC were correlated with response toCBT. These results further implicate the mPFC in the fear/anxiety extinction process and stress the importance oflateral portions of the orbitofrontal cortex in mediatingSRIs’ effectiveness in OCD.

Role of funding source

This study received financial support in the form of Grants providedby the Fundac- ~ao de Amparo �a Pesquisa do Estado de S~ao Paulo(FAPESP, Foundation for the Support of Research in the State of S~aoPaulo) to Dr. Miguel (2005/55628-8) and from FAPESP scholarships toDr. Shavitt (06/61459-7), to Dr. Diniz (06/50273-0), to Dr. Lopes(2008/10257-0), and to Ms. D’Alcante (06/58286-3). Dr. Hoexter issupported by a Ph.D. scholarship from FAPESP (2005/04206-6) andby a doctorate ‘‘sandwich’’ scholarship from the Coordenac- ~ao deAperfeic-oamento de Pessoal de Nıvel Superior (CAPES, Agency forSupport and Evaluation of Graduate Education) (4375/08-4).

Contributors

Marcelo Q. Hoexter: conception and design, analysis and interpre-tation of data, drafting the article and final approval of the versionto be published.

M.Q. Hoexter et al.578

Darin D. Dougherty: interpretation of data, revising the articlefor important intellectual content and final approval of the versionto be published.

Roseli G. Shavitt: conception and design, interpretation of data,revising the article critically for important intellectual content andfinal approval of the version to be published.

Carina C. D’Alcante: interpretation of data, revising the articlecritically for important intellectual content and final approval ofthe version to be published.

F�abio L.S. Duran: analysis and interpretation of data, revising thearticle critically for important intellectual content and finalapproval of the version to be published.

Antonio C. Lopes: interpretation of data, revising the articlecritically for important intellectual content and final approval ofthe version to be published.

Juliana B. Diniz: interpretation of data, revising the articlecritically for important intellectual content and final approval ofthe version to be published.

Marcelo C. Batistuzzo: interpretation of data, revising the articlecritically for important intellectual content and final approval ofthe version to be published.

Karleyton C. Evans: interpretation of data, revising the articlefor important intellectual content and final approval of the versionto be published.

Rodrigo A. Bressan: conception and design, interpretation ofdata, revising the article for important intellectual content andfinal approval of the version to be published.

Geraldo F. Busatto: design, analysis and interpretation of data,revising the article critically for important intellectual content andfinal approval of the version to be published.

Euripedes C. Miguel: conception and design, interpretation ofdata, revising the article critically for important intellectualcontent and final approval of the version to be published.

Conflict of interest

Drs. Hoexter, Shavitt, Duran, Lopes, Diniz, Batistuzzo and Busattodeclared no conflict of interest. Dr. Dougherty has acted as aconsultant for Medtronic, Eli Lilly, Brand Ideas, McNeil and ReedElsevier; has received research funding from Medtronic, Eli Lilly,McNeil and Cyberonics. Dr. Evans has received research funding fromPfizer and participated in research trials sponsored by Cephalon,Cyberonics and Medtronic. Dr. Bressan has received honoraria and/orconsultations fees from Novartis, Eli Lilly, Janssen and AstraZeneca.Dr. Miguel has received lecture fees from Lundbeck and Solvay.

This work was presented at the 66th Annual Meeting of the Societyof Biological Psychiatry, May 12–14, 2011, San Francisco, CA, USA.

Acknowledgments

We thank Dr. Mohammed R. Milad and Ms. Dianne M. Hezel forhelpful comments on the manuscript. We also wish to thank allpatients who participated in this study.

Appendix A. Supplementary material

Supplementary data associated with this article can befound in the online version at http://10.1016/j.euroneuro.2012.06.014.

References

Ashburner, J., Friston, K.J., 2005. Unified segmentation. Neuro-image 26, 839–851.

Baxter Jr., L.R., Schwartz, J.M., Bergman, K.S., Szuba, M.P., Guze,B.H., Mazziotta, J.C., Alazraki, A., Selin, C.E., Ferng, H.K.,Munford, P., et al., 1992. Caudate glucose metabolic ratechanges with both drug and behavior therapy for obsessive–compulsive disorder. Arch. Gen. Psychiatry 49, 681–689.

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J., 1961.An inventory for measuring depression. Arch. Gen. Psychiatry 4,561–571.

Beck, A.T., Epstein, N., Brown, G., Steer, R.A., 1988. An inventoryfor measuring clinical anxiety: psychometric properties.J. Consult. Clin. Psychol. 56, 893–897.

Belotto-Silva, C., Diniz, J.B., Malavazzi, D.M., Val �erio, C., Fossa-luza, V., Borcato, S., Seixas, A.A., Morelli, D., Migue, E.C.,Shavitt, R.G., 2012. Group cognitive-behavioral therapy versusselective serotonin reuptake inhibitors for obsessive–compulsivedisorder: a practical clinical trial. J. Anxiety Disord. 26, 25–31.

Benazon, N.R., Moore, G.J., Rosenberg, D.R., 2003. Neurochemicalanalyses in pediatric obsessive–compulsive disorder in patientstreated with cognitive-behavioral therapy. J. Am. Acad. ChildAdolesc. Psychiatry 42, 1279–1285.

Brett, M., Johnsrude, I.S., Owen, A.M., 2002. The problem offunctional localization in the human brain. Nat. Rev. Neurosci.3, 243–249.

Brody, A.L., Saxena, S., Schwartz, J.M., Stoessel, P.W., Maidment,K., Phelps, M.E., Baxter Jr., L.R., 1998. FDG-PET predictors ofresponse to behavioral therapy and pharmacotherapy in obses-sive compulsive disorder. Psychiatry Res. 84, 1–6.

Cardoner, N., Soriano-Mas, C., Pujol, J., Alonso, P., Harrison, B.J.,Deus, J., Hern�andez-Ribas, R., Menchon, J.M., Vallejo, J., 2007.Brain structural correlates of depressive comorbidity inobsessive–compulsive disorder. Neuroimage 38, 413–421.

Chamberlain, S.R., Menzies, L., Hampshire, A., Suckling, J., Fine-berg, N.A., del Campo, N., Aitken, M., Craig, K., Owen, A.M.,Bullmore, E.T., Robbins, T.W., Sahakian, B.J., 2008. Orbitofron-tal dysfunction in patients with obsessive–compulsive disorderand their unaffected relatives. Science 321, 421–422.

Cuadra, M.B., Cammoun, L., Butz, T., Cuisenaire, O., Thiran, J.P.,2005. Comparison and validation of tissue modelization andstatistical classification methods in T1-weighted MR brainimages. IEEE Trans. Med. Imaging 24, 1548–1565.

Czeh, B., Muller-Keuker, J.I., Rygula, R., Abumaria, N., Hiemke, C.,Domenici, E., Fuchs, E., 2007. Chronic social stress inhibits cellproliferation in the adult medial prefrontal cortex: hemisphericasymmetry and reversal by fluoxetine treatment. Neuropsycho-pharmacology 32, 1490–1503.

Derryberry, D., Tucker, D.M., 1992. Neural mechanisms of emotion.J. Consult. Clin. Psychol. 60, 329–338.

Diniz, J., Shavitt, R., Pereira, C., Hounie, A., Pimentel, I., Koran, L.,Dainesi, S., Miguel, E., 2009. Quetiapine versus clomipramine in theaugmentation of selective serotonin reuptake inhibitors for thetreatment of obsessive–compulsive disorder: a randomized, open-label trial. J. Psychopharmacol. 24, 297–307.

Dougherty, D.D., Weiss, A.P., Cosgrove, G.R., Alpert, N.M., Cassem,E.H., Nierenberg, A.A., Price, B.H., Mayberg, H.S., Fischman,A.J., Rauch, S.L., 2003. Cerebral metabolic correlates aspotential predictors of response to anterior cingulotomy fortreatment of major depression. J. Neurosurg. 99, 1010–1017.

el Mansari, M., Bouchard, C., Blier, P., 1995. Alteration of serotoninrelease in the guinea pig orbito-frontal cortex by selective serotoninreuptake inhibitors. Relevance to treatment of obsessive–compulsivedisorder. Neuropsychopharmacology 13, 117–127.

Evans, K.C., Dougherty, D.D., Pollack, M.H., Rauch, S.L., 2006.Using neuroimaging to predict treatment response in mood andanxiety disorders. Ann. Clin. Psychiatry 18, 33–42.

Evans, K.C., Simon, N.M., Dougherty, D.D., Hoge, E.A., Worthing-ton, J.J., Chow, C., Kaufman, R.E., Gold, A.L., Fischman, A.J.,Pollack, M.H., Rauch, S.L., 2009. A PET study of tiagabinetreatment implicates ventral medial prefrontal cortex in

GM correlates of treatment response in OCD 579

generalized social anxiety disorder. Neuropsychopharmacology34, 390–398.

First, M., Spitzer, R., Gibbon, M., Williams, J., 1997. Structuredclinical interview for DSM-IV axis I disorders: clinical version(SCID CV). American Psychiatric Press, Washington, DC.

Fossaluza, V., Diniz, J.B., Pereira Bde, B., Miguel, E.C., Pereira,C.A., 2009. Sequential allocation to balance prognostic factorsin a psychiatric clinical trial. Clinics (Sao Paulo) 64, 511–518.

Friston, K., Holmes, A., Worsley, K., Poline, J., Frith, C., Frack-owiak, R., 1994. Statistic parametric maps in functionalimaging: a general linear approach. Hum. Brain Mapp. 2,189–210.

Garner, B., Berger, G.E., Nicolo, J.P., Mackinnon, A., Wood, S.J.,Pariante, C.M., Dazzan, P., Proffitt, T.M., Markulev, C., Kerr, M.,McConchie, M., Phillips, L.J., Pantelis, C., McGorry, P.D., 2009.Pituitary volume and early treatment response in drug-naıvefirst-episode psychosis patients. Schizophr. Res. 113, 65–71.

Gilbert, A.R., Moore, G.J., Keshavan, M.S., Paulson, L.A., Narula,V., Mac Master, F.P., Stewart, C.M., Rosenberg, D.R., 2000.Decrease in thalamic volumes of pediatric patients withobsessive–compulsive disorder who are taking paroxetine. Arch.Gen. Psychiatry 57, 449–456.

Goldapple, K., Segal, Z., Garson, C., Lau, M., Bieling, P., Kennedy,S., Mayberg, H., 2004. Modulation of cortical-limbic pathways inmajor depression: treatment-specific effects of cognitive beha-vior therapy. Arch. Gen. Psychiatry 61, 34–41.

Good, C.D., Johnsrude, I.S., Ashburner, J., Henson, R.N., Friston,K.J., Frackowiak, R.S., 2001. A voxel-based morphometric studyof ageing in 465 normal adult human brains. Neuroimage 14,21–36.

Goodman, W.K., Price, L.H., Rasmussen, S.A., Mazure, C., Fleisch-mann, R.L., Hill, C.L., Heninger, G.R., Charney, D.S., 1989. TheYale–Brown Obsessive Compulsive Scale. I. Development, use,and reliability. Arch. Gen. Psychiatry 46, 1006–1011.

Greenberg, B.D., Malone, D.A., Friehs, G.M., Rezai, A.R., Kubu,C.S., Malloy, P.F., Salloway, S.P., Okun, M.S., Goodman, W.K.,Rasmussen, S.A., 2006. Three-year outcomes in deep brainstimulation for highly resistant obsessive–compulsive disorder.Neuropsychopharmacology 31, 2384–2393.

Greenberg, B.D., Rauch, S.L., Haber, S.N., 2010. Invasive circuitry-based neurotherapeutics: stereotactic ablation and deep brainstimulation for OCD. Neuropsychopharmacology 35, 317–336.

Hoexter, M.Q., Shavitt, R.G., D’Alcante, C.C., Cecconi, J.P., Diniz,J.B., Belotto-Silva, C., Hounie, A.G., Borcato, S., Moraes, I.,Joaquim, M.A., Cappi, C., Sampaio, A.S., Mathis, M.A., Batis-tuzzo, M.C., Lopes, A.C., Rosa, A.C., Muniz, R.K., Marques,A.H., Santos, L.C., Taub, A., Duran, F.L., Dougherty, D.D.,Busatto, G.F., Bressan, R.A., Miguel, E.C., 2009. The drug-naiveOCD patients imaging genetics, cognitive and treatmentresponse study: methods and sample description. Rev. Bras.Psiquiatr. 31, 349–353.

Hoexter, M.Q., de Souza Duran, F.L., D’Alcante, C.C., Dougherty, D.D.,Shavitt, R.G., Lopes, A.C., Diniz, J.B., Deckersbach, T., Batistuzzo,M.C., Bressan, R.A., Miguel, E.C., Busatto, G.F., 2012. Gray mattervolumes in obsessive–compulsive disorder before and after fluox-etine or cognitive-behavior therapy: a randomized clinical trial.Neuropsychopharmacology 37, 734–745.

Jones, D.K., Symms, M.R., Cercignani, M., Howard, R.J., 2005. Theeffect of filter size on VBM analyses of DT-MRI data. Neuroimage26, 546–554.

Koran, L.M., Hanna, G.L., Hollander, E., Nestadt, G., Simpson,H.B., 2007. Practice guideline for the treatment of patients withobsessive–compulsive disorder. Am. J. Psychiatry 164, 5–53.

MacQueen, G.M., 2009. Magnetic resonance imaging and predictionof outcome in patients with major depressive disorder. J.Psychiatry Neurosci. 34, 343–349.

Mancebo, M.C., Eisen, J.L., Pinto, A., Greenberg, B.D., Dyck, I.R.,Rasmussen, S.A., 2006. The brown longitudinal obsessive

compulsive study: treatments received and patient impressionsof improvement. J. Clin. Psychiatry 67, 1713–1720.

Martin, S.D., Martin, E., Rai, S.S., Richardson, M.A., Royall, R.,2001. Brain blood flow changes in depressed patients treatedwith interpersonal psychotherapy or venlafaxine hydrochloride:preliminary findings. Arch. Gen. Psychiatry 58, 641–648.

Mayberg, H.S., Brannan, S.K., Mahurin, R.K., Jerabek, P.A., Brickman,J.S., Tekell, J.L., Silva, J.A., McGinnis, S., Glass, T.G., Martin,C.C., Fox, P.T., 1997. Cingulate function in depression: a potentialpredictor of treatment response. Neuroreport 8, 1057–1061.

McCormick, L.M., Ziebell, S., Nopoulos, P., Cassell, M., Andreasen,N.C., Brumm, M., 2006. Anterior cingulate cortex: an MRI-basedparcellation method. NeuroImage 32, 1167–1175.

Menzies, L., Chamberlain, S.R., Laird, A.R., Thelen, S.M., Sahakian,B.J., Bullmore, E.T., 2008. Integrating evidence from neuroima-ging and neuropsychological studies of obsessive–compulsivedisorder: the orbitofronto-striatal model revisited. Neurosci.Biobehav. Rev. 32, 525–549.

Miguel, E.C., Ferrao, Y.A., Rosario, M.C., Mathis, M.A., Torres, A.R.,Fontenelle, L.F., Hounie, A.G., Shavitt, R.G., Cordioli, A.V.,Gonzalez, C.H., Petribu, K., Diniz, J.B., Malavazzi, D.M.,Torresan, R.C., Raffin, A.L., Meyer, E., Braga, D.T., Borcato,S., Valerio, C., Gropo, L.N., Prado, Hda S., Perin, E.A., Santos,S.I., Copque, H., Borges, M.C., Lopes, A.P., Silva, E.D., 2008.The Brazilian research consortium on obsessive–compulsivespectrum disorders: recruitment, assessment instruments,methods for the development of multicenter collaborativestudies and preliminary results. Rev. Bras. Psiquiatr. 30, 185–196.

Milad, M.R., Quinn, B.T., Pitman, R.K., Orr, S.P., Fischl, B., Rauch,S.L., 2005. Thickness of ventromedial prefrontal cortex inhumans is correlated with extinction memory. Proc. Natl. Acad.Sci. U. S. A. 102, 10706–10711.

Milad, M.R., Wright, C.I., Orr, S.P., Pitman, R.K., Quirk, G.J.,Rauch, S.L., 2007. Recall of fear extinction in humans activatesthe ventromedial prefrontal cortex and hippocampus in concert.Biol. Psychiatry 62, 446–454.

Phelps, E.A., Delgado, M.R., Nearing, K.I., LeDoux, J.E., 2004.Extinction learning in humans: role of the amygdala and vmPFC.Neuron 43, 897–905.

Quirk, G.J., Garcia, R., Gonz�alez-Lima, F., 2006. Prefrontalmechanisms in extinction of conditioned fear. Biol. Psychiatry60, 337–343.

Rauch, S.L., Jenike, M.A., Alpert, N.M., Baer, L., Breiter, H.C.,Savage, C.R., Fischman, A.J., 1994. Regional cerebral blood flowmeasured during symptom provocation in obsessive–compulsivedisorder using oxygen 15-labeled carbon dioxide and positronemission tomography. Arch. Gen. Psychiatry 51, 62–70.

Reitan, R., Wolfson, D., 1985. The Halstead–Reitan neuropsycholo-gical test battery: theory and clinical interpretation. Neuropsy-chology Press, Tucson, AZ.

Rosario-Campos, M.C., Miguel, E.C., Quatrano, S., Chacon, P.,Ferrao, Y., Findley, D., Katsovich, L., Scahill, L., King, R.A.,Woody, S.R., Tolin, D., Hollander, E., Kano, Y., Leckman, J.F.,2006. The Dimensional Yale–Brown Obsessive–Compulsive Scale(DY–BOCS): an instrument for assessing obsessive–compulsivesymptom dimensions. Mol. Psychiatry 11, 495–504.

Rosenberg, D.R., Benazon, N.R., Gilbert, A., Sullivan, A., Moore,G.J., 2000. Thalamic volume in pediatric obsessive–compulsivedisorder patients before and after cognitive behavioral therapy.Biol. Psychiatry 48, 294–300.

Saxena, S., Brody, A.L., Maidment, K.M., Dunkin, J.J., Colgan, M.,Alborzian, S., Phelps, M.E., Baxter Jr., L.R., 1999. Localizedorbitofrontal and subcortical metabolic changes and predictorsof response to paroxetine treatment in obsessive–compulsivedisorder. Neuropsychopharmacology 21, 683–693.

Saxena, S., Rauch, S.L., 2000. Functional neuroimaging and theneuroanatomy of obsessive–compulsive disorder. Psychiatr. Clin.North Am. 23, 563–586.

M.Q. Hoexter et al.580

Saxena, S., Brody, A.L., Ho, M.L., Zohrabi, N., Maidment, K.M.,Baxter Jr., L.R., 2003. Differential brain metabolic predictors ofresponse to paroxetine in obsessive–compulsive disorder versusmajor depression. Am. J. Psychiatry 160, 522–532.

Schwartz, J.M., Stoessel, P.W., Baxter Jr., L.R., Martin, K.M., Phelps,M.E., 1996. Systematic changes in cerebral glucose metabolicrate after successful behavior modification treatment ofobsessive–compulsive disorder. Arch. Gen. Psychiatry 53, 109–113.

Soumier, A., Banasr, M., Goff, L.K., Daszuta, A., 2009. Region- andphase-dependent effects of 5-HT(1A) and 5-HT(2C) receptoractivation on adult neurogenesis. Eur. Neuropsychopharmacol.20, 336–345.

Talairach, J., Tournoux, P., Fischer, M., 1988. Co-Planar StereotaxicAtlas of the Human Brain. Thieme Medical Publishers, New York.

Tucker, D.M., Luu, P., Pribram, K.H., 1995. Social and emotionalself-regulation. Ann. N.Y. Acad. Sci. 769, 213–239.

Valente Jr., A.A., Miguel, E.C., Castro, C.C., Amaro Jr., E., Duran,F.L., Buchpiguel, C.A., Chitnis, X., McGuire, P.K., Busatto, G.F.,2005. Regional gray matter abnormalities in obsessive–compulsive disorder: a voxel-based morphometry study. Biol.Psychiatry 58, 479–487.

Van Laere, K., Nuttin, B., Gabriels, L., Dupont, P., Rasmussen, S.,Greenberg, B.D., Cosyns, P., 2006. Metabolic imaging of anteriorcapsular stimulation in refractory obsessive–compulsive disor-der: a key role for the subgenual anterior cingulate and ventralstriatum. J. Nucl. Med. 47, 740–747.

Wechsler, D., 1999. Wechsler abbreviated scale of IQ. PsychologicalCorporation, Harcourt Brace and Company, San Antonio, TX.