Introduction
Acceptance-commitment therapy (ACT) is a psychological intervention that applies six interrelated therapeutic processes to the creation of psychological flexibility, namely: acceptance, cognitive defusion, contact with the present moment, self-as-context, values, and committed action (Luoma, Hayes, & Walser, Reference Luoma, Hayes and Walser2017). Being part of a larger group of acceptance- and mindfulness-based therapies (Hayes, Strosahl, & Wilson, Reference Hayes, Strosahl and Wilson2011a) or the ‘third wave’ of cognitive behavioral therapy (CBT) (e.g. Hayes, Reference Hayes2004; Hayes, Villatte, Levin, & Hildebrandt, Reference Hayes, Villatte, Levin and Hildebrandt2011b), ACT possesses many distinctive theoretical and practical features (Flaxman, Blackledge, & Bond, Reference Flaxman, Blackledge and Bond2011).
First, ACT is based on relational frame theory, which holds that human behavior can be dominated by linguistic relational frames without direct experience. This connection to basic behavioral principles has created a unique and empirically based model of human functioning. Thus, ACT is fundamentally transdiagnostic (Flaxman et al., Reference Flaxman, Blackledge and Bond2011). Second, unlike conventional CBT, ACT assumes that thoughts do not need to change for overt behavior to change and emphasizes on valued life even with some psychological distress than on symptom reduction. Technically, ACT uses very few cognitive challenging and restructuring; it tends to undermine problematic reasoning in many contexts. Instead, ACT focuses on the activation of value-congruent observable behavior (Ciarrochi & Bailey, Reference Ciarrochi and Bailey2008; Flaxman et al., Reference Flaxman, Blackledge and Bond2011). Third, ACT also differs from mindfulness. From an ACT perspective, mindfulness is an emergent quality of four processes: acceptance, defusion, contact with the present moment, and self-as-context (Hayes et al., Reference Hayes, Strosahl and Wilson2011a). Mindfulness is also seen in the service of potentiating valued, goal-directed action, which is the explicit function of defusion in ACT, rather than as a method of stress reduction or to facilitate decentering (Gillanders et al., Reference Gillanders, Bolderston, Bond, Dempster, Flaxman, Campbell and Remington2014).
Accordingly, ACT offers a new paradigm for understanding obsessive-compulsive disorder (OCD) and has been employed to establish its efficacy as a standard treatment (Abramowitz, Blakey, Reuman, & Buchholz, Reference Abramowitz, Blakey, Reuman and Buchholz2018; Twohig et al., Reference Twohig, Hayes, Plumb, Pruitt, Collins, Hazlett-Stevens and Woidneck2010). Obsessional thoughts are regarded as the product of excessive cognitive fusion, in which literal content is entangled with internal experiences such as thoughts and feelings. These thoughts can engender unnecessary contingencies of experiential avoidance in the form of compulsive behaviors or avoidance, which aims to avoid unwanted internal experiences (Abramowitz et al., Reference Abramowitz, Blakey, Reuman and Buchholz2018). More specifically, cognitive fusion is an overlapping yet broader concept than thought-action fusion (TAF), which is one of the major dysfunctional beliefs in OCD that thoughts and outcomes or actions are linked (Reuman, Buchholz, Blakey, & Abramowitz, Reference Reuman, Buchholz, Blakey and Abramowitz2017). In fact, cognitive fusion is correlated with TAF (Lee & Lee, Reference Lee and Lee2021) as well as obsessive-compulsive (OC) symptoms (Reuman et al., Reference Reuman, Buchholz, Blakey and Abramowitz2017; Xiong et al., Reference Xiong, Lai, Wu, Yuan, Tang, Chen and Hu2021). It added significant explanatory power, especially in the symptom dimension of unacceptable thoughts (Reuman et al., Reference Reuman, Buchholz, Blakey and Abramowitz2017; Reuman, Buchholz, & Abramowiz, Reference Reuman, Buchholz and Abramowiz2018). At this point, TAF could serve an important bridging role in the study of ACT. Based on this premise, a TAF induction task would be a good candidate for functional magnetic resonance imaging (fMRI) task to measure both TAF that underlies OC symptoms and cognitive fusion/experiential avoidance in ACT.
Until date, only a few studies have explored the neural mechanisms of ACT. All participants in these studies were patients with chronic pain. In an fMRI study, those who had completed ACT showed decreased activation during pain in the middle frontal gyrus, inferior parietal lobule, insula, cingulate gyrus, and superior temporal gyrus (STG) (Aytur et al., Reference Aytur, Ray, Meier, Campbell, Gendron, Waller and Robin2021). Two other studies discovered that following the ACT intervention, key networks in the brain that are involved in self-reflection, emotion, and cognitive control were less active both within and between them (Meier et al., Reference Meier, Ray, Waller, Gendron, Aytur and Robin2020; Smallwood, Potter, & Robin, Reference Smallwood, Potter and Robin2016). However, previous studies were preliminary in nature: two studies had a single group of nine participants (Aytur et al., Reference Aytur, Ray, Meier, Campbell, Gendron, Waller and Robin2021; Meier et al., Reference Meier, Ray, Waller, Gendron, Aytur and Robin2020) and one study had four and three participants in the ACT and control groups, respectively (Smallwood et al., Reference Smallwood, Potter and Robin2016). Similarly, there are a few well-controlled and fully implemented ACT studies exploring its neural mechanism. To the best of our knowledge, no research has examined the neural correlates of ACT for OCD.
Thus, this study aimed to investigate the neural correlates of ACT in patients with OCD using task-based and resting-state (rs) fMRI. First, hemodynamic changes were measured using a modified TAF induction task following ACT intervention. Our previous study using the same task characterized lower functional connectivity (FC) of the mid-cingulate cortex–insula in patients with OCD compared to the healthy controls (Lee et al., Reference Lee, Song, Jang, Cha, Kim, Chang and Lee2022). Changes in the insula and its associated network have been noted in recent meta-analyses of both mindfulness-based interventions and CBT (Pico-Perez et al., Reference Pico-Perez, Fullana, Albajes-Eizagirre, Vega, Marco-Pallares, Vilar and Soriano-Mas2022; Young et al., Reference Young, van der Velden, Craske, Pallesen, Fjorback, Roepstorff and Parsons2018). Therefore, we hypothesized that following ACT intervention, the aforementioned psychological treatments would all have a common effect of altering the insula and its associated network. Second, although predicting the effect of cognitive defusion is difficult due to fewer observations following ACT intervention, several experimental fMRI studies (Creswell, Way, Eisenberger, & Lieberman, Reference Creswell, Way, Eisenberger and Lieberman2007; Koenigsberg et al., Reference Koenigsberg, Fan, Ochsner, Liu, Guise, Pizzarello and Siever2010; Wang et al., Reference Wang, Vantieghem, Dong, Nemegeer, De Mey, Van Schuerbeek and Vandekerckhove2022) suggest greater activation in prefrontal cortices, including the ventrolateral prefrontal cortex (vlPFC), and lesser activation in subcortical areas as potential neural correlates of the ability to disengage from autopilot in a process similar to defusion, which thus reduces the emotional impact of verbal activity (Fletcher, Schoendorff, & Hayes, Reference Fletcher, Schoendorff and Hayes2010). Additionally, we hypothesized that changes in these areas would involve defusing affect from language processing. Third, we predicted that alterations in the cingulate gyrus would be found in rsFC analysis. Previous studies have frequently cited the effects of mindfulness and meditation on the subregions of the cingulate (Zsadanyi, Kurth, & Luders, Reference Zsadanyi, Kurth and Luders2021) in terms of local activity (Fox et al., Reference Fox, Nijeboer, Dixon, Floman, Ellamil, Rumak and Christoff2014, Reference Fox, Dixon, Nijeboer, Girn, Floman, Lifshitz and Christoff2016) and connectivity (Sezer, Pizzagalli, & Sacchet, Reference Sezer, Pizzagalli and Sacchet2022). A few previous studies have reported decreased posterior cingulate cortex (PCC) activity following ACT intervention (Smallwood et al., Reference Smallwood, Potter and Robin2016) and decreased FC between the anterior cingulate cortex (ACC) and other networks (Aytur et al., Reference Aytur, Ray, Meier, Campbell, Gendron, Waller and Robin2021) in patients with chronic pain.
Methods
Participants
The current study was part of our randomized controlled trial (RCT) to evaluate the effectiveness of the ACT in treating OCD. A total of 118 participants (age 18–40 years) were screened after referral through subway advertisements, the online boards of OOO National University, and psychiatric clinics at OOO National University Hospital. Eighty-seven participants were selected for in-person interviews. The participants were interviewed and diagnosed by two psychiatrists using the Structured Clinical Interview for DSM-5 Disorders-Clinical Version; only those with OCD as the primary diagnosis were included. Patients with acute medical or neurological, mental retardation, a history of brain trauma, or other major psychiatric illnesses were excluded. Thus, 72 patients with OCD, who met the study criteria, were randomly assigned equally to the ACT and wait-list control (WLC) groups, with a block randomization (block size = 4). An 8-week group-format ACT program was provided to the ACT group. Detailed information about our entire RCT including the program was described in our paper (Lee, Choi, & Lee, Reference Lee, Choi and Lee2023).
Among 72 participants, 48 participated in the fMRI experiment and completed the MR scan and psychological measurements twice at intervals of 8 weeks. After excluding six patients (n = 4 for head motion, n = 2 for new diagnosis), all 42 patients (21 per group) were included in the final analysis (online Supplementary Fig. S1). All participants agreed to participate in the experiments with informed consent. This study was approved by the Institutional Review Board of OOO National University Hospital (2021-04-032).
Psychological measures
Symptom measures
OCD symptoms were evaluated using the self-report version of the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS; Baer, Reference Baer1994; Seol, Kwon, & Shin, Reference Seol, Kwon and Shin2013) and the Obsessive-Compulsive Inventory-Revised (OCI-R; Foa et al. Reference Foa, Huppert, Leiberg, Langner, Kichic, Hajcak and Salkovskis2002; Woo, Kwon, Lim, & Shin, Reference Woo, Kwon, Lim and Shin2010). Depressive symptoms were measured using the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, Reference Beck, Ward, Mendelson, Mock and Erbaugh1961; Lee & Song, Reference Lee and Song1991).
Process measures
The Acceptance and Action Questionnaire-II (AAQ-II; Bond et al. Reference Bond, Hayes, Baer, Carpenter, Guenole, Orcutt and Zettle2011; Heo, Choi, & Jin, Reference Heo, Choi and Jin2009) was used to measure the psychological inflexibility and experiential avoidance emphasized in ACT. The Acceptance and Action Questionnaire for Obsessions and Compulsions (AAQ-OC; Jacoby, Abramowitz, Buchholz, & Reuman, Reference Jacoby, Abramowitz, Buchholz and Reuman2018) was used to evaluate the experiential avoidance related to OC symptoms. We translated the AAQ-OC for this study. The Cognitive Fusion Questionnaire (CFQ; Gillanders et al. Reference Gillanders, Bolderston, Bond, Dempster, Flaxman, Campbell and Remington2014; Kim & Cho, Reference Kim and Cho2015) was used to assess cognitive fusion. The Thought–Action Fusion Scale (TAFS; Lee & Lee, Reference Lee and Lee2021; Shafran, Thordarson, & Rachman, Reference Shafran, Thordarson and Rachman1996) was used to measure the degree of TAF belief.
TAF induction fMRI task
Before brain imaging, the authors received the name of two close and neutral persons each (CPs and NPs, respectively) from the participants, and these names were applied in the fMRI TAF paradigm (Fig. 1a). The participants were asked to look at the name (CP or NP) on the screen and think about the person for 4 s (name phase). They were instructed to silently read the displayed negative statement (NS), such as ‘I hope that the person (CP or NP) will soon be in a car accident,’ for 10 s (sentence phase). After the sentence phase, they were asked to rate their feelings about the sentence on a Likert scale from 1 (very little) to 4 (very much) using the button box, within 4 s (evaluation phase). After the evaluation, a fixation cross was displayed on the center of the screen for 10 s (resting phase). Thus, the modified TAF experiment included two conditions, NS/CP and NS/NP, with eight statements for each. The NS/CP and NS/NP conditions were represented in a pseudorandomized order, and the paradigm lasted for 7 min 28 s ( = 28 s for each trial × eight NSs × CP and NP conditions). The TAF paradigm was developed and modified for fMRI experiments (Lee et al., Reference Lee, Cha, Chung, Kim, Song, Chang and Lee2019) based on an original experiment (Rachman, Shafran, Mitchell, Trant, & Teachman, Reference Rachman, Shafran, Mitchell, Trant and Teachman1996).
Figure 1. Changes in task-based brain activation after acceptance-commitment therapy (ACT) intervention a. The fMRI paradigm for thought-action fusion (TAF) induction. b. Results of behavioral data, showing a significant decrease in emotional intensity in the ACT group. c. Changes in task-based brain activation after ACT intervention [false discovery rate (FDR) corrected, p < 0.05; minimum cluster size, k = 10]. d. The correlation of brain activity with the Acceptance and Action Questionnaire for Obsessions and Compulsions (AAQ-OC), Cognitive Fusion Questionnaire (CFQ), and Obsessive-Compulsive Inventory-Revised (OCI-R) across the entire participants.
Image acquisition
All magnetic resonance images were obtained on a 3.0 Tesla GE SIGNA architect scanner (GE Healthcare, Milwaukee, WI, USA) with a 48-channel head coil. Functional brain images were obtained using a T2*-weighted gradient echo planar imaging pulse sequence with a field of view (FOV) = 23 cm, matrix size = 64 × 64, slice thickness = 4 mm, repetition time (TR) = 2000 ms, echo time (TE) = 30 ms, flip angle = 90°. Structural images were obtained using a 3D brain-volume image pulse sequence with an FOV = 25.6 cm, matrix size = 256 × 256, slice thickness = 1 mm, TR = 8.5 ms, TE = 3.2 ms, flip angle = 12°.
Imaging data analysis
Preprocessing
All functional images were preprocessed using Statistical Parametric Mapping (SPM12; http://www.fil.ion.ucl.ac.uk/spm) and Computational Anatomy Toolbox (CAT12; https://www.nitrc.org/projects/cat). The preprocessing steps included slice timing, realignment, co-registration, segmentation, normalization to the Montreal Neurological Institute coordinate space, and smoothing (full width at half maximum = 6 mm). The head motion and outlier factors were detected according to the intermediate setting using Artifact Detection Tools (https://www.nitrc.org/projects/artifact_detect), and removed in the first-level analysis.
Task-evoked fMRI
Individual task-evoked functional images were analyzed based on the general linear model with the TAF task design matrix and temporal high-pass filtering (0.008 Hz) was applied. Changes in task-evoked brain activity were analyzed in the NS/CP and NS/NP conditions. Furthermore, a generalized psycho-physiological interaction (PPI) was conducted to investigate task-modulated connectivity on significant brain activation regions using the CONN toolbox (https://web.conn-toolbox.org).
Resting-state fMRI
Independent component analysis (ICA) was carried out on rs-fMRI data to examine the alteration of spontaneous activity in the functional brain network using the CONN toolbox. All rs-fMRI data were subjected to temporal band-pass filtering (0.008–0.09 Hz). In the first-level analysis, G1 fast ICA (Hyvarinen, Reference Hyvarinen1999) and GICA3 back-reconstruction (Calhoun, Adali, Pearlson, & Pekar, Reference Calhoun, Adali, Pearlson and Pekar2001) were applied to estimate independent spatial components and separation of spatial components at the individual level. The subject-level reduction dimension applied was 64, the target component number was 57, and the average dimension for individual component factors was calculated using the minimum description length criteria (Li, Adali, & Calhoun, Reference Li, Adali and Calhoun2006). Our previous research suggested that components of the default mode network (DMN) and salience network (SN) were important for processing the TAF stimuli (Lee et al., Reference Lee, Cha, Chung, Kim, Song, Chang and Lee2019, Reference Lee, Song, Jang, Cha, Kim, Chang and Lee2022). Therefore, alterations in interconnectivity within the DMN and SN as well as among the large-scale networks (Yeo et al., Reference Yeo, Krienen, Sepulcre, Sabuncu, Lashkari, Hollinshead and Buckner2011) were evaluated in the resting-state fMRI analyses.
Group analysis
The group-by-time interaction of brain activity (ACT v. WLC × pre- v. post-treatment) was determined by a 2-by-2 full factorial analysis of variance (ANOVA). Additionally, a paired t test comparing pre- and post-treatment was carried out within the ACT group to evaluate the alterations in brain activity brought about by the treatment. The task-evoked activation maps were thresholded at p < 0.05, with false discovery rate (FDR) correction across the whole brain. The gPPI and ICA maps were thresholded at cluster-level FDR corrected, p < 0.05, and voxel threshold uncorrected, p < 0.001 (Worsley et al., Reference Worsley, Marrett, Neelin, Vandal, Friston and Evans1996). All regions of interest (ROIs) were created as spheres with a 4 mm radius located at the maximum T value of multiplied significant regions in both post-treatment minus pre-treatment and group-by-time interaction maps. The REX toolbox was used to extract all beta and connectivity values (https://www.nitrc.org/projects/rex).
Statistical analysis
The Jeffreys's Amazing Statistics Program (JASP; https://jasp-stats.org) was used for normality test, t test between the demographic and clinical characteristics of the groups, task performance, and the correlation analysis between the changes in brain and psychological measures. Depending on assumptions of normality, the correlation coefficient was calculated using Pearson or Spearman correlation analysis, controlling out the depressive score. Because the TAFS and AAQ-OC, the two process measures, had an outlier (>3 standard deviations, SDs), only the results that maintained significance even after removing outliers were reported as meaningful.
Results
Baseline demographic and clinical characteristics of the participants
The mean (±s.d.) age of the ACT and WLC groups was 24.9 ± 5.1 and 26.5 ± 6.4 years, respectively. No group differences were found in age, sex, education level, or duration of illness. The participants reported moderate to severe OC symptoms measured by Y-BOCS at baseline, with no group differences (22.5 ± 7.0 and 23.5 ± 6.8 for the ACT and WLC groups, respectively). There were no group differences in the other baseline symptoms and process measures. Table 1 presents the detailed demographic and clinical characteristics.
Table 1. Baseline demographic, clinical and pharmacological characteristics of patients with obsessive-compulsive disorder (mean ± s.d.)
ACT, acceptance-commitment therapy; WLC, wait-list control; Y-BOCS, Yale-Brown Obsessive-Compulsive Scale; OCI-R, Obsessive-Compulsive Inventory-Revised; BDI, Beck Depression Inventory; AAQ-II, -OC, Acceptance and Action Questionnaire-II, -for Obsessions and Compulsions; CFQ, Cognitive Fusion Questionnaire; TAFS, Thought-Action Fusion Scale.
a Among those who took the medication, all participants remained on the same regimen during the study except for 3 in the WLC who had minor changes in dosage or deletion of previous medication, or addition of new adjunctive drugs.
Effects of ACT on clinical symptoms, process measures, and behavioral task
Online Supplementary Table S1 presents the effects of the ACT on clinical and process measures. All variables in the repeated measures ANOVA except BDI showed significant group-by-time interactions (all p values < 0.05). Within the ACT group, paired t tests showed significant reductions in OC symptoms post-treatment (t = 3.9, p < 0.001 for Y-BOCS; t = 3.7, p = 0.001 for OCI-R). The ACT group also showed significant decreases post-treatment in all ACT-related process measures, such as the AAQ-OC (t = 4.0, p < 0.001), CFQ (t = 3.5, p = 0.002), and TAFS (t = 2.4, p = 0.029) scores, except for the AAQ-II scores. However, after treatment, the WLC group did not show any positive changes in outcome measures. Independent t test revealed significant group differences in the post-treatment scores of Y-BOCS (t = 2.4, p = 0.024) and AAQ-OC (t = −3.7, p < 0.001).
There were significant changes in the TAF induction task performance following the ACT intervention (Fig. 1b). In the repeated measures ANOVA, there were no significant group-by-time interactions. However, paired t test revealed that mean (±s.d.) emotional intensity induced by negative statements in the task significantly decreased in the ACT group (3.7 ± 0.3 v. 3.3 ± 0.9, t = 2.3, p = 0.031) whereas no significant change was found in the WLC group (3.6 ± 0.7 v. 3.4 ± 0.8, t = 1.4, p = 0.182). Furthermore, both groups showed no significant change in reaction times between pre- and post-treatment.
Effects of ACT on TAF-related brain activity
Four participants (two in each group) were excluded from the analyses due to no (two participants) or unreliable (two participants) behavioral responses on the TAF task during fMRI. An unreliable response was defined as a ‘very little’ response of emotional intensity over 3 SDs to all NSs at the pre-treatment scan.
Significant group-by-time interactions were observed in the bilateral insula, STG, and middle cingulate gyrus in the NS/CP condition (online Supplementary Table S2). In the post-hoc paired t test, the ACT group demonstrated a significant increase in activity in the bilateral insula and STG posttreatment (FDR-corrected p < 0.05) (online Supplementary Tables S3, S4, Figs 1c and 4). The NS/NP condition had no significant findings, and the results with an uncorrected statistical threshold are shown in online Supplementary Fig. S2.
Correlation analyses for all participants revealed that increases in the left insular activity were significantly related to decreases (i.e. improvements) in the AAQ-OC (rho = −0.346, p = 0.036) and CFQ (r = −0.349, p = 0.034) scores. In addition, right insular and STG activity was related to the reduction in the OCI-R (r = −0.371, p = 0.024) and CFQ (r = −0.436, p = 0.007), respectively. Detailed results are presented in online Supplementary Table S5 and Fig. 1d. However, no significant correlation was found within the ACT group only (online Supplementary Table S6).
Effects of ACT on TAF-related brain connectivity
PPI analyses revealed that the connectivity between the left insular–left inferior frontal gyrus (IFG) showed a significant group-by-time interaction (FDR-corrected p < 0.05) (online Supplementary Table S2), and this connectivity was strengthened after the ACT intervention (FDR-corrected p < 0.05) (Figs 2a and 4). Moreover, changes in the left IFG connectivity were significantly related to changes in the CFQ (r = −0.375, p < 0.05) scores for all participants (online Supplementary Table S5 and Fig. 2b). However, no significant correlations were found within the ACT group (online Supplementary Table S6).
Figure 2. Psycho-physiological interaction (PPI) results for task-based brain activation after acceptance-commitment therapy (ACT) intervention a. PPI analysis revealed a significant increase in functional connectivity between the left insula (seed; x = −46, y = −4, z = 14) and left inferior frontal gyrus (IFG; x = −40, y = 16, z = 24) [false discovery rate (FDR) corrected, p < 0.05; minimum cluster size, k = 10]. b. Correlation between the functional connectivity of the left insula and left IFG and the Cognitive Fusion Questionnaire (CFQ) across the entire participants.
Additionally, there was no significant task-dependent FC from other seed regions, such as the right insula and bilateral STG.
Effects of the ACT treatment on resting brain connectivity
No significant changes were found in interconnectivities among large-scale networks (online Supplementary Fig. S3), and within the SN, therefore, the analyses within the DMN were described. The components of the DMN were selected based on the results of correlation for matching (r = 0.22). Within the DMN, the bilateral PCC, left lingual gyrus, and bilateral precuneus demonstrated a significant group-by-time interaction; in addition to these regions, connectivity in the middle temporal gyrus was enhanced after ACT intervention in paired t test (FDR-corrected p < 0.05) (online Supplementary Table S2, S4, Figs 3a and 4).
Figure 3. Brain regions showing differences in resting state (rs) functional connectivity (FC) after acceptance-commitment therapy (ACT) intervention a. Brain regions showing differences in rsFC between pre- and post-treatment in the ACT group [false discovery rate (FDR) corrected, p < 0.05; minimum cluster size, k = 10]. b. The correlation of resting-state FC with the Acceptance and Action Questionnaire for Obsessions and Compulsions (AAQ-OC), Cognitive Fusion Questionnaire (CFQ), and Thought-Action Fusion Scale (TAFS) across the entire participants.
Figure 4. Summary of neural correlates of the effect of acceptance-commitment therapy (ACT) Thought-action fusion (TAF) task-based signal changes in the bilateral insula (INS, red) and superior temporal gyrus (STG, pink), strengthened the psycho-physiological interaction between the left insula (seed) and left inferior frontal gyrus (IFG, green), and the resting-state functional connectivity between the bilateral posterior cingulate gyrus (PCG), precuneus (PC), and left lingual gyrus (LG) (in blue).
Correlation analyses for all participants showed that changes in right PCC connectivity were related to the reduction in AAQ-OC (rho = −0.488, p = 0.001) and CFQ (r = −0.322, p = 0.040) scores; the precuneus connectivity change was associated with changes in TAFS scores (rho = −0.321, p = 0.041) (online Supplementary Table S5 and Fig. 3b). However, no significant correlations were found within the ACT group (online Supplementary Table S6).
Overall, all regions that were significant in the analyses above are depicted together in Fig. 4.
Discussion
The current fMRI study demonstrated that patients with OCD show significantly increased activation in the bilateral insula and STG in the TAF task after ACT intervention. Further PPI analyses with these regions as seed ROIs revealed that the left insular–left IFG connectivity was strengthened post-treatment in the ACT group. Differences in rsFC were also found in the PCC, precuneus, and lingual gyrus post-treatment in the ACT group. Some of these regions demonstrated significant correlations with ACT process measures, although our whole group correlation analyses could not confirm that these changes were ACT specific.
Increased TAF task-induced activation of the insula and STG
Bilateral activation of the insula
In this study, patients with OCD demonstrated robustly increased activation of the bilateral insula during the TAF task after an 8-week ACT intervention. Mindfulness-based interventions, overlapping with the ACT, have revealed increased insular cortex activity as the most consistent longitudinal effect (Young et al., Reference Young, van der Velden, Craske, Pallesen, Fjorback, Roepstorff and Parsons2018). The insula is considered to support interoceptive awareness, such as attending to bodily sensations in the present moment (Marchand, Reference Marchand2014; Tang et al., Reference Tang, Lu, Geng, Stein, Yang and Posner2010). Interestingly, change in insula activity is closely related to the efficacy of CBT (Pico-Perez et al., Reference Pico-Perez, Fullana, Albajes-Eizagirre, Vega, Marco-Pallares, Vilar and Soriano-Mas2022) and psychotherapy (Marwood, Wise, Perkins, & Cleare, Reference Marwood, Wise, Perkins and Cleare2018). A recent meta-analysis of task-based fMRI studies demonstrated that brain response in a network of regions involved in salience and interoception processing, encompassing fronto-insular and fronto-limbic cortices was associated with a positive CBT outcome (Pico-Perez et al., Reference Pico-Perez, Fullana, Albajes-Eizagirre, Vega, Marco-Pallares, Vilar and Soriano-Mas2022). Based on our results on the neural effect of ACT, the strengthening of insula-related networks may be a pivotal region of trans-therapeutic mechanisms across different psychosocial treatments.
Our finding of insula activation also had OCD-specific implications. The right insula was the only region correlated with the symptom measure of OCI-R in this study. Anxiety has been linked to altered interoceptive sensitivity, which focuses on insular cortex dysfunction (Paulus & Stein, Reference Paulus and Stein2006). Patients with OCD exhibit impaired interoception, which may be maintained during the course of CBT (Schultchen, Zaudig, Krauseneck, Berberich, & Pollatos, Reference Schultchen, Zaudig, Krauseneck, Berberich and Pollatos2019), and are differentially associated with rsFC of the insula (Eng et al., Reference Eng, Collins, Brown, Ludlow, Tobe, Iosifescu and Stern2022). Specifically, in OCD, disgust sensitivity has been focused on as a specific psychological vulnerability for the development of contamination-based OCD (Mataix-Cols et al., Reference Mataix-Cols, Wooderson, Lawrence, Brammer, Speckens and Phillips2004; Olatunji, Tart, Ciesielski, McGrath, & Smits, Reference Olatunji, Tart, Ciesielski, McGrath and Smits2011). Another type of OCD associated with abnormal interoception appears in individuals who focus on the physical sensations (i.e. sensation of skin contact with clothing or swallowing saliva), rather than primary disturbing thoughts. Although we consider the increase in insular activation to be the ‘typical’ response to ACT and mindfulness intervention, future studies may need to investigate the differential hemodynamic changes in these OC symptom dimensions after ACT.
Bilateral activation of STG
The STG is the site of language processing and higher-order multimodal integration. It also regulates amygdala activity (Muller, Cieslik, Turetsky, & Eickhoff, Reference Muller, Cieslik, Turetsky and Eickhoff2012), reflecting its association with social cognition. Considering that our task stimuli were socio-affective written sentences, increased STG activation in the ACT group compared to the WLC group may help the patients with OCD reinterpret negative statements in other contexts, such as an experimental situation rather than interpreting them literally. This may modulate affective arousal via effective connectivity to the amygdala (Kohn et al., Reference Kohn, Eickhoff, Scheller, Laird, Fox and Habel2014). Abnormalities in the STG have also been found with other neuroimaging modalities, such as decreased white matter volume (Goncalves et al., Reference Goncalves, Sousa, Carvalho, Leite, Ganho, Fernandes-Goncalves and Sampaio2017b) and increased FC in the right STG (Peng et al., Reference Peng, Xu, He, Shi, Wei, Miao and Chan2014), and increased fractional anisotropy in the bilateral STG (Shaw et al., Reference Shaw, Sharp, Sudre, Wharton, Greenstein, Raznahan and Rapoport2015; Yoo et al., Reference Yoo, Jang, Shin, Kim, Park, Moon and Kwon2007).
Task-specific increase in the relationship between the left insula and left IFG
Another novel finding of the current study was the increase in left IFG connectivity with the seed region, left insula, during the TAF induction task. Similar findings have been observed in previous experimental studies, such as the mere labeling of affective stimuli (Creswell et al., Reference Creswell, Way, Eisenberger and Lieberman2007) and the effect of distancing to regulate emotional responses to aversive images (Koenigsberg et al., Reference Koenigsberg, Fan, Ochsner, Liu, Guise, Pizzarello and Siever2010). As these evidenced, Fletcher et al. (Reference Fletcher, Schoendorff and Hayes2010) proposed a pattern of increased IFG activity (in their paper, vlPFC) and decreased amygdala activity as possible neural correlates of cognitive defusion. The current study is the first to report IFG involvement after full-scale and real ACT intervention in OCD.
In this study, an increase in left insular–left IFG connectivity correlated with a decrease in cognitive fusion scores. Defusion refers to processes that undermine the domination of language primarily by becoming aware of the process of thinking itself and being able to become aware of thoughts, emotions, and memories as passing events rather than ‘things’ that are literally true or false (Fletcher et al., Reference Fletcher, Schoendorff and Hayes2010). We also assume that the left IFG may play a significant role in processing cognitive defusion because of its fundamental role in human thinking. Specifically, first, the left IFG performs the actual integration of information from speech and gestures, resulting in a new representation with abstract and non-literal meaning (Steines, Nagels, Kircher, & Straube, Reference Steines, Nagels, Kircher and Straube2021). Second, the left IFG is actively involved in both declarative and procedural systems. The anteroventral IFG is associated with the encoding, selection, and retrieval of declarative memories (Ullman, Reference Ullman2004) while the posterior-dorsal left IFG processes rule-governed grammar and automatized procedural memories (Ullman, Reference Ullman, G. and S. L.2016). Third, the IFG is the most distinguished prefrontal region for cognitive control processes in cognitive reappraisal, which regulates emotions by manipulating semantic representation of affective scenarios (Buhle et al., Reference Buhle, Silvers, Wager, Lopez, Onyemekwu, Kober and Ochsner2014; Ochsner, Silvers, & Buhle, Reference Ochsner, Silvers and Buhle2012). Taken together, these functions are thought to be involved in suppressing the excessively learned, rule-based literal interpretation, distinguishing formal and arbitrary stimulus properties, and forming a new response (i.e. representation), which focus on the process of contextual thinking as mere private events passing by (Ciarrochi & Bailey, Reference Ciarrochi and Bailey2008).
Changes in the rs connectivity of DMN
After ACT, participants reported increased intrinsic connectivity of the PCC and precuneus within the DMN. The DMN is activated during passive resting rather than goal-oriented tasks (Raichle & Snyder, Reference Raichle and Snyder2007) and is related to self-referential processing, focusing on one's own feelings or thoughts (Whitfield-Gabrieli & Ford, Reference Whitfield-Gabrieli and Ford2012), or autobiographical memory processing (Svoboda, McKinnon, & Levine, Reference Svoboda, McKinnon and Levine2006). The PCC and precuneus appear to be important functional hubs of the posterior part of the DMN (Washington & VanMeter, Reference Washington and VanMeter2015), which involves episodic memory, self-referential processing, and theory of mind (Barnett et al., Reference Barnett, Reilly, Dimsdale-Zucker, Mizrak, Reagh and Ranganath2021).
Previous studies using rs-fMRI have reported alterations in the DMN in patients with OCD. The magnitude of FC in the PCC and precuneus was reduced in these patients (Cheng et al., Reference Cheng, Xu, Nie, Luo, Yang, Li and Xu2013). Patients with OCD and their siblings also demonstrated a decrease in FC in the PCC (Peng et al., Reference Peng, Xu, He, Shi, Wei, Miao and Chan2014). In addition, in a task fMRI study, the OCD patients showed difficulties in reducing DMN activation during non-rest emotional conditions (Goncalves et al., Reference Goncalves, Soares, Carvalho, Leite, Ganho-Avila, Fernandes-Goncalves and Sampaio2017a). In an electroencephalogram study using the TAF task, a group with high OC symptoms showed greater precuneus activity, which was correlated with the TAF scores (Jones & Bhattacharya, Reference Jones and Bhattacharya2014). In our previous study using the TAF paradigm, over-engagement of the SN-DMN connection was found to be related to an increased TAF response (Lee et al., Reference Lee, Song, Jang, Cha, Kim, Chang and Lee2022). In summary, (1) decreased PCC/precuneus FC during resting and (2) aberrant functional involvement of the DMN during the task have indicated alterations in the DMN in OCD. Therefore, increased PCC/precuneus FC during rs-fMRI could be related to the therapeutic effects of the ACT in patients with OCD. This is corroborated by the associations between increased PCC/precuneus FC and decreased scores of psychological inflexibility measures.
DMN dysfunction could be a transdiagnostic factor, as it has been found in many psychiatric illnesses including schizophrenia, major depressive disorder, and OCD (Doucet et al., Reference Doucet, Janiri, Howard, O'Brien, Andrews-Hanna and Frangou2020). Since DMN is involved in several mental activities (Kleckner et al., Reference Kleckner, Zhang, Touroutoglou, Chanes, Xia, Simmons and Barrett2017), it is challenging to determine how its increased FC affects ACT. However, among the various mental activities related to PCC/precuneus function, we believe that improved self-referential processing (Koban, Gianaros, Kober, & Wager, Reference Koban, Gianaros, Kober and Wager2021) may help patients with OCD to view themselves contextually. The PCC/precuneus function can be modulated by the self-relatedness of stimuli (Knyazev, Savostyanov, Bocharov, Levin, & Rudych, Reference Knyazev, Savostyanov, Bocharov, Levin and Rudych2020) and whether the self-judgment is context-dependent or -independent (Martial, Stawarczyk, & D'Argembeau, Reference Martial, Stawarczyk and D'Argembeau2018). Further fMRI studies on ACT can clarify our preliminary suggestions.
There are several limitations to our study. First, the results of a task-based fMRI study largely depend on the task. Although we believe that our paradigm could sensitively evaluate the response to verbal stimuli, which is associated with ACT effects (i.e. acceptance and cognitive defusion), it could hardly measure other ACT components, such as value and committed action. In the same context, it is challenging to draw any conclusion about the involvement of the prefrontal cortex or ACC, which have been well elucidated as an effect of CBT, since our task is simple and passive. It is necessary to create novel challenges to identify more specific ACT constructs. Second, although two-thirds of the ACT group was drug-naïve or drug-free during the study period, we could not completely exclude the possibility that medication affected our results. Third, unlike AAQ-OC, AAQ-II did not correlate with neural correlates in this study. This observation may support that AAQ-II does not capture experiential avoidance as it specifically relates to obsessional thoughts (Jacoby et al., Reference Jacoby, Abramowitz, Buchholz and Reuman2018) and underestimates treatment effects concerning psychological flexibility (Benoy et al., Reference Benoy, Knitter, Schumann, Kader, Walter and Gloster2019). Fourth, although the WLC group is frequently employed as a control (Öst, Havnen, Hansen, & Kvale, Reference Öst, Havnen, Hansen and Kvale2015), this study cannot demonstrate the specific characteristics of ACT that set it apart from CBT. Further studies are required to directly compare CBT and ACT in order to find an answer to the question. Fifth, our study cannot find any significant correlation results within the ACT group. Although we used changes between post-intervention and pre-intervention, the correlation results cannot be determined as ACT specific. Furthermore, the significance of our correlation results disappeared after multiple comparison corrections, and some of the significant results in uncorrected levels decreased after removing outliers or controlling baseline symptoms. Therefore, it is important to carefully assess our preliminary correlation results, and it is preferable to use them as insightful cornerstones for subsequent research.
In conclusion, this is the first fMRI study to report the neural mechanisms underlying ACT in OCD treatment. Our task-based and rs-fMRI analyses suggest that the therapeutic effect of ACT on OCD may involve the salience and interoception process (i.e. insula), multisensory integration (i.e. STG), language (i.e. IFG), and self-referential processes (i.e. PCC and precuneus). These areas or their interactions may be relevant to the psychological processes of ACT, in particular acceptance, contact with the present moment, self-as-context, and cognitive defusion.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291723001769
Acknowledgements
We thank Mina Choi and the members of the BMRlab for supporting our experiments.
Financial support
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIP) (SJL, grant number 2021R1A2C2004720); the basic research program through the Korean Brain Research Institute funded by the Ministry of Science and Information Communications Technology (ICT) (YC, grant nubmer 23-BR-03-01).
Conflict of interest
The authors declare none.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Declaration of Helsinki of 1975, as revised in 2008
Open access
