Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 1.
Published in final edited form as: J Child Psychol Psychiatry. 2019 Dec 31;61(12):1299–1308. doi: 10.1111/jcpp.13191

Structural neural markers of response to cognitive behavioral therapy in pediatric obsessive-compulsive disorder

David Pagliaccio 1,2, Jiook Cha 1,2, Xiaofu He 1,2, Marilyn Cyr 1,2, Paula Yanes-Lukin 1,2, Pablo Goldberg 1,2, Martine Fontaine 1,2, Moira A Rynn 3, Rachel Marsh 1,2
PMCID: PMC7326644  NIHMSID: NIHMS1063614  PMID: 31889307

Abstract

Background:

Cognitive behavioral therapy (CBT) is an effective, first-line treatment for pediatric obsessive-compulsive disorder (OCD). While neural predictors of treatment outcomes have been identified in adults with OCD, robust predictors are lacking for pediatric patients. Herein, we sought to identify brain structural markers of CBT response in youth with OCD.

Methods:

Twenty-eight children/adolescents with OCD and 27 matched healthy participants (7–18-year-olds, M=11.71 years, SD=3.29) completed high-resolution structural and diffusion MRI (all unmedicated at time of scanning). Patients with OCD then completed 12–16 sessions of CBT. Subcortical volume and cortical thickness were estimated using FreeSurfer. Structural connectivity (streamline counts) was estimated using MRtrix.

Results:

Thinner cortex in nine fronto-parietal regions significantly predicted improvement in Children’s Yale-Brown Obsessive Compulsive Scale (CY-BOCS) scores (all ts>3.4, FDR-corrected ps<.05). These included middle and superior frontal, angular, lingual, precentral, superior temporal, and supramarginal gyri (SMG). Vertex-wise analyses confirmed a significant left SMG cluster, showing large effect size (Cohen’s d=1.42) with 72.22% specificity and 90.00% sensitivity in predicting CBT response. Ten structural connections between cingulo-opercular regions exhibited fewer streamline counts in OCD (all ts>3.12, Cohen’s ds>0.92) compared to healthy participants. These connections predicted post-treatment CY-BOCS scores, beyond pre-treatment severity and demographics, though not above and beyond cortical thickness.

Conclusions:

The current study identified group differences in structural connectivity (reduced among cingulo-opercular regions) and cortical thickness predictors of CBT response (thinner fronto-parietal cortices) in unmedicated children/adolescents with OCD. These data suggest, for the first time, that cortical and white matter features of task control circuits may be useful in identifying which pediatric patients respond best to individual CBT.

Keywords: Magnetic resonance imaging, obsessive compulsive disorder, cognitive therapy, structural MRI (sMRI), child development

Introduction

Obsessive-compulsive disorder (OCD) is characterized by the presence of obsessions (i.e. intrusive thoughts, images, or urges) and compulsions (i.e. repetitive actions aimed at preventing or reducing distress). Cognitive behavioral therapy (CBT) with exposure and ritual prevention (EX/RP) is a first-line treatment for OCD (Koran et al., 2007). While CBT generally elicits more clinically significant improvement than pharmacotherapy in pediatric OCD (Watson and Rees, 2008, Ost et al., 2016), remission rates are still relatively low for both treatment options (e.g. in the Pediatric Obsessive Compulsive Treatment Study: 39.3% remission with CBT vs. 21.4% with sertraline (March et al., 2004)). Remission rates could potentially be improved through personalized medicine approaches, i.e. identifying patients likely to respond to CBT, pharmacotherapy, or combination therapy and guiding them to appropriate treatment.

Some factors have been identified to predict CBT or pharmacological treatment outcomes, e.g. functional impairment, comorbidities, and family accommodation (Garcia et al., 2010). Neuroscience could potentially aid in identifying robust pre-treatment markers of treatment responsivity above and beyond severity/impairment and familial factors. However, extant studies have mainly focused on adults with OCD, e.g. identifying task-based fMRI (Pagliaccio et al., 2019) or structural (orbitofrontal and cingulate cortex) predictors of treatment outcomes (Hoexter et al., 2015, Hoexter et al., 2013, Fullana et al., 2014). Data from youth suggest potential increases in orbitofrontal volume following CBT (Huyser et al., 2013, Huyser et al., 2014). FMRI findings suggest functional alterations in fronto-parietal and cingulo-opercular circuits in adults (Maia et al., 2008, Menzies et al., 2008, Del Casale et al., 2016, van Velzen et al., 2014) and youth with OCD during cognitive control paradigms (Fitzgerald et al., 2013, Huyser et al., 2010, Woolley et al., 2008). Functioning of these circuits is likely critical for engagement of control over obsessions and compulsions and, presumably, success abstaining from performing rituals during exposure-based CBT (Kalanthroff et al., 2017). More work, however, is needed to identify neural predictors of CBT in pediatric OCD, which could aid in treatment decisions and potentially lessen family burden, such as delayed treatment response, cost of treatments, and the complex decision to begin pharmacotherapy for young children.

Meta- and mega-analysis of structural abnormalities in pediatric OCD by the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) consortium examine the largest samples to date. ENIGMA highlighted larger thalamic volumes (Boedhoe et al., 2017b), but no significant differences in cortical thickness (Boedhoe et al., 2017a) in unmedicated youth with OCD compared to healthy participants. Diffusion MRI findings examining white matter structure have been mixed, identifying either no differences (Fitzgerald et al., 2014, Jayarajan et al., 2012, Silk et al., 2013), greater (Zarei et al., 2011, Gruner et al., 2012) or lower fractional anisotropy (FA)(Rosso et al., 2013, Lazaro et al., 2014a), or lower white matter volume (Chen et al., 2013, Lazaro et al., 2014b) in pediatric OCD across mainly corpus callosum, frontal, and cingulate areas. One pediatric study suggested that caudate volume predicts response to group CBT or pharmacotherapy (Vattimo et al., 2019), but no study, to our knowledge, has assessed brain structure or structural connectivity as predictors of individual CBT in unmediated pediatric OCD.

Herein, we investigated alterations in brain structure and structural connectivity in unmedicated children/adolescents with OCD and tested whether these structural measures predicted CBT outcomes. High-resolution structural and diffusion-weighted MRI data were acquired before patients completed 12–16 CBT sessions. Given small effect sizes observed by ENIGMA OCD, we did not expect significant group differences in subcortical volumes or cortical thickness. Instead, we hypothesized that variation in fronto-parietal and cingulo-opercular network regions would predict CBT outcomes. We also hypothesized altered structural connectivity in the corpus callosum and in tracts within fronto-partial and cingulo-opercular networks that are functionally altered in OCD. We hypothesized that such alterations in structural connectivity would also predict CBT outcomes. Finally, we took steps to characterize effect size and sensitivity/specificity of associations between treatment outcomes and brain structure.

Methods

Participants

Fifty-five unmedicated 7–18-year-olds participated in the current study from 2014–2018, including n=28 patients with OCD and n=27 healthy participants (group-matched on age and sex). Participants were recruited from the New York City area using flyers, internet advertisements, and word-of-mouth. The Institutional Review Board of the New York State Psychiatric Institute approved this study. Children provided assent and parents/guardians provided written informed consent. Participants with MRI contraindications, history of neurological illness, past seizures, head trauma with loss of consciousness, or pervasive developmental disorder, or Wechsler Abbreviated Scale of Intelligence IQ score <80 were excluded.

Participants were clinically assessed using the Anxiety and Related Disorders Interview Schedule (ADIS) for DSM-IV (Albano and Silverman, 1996). A trained independent evaluator assessed patients’ OCD symptom severity using the Children’s Yale-Brown Obsessive Compulsive Scale (CY-BOCS; (Scahill et al., 1997). The CY-BOCS includes 10 items rated on an ordinal 0–4 scale; total scores were calculated. Participants were included in the OCD group if they met DSM-IV criteria for OCD and exhibited clinically significant obsessive-compulsive symptoms (CY-BOCS≥16). Comorbid anxiety disorders, but no other lifetime psychiatric diagnoses, were permitted in the OCD group as long as OCD was the primary diagnosis. Healthy participants had no lifetime psychiatric disorders. See Appendix S1 for details.

Treatment

Patients diagnosed with OCD received standardized CBT with EX/RP (March and Mulle, 1998) with a clinical psychologist or advanced graduate student in clinical psychology. Patients received 12–16 hour-long sessions (see Appendix S1). As part of clinical best practice, patients (n=4) exhibiting no change or worsening of symptoms (CY-BOCS) after 6 weeks of treatment were offered the option of psychopharmacological intervention (selective serotonin reuptake inhibitor; SSRI). CY-BOCS scores were the primary outcome measure assessed at pre-, mid-, and post-treatment. CBT “responders” were defined by a post-treatment CY-BOCS≤12, given prior work (Simpson et al., 2006).

MRI acquisition

Neuroimaging acquisition (see Appendix S1) was adapted from the Human Connectome Project protocols (Van Essen et al., 2012) for a GE Signa 3-Tesla MR750 scanner (Milwaukee, WI) with a 32-channel Nova head coil. Two high-resolution T1-weighted BRAVO structural scans (0.8mm isotropic voxels) and a high-resolution T2-weighted scan were acquired for each participant. Two runs of diffusion MRI data were collected (single-shot EPI; b value=1000s/mm2; in-plane resolution=1.5mm, slice thickness=2.0mm), along 32 non-collinear directions.

Structural image processing

Structural images were averaged and bias field corrected (see Appendix S1).

Data were then processed in FreeSurfer v6.0 (Fischl et al., 2002). This included subcortical segmentation (left and right thalamus, caudate, putamen, pallidum, hippocampus, amygdala, and nucleus accumbens) and cortical parcellation (74 regions per hemisphere) based on the Destrieux atlas (Destrieux et al., 2010). Output images were visually inspected for quality; no manual edits were needed.

Diffusion MRI processing

Diffusion MRI data were assessed for data quality and head motion. A structural connectivity matrix was estimated for each participant using the standard MRtrix3 preprocessing pipeline (Tournier et al., 2004). This yielded 164×164 symmetrical streamline count matrix for each participant using the FreeSurfer segmentations/parcellations noted above (148 cortical, 14 subcortical regions, and the left and right cerebellum). See Appendix S1 for details. Using this approach, streamline count between brain regions is proportional to the cross-sectional area of white matter fibers connecting those regions. Thus, streamline count is a biologically plausible metric of “structural connectivity” proposed to represent the communication ‘bandwidth’ between regions, though it may indicate either sparse but fast (i.e. few large diameter axons) or dense but slow (i.e. many small diameter axons) connections (Smith et al., 2015). Furthermore, streamline counts in macaques significantly correlate with connection strength from anatomical tract-tracing, supporting its validity as an indicator of fiber connection strength (van den Heuvel et al., 2015).

Statistical analysis

Participant characteristics.

Data analysis was performed in R v3.5.0 (Team, 2015). Group differences in demographic and clinical characteristics were tested using t-tests for continuous variables and chi-squared tests for categorical variables.

Treatment outcomes.

Linear regression was used to examine whether demographic factors or pre-treatment CY-BOCS related to the number of treatment sessions completed. CY-BOCS scores were examined with an intent-to-treat approach across all three assessment points (and allowing for missing data) using linear mixed-effects (LME) models with the R lme4 package (Bates et al., 2014). Specifically, all available CY-BOCS data (repeated measure: pre-, mid-, and post-treatment) were included as the dependent variable with a random effect for participant and fixed effects for time (pre-, mid-, post-treatment), and controlled for demographic characteristics: age, sex, and IQ.

Structural analysis

Group differences.

Group differences in regional subcortical volume and cortical thickness were assessed using linear regressions controlling for age, sex, and IQ (intra-cranial volume [ICV] covaried in subcortical analyses). False discovery rate (FDR) was used to correct for multiple comparisons (14 subcortical analyses and 148 cortical analyses). As in the ENIGMA studies (Boedhoe et al., 2017a, Boedhoe et al., 2017b), effect size was calculated as Cohen’s d (tn1+n2n1n2df) from the group difference regression t-statistic. A follow-up vertex-wise analysis (FreeSurfer QDEC) probed group differences in cortical thickness (smoothed at 20mm), controlling for age, sex, and IQ. Results were corrected for multiple comparisons using Monte Carlo simulations with a vertex-wise threshold of p<.001 and an analysis-level threshold of p<.05 correcting for tests in each hemisphere (mri_glmfit-sim --cache 3 abs --2spaces).

Treatment outcomes.

To examine whether treatment was associated with subcortical volumes or with cortical thickness, LME models were conducted with CY-BOCS scores as the repeated measure dependent variable, a random effect for participant, and fixed effects for time (pre-, mid-, and post-treatment), regional volume/thickness, and the interaction between time and volume/thickness, controlling for age, sex, and IQ (ICV covaried in subcortical analyses). The structure × time interaction (FDR-corrected across regions as above) was the predictor of interest, indicating that the slope of change in CY-BOCS over treatment differed as a function of volume/thickness.

A follow-up vertex-wise analysis was conducted in QDEC. As this LME model is not currently implemented in FreeSurfer, symptom improvement (pre- to post-treatment CY-BOCS difference scores) was examined as a predictor of vertex-wise thickness, controlling for age, sex, and IQ. Results were corrected for multiple comparisons as above (vertex-wise p<.001, analysis-level p<.05). Follow-up regressions confirmed that resultant clusters predicted post-treatment CY-BOCS scores, controlling for age, sex, IQ, and pre-treatment CY-BOCS scores.

Sensitivity analyses ensured that medication augmentation or comorbid anxiety did not influence the results. First, LME models were re-run for regions identified above excluding patients receiving mid-treatment psychopharmacological augmentation (n=4). Second, LME models were re-run controlling for comorbid anxiety in the full sample. Finally, analyses characterized effect size using logistic regression to predict response to CBT (post-treatment CY-BOCS≤12).

Diffusion analysis

Two models were used to examine group differences in structural connectivity (streamline count) and associations with treatment outcome (pre- to post-treatment CY-BOCS difference scores), controlling for age, sex, and IQ. Structural connectivity matrices were analyzed using the network-based statistics (NBS) toolbox (Zalesky et al., 2010) in Matlab v2018a. NBS controls for family-wise error rate using permutation testing to identify components or “clusters” of contiguous region-to-region connections. Each model was tested using two one-tailed contrasts (OCD>healthy and healthy>OCD; positive and negative associations with CY-BOCS change), both corrected to an analysis-level p<.025 (10,000 permutations). To avoid examining very sparse connections, those with median streamline count<100 were zeroed out for exclusion from analysis. See Appendices S1 and S2 for details and convergent results using unthresholded matrices. As above, sensitivity analyses examined the influence of medication augmentation or comorbid anxiety.

Results

Participant characteristics

Fifty-five children and adolescents (27 healthy; 28 OCD) participated in the current study. Diffusion data was available for 50 participants (25 healthy; 25 OCD: not acquired n=2 OCD, poor quality/excessive head motion n=1 OCD, n=2 healthy). No demographic differences were noted between groups (Table 1). Seventeen of the 28 OCD patients (60.7%) had a secondary anxiety diagnosis (12 generalized, 7 social, 3 separation, 2 specific phobia; Table S1). Nearly all patients were treatment naïve (see Appendix S1).

Table 1:

Demographic and Clinical Characteristics

Healthy (n=27) OCD (n=28) Difference
Age (years) 11.26 (3.23) 12.14 (3.34) t=1.00
Sex (n/% female) 13 (48.1 %) 14 (50.0 %) χ2=0.00
IQ 109.59 (12.14) 106.96 (16.20) t=−0.68
CY-BOCS Pre-Treatment 24.32 (5.14)
CY-BOCS Post-Treatment 15.72 (7.73)#
CY-BOCS Improvement 8.52 (6.66)#
N CBT Sessions Completed 13.14 (3.29)#
Intracranial Volume (× 10,000 mm3) 13.88 (1.66) 14.49 (1.50) t=1.44
Subcortical Gray Volume (× 1000 mm3) 57.41 (4.24) 59.01 (47.43) t=1.31
Mean Cortical Thickness (mm) 2.66 (0.09) 2.65 (0.08) t=−0.42
Open in a new tab

Demographic and clinical characteristics of the sample are presented here for the healthy youth and the youth with OCD. Global brain structure measures are additionally displayed by group. No significant group differences were found for any of these variables (Difference: t- or χ2 test). CY-BOCS improvement indicates difference scores (pre-treatment –post-treatment) i.e., high values indicate improvement with treatment. OCD=Obsessive Compulsive Disorder; CY-BOCS=Children’s Yale-Brown Obsessive Compulsive Scale;

#

n=25 patients with OCD.

Treatment outcomes

Most patients completed the full course of CBT (23 of 28 completed 12+ sessions; M=13.14 sessions, SD=3.29, range=5–16). Number of sessions completed was not associated with any demographic variable or pre-treatment CY-BOCS (all ps>.05). Three participants dropped out of the study before the post-treatment assessment (see Appendix S2) and thus did not complete all CY-BOCS assessments. Four patients initiated SSRI augmentation (sertraline) mid-treatment as they did not show meaningful improvement in symptoms from pre- to mid-treatment; these patients also did not show symptom improvement from mid- to post-treatment timepoints (Figure S1).

CY-BOCS scores significantly decreased from pre- to post-treatment (paired t-test: t(24)=6.40, p<.001). In a LME model (fixed effects marginal R2=.34), CY-BOCS scores declined significantly across assessment timepoints (b=−4.20, t=−7.63, p<.001; Figure S1). In regression analyses, there were no significant effects of age, sex, IQ, or comorbid anxiety predicting post-treatment CY-BOCS, controlling for pre-treatment CY-BOCS (all ps>.05). Ten patients were classified as “responders” (post-treatment CY-BOCS≤12).

Structural analyses

Group differences.

No group differences in global structural measures were found (Table 1). No FDR-corrected group differences were detected in subcortical volumes (Cohen’s d=−0.37–0.44; Table S2) or regional cortical thickness (Cohen’s d=−0.94–0.89; Table S3; Figure S2). Effect sizes are compared to ENIGMA in Tables S2 and S4 (Desikan-Killiany atlas (Desikan et al., 2006)). No clusters passed multiple comparisons correction in vertex-wise analysis.

Treatment outcomes.

No FDR-corrected subcortical volume × time (pre-, mid-, post-treatment assessment) interactions were detected (see Appendix S2; Table S2). Nine cortical regions (Destrieux atlas) showed significant FDR-corrected thickness × time interactions predicting CY-BOCS scores (Figure 1; Table 2; Table S3). These included the left parietal lobe (angular and supramarginal gyri [SMG]), right insular cortex, left middle and superior frontal, left precentral, left superior temporal, and left and right lingual gyri. For all nine regions, patients with the thinnest regional cortex showed the greatest symptom decline with treatment. Simultaneously entered into a linear regression, thickness in these nine regions jointly predicted a large portion of variance in post-treatment CY-BOCS scores—significantly improving prediction above and beyond pre-treatment CY-BOCS scores, age, sex, IQ (model change F=13.66, p<.001; model R2=.96, adjusted R2=.90; R2 change=.50, adjusted R2 change=.55).

Figure 1:

Figure 1:

Open in a new tab

All cortical parcellations from the Destrieux et al., 2010 atlas are outlined in gray on this inflated cortical surface representation. Thickness × time interactions in linear-mixed effects models predicting CY-BOCS scores passing FDR correction (Table 2) are shown in yellow-red (t-stat of interaction effect). The left supramarginal gyrus cluster identified in the vertex-wise QDEC analysis is overlaid in black. For all identified regions, patients with the thinnest regional cortex showed the greatest symptom decline over the course of treatment.

Table 2:

Cortical Thickness Predicting Treatment Response

Index Hemi Region t q Response d
25 L Angular Gyrus 3.43 .021 0.88
26 L Supramarginal Gyrus 5.00 .001 1.01
39 R Horizontal ramus of the anterior segment of the lateral sulcus (or fissure) 4.83 .001 0.60
15 L Middle Frontal Gyrus 3.52 .017 0.84
16 L Superior Frontal Gyrus 3.80 .012 0.89
29 L Precentral Gyrus 4.69 .001 1.25
34 L Lateral aspect of the superior temporal gyrus 3.59 .016 0.95
61 L Medial occipito-temporal sulcus and lingual sulcus 3.71 .013 0.40
61 R Medial occipito-temporal sulcus and lingual sulcus 4.36 .002 0.37
Open in a new tab

The nine regions showing significant cortical thickness × time interactions predicting change in CY-BOCS scores over treatment are summarized here. These regions all passed FDR correction for multiple comparisons. The Index number corresponds to the list of anatomical parcellations originally described in Table 1 of Destrieux et al., 2010. The hemisphere of each region is noted in the Hemi column (L=left, R=right). The t-statistic of the interaction effect and its FDR-corrected p-value are denoted in the t and q columns respectively. Positive interaction t-statistics indicate that thinner cortices associated with greater reduction in symptoms over the course of treatment. The Response d column indicates the Cohen’s d effect size for differences in regional thickness by non-responders versus responders (positive d indicates thinner cortices in responders). All interaction effects were significant at p<.00125 uncorrected. Results for all cortical parcellations are found in Table S3.

In a follow-up vertex-wise analysis, one left SMG cluster (Figure 1; cluster-wise p=.005, 1264 voxels, 534.58mm2, peak co-ordinates = −55.1, −41.7, 44.5) was significantly related to CY-BOCS improvement (difference scores). A post-hoc regression confirmed that mean thickness in this cluster significantly predicted post-treatment CY-BOCS scores (b=24.35, B=0.57, t=−4.76, p<.001; Figure 2a), controlling for age, sex, IQ, and pre-treatment CY-BOCS, such that 0.04mm thinner cortex (1.49% difference relative to mean thickness [2.75mm], SD=0.18mm) was associated with 1 point lower post-treatment CY-BOCS.

Figure 2:

Figure 2:

Open in a new tab

A) Left SMG thickness from the vertex-wise analysis identified cluster associated with improvement in CY-BOCS scores (pre- to post-CBT), such that thinner cortex related to greater improvement. Treatment responders are denoted by filled green circles, treatment non responders are denoted by open red diamonds, and non-responders who recieved SSRI augmentation mid-treatment are denoted by blue diamonds with a plus-sign. B) Group differences in left SMG thickness are shown as a function of response after CBT with non-responders in red and responders in green. C) ROC analysis of left SMG thickness predicting treatment responder/non-responder status (blue line) relative to baseline CY-BOCS predicting treatment responder/non-responder status (green line). Sensitivity/specificity is denoted at a threshold of 2.77mm, including 95% confidence intervals.

Assessing potential confounds, cortical thickness × time interactions for all nine atlas regions and the left SMG cluster remained significant excluding n=4 patients that received psychopharmacological augmentation mid-treatment (ts>2.17; ps<.04) and when including comorbid anxiety as a binary covariate (ts>3.21; ps<.003). Most of the identified regions exhibited expected negative correlations between age and cortical thickness, i.e. thinner cortex in older youths. Exploratory analyses of thickness by age interactions predicting treatment response are presented in Appendix S2.

Further characterizing effect size of the vertex-wise analysis, thickness in the left SMG cluster significantly predicted CBT response (post-treatment CY-BOCS<=12) in a logistic regression (z=−2.22, p=.026), controlling for age, sex, IQ, and pre-treatment CY-BOCS, such that 0.01mm thinner cortex associated with 1.19 times greater odds of response (95% CI=1.02–1.39). Comparing left SMG thickness between patients who did vs. did not respond to CBT indicated a large effect size (Cohen’s d=1.42; Figure 2b). Additionally, in a ROC analysis (Figure 2c), thresholding left SMG thickness at 2.77mm was 72.22% specific and 90.00% sensitive in predicting response (AUC=86.67), which was significantly more discriminating than pre-treatment CY-BOCS scores (AUC=63.06).

Diffusion analyses

Group differences.

NBS analyses identified one component (p<.002) of 10 connections between 11 regions showing significant healthy>OCD differences in streamline count (all connections ts<−3.12, ps<.003, Cohen’s ds<−0.92; Table S5, Figure 3a). Regions included the left anterior cingulate cortex (ACC), insular cortex, thalamus, putamen, and inferior, middle, and superior frontal sulci. Mean streamline count across these connections differed by group (b=−102.67, t=−7.06, p<.001, Cohen’s d=−2.10; Figure 3b). Supplementary analysis of unthresholded matrices identified an overlapping component (p=.018; Table S6, Figure S3). Given prior work, we also assessed group differences in FA within this main component and no significant findings were detected.

Figure 3:

Figure 3:

Open in a new tab

A) Group differences in structural connectivity (streamline count) for the component identified in the NBS analysis (Table S5) are presented here. A sphere is placed at the centeroid of each region, where the size of the sphere is proportion to number of significant connections to which it contributes. The size of each connection reflects the strength of the group difference effect. B) Group differences in average streamline count across all connections in the identified component are presented here (healthy M=328.22, SD=52.38 vs. OCD M=227.69, SD=46.83). Individual points on the plot represent the values for each participant in the study. mACC = middle-anterior part of the cingulate gyrus and sulcus; MFG=middle frontal gyrus; InsCircSup= superior segment of the circular sulcus of the insula, InsCircInf= inferior segment of the circular sulcus of the insula, ACC= anterior part of the cingulate gyrus and sulcus; IFS=inferior frontal sulcus; MFS=middle frontal sulcus; SFS=superior frontal sulcus.

Treatment outcomes.

While NBS analyses did not reveal additional components that predicted change in CY-BOCS, follow-up analyses tested whether the above component also predicted CY-BOCS change. In a LME model, mean streamline count across connections interacted with time to predict change in CY-BOCS (b=0.03, B=1.91, t=2.02, p=.05), controlling for age, sex, and IQ. Of the 10 connections, medial ACC-putamen showed a significant count × time interaction (t=2.16, p=.04) such that fewer streamline counts predicted greater decline in symptoms over treatment, but this did not pass FDR (Table S5). Further, having fewer medial ACC-putamen streamlines predicted greater reductions in post-treatment CY-BOCS (b=0.07, B=0.95, t=3.07, p=.007) above and beyond pre-treatment CY-BOCS and demographics. Entered simultaneously in a linear regression, all component connections jointly predicted a large portion of variance in post-treatment CY-BOCS—significantly improving prediction above and beyond pre-treatment CY-BOCS, age, sex, IQ (model change F=5.62, p=.02; model R2=.93, adjusted R2=.81; R2 change=.51, adjusted R2 change=.50). Examining a multi-modal combination of measures, SMG cortical thickness predicted significant additional variance in post-treatment CY-BOCS (b=23.66, B=0.20, t=2.72, p=.02) above and beyond medial ACC-putamen streamline count, controlling for pre-treatment CY-BOCS and demographics. Streamline count was no longer significant in this concurrent model (b=0.007, B=0.09, t=0.23, p=.82). Given the co-linearity between these structural measures (r=.47, t(23)=2.53, p=.01), they may be capturing overlapping variance in treatment response.

Excluding n=4 patients who received psychopharmacological augmentation mid-treatment minimally affected the time × mean streamline count LME result (b=0.03, B=1.31, t=1.98, p=.055) but the medial ACC-putamen effect was no longer significant (b=0.01, B= 0.91, t=1.10, p=.28). Neither effect remained significant when including comorbid anxiety as a binary covariate (ts<1.62; ps>.05).

Discussion

We examined alterations in brain structure and structural connectivity in children and adolescents with OCD and healthy controls and tested whether structural markers predicted outcomes following CBT with EX/RP. Significant group differences in cortical thickness were not detected, but thinner cortex in fronto-parietal regions, particularly left SMG, robustly predicted greater improvement in OCD symptoms after CBT. Compared to healthy youth, patients with OCD exhibited less connectivity (fewer streamline counts) between cingulo-opercular regions that were somewhat predictive of their CBT outcome. These data suggest, for the first time, that structural proprieties (cortical thickness and connectivity) of task control circuits may aid in identifying which pediatric patients respond best to individual CBT.

In contrast to ENIGMA OCD findings, we did not detect group differences in thalamic volumes in unmedicated children/adolescents (Boedhoe et al., 2017b), perhaps reflecting our smaller, more homogenous, and slightly younger sample. In addition, our OCD sample was primarily treatment naïve whereas the ENIGMA sample included previously medicated participants. Prior data suggest that caudate volume in children/adolescents predicts treatment outcomes, but this was not specific to CBT (n=17 in group CBT and n=12 receiving fluoxetine) (Vattimo et al., 2019). Herein, thinner cortex in fronto-parietal regions robustly predicted greater CY-BOCS improvement after individual CBT among pediatric patients with OCD. These regions included the left supramarginal, angular, middle frontal, superior frontal, precentral, and superior temporal gyri, right anterior insula, and the left and right medial occipito-temporal and lingual sulci. Vertex-wise analysis (CY-BOCS difference scores predicting thickness rather than LME models) with stringent multiple comparisons correction, revealed a strong effect in the left supramarginal gyrus. Although effects were strongest in this region, we cannot ascribe a single functional consequence to cortical thickness variability in a specific area; instead, the current results more likely point to the involvement of broader fronto-parietal control circuitry in OCD treatment response. As these fronto-parietal regions typically thin over development, we speculate that advanced cortical maturation might relate to better treatment outcomes. Previous findings suggest reduced SMG thickness or gray matter density in OCD (Fallucca et al., 2011, Valente et al., 2005, Yoo et al., 2008, Rotge et al., 2010, Narayan et al., 2008) and that smaller parietal gray matter volumes, including SMG, may remediate with SSRI treatment in pediatric OCD (Lázaro et al., 2009). Critically, prior studies have not examined whether pre-treatment differences in brain structure and connectivity can predict which pediatric patients responded to individual CBT, an important step towards personalized medicine efforts.

Passing criteria for statistical significance does not ensure that a neural marker can be clinically useful in predicting outcomes, but rather this depends on the predictor’s effect size and sensitivity/specificity. Herein, we found a large effect size for SMG thickness between patients who did versus did not respond; moreover, SMG thickness predicted response with >70% specificity and 90% sensitivity. In fact, SMG thickness had more predictive power than pre-treatment symptom severity alone. The specific threshold identified may help guide future work but should replicated and validated before being utilized clinically. Additionally, thickness in the nine identified regions together improved model fit, predicting post-treatment symptom severity significantly above and beyond pre-treatment severity and demographic factors. Importantly, structural predictors could be combined with additional measures to improve treatment prediction in the future. For example, data suggest that baseline glutamate in the ACC predicts CBT response in youth (O’Neill et al., 2017). The combination of our multi-modal structural measures did not improve prediction, likely because these measures were themselves colinear. Perhaps combining such spectroscopy measures with one of our structural measures would lead to even better predictive capacity.

We also identified altered structural connectivity in pediatric OCD with fewer streamline counts between left hemisphere cingulo-opercular regions (e.g., cingulate and insula) compared to healthy participants. Group differences were of large effect size (Cohen’s d>0.92). Furthermore, medial ACC-putamen streamline count and entering all ten connections in a simultaneous regression predicted CBT outcomes. These findings build on a sparse literature examining diffusion MRI in pediatric OCD primarily employing voxel-wise methods (e.g., Tract-Based Spatial Statistics). The largest study to date (>60 pediatric patients with OCD [>80% on SSRIs]) identified less FA in the corpus callosum (Lazaro et al., 2014a) and less white matter volume in the left superior and middle frontal gyri and anterior cingulate gyrus (Lazaro et al., 2014b). Although our findings also implicate connections between frontal and cingulate regions, our methodology was indeed different and applied to an unmedicated sample. Other findings have been mixed, indicating either greater FA across the brain (Zarei et al., 2011), including the corpus callosum, cingulum and other tracts (Gruner et al., 2012), lower FA (Rosso et al., 2013) or less white matter volume (Chen et al., 2013) in cingulate, callosal and other regions, or no differences (Fitzgerald et al., 2014, Jayarajan et al., 2012, Silk et al., 2013) compared to healthy youth. These prior studies typically examined small sample sizes (8–36 patients) with most taking psychiatric medications and/or exhibiting psychiatric comorbidities. In contrast, our analyses tested structural connectomes estimated using probabilistic tractography, permitting direct interpretation of regional connectivity, rather than voxel-wise differences in FA. Voxel-wise FA can reflect several different microstructural properties, e.g. myelin, axon density, and fiber orientation spread, making interpretation regarding white matter ‘integrity’ difficult without additional evidence (Jones et al., 2013).

The current study had several strengths. First, our OCD sample was free of psychotropic medications at the MRI scan, which is important given the effects of SSRIs on brain structure and connectivity (e.g. (Lugo-Candelas et al., 2018, Bernstein et al., 2018). Second, this study used high-resolution data and the most up-to-date analysis methods (FreeSurfer v6.0 and MRtrix3), allowing for improved data quality and segmentation/parcellation. Third, our structural measures showed high test-retest reliability (see Appendix S2), suggesting that our findings are likely reproducible. Finally, our analyses examined all available CY-BOCS data rather than excluding participants who do not finish treatment and utilized several methods to quantify treatment prediction effect size.

Several limitations should also be noted. First, although this is an important first step towards identifying potential pre-treatment markers of CBT response in pediatric OCD, independent sample replication was not available in the current study and is critical to advance this work. Second, while larger than many prior studies, our sample size was still relatively modest, likely contributing to non-significant group differences in brain structure. Additionally, effects that pass stringent correction for multiple tests could be inflated estimates of true underlying effect sizes. We provide thorough information regarding effect sizes of our significant and non-significant results to help guide future work, benchmark initial effect size estimates, and aid future meta-analyses. Third, over half of patients with OCD exhibited secondary anxiety disorders, and four patients began pharmacotherapy mid-CBT. Several post-hoc analyses confirmed that our results of cortical thickness predictors of treatment response were not driven by these factors. Fourth, the current study design precludes determining the specificity of these predictors to CBT versus placebo or other types of interventions. Future work using a multi-arm treatment study should test whether these structural factors predict response to CBT specifically or mark treatment responsivity more generally.

In summary, children and adolescents with OCD exhibited altered cingulo-opercular structural connectivity compared to healthy youth and, importantly, thinner fronto-parietal cortices predicted CBT outcomes. These were of large effect sizes, particularly associations with left supramarginal gyrus thickness, which predicted response with relatively high sensitivity and specificity. Future work should establish the replicability of these findings in independent samples. These findings represent an important step towards understanding the neural circuits underlying pediatric OCD as well as identifying robust neural markers that predict response to treatment.

Supplementary Material

Supp info

Key points.

  • While CBT is an effective, first-line treatment, not all pediatric patients with OCD respond. Neuroscience research may elucidate pre-treatment markers to identify patients most likely to respond well.

  • The current study identified that cortical thickness in nine fronto-parietal brain regions, particularly supramarginal gyrus, predicted response to CBT.

  • Novel alterations in cingulo-opercular structural connectivity were also identified in patients with OCD compared to healthy children.

  • Structural features of task control circuits may thus aid in identifying which pediatric patients respond best to CBT.

  • Future studies should replicate and extend this work in independent samples. Easy-to-acquire pre-treatment neural markers could aid precision medicine efforts and help guide children suffering with OCD to the most effective treatments.

Acknowledgements

This work was supported by a grant from the National Institute of Mental Health (1R21MH101441-01A1; MPIs: Marsh and Rynn). The study was registered with clinicaltrials.gov ( NCT02421315). The authors would like to thank the families who participated in this study. The authors have declared that they have no competing or potential conflicts of interest.

Footnotes

Conflict of interest statement: No conflicts declared.

Supporting information

Additional supporting information may be found online in the Supporting Information section at the end of the article:

Appendix S1. Supplementary methods.

Appendix S2. Supplementary results.

Figure S1. Patient-level change in symptom severity.

Figure S2. Group differences in cortical thickness.

Figure S3. Group differences in unthresholded structural connectivity matrices.

Table S1. Demographic and clinical characteristics comparing patients with and without comorbid anxiety.

Table S2. Group differences in subcortical volumes and associations with treatment.

Table S3. Group differences in cortical thickness and associations with treatment (destrieux atlas parcellation).

Table S4. Group differences in cortical thickness and associations with treatment (desikan atlas parcellation).

Table S5. Group differences in structural connectivity and associations with treatment response and remission.

Table S6. Group differences in structural connectivity (unthresholded matrices).

References

  1. ALBANO A & SILVERMAN W (1996). The Anxiety Disorders Interview Schedule for Children for DSM-IV: Clinician manual (child and parent versions). San Antonio, TX: Psychological Corporation. [Google Scholar]
  2. BATES D, MÄCHLER M, BOLKER B & WALKER S (2014). Fitting linear mixed-effects models using lme4. arXiv preprint arXiv:14065823. [Google Scholar]
  3. BERNSTEIN GA, CULLEN KR, HARRIS EC, CONELEA CA, ZAGOLOFF AD, CARSTEDT PA, LEE SS & MUELLER BA (2018). Sertraline Effects on Striatal Resting-State Functional Connectivity in Youth With OCD: A Pilot Study. Journal of the American Academy of Child & Adolescent Psychiatry. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. BOEDHOE PS, SCHMAAL L, ABE Y, ALONSO P, AMEIS SH, ANTICEVIC A, ARNOLD PD, BATISTUZZO MC, BENEDETTI F, BEUCKE JC, BOLLETTINI I, BOSE A, BREM S, CALVO A, CALVO R, CHENG Y, CHO KIK, CIULLO V, DALLASPEZIA S, DENYS D, FEUSNER JD, FITZGERALD KD, FOUCHE JP, FRIDGEIRSSON EA, GRUNER P, HANNA GL, HIBAR DP, HOEXTER MQ, HU H, HUYSER C, JAHANSHAD N, JAMES A, KATHMANN N, KAUFMANN C, KOCH K, KWON JS, LAZARO L, LOCHNER C, MARSH R, MARTINEZ-ZALACAIN I, MATAIX-COLS D, MENCHON JM, MINUZZI L, MORER A, NAKAMAE T, NAKAO T, NARAYANASWAMY JC, NISHIDA S, NURMI E, O’NEILL J, PIACENTINI J, PIRAS F, PIRAS F, REDDY YCJ, REESS TJ, SAKAI Y, SATO JR, SIMPSON HB, SORENI N, SORIANO-MAS C, SPALLETTA G, STEVENS MC, SZESZKO PR, TOLIN DF, VAN WINGEN GA, VENKATASUBRAMANIAN G, WALITZA S, WANG Z, YUN JY, GROUP E-OW, THOMPSON PM, STEIN DJ, VAN DEN HEUVEL OA & GROUP EOW (2017a). Cortical Abnormalities Associated With Pediatric and Adult Obsessive-Compulsive Disorder: Findings From the ENIGMA Obsessive-Compulsive Disorder Working Group. Am J Psychiatry, appiajp201717050485. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. BOEDHOE PS, SCHMAAL L, ABE Y, AMEIS SH, ARNOLD PD, BATISTUZZO MC, BENEDETTI F, BEUCKE JC, BOLLETTINI I, BOSE A, BREM S, CALVO A, CHENG Y, CHO KI, DALLASPEZIA S, DENYS D, FITZGERALD KD, FOUCHE JP, GIMENEZ M, GRUNER P, HANNA GL, HIBAR DP, HOEXTER MQ, HU H, HUYSER C, IKARI K, JAHANSHAD N, KATHMANN N, KAUFMANN C, KOCH K, KWON JS, LAZARO L, LIU Y, LOCHNER C, MARSH R, MARTINEZ-ZALACAIN I, MATAIX-COLS D, MENCHON JM, MINUZZI L, NAKAMAE T, NAKAO T, NARAYANASWAMY JC, PIRAS F, PIRAS F, PITTENGER C, REDDY YC, SATO JR, SIMPSON HB, SORENI N, SORIANO-MAS C, SPALLETTA G, STEVENS MC, SZESZKO PR, TOLIN DF, VENKATASUBRAMANIAN G, WALITZA S, WANG Z, VAN WINGEN GA, XU J, XU X, YUN JY, ZHAO Q, GROUP EOW, THOMPSON PM, STEIN DJ & VAN DEN HEUVEL OA (2017b). Distinct Subcortical Volume Alterations in Pediatric and Adult OCD: A Worldwide Meta- and Mega-Analysis. Am J Psychiatry, 174, 60–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. CHEN J, SILK T, SEAL M, DALLY K & VANCE A (2013). Widespread decreased grey and white matter in paediatric obsessive-compulsive disorder (OCD): a voxel-based morphometric MRI study. Psychiatry Res, 213, 11–17. [DOI] [PubMed] [Google Scholar]
  7. DEL CASALE A, RAPINESI C, KOTZALIDIS GD, DE ROSSI P, CURTO M, JANIRI D, CRISCUOLO S, ALESSI MC, FERRI VR & DE GIORGI R (2016). Executive functions in obsessive–compulsive disorder: An activation likelihood estimate meta-analysis of fMRI studies. The World Journal of Biological Psychiatry, 17, 378–393. [DOI] [PubMed] [Google Scholar]
  8. DESIKAN RS, SEGONNE F, FISCHL B, QUINN BT, DICKERSON BC, BLACKER D, BUCKNER RL, DALE AM, MAGUIRE RP, HYMAN BT, ALBERT MS & KILLIANY RJ (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage, 31, 968–980. [DOI] [PubMed] [Google Scholar]
  9. DESTRIEUX C, FISCHL B, DALE A & HALGREN E (2010). Automatic parcellation of human cortical gyri and sulci using standard anatomical nomenclature. Neuroimage, 53, 1–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. FALLUCCA E, MACMASTER FP, HADDAD J, EASTER P, DICK R, MAY G, STANLEY JA, RIX C & ROSENBERG DR (2011). Distinguishing Between Major Depressive Disorder and Obsessive-Compulsive Disorder in Children by Measuring Regional Cortical Thickness. Archives of General Psychiatry, 68, 527–533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. FISCHL B, SALAT DH, BUSA E, ALBERT M, DIETERICH M, HASELGROVE C, VAN DER KOUWE A, KILLIANY R, KENNEDY D & KLAVENESS S (2002). Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron, 33, 341–355. [DOI] [PubMed] [Google Scholar]
  12. FITZGERALD KD, LIU Y, REAMER EN, TAYLOR SF & WELSH RC (2014). Atypical frontal-striatal-thalamic circuit white matter development in pediatric obsessive-compulsive disorder. J Am Acad Child Adolesc Psychiatry, 53, 1225–1233, 1233 e1221–1229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. FITZGERALD KD, LIU Y, STERN ER, WELSH RC, HANNA GL, MONK CS, PHAN KL & TAYLOR SF (2013). Reduced error-related activation of dorsolateral prefrontal cortex across pediatric anxiety disorders. J Am Acad Child Adolesc Psychiatry, 52, 1183–1191 e1181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. FULLANA MA, CARDONER N, ALONSO P, SUBIRA M, LOPEZ-SOLA C, PUJOL J, SEGALAS C, REAL E, BOSSA M, ZACUR E, MARTINEZ-ZALACAIN I, BULBENA A, MENCHON JM, OLMOS S & SORIANO-MAS C (2014). Brain regions related to fear extinction in obsessive-compulsive disorder and its relation to exposure therapy outcome: a morphometric study. Psychol Med, 44, 845–856. [DOI] [PubMed] [Google Scholar]
  15. GARCIA AM, SAPYTA JJ, MOORE PS, FREEMAN JB, FRANKLIN ME, MARCH JS & FOA EB (2010). Predictors and moderators of treatment outcome in the Pediatric Obsessive Compulsive Treatment Study (POTS I). J Am Acad Child Adolesc Psychiatry, 49, 1024–1033; quiz 1086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. GRUNER P, VO A, IKUTA T, MAHON K, PETERS BD, MALHOTRA AK, ULUG AM & SZESZKO PR (2012). White matter abnormalities in pediatric obsessive-compulsive disorder. Neuropsychopharmacology, 37, 2730–2739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. HOEXTER MQ, DINIZ JB, LOPES AC, BATISTUZZO MC, SHAVITT RG, DOUGHERTY DD, DURAN FL, BRESSAN RA, BUSATTO GF, MIGUEL EC & SATO JR (2015). Orbitofrontal Thickness as a Measure for Treatment Response Prediction in Obsessive-Compulsive Disorder. Depress Anxiety, 32, 900–908. [DOI] [PubMed] [Google Scholar]
  18. HOEXTER MQ, DOUGHERTY DD, SHAVITT RG, D’ALCANTE CC, DURAN FL, LOPES AC, DINIZ JB, BATISTUZZO MC, EVANS KC, BRESSAN RA, BUSATTO GF & MIGUEL EC (2013). Differential prefrontal gray matter correlates of treatment response to fluoxetine or cognitive-behavioral therapy in obsessive-compulsive disorder. Eur Neuropsychopharmacol, 23, 569–580. [DOI] [PubMed] [Google Scholar]
  19. HUYSER C, VAN DEN HEUVEL OA, WOLTERS L, DE HAAN E, LINDAUER R & VELTMAN DJ (2014). A longitudinal VBM study in paediatric obsessive-compulsive disorder at 2-year follow-up after cognitive behavioural therapy. World J Biol Psychiatry, 15, 443–452. [DOI] [PubMed] [Google Scholar]
  20. HUYSER C, VAN DEN HEUVEL OA, WOLTERS LH, DE HAAN E, BOER F & VELTMAN DJ (2013). Increased orbital frontal gray matter volume after cognitive behavioural therapy in paediatric obsessive compulsive disorder. World J Biol Psychiatry, 14, 319–331. [DOI] [PubMed] [Google Scholar]
  21. HUYSER C, VELTMAN DJ, WOLTERS LH, DE HAAN E & BOER F (2010). Functional Magnetic Resonance Imaging During Planning Before and After Cognitive-Behavioral Therapy in Pediatric Obsessive-Compulsive Disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 49, 1238–1248. [DOI] [PubMed] [Google Scholar]
  22. JAYARAJAN RN, VENKATASUBRAMANIAN G, VISWANATH B, JANARDHAN REDDY YC, SRINATH S, VASUDEV MK & CHANDRASHEKAR CR (2012). White matter abnormalities in children and adolescents with obsessive-compulsive disorder: a diffusion tensor imaging study. Depress Anxiety, 29, 780–788. [DOI] [PubMed] [Google Scholar]
  23. JONES DK, KNOSCHE TR & TURNER R (2013). White matter integrity, fiber count, and other fallacies: the do’s and don’ts of diffusion MRI. Neuroimage, 73, 239–254. [DOI] [PubMed] [Google Scholar]
  24. KALANTHROFF E, HENIK A, SIMPSON HB, TODDER D & ANHOLT GE (2017). To Do or Not to Do? Task Control Deficit in Obsessive-Compulsive Disorder. Behav Ther, 48, 603–613. [DOI] [PubMed] [Google Scholar]
  25. KORAN LM, HANNA GL, HOLLANDER E, NESTADT G, SIMPSON HB & AMERICAN PSYCHIATRIC A (2007). Practice guideline for the treatment of patients with obsessive-compulsive disorder. Am J Psychiatry, 164, 5–53. [PubMed] [Google Scholar]
  26. LÁZARO L, BARGALLÓ N, CASTRO-FORNIELES J, FALCÓN C, ANDRÉS S, CALVO R & JUNQUÉ C (2009). Brain changes in children and adolescents with obsessive–compulsive disorder before and after treatment: a voxel-based morphometric MRI study. Psychiatry Research: Neuroimaging, 172, 140–146. [DOI] [PubMed] [Google Scholar]
  27. LAZARO L, CALVO A, ORTIZ AG, ORTIZ AE, MORER A, MORENO E, CALVO R & BARGALLO N (2014a). Microstructural brain abnormalities and symptom dimensions in child and adolescent patients with obsessive-compulsive disorder: a diffusion tensor imaging study. Depress Anxiety, 31, 1007–1017. [DOI] [PubMed] [Google Scholar]
  28. LAZARO L, ORTIZ AG, CALVO A, ORTIZ AE, MORENO E, MORER A, CALVO R & BARGALLO N (2014b). White matter structural alterations in pediatric obsessive-compulsive disorder: relation to symptom dimensions. Prog Neuropsychopharmacol Biol Psychiatry, 54, 249–258. [DOI] [PubMed] [Google Scholar]
  29. LUGO-CANDELAS C, CHA J, HONG S, BASTIDAS V, WEISSMAN M, FIFER WP, MYERS M, TALATI A, BANSAL R, PETERSON BS, MONK C, GINGRICH JA & POSNER J (2018). Associations Between Brain Structure and Connectivity in Infants and Exposure to Selective Serotonin Reuptake Inhibitors During Pregnancy. JAMA Pediatr, 172, 525–533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. MAIA TV, COONEY RE & PETERSON BS (2008). The neural bases of obsessive-compulsive disorder in children and adults. Dev Psychopathol, 20, 1251–1283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. MARCH JS, FOA E, GAMMON P, CHRISMAN A, CURRY J, FITZGERALD D, SULLIVAN K, FRANKLIN M, HUPPERT J, RYNN M, ZHAO N, ZOELLNER L, LEONARD H, GARCIA A, FREEMAN J, TU X & TEAM P (2004). Cognitive-behavior therapy, sertraline, and their combination for children and adolescents with obsessive-compulsive disorder - The Pediatric OCD Treatment Study (POTS) randomized controlled trial. Jama-Journal of the American Medical Association, 292, 1969–1976. [DOI] [PubMed] [Google Scholar]
  32. MARCH JS & MULLE K (1998). OCD in children and adolescents: A cognitive-behavioral treatment manual: Guilford Press. [Google Scholar]
  33. MENZIES L, CHAMBERLAIN SR, LAIRD AR, THELEN SM, SAHAKIAN BJ & BULLMORE ET (2008). Integrating evidence from neuroimaging and neuropsychological studies of obsessive-compulsive disorder: the orbitofronto-striatal model revisited. Neurosci Biobehav Rev, 32, 525–549. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. NARAYAN VM, NARR KL, PHILLIPS OR, THOMPSON PM, TOGA AW & SZESZKO PR (2008). Greater regional cortical gray matter thickness in obsessive-compulsive disorder. Neuroreport, 19, 1551–1555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. O’NEILL J, PIACENTINI J, CHANG S, LY R, LAI TM, ARMSTRONG CC, BERGMAN L, ROZENMAN M, PERIS T, VREELAND A, MUDGWAY R, LEVITT JG, SALAMON N, POSSE S, HELLEMANN GS, ALGER JR, MCCRACKEN JT & NURMI EL (2017). Glutamate in Pediatric Obsessive-Compulsive Disorder and Response to Cognitive-Behavioral Therapy: Randomized Clinical Trial. Neuropsychopharmacology, 42, 2414–2422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. OST LG, RIISE EN, WERGELAND GJ, HANSEN B & KVALE G (2016). Cognitive behavioral and pharmacological treatments of OCD in children: A systematic review and meta-analysis. J Anxiety Disord, 43, 58–69. [DOI] [PubMed] [Google Scholar]
  37. PAGLIACCIO D, MIDDLETON R, HEZEL D, STEINMAN S, SNORRASON I, GERSHKOVICH M, CAMPEAS R, PINTO A, VAN METER P, SIMPSON HB & MARSH R (2019). Task-based fMRI predicts response and remission to exposure therapy in obsessive-compulsive disorder. Proc Natl Acad Sci U S A, 116, 20346–20353. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. ROSSO IM, BRITTON JC, STEWART SE, PAPAMIDITRIOU G, KILLGORE WDS, MAKRIS N, WILHELM S, JENIKE MA & RAUCH SL (2013). Brain White Matter Integrity and Association with Age at Onset in Pediatric Obsessive-Compulsive Disorder. Biological Psychiatry, 73, 75s–75s. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. ROTGE JY, LANGBOUR N, GUEHL D, BIOULAC B, JAAFARI N, ALLARD M, AOUIZERATE B & BURBAUD P (2010). Gray matter alterations in obsessive-compulsive disorder: an anatomic likelihood estimation meta-analysis. Neuropsychopharmacology, 35, 686–691. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. SCAHILL L, RIDDLE MA, MCSWIGGIN-HARDIN M, ORT SI, KING RA, GOODMAN WK, CICCHETTI D & LECKMAN JF (1997). Children’s Yale-Brown Obsessive Compulsive Scale: reliability and validity. J Am Acad Child Adolesc Psychiatry, 36, 844–852. [DOI] [PubMed] [Google Scholar]
  41. SILK T, CHEN J, SEAL M & VANCE A (2013). White matter abnormalities in pediatric obsessive-compulsive disorder. Psychiatry Res, 213, 154–160. [DOI] [PubMed] [Google Scholar]
  42. SIMPSON HB, HUPPERT JD, PETKOVA E, FOA EB & LIEBOWITZ MR (2006). Response versus remission in obsessive-compulsive disorder. J Clin Psychiatry, 67, 269–276. [DOI] [PubMed] [Google Scholar]
  43. SMITH RE, TOURNIER JD, CALAMANTE F & CONNELLY A (2015). The effects of SIFT on the reproducibility and biological accuracy of the structural connectome. Neuroimage, 104, 253–265. [DOI] [PubMed] [Google Scholar]
  44. TEAM RC (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
  45. TOURNIER JD, CALAMANTE F, GADIAN DG & CONNELLY A (2004). Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage, 23, 1176–1185. [DOI] [PubMed] [Google Scholar]
  46. VALENTE AA JR., MIGUEL EC, CASTRO CC, AMARO E JR., DURAN FL, BUCHPIGUEL CA, CHITNIS X, MCGUIRE PK & BUSATTO GF (2005). Regional gray matter abnormalities in obsessive-compulsive disorder: a voxel-based morphometry study. Biol Psychiatry, 58, 479–487. [DOI] [PubMed] [Google Scholar]
  47. VAN DEN HEUVEL MP, DE REUS MA, FELDMAN BARRETT L, SCHOLTENS LH, COOPMANS FM, SCHMIDT R, PREUSS TM, RILLING JK & L. LI (2015). Comparison of diffusion tractography and tract-tracing measures of connectivity strength in rhesus macaque connectome. Hum Brain Mapp, 36, 3064–3075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. VAN ESSEN DC, UGURBIL K, AUERBACH E, BARCH D, BEHRENS TE, BUCHOLZ R, CHANG A, CHEN L, CORBETTA M, CURTISS SW, DELLA PENNA S, FEINBERG D, GLASSER MF, HAREL N, HEATH AC, LARSON-PRIOR L, MARCUS D, MICHALAREAS G, MOELLER S, OOSTENVELD R, PETERSEN SE, PRIOR F, SCHLAGGAR BL, SMITH SM, SNYDER AZ, XU J, YACOUB E & CONSORTIUM WU-MH (2012). The Human Connectome Project: a data acquisition perspective. Neuroimage, 62, 2222–2231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. VAN VELZEN LS, VRIEND C, DE WIT SJ & VAN DEN HEUVEL OA (2014). Response inhibition and interference control in obsessive-compulsive spectrum disorders. Front Hum Neurosci, 8, 419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. VATTIMO EFQ, BARROS VB, REQUENA G, SATO JR, FATORI D, MIGUEL EC, SHAVITT RG, HOEXTER MQ & BATISTUZZO MC (2019). Caudate volume differences among treatment responders, non-responders and controls in children with obsessive-compulsive disorder. Eur Child Adolesc Psychiatry. [DOI] [PubMed] [Google Scholar]
  51. WATSON HJ & REES CS (2008). Meta-analysis of randomized, controlled treatment trials for pediatric obsessive-compulsive disorder. J Child Psychol Psychiatry, 49, 489–498. [DOI] [PubMed] [Google Scholar]
  52. WOOLLEY J, HEYMAN I, BRAMMER M, FRAMPTON I, MCGUIRE PK & RUBIA K (2008). Brain activation in paediatric obsessive compulsive disorder during tasks of inhibitory control. Br J Psychiatry, 192, 25–31. [DOI] [PubMed] [Google Scholar]
  53. YOO SY, ROH MS, CHOI JS, KANG DH, HA TH, LEE JM, KIM IY, KIM SI & KWON JS (2008). Voxel-based morphometry study of gray matter abnormalities in obsessive-compulsive disorder. J Korean Med Sci, 23, 24–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. ZALESKY A, FORNITO A & BULLMORE ET (2010). Network-based statistic: identifying differences in brain networks. Neuroimage, 53, 1197–1207. [DOI] [PubMed] [Google Scholar]
  55. ZAREI M, MATAIX-COLS D, HEYMAN I, HOUGH M, DOHERTY J, BURGE L, WINMILL L, NIJHAWAN S, MATTHEWS PM & JAMES A (2011). Changes in gray matter volume and white matter microstructure in adolescents with obsessive-compulsive disorder. Biol Psychiatry, 70, 1083–1090. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp info
NIHMS1063614-supplement-Supp_info.docx (4.1MB, docx)

RESOURCES