Introduction
Obsessive–compulsive disorder (OCD) is characterized by recurrent and intrusive thoughts (obsessions) accompanied by anxiety and repetitive behaviors (compulsions) to relieve the obsessional distress. Patients with OCD report inability to regulate such disturbing thoughts and feelings with anxiety, leading to compulsive behaviors (Milad & Rauch, Reference Milad and Rauch2012). Although OCD patients are aware of the irrationality of compulsive habits, overwhelming anxiety prevents the patients from resisting the repetitive (compulsive) acts (Taylor & Liberzon, Reference Taylor and Liberzon2007). In this case, inexorable thoughts combined with the feelings of anxiety can be associated with an abnormal interaction between cognition and emotion. This abnormal cognitive–affective interaction is further manifested in OCD patients by inflexible adoption of efficient learning strategies in an OCD-specific context (Zetsche et al. Reference Zetsche, Rief, Westermann and Exner2015). On a neurobiological level, the cognitive–affective dysfunction in OCD patients may relate to the unsuccessful neural interactions within brain circuits associated with emotion regulation.
The interplay between the amygdala and the prefrontal cortex such as the dorsolateral prefrontal cortex (DLPFC), ventrolateral prefrontal cortex (VLPFC), dorsomedial prefrontal cortex (dmPFC), ventromedial prefrontal cortex (vmPFC) and orbitofrontal cortex (OFC) is involved in emotion regulation including cognitive control, inhibitory control or voluntary down-regulating of emotion, especially negative emotion (Ochsner et al. Reference Ochsner, Ray, Cooper, Robertson, Chopra, Gabrieli and Gross2004; Ochsner & Gross, Reference Ochsner and Gross2005; Phillips et al. Reference Phillips, Ladouceur and Drevets2008). The amygdala is assumed to mediate identification of potential threat and evaluation of affective values (Davis, Reference Davis1992; Davis & Whalen, Reference Davis and Whalen2001). In OCD patients, previous studies have observed abnormal responsiveness in the amygdala to negative emotional stimuli or symptom-provoking stimuli (Cannistraro et al. Reference Cannistraro, Wright, Wedig, Martis, Shin, Wilhelm and Rauch2004; Lawrence et al. Reference Lawrence, An, Mataix-Cols, Ruths, Speckens and Phillips2007; Simon et al. Reference Simon, Kaufmann, Musch, Kischkel and Kathmann2010; Cardoner et al. Reference Cardoner, Harrison, Pujol, Soriano-Mas, Hernandez-Ribas, Lopez-Sola, Real, Deus, Ortiz, Alonso and Menchon2011; de Wit et al. Reference de Wit, van der Werf, Mataix-Cols, Trujillo, van Oppen, Veltman and van den Heuvel2015). However, the exaggerated amygdala activation during OCD symptom provocation was dampened by attentional distraction (Simon et al. Reference Simon, Adler, Kaufmann and Kaufmann2014). Besides the amygdala, the DLPFC is associated with cognitive control to maintain task-related requirements in the presence of task-irrelevant negative emotional distraction (Dolcos & McCarthy, Reference Dolcos and McCarthy2006; Dolcos et al. Reference Dolcos, Iordan and Dolcos2011) and effortful processing that accompanies reappraisal (Phillips et al. Reference Phillips, Ladouceur and Drevets2008). In studies with OCD patients, increased prefrontal engagements including the DLPFC were observed during symptom-provoking picture presentations, reflecting the prefrontal hyperactivation as the top-down cognitive control over affective responses in the amygdala (Simon et al. Reference Simon, Kaufmann, Musch, Kischkel and Kathmann2010, Reference Simon, Kaufmann, Kniesche, Kischkel and Kathmann2013). In a working memory (WM) task with no emotional stimuli, however, OCD patients exhibited no significantly different task-related DLPFC activation relative to healthy controls (van der Wee et al. Reference van der Wee, Ramsey, Jansma, Denys, van Megen, Westenberg and Kahn2003). Within the fronto-limbic interplay, patients with OCD showed enhanced WM task-related prefrontal demands and increased functional coupling between the prefrontal regions including the DLPFC and amygdala (de Vries et al. Reference de Vries, de Wit, Cath, van der Werf, van der Borden, van Rossum, van Balkom, van der Wee, Veltman and van den Heuvel2014). However, a recent functional connectivity study revealed that OCD patients exhibited less DLPFC engagement and dmPFC–amygdala connectivity during down-regulation of negative affect (de Wit et al. Reference de Wit, van der Werf, Mataix-Cols, Trujillo, van Oppen, Veltman and van den Heuvel2015).
Moreover, the interplay between the DLPFC and OFC is critical in cognitive–affective interaction and its disruption can be a component of the pathophysiology in psychiatric disorders (Moghaddam & Homayoun, Reference Moghaddam and Homayoun2008). In conjunction with the DLPFC, the OFC has an important role for effective inhibitory control in a delayed WM task (Petrides, Reference Petrides2000). Also, the OFC plays an integral role, acting as a hub to integrate and modulate brain activation in order to regulate the cognitive–affective responses (Rule et al. Reference Rule, Shimamura and Knight2002; Evans et al. Reference Evans, Lewis and Lobst2004). Anatomically the OFC has extensive and reciprocal connections with the DLPFC and amygdala so that it may help mediate the interaction between the DLPFC and amygdala during emotion regulation (Phillips et al. Reference Phillips, Ladouceur and Drevets2008). As a cytoarchitecturally or functionally heterogeneous region, the OFC is characterized by its subregions. The anterior part of the OFC is interconnected to the DLPFC and is involved in cognitive processing, whereas the posterior part is interconnected to the amygdala and is associated with emotional functions (Zald & Kim, Reference Zald and Kim1996; Choi et al. Reference Choi, Kang, Kim, Ha, Lee, Youn, Kim, Kim and Kwon2004; Kwon et al. Reference Kwon, Jang, Choi and Kang2009). A previous study showed that the OFC was engaged in cognitive reappraisal of negative emotion, whereas the DLPFC was more generally recruited for cognitive control regardless of emotion types in emotion regulation (Golkar et al. Reference Golkar, Lonsdorf, Olsson, Lindstrom, Berrebi, Fransson, Schalling, Ingvar and Ohman2012).
Despite a number of studies on abnormal neural responses underlying cognitive and emotional processing in OCD patients, it remains unclear how to define the effective connectivity and causal relationship of the aberrant neural interplay within the single components of emotion regulation circuits. In the current study, dynamic causal modeling (DCM) was employed to examine and model the effective connectivity on functional magnetic resonance imaging (fMRI) data of OCD patients and healthy controls during a delayed WM task under negative emotional distraction. In univariate findings, we hypothesized that OCD patients would exhibit dysfunctions within emotion regulation circuits. More precisely, we expected that the task-irrelevant negative emotional distraction during a WM task causes hyperactivations in the prefrontal regions, especially in the DLPFC for cognitive control and the amygdala activation for emotional processing in patients with OCD. Concerning the DCM analysis, we hypothesized that consistent with an altered cognitive control exerted by the DLPFC, exaggerated DLPFC engagement in patients with OCD is related to altered connectivity between the DLPFC and the amygdala or other frontal regions underlying emotion regulation. In particular, we predicted that OCD patients would have the abnormal modulation effect by negative emotional distraction on effective connectivity between the DLPFC and OFC, as these regions are critical for cognitive control to inhibit emotional responses.
Method
Participants
A total of 24 OCD patients were recruited from the OCD clinic at Seoul National University Hospital. The diagnosis and co-morbidity were established by board-certified psychiatrists with Structured Clinical Interview for DSM-IV Axis-Ι and ΙΙ disorders (SCID-I and -II). Also 23 control subjects were recruited and matched for sex, age, intelligence quotient (IQ) and handedness through an Internet advertisement. All healthy controls were pre-screened by the Structured Clinical Interview for DSM-IV Axis I Disorders Non-Patient Edition (First et al. Reference First, Spitzer, Gibbon and Williams1996). Exclusion criteria for all participants included a history of psychosis, bipolar disorder, Tourette's disorder or other tic-related conditions, traumatic brain injury, epilepsy, alcohol or substance abuse, intellectual disability (IQ < 70) and any other neurological diseases. Four OCD patients and two control subjects were excluded from the final analysis due to excessive head movements. Among the remaining 20 OCD patients, 15 OCD patients were not been diagnosed with any co-morbid Axis Ι and ΙΙ disorders, and five had the following disorders: major depressive disorder (n = 3), parasomnia not otherwise specified (n = 1), obsessive–compulsive personality disorder (n = 1) and schizoid personality disorder (n = 1). There were five non-medicated (drug-free for at least 4 weeks) OCD (UMO) patients and four drug-naive OCD (DNO) patients at the time of study. Of the OCD patients, 11 were medicated with a stable dosage at least 4 weeks before scanning. They were taking at least one selective serotonin reuptake inhibitor (fluoxetine, n = 7; escitalopram, n = 2; sertraline, n = 1; paroxetine, n = 1), dopamine and noradrenaline reuptake inhibitor (bupropion, n = 1), and anxiolytics or sedatives (clonazepam, n = 5; zolpidem, n = 1). Among them, three patients received low-dose atypical antipsychotics for adjuvant treatment (risperidone, n = 1; aripiprazole, n = 2).
Experienced psychiatrists performed the Yale–Brown Obsessive Compulsive Scale (Goodman et al. Reference Goodman, Price, Rasmussen, Mazure, Delgado, Heninger and Charney1989a , Reference Goodman, Price, Rasmussen, Mazure, Fleischmann, Hill, Heninger and Charney b ) to assess the severity of the OCD symptoms in each OCD patient. All patients completed a measure of depression and anxiety levels using the Beck Depression Inventory (BDI) (Beck et al. Reference Beck, Ward, Mendelson, Mock and Erbaugh1961) and the Beck Anxiety Inventory (BAI) (Beck et al. Reference Beck, Epstein, Brown and Steer1988), respectively. We classified the OCD patients according to the five clinical dimensions (Mataix-Cols et al. Reference Mataix-Cols, Rauch, Manzo, Jenike and Baer1999) and excluded patients with hoarding symptoms due to their different neural involvement from non-hoarding OCD patients (Saxena et al. Reference Saxena, Brody, Maidment, Smith, Zohrabi, Katz, Baker and Baxter2004; Lochner et al. Reference Lochner, Kinnear, Hemmings, Seller, Niehaus, Knowles, Daniels, Moolman-Smook, Seedat and Stein2005). Predominant obsessions/compulsions were as follows: contamination/cleaning (n = 9), aggressive/checking (n = 3), miscellaneous (n = 6), and symmetry/ordering (n = 2).
All participants had normal or corrected-to-normal vision. This study was approved by the Seoul National University Hospital institutional review board (H-1112-050-389) and the protocols were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants and from the parents of those who were under 18 years old.
Procedure
The study design modified the previous design from Dolcos & McCarthy (Reference Dolcos and McCarthy2006; see online Supplementary Fig. S1 for study design). All subjects performed a modified delayed WM task with distracters presented during the delay interval. Three similar faces (female faces for 50% of trials) were presented as the memoranda. The visual distracters consisted of negative emotional scenes, neutral scenes and digitally scrambled pictures. The negative emotional and neutral pictures were selected from the International Affective Pictures System (Lang et al. Reference Lang, Bradley and Cuthberg1997). A pool of 108 trials was divided into six runs, which consisted of 18 trials each (six negative emotional, six neutral and six scrambled). The trials in each run were presented in a pseudo-randomized order, and no more than two consecutive trials of the same type were presented. Each trial began with the presentation of three similar faces of memoranda for 3 s, which subjects were required to encode and then maintain them into WM during the delay interval between the offset of the memoranda and the onset of the memory probe (7 s). After the offset of the memoranda, two pictures of the same distracter type were consecutively presented for 5 s (2.5 s each), and subjects were instructed to look at these distracters while maintaining focus on the previously presented memoranda. Lastly, a single face as a probe was presented for 1.5 s, and participants were asked to respond whether the single face was one of the three faces in the memoranda or a new face (old faces for 50% of the trials) as quickly and accurately as possible while the probe was on the screen. In order to allow the hemodynamic response to return to baseline, a fixation was presented for 10.5 s.
After fMRI scanning, healthy controls and OCD patients performed two and three consecutive rating tasks, respectively. All subjects were required to record subjective reports on the meaningful pictures for intensity and distractibility (as perceived during the WM task). The OCD patients had one more rating task for symptom provocation. All rating tasks used a four-point Likert scale (1, lowest; 4, highest; 0, none for symptom provocation). These subjective reports were averaged for all the participants and further used as covariates to investigate the link between behavioral responses and brain activity.
Image acquisition
Blood oxygenation level-dependent (BOLD) contrast functional images were acquired with echo-planar T2*-weighted imaging (EPI) using a 3-T scanner (Siemens Magnetom Trio, Germany). Each image volume consisted of a series of 27 functional slices with a 1 mm inter-slice gap (axial plane; repetition time = 2 s; echo time = 30 ms; flip angle = 90°; field of view = 220 mm; voxel size = 3.4 × 3.4 × 4 mm3). Three-dimensional T1-weighted magnetization-prepared rapid-acquisition gradient echo (MPRAGE) images were acquired in 208 slices (repetition time = 1.67 s; echo time = 1.89 ms; flip angle = 9°; field of view = 250 mm; voxel size = 0.9 × 0.9 × 1 mm3).
Behavior data analysis
Demographic and clinical data were compared across groups using independent-sample t tests and χ 2 tests. Behavioral data were analysed with the three different distracter types (negative emotional v. neutral v. scrambled) as the within-subject variable and groups (healthy controls v. OCD) as the between-subject variable using repeated-measures analyses of covariance with BDI scores, age and sex as covariates. If data did not meet parametric assumptions, non-parametric tests were used.
fMRI pre-processing and analysis
Functional imaging analysis was conducted by the following pre-processing steps using SPM8 (http://www.fil.ion.ucl.ac.uk/spm) after discarding the first three volumes of each session: slice timing, motion correction, co-registration, normalization (3 mm3 resampling voxel size) and smoothing (8 mm3 kernel). A total of 206 volumes were acquired in each session. For the first-level analysis, five experimental conditions were included in each run: memoranda, the three types of distracters during the delay interval (negative emotional, neutral, scrambled), and probe, as well as six motion regressors for each session. Contrast images obtained from the first-level analysis were entered into the second-level 3 × 2 full factorial model with the distracter types (negative emotional, neutral, scrambled) and group (healthy controls, OCD) including BDI scores, age and sex as covariates. This factorial design included the ‘main effect of distracter types’ as the within group comparisons, the ‘main effect of groups’ as the between-group comparisons, and the ‘interaction effect of distracter types x groups’. Our contrast of interest was the effect of negative emotional distraction on ongoing WM task (negative emotional > scrambled). We predefined the following regions of interest (ROIs): DLPFC [Brodmann area (BA) 9 and 46], OFC (BA 11 and 12), VLPFC (BA 47), dmPFC (BA 9 and 10), and vmPFC (BA 11) and amygdala using a predefined anatomical mask or the automatic anatomic labels implemented in the WFU PickAtlas (Wake Forest University School of Medicine; http://www.fmri.wfubmc.edu/cms/software). We determined a significant threshold level of p < 0.05, whole-brain family-wise error (FWE) corrected for multiple comparisons, as well as Bonferroni-corrected for the number of ROIs (small volume correction, SVC; p FWE-SVC < 0.008 as significant; 0.008 < p FWE-SVC > 0.05 as trend significant; Worsley et al. Reference Worsley, Marrett, Neelin, Vandal, Friston and Evans1996). To specifically assess the pattern of activity, we used the MarsBar tool box (http://marsbar.sourceforge.net) to extract percentage signal change data from peak coordinates of ROIs with a 6 mm radius sphere. Additional correlation analyses between the extracted percentage signal change data and behavioral data were performed using PASW Statistics 18 (SPSS, USA). For non-parametric variables, Spearman's ρ was used. All analyses included only correct trials.
DCM
To investigate if there were group differences on effective connectivity in the neural circuits associated with emotion regulation, DCM 10 as implemented in SPM8 was used. In DCM, a Bayesian model comparison procedure was used to estimate hidden neuronal effective (causal) connectivity and its modulation effect by experimental manipulations. DCM allows modeling of the task-independent intrinsic connectivity (DCM.A), of the task-dependent modulatory effect (DCM.B) by experimental manipulation on the endogenous coupling, and of the direct influence on individual or groups of regions (DCM.C).
Time series extraction
The ROIs with between-group differences (right DLPFC, x/y/z = 39/44/37; OFC, x/y/z = 27/50/−14) were chosen as seeds (see Results). The right amygdala (x/y/z = 27/−1/−26) as negative emotional effects and the right visual cortex (V1, x/y/z = 45/−79/−8) as a direct driving input were derived from the whole-brain main effect of the distracter types across all subjects. All regions were right-lateralized, where the strongest group activations were found. The observed lateralization is also in accordance with previous findings, which detected a greater impact on the right prefrontal regions by negative emotional distracters during the delay period (Dolcos & McCarthy, Reference Dolcos and McCarthy2006; Dolcos et al. Reference Dolcos, Diaz-Granados, Wang and McCarthy2008). Additionally, previous studies suggested a functional specialization of the right amygdala for the processing and encoding of non-verbal affective stimuli (Anderson et al. Reference Anderson, Christoff, Panitz, De Rosa and Gabrieli2003; Ochsner et al. Reference Ochsner, Ray, Cooper, Robertson, Chopra, Gabrieli and Gross2004). Each subject's activation maxima within a sphere of 6 mm at a single-subject significance threshold (p < 0.1) was used to center and then extracted the first eigenvariate. One subject from each group was excluded for no significant activations in the three regions.
Model space
In a previous study (Sladky et al. Reference Sladky, Hoflich, Kublbock, Kraus, Baldinger, Moser, Lanzenberger and Windischberger2013), Bayesian model averaging (BMA) was used, which provides averages of parameter estimates within the entire model space weighted by the posterior probability for each model (Hoeting et al. Reference Hoeting, Madigan, Raftery and Volinsky1999; Penny et al. Reference Penny, Stephan, Daunizeau, Rosa, Friston, Schofield and Leff2010). The inference on model structure can be one method to compare the winning models of the OCD patients and healthy controls; however, the method can be insufficient due to a possibility that the OCD-related deficits are mediated by abnormal modulation effects rather than disruption on model structure (Sladky et al. Reference Sladky, Hoflich, Kublbock, Kraus, Baldinger, Moser, Lanzenberger and Windischberger2013). Therefore, BMA is an alternative approach for comparing parameter estimates across groups, and therefore can explain the uncertainty regarding model structure (Stephan et al. Reference Stephan, Penny, Moran, den Ouden, Daunizeau and Friston2010).
We created DCM models as an initial model space to select connections among the right DLPFC, OFC and amygdala with significant posterior evidence and to remove improper connections for the sake of computational complexity. Based on the previous neurobiological evidence (Kringelbach & Rolls, Reference Kringelbach and Rolls2004; Diwadkar et al. Reference Diwadkar, Wadehra, Pruitt, Keshavan, Rajan, Zajac-Benitez and Eickhoff2012), we set bidirectional connections between the amygdala and the OFC and from the amygdala to the DLPFC. Then, DCM models were generated with possible anatomical connectivity configurations between the DLPFC and the OFC, and from the DLPFC to the amygdala, yielding an initial model space of eight models. A direct visual input entered the V1 as a driving input, which has a unidirectional connection both to the DLPFC and the amygdala in all DCM models (Ongur & Price, Reference Ongur and Price2000).
Based on the chosen connectivity from the initial model space in each group, connections that were not significant in both groups were removed in further DCM analysis. All other significant connections were used as a revised model space. In this revised model space, modulation effects by negative emotional distraction were additionally introduced. According to possible modulatory effects by negative emotional distraction on each connection, a total of 25 model configurations were created (online Supplementary Fig. S2). BMA averaged the connectivity parameter estimates and their modulations within each subject's 25 models, and these results were analysed using one-sample and two-sample t tests at a threshold of p < 0.05, both Bonferroni-corrected for multiple comparisons and uncorrected. For non-parametric data, non-parametric tests were used.
Results
Demographic and behavioral results
Demographic and clinical data for each group are shown in Table 1. Friedmann's analysis of variance showed that response times (RTs) for each separate group showed no significant differences (OCD patients: χ 2 = 2.80, p = 0.247; healthy controls: χ 2 = 0.286, p = 0.867). Between-group comparisons did not reveal any significant differences on RTs (Mann–Whitney U test; negative: p = 0.118; neutral: p = 0.112; scrambled: p = 0.192). For the correct rates, there were no significant main effects of distracter type, group and interaction effects (F = 0.228, p = 0.797; F = 0.019, p = 0.891; F = 0.578, p = 0.564, respectively). The average scores for emotional intensity and distractibility did not differ between groups (intensity for negative: controls, 2.56; patients, 2.30, p = 0.137; intensity for neutral: controls, 1.43; patients, 1.43, p = 0.997; distractibility for negative: controls, 2.20; patients, 1.88, p = 0.113; distractibility for neutral: controls, 1.32; patients, 1.45, p = 0.475). In OCD patients, the average symptom provocation scores were 1.05 (s.d. = 1.05) for negative and 0.61 (s.d. = 0.90) for neutral.
Table 1. Demographic characteristics of the subjects
Data are given as mean (standard deviation) unless otherwise indicated.
OCD, Obsessive–compulsive disorder; IQ, intelligence quotient; BDI, Beck Depression Inventory; BAI, Beck Anxiety Inventory; Y-BOCS, Yale–Brown Obsessive Compulsive Scale; HAM-D, Hamilton Rating Scale for Depression; HAM-A, Hamilton Rating Scale for Anxiety.
fMRI results
The main effect of the distracter types was found in the bilateral VLPFC (BA 47), amygdala, DLPFC (BA 9/46), lateral parietal cortex (BA 40), inferior temporal cortex (BA 20), OFC (BA 11/12), anterior prefrontal cortex (BA10), occipital cortex (BA 19), dmPFC (BA 9/10) and vmPFC (BA 11) across groups, which included our ROIs (p < 0.05, either whole-brain FWE-corrected or FWE-SVC; see Fig. 1a , Table 2 and online Supplementary Table S1). In the within-group analyses, the healthy controls showed dmPFC (BA 10), vmPFC (BA 11), VLPFC (BA 47), amygdala, middle temporal cortex (BA 21) and occipital cortex (BA 19) activations while performing a WM task under negative emotion distraction (i.e. negative emotion > scrambled condition; p < 0.05, either whole brain FWE-corrected or FWE-SVC; see online Supplementary Table S2). The patients with OCD exhibited DLPFC (BA 9/46), OFC (BA 11/12), dmPFC (BA 10), vmPFC (BA 11), VLPFC (BA 47), amygdala, superior parietal cortex (BA 5/7) and occipital cortex (BA 19) activations (i.e. negative emotion > scrambled condition; p < 0.05, either whole brain FWE-corrected or FWE-SVC; see online Supplementary Table S2). For the main effect of group over all conditions, there were significant differences between groups in the OFC (BA 11/12), DLPFC (BA 9/46), superior temporal cortex (BA 22) and occipital cortex (BA 19) (p < 0.05, either whole brain FWE-corrected or FWE-SVC; see Table 2), which included our main ROIs, DLPFC and OFC (DLPFC at p FWE-SVC = 0.029; OFC at p FWE-SVC = 0.004). There was no significant group x task interaction effect. The analysis to test the pattern of activity revealed that OCD patients exhibited DLPFC (BA 9/46) and OFC (BA 11/12) activations under negative emotional distraction, whereas healthy controls showed deactivations in these regions (Fig. 1c, d ). Plus, a similar pattern of activity was observed in the bilateral amygdala in both groups (Fig. 1b ).
Fig. 1. Main effect of task (distracter types) and group. (a) Whole-brain analysis of the main effect of distracter types [p < 0.05, whole-brain family-wise error (FWE)-corrected]. (b) Pattern of activity in the right amygdala (p FWE-SVC < 0.05). The right amygdala activation was significantly greater in negative emotional distraction in both healthy controls and obsessive–compulsive disorder (OCD) patients, and there was no significant group difference. The left amygdala had a similar pattern of activity to the right amygdala. (c) OCD patients exhibited increased dorsolateral prefrontal cortex activation (Brodmann area 9/46) (at trend level, 0.008 < p FWE-SVC > 0.05) in negative emotional distraction, whereas healthy controls showed deactivations in this region. (d) The right orbitofrontal cortex (Brodmann area 11/12) showed a significant group difference under negative emotional distraction (p FWE-SVC < 0.05). SVC, Small volume correction.
Table 2. Main effect of task (distracter types) and group
H, Hemisphere; BA, Brodmann area; MNI, Montreal Neurological Institute; k e , cluster size; p-FWE, p value with family-wise error correction for the search volume; L, left; R, right; Inf., infinite; SVC, small volume correction.
* Significant at p < 0.05, whole-brain family-wise error corrected.
** Significant at trend-level (0.008 < p FWE-SVC > 0.05).
Correlations with BOLD activation data
The present study identified that we found a negative correlation between left VLPFC activation and RTs under the negative emotional distraction only in healthy controls (Spearman's ρ = −0.579, p = 0.006 in healthy controls; ρ = 0.084, p = 0.724 in OCD patients). Additionally, a positive correlation between the bilateral amygdala and distractibility rating scores on negative emotional pictures was found only in OCD patients (OCD patients, left: ρ = 0.592, p = 0.006; right: ρ = 0.721, p < 0.0001; healthy controls, left: ρ = −0.052, p = 0.823; right: ρ = 0.056, p = 0.811). The right amygdala in OCD patients also had positive correlations with the symptom provocation scores on negative emotional distracters (ρ = 0.471, p = 0.036).
DCM results
Model structure
In the initial model space, healthy controls displayed no significant unidirectional connection from the DLPFC to the amygdala; however, OCD patients showed all connections significantly. Therefore, an interconnected model with all three regions, which was found to be significant, was selected for further DCM analyses (online Supplementary Fig. S3).
Intrinsic connectivity and modulatory effect by negative emotional distraction
Again, BMA was applied within each subject's 25 models and one-sample and two-sample t tests were used for comparisons. The results of the intrinsic connectivity revealed no significant group differences (Table 3). As a result of the modulatory effect by negative emotional distraction, OCD patients showed a reduced modulation on the connection from the DLPFC to the OFC (p < 0.05, uncorrected). However, no significant modulatory effects were found on any connections in healthy controls. Comparison between groups showed that relative to healthy controls, OCD patients showed reduced modulation effects by negative emotional distraction on the DLPFC to the OFC connection, which was endogenously coupled in the positive direction (p < 0.05, Bonferroni-corrected; Fig. 2 and Table 3).
Fig. 2. Bayesian model averaging results. The significantly reduced modulatory effects by negative emotional distraction on the connections from the dorsolateral prefrontal cortex (DLPFC) to the orbitofrontal cortex (OFC) were found only in obsessive–compulsive disorder (OCD) patients compared with healthy controls (p < 0.05, Bonferroni corrected). V1, Primary visual cortex. * p < 0.05, uncorrected. ** p < 0.05, Bonferroni corrected.
Table 3. Bayesian model averaging results for connectivity parameters
Data are given as mean (standard deviation).
OCD, Obsessive–compulsive disorder; DLPFC, dorsolateral prefrontal cortex; n.s., not significant; OFC, orbitofrontal cortex; V1, primary visual cortex.
a Bonferroni corrected.
* p < 0.05, uncorrected.
Discussion
This study demonstrates the disrupted neural interactions within brain circuits associated with emotion regulation in OCD patients. During the delay interval under negative emotional distraction, the OCD patients showed significant differences between groups in the DLPFC and OFC activations. However, both groups activated the amygdala the greatest. Additionally, modulation effects by negative emotional distraction on the connection from the DLPFC to the OFC, particularly in the anterior part, exhibited reduced connectivity levels in OCD patients, compared with healthy controls.
The present fMRI study reveals that both healthy controls and OCD patients show similar patterns in the amygdala in response to general negative emotional stimuli, indicating strong disturbances by negative emotional pictures during the WM maintenance phase (Dolcos & McCarthy, Reference Dolcos and McCarthy2006; Denkova et al. Reference Denkova, Wong, Dolcos, Sung, Wang, Coupland and Dolcos2010; Dolcos et al. Reference Dolcos, Iordan and Dolcos2011). The increased amygdala under negative emotional distraction is consistent with a previous finding on higher level of fear of negative emotions in people with heightened obsessive–compulsive symptoms (Stern et al. Reference Stern, Nota, Heimberg, Holaway and Coles2014). Interestingly, the bilateral amygdala was positively correlated with distractibility rating scores on negative emotional distraction only in OCD patients. Also, OCD patients had positive correlations between the right amygdala and symptom provocation rating scores on negative emotional distraction. The human amygdala has been considered a core brain structure responsible for emotional processing, especially for aversive stimuli, and triggers interference on cognitive tasks at hand (Etkin et al. Reference Etkin, Egner, Peraza, Kandel and Hirsch2006; Dolcos et al. Reference Dolcos, Iordan and Dolcos2011; Han et al. Reference Han, Lee, Kim and Kim2013). Thus, the amygdala may be susceptible to subjective distractibility and symptom provocation in OCD patients.
OCD patients also exhibited between-group differences in the DLPFC and OFC during WM maintenance under negative emotional distraction, compared with healthy controls. The OFC has been considered to play a role in inhibitory cognitive processing and is associated with the DLPFC (Savage et al. Reference Savage, Baer, Keuthen, Brown, Rauch and Jenike1999; Kwon et al. Reference Kwon, Jang, Choi and Kang2009). In previous studies, increased frontal and parietal activations including the DLPFC and presupplementary motor area were found in OCD patients during a cognitive task (Ciesielski et al. Reference Ciesielski, Hamalainen, Lesnik, Geller and Ahlfors2005; Henseler et al. Reference Henseler, Gruber, Kraft, Krick, Reith and Falkai2008; de Wit et al. Reference de Wit, de Vries, van der Werf, Cath, Heslenfeld, Veltman, van Balkom, Veltman and van den Heuvel2012; de Vries et al. Reference de Vries, de Wit, Cath, van der Werf, van der Borden, van Rossum, van Balkom, van der Wee, Veltman and van den Heuvel2014). Some of these studies found no group differences on behavioral task performances like our behavioral result (Ciesielski et al. Reference Ciesielski, Hamalainen, Lesnik, Geller and Ahlfors2005; Henseler et al. Reference Henseler, Gruber, Kraft, Krick, Reith and Falkai2008; de Wit et al. Reference de Wit, de Vries, van der Werf, Cath, Heslenfeld, Veltman, van Balkom, Veltman and van den Heuvel2012) and, more interestingly, the other study found task-related hyperactivation in OCD patients with normal WM performance, in contrast to those with behavioral WM impairment (de Vries et al. Reference de Vries, de Wit, Cath, van der Werf, van der Borden, van Rossum, van Balkom, van der Wee, Veltman and van den Heuvel2014). All of these studies explain that hyperactivations in the cognitive task-related regions may relate to a compensatory neural recruitment (Ciesielski et al. Reference Ciesielski, Hamalainen, Lesnik, Geller and Ahlfors2005; Henseler et al. Reference Henseler, Gruber, Kraft, Krick, Reith and Falkai2008; de Wit et al. Reference de Wit, de Vries, van der Werf, Cath, Heslenfeld, Veltman, van Balkom, Veltman and van den Heuvel2012; de Vries et al. Reference de Vries, de Wit, Cath, van der Werf, van der Borden, van Rossum, van Balkom, van der Wee, Veltman and van den Heuvel2014). Therefore, it can be suggested that enhanced recruitments in the WM-related regions in OCD patients compensate for a detrimental effect of negative emotional distraction on WM performance to reach a similar level of performance in healthy controls. Thus, the behavioral WM deficits may not be detected in OCD patients; however, the OCD patients may possess latent deficits (Pujol et al. Reference Pujol, Torres, Deus, Cardoner, Pifarre, Capdevila and Vallejo1999; Henseler et al. Reference Henseler, Gruber, Kraft, Krick, Reith and Falkai2008).
The novel aspect of the present findings was that the connection from the DLPFC to the OFC is negatively modulated by negative emotional distraction in OCD patients, compared with that in healthy controls. From this result, it can be inferred that the negative emotional distraction induced a dampening influence on the cortico-cortical interaction, which was not found in healthy controls. A convergence of OCD research has pointed to the dysfunction of cortico-striato-thalamo-cortical circuitry (Saxena et al. Reference Saxena, Brody, Schwartz and Baxter1998; Saxena & Rauch, Reference Saxena and Rauch2000). However, a review of OCD pathophysiology points out that this circuitry is insufficient to explain the pathophysiology of OCD (Milad & Rauch, Reference Milad and Rauch2012). In fact, for instance, recent studies identified altered functional connectivity on the frontal–limbic circuitry in OCD patients (de Vries et al. Reference de Vries, de Wit, Cath, van der Werf, van der Borden, van Rossum, van Balkom, van der Wee, Veltman and van den Heuvel2014; de Wit et al. Reference de Wit, van der Werf, Mataix-Cols, Trujillo, van Oppen, Veltman and van den Heuvel2015; van Velzen et al. Reference van Velzen, de Wit, Ćurĉić-Blake, Cath, de Vries, Veltman, van der Werf and van den Heuvel2015). Therefore, the present study validates this notion, and further reveals a dysfunction of cortico-cortical interaction in OCD patients. The OFC has been considered one of the key regions in the pathophysiology of OCD (Menzies et al. Reference Menzies, Chamberlain, Laird, Thelen, Sahakian and Bullmore2008; Kwon et al. Reference Kwon, Jang, Choi and Kang2009). The OFC promotes cognitive–affective interaction through its essential role to integrate and modulate neural activation (Rule et al. Reference Rule, Shimamura and Knight2002). It is also important to interact the DLPFC–OFC connection in cognitive functions; thus the OFC's disruption contributes to some mental disorders (Moghaddam & Homayoun, Reference Moghaddam and Homayoun2008). Therefore, our finding could be interpreted that general negative emotional distraction triggers exaggerated recruitment of the DLPFC in OCD patients when compared with healthy controls. This DLPFC hyperactivation may lead to reduce top-down input to the OFC and may further interrupt the integrations of cognitive control for inhibiting the detrimental effects of negative emotion.
This study had several limitations. First, medicated OCD patients were included in all analyses, so pharmacological factors cannot be excluded in this study. However, our findings were not significantly different from cognitive performance, BOLD-fMRI, or DCM analyses between medicated OCD and DNO/UMO patients. Second, OCD patients with other Axis I and II disorders were included; however, findings from both univariate and DCM analyses remain after excluding those five OCD patients.
In summary, the present study reveals altered effective connectivity from inputs in the DLPFC to the OFC under negative emotional distraction in OCD patients, as well as abnormal activations in these brain regions. The exaggerated top-down signals in the DLPFC further reduce cortico-cortical interactions with the OFC, which may be responsible for dysfunctions of cognitive and emotional processing in OCD patients. The disrupted DLPFC–OFC connectivity under negative emotional distraction thus can be a neurobiological model in OCD.
Supplementary material
For supplementary material accompanying this paper visit http://dx.doi.org/10.1017/S0033291715002391
Acknowledgements
This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2013R1A2A1A03071089).
Declaration of Interest
None.
