Steward et al. Translational Psychiatry (2019)9:194 https://doi.org/10.1038/s41398-019-0533-3 Translational Psychiatry ARTICLE Open Access A multimodal MRI study of the neural mechanisms of emotion regulation impairment in women with obesity Trevor Steward1,2,3, Maria Picó-Pérez1,4,5, Gemma Mestre-Bach1,2, Ignacio Martínez-Zalacaín1,6, Maria Suñol1,6,7, Susana Jiménez-Murcia1,2,6, Jose A. Fernández-Formoso2, Nuria Vilarrasa8,9, Amador García-Ruiz-de-Gordejuela10, Misericordia Veciana de las Heras11, Nuria Custal1, Nuria Virgili8, Rafael Lopez-Urdiales8, José M. Menchón 1,6,7, Roser Granero2,12, Carles Soriano-Mas 1,7,12 and Fernando Fernandez-Aranda 1,2,6 Abstract Maladaptive emotion regulation contributes to overeating and impedes weight loss. Our study aimed to compare the voluntary downregulation of negative emotions by means of cognitive reappraisal in adult women with obesity (OB) and female healthy controls (HC) using a data-driven, multimodal magnetic resonance imaging (MRI) approach. Women with OB (n = 24) and HC (n = 25) carried out an emotion regulation task during functional MRI scanning. Seed-to-voxel resting-state connectivity patterns derived from activation peaks identified by this task were compared between groups. Diffusion tensor imaging (DTI) was used to examine white matter microstructure integrity between regions exhibiting group differences in resting-state functional connectivity. Participants in the OB group presented 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; reduced activation in the ventromedial prefrontal (vmPFC) cortex in comparison to the HC group when downregulating negative emotions, along with heightened activation in the extrastriate visual cortex (p < 0.05, AlphaSim-corrected). Moreover, vmPFC peak activity levels during cognitive reappraisal were negatively correlated with self-reported difficulties in emotion regulation. OB patients exhibited decreased functional connectivity between the vmPFC and the temporal pole during rest (peak-pFWE = 0.039). Decreased fractional white-matter track volume in the uncinate fasciculus, which links these two regions, was also found in participants with OB. Taken together, our findings are indicative of emotion regulation deficits in OB being underpinned by dysfunctional hypoactivity in the vmPFC and hyperactivity in the extrastriate visual cortex. Our results provide a potential target circuit for neuromodulatory interventions to improve emotion regulation skills and weight-loss intervention outcomes. Introduction Emotional distress is increasingly understood as one of Obesity (OB) has become a major global health chal- the root causes of weight gain3 and recent research found lenge1 and a significant contributor to a reduction in affective negative states to be a predictor of lapses during improving mortality rates in the United States2. a 12-month weight-loss program4. Emotion regulation pertains to the implementation of a conscious or nonconscious strategy to start, stop, or Correspondence: Carles Soriano-Mas (csoriano@idibell.cat) or otherwise modulate the trajectory of an emotion5. When Fernando Fernandez-Aranda (ffernandez@bellvitgehospital.cat) 1 Department of Psychiatry, Bellvitge University Hospital-IDIBELL, C/Feixa Llarga compared with control groups, meta-analyses have s/n, 08907 Barcelona, Spain 2 demonstrated that individuals with OB are more likely to Ciber Fisiopatología Obesidad y Nutrición (CIBERObn), Instituto Salud Carlos endorse higher scores of alexithymia, difficulty in identi- III, C/Feixa Llarga s/n, 08907 Barcelona, Spain Full list of author information is available at the end of the article. fying feelings, and an externally oriented thinking style, These authors contributed equally: Trevor Steward, Maria Picó-Pérez © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Steward et al. Translational Psychiatry (2019)9:194 Page 2 of 10 which, in turn, may interfere with subsequent emotion explained by diminished functional integration22. Overall, regulation responses6. Relatedly, a stressful environment these results uphold the benefits of utilizing multimodal in combination with ineffective emotion regulation has imaging approaches to reveal the neurobiological been linked to abnormal cortisol patterns7, emotional abnormalities underpinning the behaviors found in OB. eating8, a sedentary lifestyle9, and sleep problems10. Taking into account the relevance of emotion regulation Indeed, the latest models of OB have stressed the need to in OB interventions23, the aim of the present study was to incorporate emotion regulation training into interven- use multimodal MRI methods to examine emotion reg- tions in order to improve treatment outcomes11–13. ulation in adult women with OB and healthy weight Based on the findings of numerous neuroimaging stu- controls. As opposed to selecting which brain regions to dies, mechanistic neural models of emotion regulation explore a priori, the study at hand employed a data-driven coincide in highlighting the role of prefrontal and cin- approach in which the findings obtained from the cog- gulate control systems in modulating activity in percep- nitive reappraisal paradigm informed where the sub- tual, semantic, and affect systems14. For example, one sequent MRI modality would be used. Moreover, we pathway involved in the downregulation of negative sought to investigate associations between fMRI activa- emotions consists of the recruitment of dorsomedial or tion patterns and the structural properties of white matter ventrolateral prefrontal regions so as to diminish amyg- fiber tracts with self-reported emotion regulation deficits dala responses via their impact on the ventromedial pre- and anthropometric variables. frontal cortex (vmPFC15,16). Other studies using functional magnetic resonance imaging (fMRI) in patients Materials/subjects and methods with psychiatric disorders, including those with dis- Participants ordered eating, have described abnormal neural activation Our sample was made up of 24 adult women with OB patterns encompassing, among others, the above- recruited from the Bariatric and Metabolic Surgery Unit mentioned regions during emotion regulation tasks17,18. and the Endocrinology and Nutrition Unit at Bellvitge Likewise, the only fMRI study to date using a cognitive University Hospital (Barcelona, Spain). Participants with reappraisal task in individuals with OB found that, when OB were compared to 25 healthy-weight controls (HC) instructed to regulate negative emotions by means of from the same hospital catchment area. As part of the cognitive reappraisal, young adults with excess weight evaluation procedure, all participants underwent the displayed heightened activation in the right anterior Mini-International Neuropsychiatric Interview (M.I.N.I.) insula, as well as decreased functional coupling between with staff psychologists from the Department of Psy- the right anterior insula and dorsal regions of the pre- chiatry at Bellvitge University Hospital24. Complete frontal cortex19. However, further research efforts are inclusion and exclusion criteria for this are included in the needed to fully characterize the neurobiological under- Supplementary Information. Patients exhibiting binge- pinnings of altered emotion regulation in OB. eating behavior were not included in the study sample. In parallel, current MRI assessments grant researchers The present study was carried out in accordance with the ability to delineate the neurobiological correlates of the latest version of the Declaration of Helsinki. The behavior in multiple units of analysis, which, when Bellvitge University Hospital Clinical Research Ethics properly integrated, permit a deeper understanding of the Committee and Institutional Review Board approved the mechanistic effects accounting for disrupted behaviors. study (PR146/14). Signed informed consent was obtained This has led to the combination of different imaging from all participants. methods in order to investigate the relationship between task-related local brain activations and network-level Clinical measures connectivity assessments, as well as between functional All participants completed the Difficulties in Emotion and structural measurements20. Within the context of OB, Regulation Scale (DERS) (Spanish validation by Wolz one multimodal study obtained results supporting an et al. 25). This 36-item self-report measure assesses association between higher body mass index (BMI) and emotion regulation difficulties using six separate sub- decreased fractional anisotropy (FA) in white matter scales. Higher scores on the DERS indicate greater emo- fibers connecting brain regions that support working tion regulation impairment. Cronbach’s alphas to memory, though no link with blood-oxygen-level depen- determine the internal consistency of the DERS subscales dent (BOLD) activity was found21. This finding may are included in Table S1 of the Supplementary indicate that changes in brain structure precede devia- Information. tions in function. Other studies have opted to combine both intrinsic resting-state and task-based activations, Anthropometric measures identifying alterations in the middle frontal gyrus and A Tanita BC-420MA was utilized to assess body com- occipital areas during perceptual processes that may be position and to calculate BMI. This noninvasive and Steward et al. Translational Psychiatry (2019)9:194 Page 3 of 10 validated device (Tanita BC-420MA, Tanita Corp., Tokyo, response at each voxel was convolved with the SPM12 Japan) uses bioelectrical impedance analysis to measure canonical hemodynamic response function using a 128-s weight and body composition variables (i.e. body fat high-pass filter. percentage26). Height was measured via a stadiometer. Second-level analyses Statistical analyses involving clinical and anthropometric Between-group comparisons in task-induced activations measures were conducted via two-sample t-tests using group (HC Statistical analysis was carried out with SPSS 23 (IBM vs. OB) as the main factor. Analyses were carried out Corp., Armonk, NY, USA). Between-group comparisons using masks generated by extracting and conjoining areas were carried out using independent sample t-tests, and from one-sample (OB and HC) activations for each con- linear associations were estimated using Pearson’s corre- trast. The peak activation differences derived were lations. Finner’s method was used to control for Type-I extracted and entered into an SPSS data matrix to assess error stemming from multiple comparisons27, Levene’s their relationship with clinical and anthropometric test assessed the equality of variances between the groups, measures. and Shapiro–Wilk tests were performed to confirm nor- mality. Effect size for mean differences was measured Psychophysiological interactions (PPI) analysis using Cohen’s-d coefficient (|d| > 0.2–0.5 was considered In order to explore differences in task-induced con- low, |d| > 0.5–0.8 moderate, and |d| > 0.8 large28. For nectivity between the brain regions activated during the Pearson’s correlations, and due to the strong association emotion regulation task, PPI analyses were conducted using between this coefficient and sample size, significance was SPM 12. Here, we explored the impact of the contrast of based on effect size (|r| > 0.10–0.24 was considered low, | interest (the ‘psychological’ factor) on the strength of time- r| > 0.24–0.37 moderate, and |r| > 0.37 large)29. Estimated course correlations between our empirically obtained power analysis for mean comparison t-tests are reported region of interest (ROI, the ‘physiological’ factor) with the in Table S1, whereas estimated power for the one-sample conjoining areas from one-sample activations. To perform correlation tests were between 0.60 and 0.95. first-level analyses, ROIs were drawn from the regions showing group differences during the emotion regulation fMRI cognitive reappraisal task task. Specifically, the following regions stemming from the See the Supporting Information for additional details on results of the Regulate>Maintain contrast of the emotion MRI image acquisition and preprocessing. regulation task (see below) were used as seeds: the left A modified version of the cognitive reappraisal task vmPFC (x = −10, y = 64, z = −18), and the right extra- designed by Phan et al.30 was used to evaluate emotion striate visual cortex (x = 56, y = −72, z = 10), which were regulation during fMRI scanning. This task has been used defined with 3-mm radial spheres (see Table 2). Functional in different psychiatric populations31 and in patients with connectivity maps were estimated for the selected seeds excess weight19. The task consists of presenting a series of by including our signal of interest (seed) in interaction blocks displaying neutral or negative picture stimuli that with the task blocks, controlling for the raw signal of the participants were instructed to (1) “Observe” (to passively seed and the task blocks. Resulting images were then observe neutral images); (2) “Maintain” (to actively sustain included in a two-sample t-test model for each seed the negative emotions elicited by the images); or (3) (second-level) to assess between-group effects. “Regulate” (to reappraise and reduce the intensity of negative emotions by means of previously trained cogni- Significance thresholding tive reappraisal techniques). Further descriptions of the Statistical significance was determined by a combination task have been reported elsewhere30 and can be found in of voxel-level and cluster-extent thresholds using the the Supporting Information. AlphaSim algorithm as implemented in the SPM REST- plus V1.2 toolbox33. The minimum spatial cluster extent fMRI emotion regulation task effects (KE) to satisfy a family-wise error (FWE) rate correction First-level analyses of pFWE < 0.05 varied over the different contrasts. Input The two contrasts of interest defined for first-level parameters to AlphaSim included a voxel-level probability (single-subject) analysis were Maintain vs. Observe and of p < 0.001, a rmm (edge connected for cluster) of 5, a full Regulate vs. Maintain. The former assesses brain activa- width at half maximum (FWHM) corresponding to the tions associated with negative emotion generation, actual smoothing of the data after model estimation, and a whereas the latter delineates brain activations associated mask volume contingent on the contrast of interest. As a with cognitive reappraisal18,32. Conditions were modeled result, for task activation analyses we obtained a cluster- for the 20 s that the images were on the screen and did not extent threshold of 33 voxels and thresholds of 10 and 51 include instruction and rating periods. The BOLD voxels for PPI analyses to satisfy a pFWE < 0.05. Steward et al. Translational Psychiatry (2019)9:194 Page 4 of 10 Resting-state analysis using 50 iterations37. Parameters for probabilistic tracking CONN Toolbox (Version 17, McGovern Institute for were five seed points per voxel, 1000 Monte Carlo itera- Brain Research, Massachusetts Institute of Technology, tions, and a threshold fraction of 0.001, and, at each Cambridge, USA, http://www.nitrc.org/projects/conn) iteration, the algorithm found locations of tracts con- was used for resting-state, seed-to-voxel connectivity necting pairs of targets. Only white-matter tracts con- analyses. A band-pass filter with cut-off frequencies of necting targets in at least 85% of the total sample were 0.008 and 0.09 Hz was applied to exclude drifts and high- included. Standard propagation parameters for this algo- frequency activations. ROIs were generated as 5-mm rithm were used: 60° maximum angle of propagation radius spheres centered on the peak coordinates from confined to voxels with FA > 0.2. Here we report FA and significant task activations acquired from the emotion mean diffusivity (MD) for each WM-ROI, as calculated by regulation task (i.e. the vmPFC and the right extrastriate FATCAT, as well as tract volume standardized to whole- visual cortex). Whole-brain functional connectivity maps brain volume (fNV). After exporting the DTI measures to were estimated for these two seeds, and resulting images SPSS, a multivariate analysis was used to compare FA, were included in second-level two-sample t-tests to assess MD and fNV between HC and OB groups. between-group effects. Results Diffusion-weighted imaging (DWI) analyses Sociodemographic and clinical results DWI data were visually inspected for motion and Sociodemographic information on the study sample is dropout slices, with individual volumes discarded (sub- summarized in Table 1. No significant differences were jects were excluded when <33 volumes remained). The found between groups with regards to age and years of DIFFPREP module in TORTOISE V3.1.034 was used to education (p > 0.05). As expected, the OB group had sig- compute distortion corrections for subject motion, eddy nificantly higher body fat percentage and BMI than the currents, and basic EPI distortions. Fat_proc* programs in HC group (p < 0.001). AFNI were used to invert the contrast of the T1w images Participants in the OB group endorsed significantly to imitate T2-weighted images in order to provide an greater difficulty in emotion regulation on the following anatomical reference volume within TORTOISE35. Lastly, DERS subscales (see Table S1): non-acceptance of emo- preprocessed DWIs were exported to AFNI for diffusion tional responses (p = 0.009), lack of emotion awareness tensor (DT) and associated parameter fitting. (p = 0.001), and lack of emotional clarity (p = 0.049). Total DERS scores were also significantly higher in the Target placement OB group in comparison to the HC group (p = 0.048). Pairs of regions displaying significant between-group differences in intrinsic resting-state functional con- Emotion regulation task results nectivity were selected as targets for tractography analysis In-scanner behavioral results given that previous studies have established that positive The analysis of in-scanner ratings for each condition relationships between BOLD signal and white matter (Observe, Maintain, and Regulate) revealed a main effect measures are largely found in resting state data36. One of condition (F(2, 94) = 60.94, p < 0.001). Post-hoc com- pair of targets was placed within the left vmPFC and the parisons showed that Maintain differed from Observe, left temporal pole. The other included the right extra- indicating successful negative emotion induction during striate visual cortex and the left inferior temporal lobe. These targets were generated in MNI space after which Table 1 Sample characteristics they were transformed to each subject’s DW space and inflated in volume to ensure adequate WM coverage. Healthy Obese weight n = 24 Probabilistic tractography n = 25 White-matter regions of interest (referred to hereafter as WM-ROIs) connecting pairs of resting-state targets Mean SD Mean SD p were identified using probabilistic tractography, as Age (years) 31.92 12.95 38.08 10.43 0.099 implemented in AFNI’s FATCAT utility toolkit37. FAT- CAT efficiently finds white-matter connections within Body mass index (BMI, kg/m2) 20.89 1.87 42.67 7.11 <0.001* networks and provides quantitative measures for all Body fat (%) 24.20 4.99 47.54 5.41 <0.001* identified WM-ROIs. DTI parameter (FA and the direc- Years of education 15.68 1.60 14.92 2.00 0.179 tion of the first eigenvector of the DTI tensor relative to SD standard deviation the mean direction) uncertainty maps for probabilistic *Bold: significant difference (p < 0.05). p-values include Finner’s correction for tracking were calculated with FATCAT 3dDWUncert multiple comparisons Steward et al. Translational Psychiatry (2019)9:194 Page 5 of 10 this condition for both groups (Maintain > Observe: F(1, the HC group, as compared to the OB group, presented 47) = 95.84, p < 0.001). Likewise, comparisons between increased intrinsic functional connectivity with the left Regulate and Maintain contrasts showed significant dif- temporal pole (peak-pFWE = 0.039, see Table S3, Fig. 3) ferences between the two conditions (Maintain > Regulate F(1, 47) = 18.95, p < 0.001), indicating participants Resting-state extrastriate visual cortex connectivity reduced distress levels when instructed to use cognitive Participants in the OB group displayed increased func- reappraisal. No significant effects of group (F(1, 47) = tional connectivity during rest between the right extra- 1.57, p > 0.05) or interactions (p > 0.05) were found. striate visual cortex and the left inferior temporal cortex compared to HC (cluster-pFWE = 0.016, Table S3, Fig. Task activations and associations S1). No significant differences were found between groups during the Maintain > Observe contrast. Conversely, DTI results between-group comparisons showed that OB participants Distribution of tracts presented reduced activity in the left vmPFC compared to White-matter tracts connecting the left vmPFC and the HC participants during the Regulate > Maintain contrast left temporal pole from the abovementioned resting-state (Table 2, Fig. 1). In contrast, OB participants displayed analyses were identified in at least 85% of the total sample increased activation in the right extrastriate visual cortex (n = 45). After visual inspection, and taking into account in comparison to HC in this same comparison (Table 2, that our probabilistic tractography approach was imple- Fig. 2). mented to isolate fibers between target ROIs in the A negative association between the peak activation vmPFC and temporal pole, individual white-matter tracts eigenvalues extracted from the vmPFC during Regulate vs. connecting these two regions were identified as belonging Maintain and DERS total scores was found in the study to the left uncinate fasciculus38. An example of identified sample (r(47) = −0.397, p = 0.005) and in the OB group (r uncinate fasciculus tracts from a representative subject (22) = −0.466, p = 0.022). Likewise, a significant positive can be found in Fig. 3. White-matter tracts connecting the association was found between right extrastriate visual right extrastriate visual cortex and the left inferior tem- cortex peak eigenvalues during Regulate vs. Maintain and poral lobe were found in an insufficient number of par- body fat percentage in the study sample (r(47) = 0.494, p < ticipants to be included in subsequent analyses. 0.005) and in the HC group (r(23) = 0.431, p = 0.031). Differences in DTI measures PPI connectivity A multivariate analysis comparing FA, MD and the In comparison to HC, OB participants presented fractional volume of tracks (fNV) between HC and OB decreased connectivity between the left vmPFC and the participants was performed. Groups significantly differed left globus pallidus and increased connectivity between in the measures examining left uncinate fasciculus the vmPFC and the right crus cerebellum II (Table S2, microstructure (OB patients n = 21, HC n = 24, Fig. 1). Increased connectivity was found between the F(1, 42) = 3.56, p = 0.012). According to post-hoc tests, right extrastriatal visual cortex and the left inferior tem- group differences were driven by lower fNV in the OB poral lobe, the left precuneus and the left supramarginal group compared to the HC group (p = 0.02) (Table 3). gyrus in the OB group compared to HC (Table S2, Fig. 2). Associations between DTI and clinical measures Resting-state functional connectivity results Within the whole sample [r(42) = −0.401, p = 0.006] Resting-state vmPFC connectivity and within the OB group [r(18) = −0.624, p = 0.002], fNV Using the left vmPFC peak coordinate from the emotion in the left uncinate fasciculus was negatively associated regulation task for seed-to-voxel analysis, participants in with DERS Lack of Emotional Clarity scores (Fig. 3). Table 2 fMRI emotion regulation task results Contrast Peak region MNI coordinates (x, y, z) Kea t-statistic HC > OB Left ventromedial prefrontal cortex −10, −64, −18 44 4.21 OB > HC Right extrastriate visual cortex 56, −72, 10 35 4.50 Regions showing between-group differences during the Regulate > Maintain contrast of the emotion regulation task (AlphaSim voxel-level probability = p < 0.001, p < 0.05 FWE-cluster corrected) HC healthy controls, OB obesity, MNI Montreal Neurological Institute a Cluster extent in voxels Steward et al. Translational Psychiatry (2019)9:194 Page 6 of 10 Fig. 1 a Increased activation was found in the left ventromedial prefrontal cortex (vmPFC) in the healthy control group in comparison to the obese group (p < 0.05, AlphaSim cluster-extent corrected) during the emotion regulation task (Regulate vs. Maintain). Color bar represents t-values. b A scatterplot depicting the negative association between extracted activation eigenvalues from the vmPFC peak during the emotion regulation task (Regulate vs. Maintain) and Difficulties in Emotion Regulation Scale (DERS) total scores in the obesity group [n = 24, r(22) = −0.466, p = 0.022)]. Error bars show the 95% confidence interval. c Increased task-induced connectivity between the vmPFC and the left globus pallidus was found in the healthy control group in comparison to the obesity group using psychophysiological interactions (PPI) analysis (p < 0.05, AlphaSim cluster-extent corrected). Color bar represents t-values Fig. 2 a Brain regions showing increased activation in the obese group during the emotion regulation task (Regulate vs. Maintain). Increased activation was found in the right extrastriatal visual cortex in the obese group in comparison to the HC group (p < 0.05, AlphaSim cluster-extent corrected). Color bar represents t-values. b A scatterplot depicting the positive correlation between extracted activation eigenvalues from the extrastriatal visual cortex peak during the emotion regulation task (Regulate vs. Maintain) and total body fat percentage in the control group [n = 25, (r(23) = 0.431, p = 0.031)]. Error bars show the 95% confidence interval (CI). c Increased psychophysiological interactions (PPI) connectivity using the extrastriatal visual cortex as a seed was found in the left inferior temporal lobe, the left precuneus, and the left supramarginal gyrus in the obese group in comparison to the healthy control group. Color bar represents t-values Steward et al. Translational Psychiatry (2019)9:194 Page 7 of 10 Fig. 3 a Targets for probabilistic fiber tracking, derived from the resting-state analyses results (featured in the box), included the left ventromedial prefrontal cortex (vmPFC) and the left temporal pole (marked in red). An example of the uncinate fasciculus tracts connecting the vmPFC target with the temporal pole target is shown (marked in yellow). SUMA/AFNI Surface Mapper software was used for DTI image preparation. b A scatterplot depicting the negative correlation between the fractional volume of the left uncinate fasciculus and residual Difficulties in Emotion Regulation Scale (DERS) Lack of Clarity scores in the obese group controlling for age [n = 21, r(19) = −0.624, p = 0.002]. Error bars show the 95% confidence interval (CI). *This correlation remained significant after excluding the highlighted subject (p = 0.003 vs. p = 0.010) Discussion Table 3 Comparison between groups in DTI white-matter The study at hand used multimodal MRI to identify the microstructure measures neurobiological foundation of altered emotion regulation in WM-ROI DTI F-statistic p-value Direction OB. During the completion of a cognitive reappraisal task, parameter participants with OB displayed reduced activation of the vmPFC in comparison to HC participants. Additionally, Left uncinate FA 0.11 0.744 − activation levels in the vmPFC were negatively associated fasciculus MD 3.27 0.077 − with self-reported difficulties in emotion regulation, sug- fNV 11.07 0.002 HC > OB gesting that, despite being detected in an experimental context, the effects of vmPFC hypoactivation translate into WM-ROI white-matter region of interest, DTI diffusion tensor image, FA fractional anisotropy, MD mean diffusivity, fNV fractional volume of tracts, HC healthy real-life contexts. Human and animal studies have demon- controls, OB obese strated that the vmPFC is a key node of cortical and Steward et al. Translational Psychiatry (2019)9:194 Page 8 of 10 subcortical networks subserving the generation and reg- Emotional Clarity scores, indicating that lower uncinate ulation of negative emotions via interactions with the tract volume could underlie the dispositional difficulties amygdala, the dorsal anterior cingulate cortex and other in this specific dimension of emotion regulation. One of regions38,39. By directly acting upon inhibitory interneurons the overarching functions of the uncinate fasciculus is to within subcortical structures, the vmPFC mediates the allow mnemonic representations originating in the tem- inhibition of negative emotion and acts as a monitor for poral lobe to modify behavior through interactions with assessing the effectiveness of emotion regulation strate- prefrontal regions supplying valence-based biasing of gies38–40. Altered vmPFC activity during emotion regulation decisions38. Similarly, uncinate fasciculus tracts between is commonly reported in populations with substance use the anterior temporal lobe and the prefrontal cortex also disorders41, and with psychiatric disorders, such as mood pass through the amygdala and are understood to con- and anxiety disorders18,42. Other work has implicated a tribute to the emotional tone (i.e. positive or negative failure to engage the vmPFC with loss-of-control eating in feelings, and personal significance) of representations overweight and obese girls in response to distress, suggest- stored in the anterior temporal lobe51. Disruption of the ing that perturbations in such neural circuits may lead to uncinate is believed to cause problems in guiding deci- overeating in order to cope with negative affect43. More sions and to underlie numerous psychiatric disorders52,53. recent research using stress induction, food cues and eco- It is likely that uncinate perturbation could potentially be logical momentary assessment found that changes in acti- problematic in domains that extend beyond memory to vation in the vmPFC significantly moderated the include social–emotional functioning and higher-level relationship of affect to binge eating44, thereby lending motivation38. Overall, our findings lend support to the support to models positing negative urgency as being a risk notion that, beyond the impairments demonstrated by factor for bulimic symptoms45. individuals with OB when confronting negative scenarios, In this study, the specific subcortical structure with less specific trait-like features leading to impaired emotion functional connectivity from the vmPFC in the OB group regulation capacities, such as difficulties correctly identi- when instructed to regulate negative emotions was the fying emotions6,54,55, may be at least partly underpinned globus pallidus. As part of the cortico-striato-thalamo- by alterations in circuits not necessarily engaged during cortical (CSTC) limbic loop, the globus pallidus plays a attempts of emotion regulation in experimental settings. pivotal role in maintaining and regulating motivation by The OB group presented differential activation patterns acting as a relay with the dorsomedial nucleus of the tha- in the extrastriate visual cortex in comparison to controls lamus46. Disengagement of this loop in patients with OB during both cognitive reappraisal and at rest. The extra- when instructed to cognitively reappraise negative emotions striate visual cortex is highly implicated in the conscious may be suggestive of an inability to sustain an emotion perception of emotional signals and its response is known regulation strategy for a sufficient length of time to effec- to be amplified by stress in a manner consistent with tively reduce negative emotions47. Moreover, this finding vigilance for threat56,57. For example, the extrastriate body dovetails with the results of research showing that con- area (EBA), which is part of our pattern of findings, is nectivity between the vmPFC and the striatum is crucial to known to be involved in the processing of emotionally cognitive reappraisal by shifting experienced emotional salient stimuli58. Hyperactivation of extrastriate visual valence from negative to positive48. A disordered reward- areas in OB patients, and increased connectivity of this attribution system, wherein the vmPFC fails to accurately region with other high-order visual processing areas, attach value to stimuli, could constitute a general could be indicative of diminished control in the visual mechanism underpinning the difficulties reported by people regions while processing emotionally salient stimuli59. In with OB in resisting rewarding foods and wielding negative this sense, bidirectional feedback connections between the emotions49. Other studies including patients with OB and amygdala and extrastriatal visual areas are involved in the patients with binge eating disorder (BED) found dietary perceptual enhancement of emotionally relevant stimuli60 restraint scores to be negatively correlated with and vmPFC and failure to properly engage this visual processing activity, supporting that BED individuals’ diminished ability stream is likely to underlie in some measure the atten- to recruit impulse-control-related brain regions could be tional biases found in OB60–62. Importantly, activation in associated with impaired dietary restraint50. the extrastriate visual cortex during emotion regulation Participants with OB presented decreased resting-state was positively associated with body fat percentage in the connectivity between the vmPFC and the left temporal control group, which indicates that increased activation of pole, together with lower fractional volume in the white- this region is not exclusively observed in obese individuals matter tracts connecting these regions (the left uncinate but also in those lean controls with a higher body fat fasciculus). Moreover, fNV values from the uncinate fas- percentage. Further research should aim at ascertaining ciculus were negatively associated with DERS Lack of the putative mechanisms linking adiposity to overreaction Steward et al. Translational Psychiatry (2019)9:194 Page 9 of 10 to affective stimuli. Body fat has indeed been shown to Medicine, University of Barcelona, C/Feixa Llarga s/n, 08907 Barcelona, Spain. 7 modulate activity in other brain networks, such as those Ciber Salud Mental (CIBERSAM), Instituto Salud Carlos III, C/Feixa Llarga s/n, 08907 Barcelona, Spain. 8Department of Endocrinology and Nutrition, subserving homeostatic and reward signal processing63. University Hospital of Bellvitge-IDIBELL, C/Feixa Llarga s/n, 08907 Barcelona, Interpretations from the presented findings should be Spain. 9CIBERDEM-CIBER de Diabetes y Enfermedades Metabólicas Asociadas, made bearing in mind the limitations of this study. First, Instituto de Salud Carlos III, C/Feixa Llarga s/n, 08907 Barcelona, Spain. 10 Bariatric and Metabolic Surgery Unit, Service of General and Gastrointestinal its cross-sectional design does not allow for inferences Surgery, Bellvitge University Hospital-IDIBELL, C/Feixa Llarga s/n, 08907 about causality. In this sense, it would be of great interest Barcelona, Spain. 11Neurology Department, Bellvitge University Hospital- to examine how emotion regulation deficits may pose a IDIBELL, C/Feixa Llarga s/n, 08907 Barcelona, Spain. 12Departament de Psicobiologia i Metodologia de les Ciències de la Salut, Universitat Autònoma risk factor for developing OB64. Second, we recruited a de Barcelona, 08193 Barcelona, Spain moderately sized sample of individuals with OB, though we did aim for homogeneity in that the sample was made Conflict of interest The authors declare that they have no conflict of interest. up of only women, since sex has been shown to be a significant confounder associated with emotion dysregu- lation65. As a final limitation, no quantitative data on the Publisher’s note nutritional aspects of the participants’ eating behaviors Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. were collected. Future studies would stand to benefit from assessing the extent to which neural activation patterns Supplementary Information accompanies this paper at (https://doi.org/ during emotion regulation are linked to the consumption 10.1038/s41398-019-0533-3). of highly palatable “comfort” foods.23 Received: 9 March 2019 Revised: 16 May 2019 Accepted: 31 May 2019 Conclusion This study advances our understanding of the neural foundation of emotion regulation impairments in OB. References Using a multimodal MRI approach, we demonstrated that 1. Ng, M. et al. Global, regional, and national prevalence of overweight and hypoactivation in the vmPFC during emotion regulation is obesity in children and adults during 1980-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 384, 766–781 (2014). associated with self-reported emotion regulation difficul- 2. Preston, S. H., Vierboom, Y. C. & Stokes, A. The role of obesity in exceptionally ties in patients with OB. Furthermore, we identified how slow US mortality improvement. Proc. Natl Acad. Sci. USA 115, 957–961 (2018). structural and functional connectivity alterations origi- 3. Hemmingsson, E. A new model of the role of psychological and emotional distress in promoting obesity: Conceptual review with implications for treat- nating in the vmPFC and spanning the uncinate fasciculus ment and prevention. Obes. Rev. 15, 769–779 (2014). are linked to excess weight and emotion regulation defi- 4. Forman, E. M. et al. Ecological momentary assessment of dietary lapses across cits, thereby providing researchers and clinicians alike behavioral weight loss treatment. Ann. Behav. Med. 51, 741–753 (2017). 5. Etkin, A., Buechel, C. & Gross, J. J. The neural bases of emotion regulation. Nat. with a potential target circuit for neuromodulatory Rev. Neurosci. 16, 693 (2015). interventions66. 6. Fernandes, J., Ferreira-Santos, F., Miller, K. & Torres, S. Emotional processing in obesity: a systematic review and exploratory meta-analysis. Obes. Rev. 19, Acknowledgements 111–120 (2018). This manuscript and research was supported by a grant from the Instituto de 7. Pervanidou, P. & Chrousos, G. P. Metabolic consequences of stress during Salud Carlos III (ISCIII) (FIS PI14/00290, PI17/ 01167; PI13/01958) and co-funded childhood and adolescence. Metab.: Clin. Exp. 61, 611–619 (2012). by FEDER funds/European Regional Development Fund (ERDF), a way to build 8. Nicholls, W., Devonport, T. J. & Blake, M. The association between emotions Europe. CIBERobn and CIBERsam are both initiatives of ISCIII. M.P.-P. and G.M.-B. and eating behaviour in an obese population with binge eating disorder. are supported by predoctoral AGAUR grants (2015 FI_B 00839 and 2018 FI_B2 Obes. Rev. 17, 30–42 (2016). 00174), grants co-funded by the European Social Fund (ESF) “ESF”, investing in 9. Isasi, C. R., Ostrovsky, N. W. & Wills, T. A. The association of emotion regulation your future. I.M.-Z. receives support from a P-FIS Carlos III grant (FI17/00294). M. with lifestyle behaviors in inner-city adolescents. Eat. Behav. 14, 518–521 S. is supported by a CIBERSAM PhD grant (CNV665/914). C.S.-M. is supported by (2013). a Miguel Servet contract from the Carlos III Health Institute (CPII16/00048). 10. Kahn, M., Sheppes, G. & Sadeh, A. Sleep and emotions: bidirectional links and Study resulting from the SLT006/17/00246 grant, funded by the Department of underlying mechanisms. Int. J. Psychophysiol. 89, 218–228 (2013). Health of the Generalitat de Catalunya from the call “Acció instrumental de 11. Raman, J., Smith, E. & Hay, P. The Clinical Obesity Maintenance Model: an programes de recerca orientats en l'àmbit de la recerca i la innovació en salut”. integration of psychological constructs including mood, emotional regulation, We thank CERCA Program/Generalitat de Catalunya for institutional support. disordered overeating, habitual cluster behaviours, health literacy and cogni- Additional support received from EU Grant Eat2beNice (H2020-SFS-2016-2; Ref tive function. J. Obes. 2013, 1–9 (2013). 728018). 12. Aparicio, E., Canals, J., Arija, V., De Henauw, S. & Michels, N. The role of emotion regulation in childhood obesity: implications for prevention and treatment. Nutr. Res. Rev. 29, 17–29 (2016). Author details 13. Leehr, E. J. et al. Emotion regulation model in binge eating disorder and 1 Department of Psychiatry, Bellvitge University Hospital-IDIBELL, C/Feixa Llarga obesity—a systematic review. Neurosci. Biobehav. Rev. 49, 125–134 (2015). s/n, 08907 Barcelona, Spain. 2Ciber Fisiopatología Obesidad y Nutrición 14. Langner, R., Leiberg, S., Hoffstaedter, F. & Eickhoff, S. B. Towards a human self- (CIBERObn), Instituto Salud Carlos III, C/Feixa Llarga s/n, 08907 Barcelona, Spain. regulation system: common and distinct neural signatures of emotional and 3 Melbourne School of Psychological Sciences, Faculty of Medicine, Dentistry behavioural control. Neurosci. Biobehav. Rev. 90, 400–410 (2018). and Health Sciences, University of Melbourne, Parkville, VIC 3010, Australia. 4Life 15. Diekhof, E. K., Geier, K., Falkai, P. & Gruber, O. Fear is only as deep as the mind and Health Sciences Research Institute (ICVS), School of Medicine, University of allows. A coordinate-based meta-analysis of neuroimaging studies on the Minho, Braga, Portugal. 5ICVS/3B’s—PT Government Associate Laboratory, regulation of negative affect. Neuroimage 58, 275–285 (2011). Braga/Guimarães, Portugal. 6Department of Clinical Sciences, School of Steward et al. Translational Psychiatry (2019)9:194 Page 10 of 10 16. Ochsner, K. N., Silvers, J. A. & Buhle, J. T. Functional imaging studies of emotion 42. Stephanou, K., Davey, C. G., Kerestes, R., Whittle, S. & Harrison, B. J. Hard regulation: a synthetic review and evolving model of the cognitive control of to look on the bright side: neural correlates of impaired emotion reg- emotion. Ann. N. Y. Acad. Sci. 1251, E1–E24 (2012). ulation in depressed youth. Soc. Cogn. Affect. Neurosci. 12, 17. Seidel, M. et al. Processing and regulation of negative emotions in anorexia 1138–1148 (2017). nervosa: an fMRI study. NeuroImage Clin. 18, 1–8 (2018). 43. Jarcho, J. M. et al. Neural activation during anticipated peer evaluation and 18. Picó-Pérez, M., Radua, J., Steward, T., Menchón, J. M. & Soriano-Mas, C. Emotion laboratory meal intake in overweight girls with and without loss of control regulation in mood and anxiety disorders: a meta-analysis of fMRI cognitive eating. Neuroimage 108, 343–353 (2015). reappraisal studies. Prog. Neuro-Psychopharmacol. Biol. Psychiatry 79, 96–104 44. Wonderlich, J. A. et al. The impact of neural responses to food cues following (2017). stress on trajectories of negative and positive affect and binge eating in daily 19. Steward, T. et al. Emotion regulation and excess weight: impaired affective life. J. Psychiatr. Res. 102, 14–22 (2018). processing characterized by dysfunctional insula activation and connectivity. 45. Puccio, F., Fuller-Tyszkiewicz, M., Buck, K. & Krug, I. Negative urgency and the PLoS ONE 11, e0152150 (2016). dual pathway model of bulimic symptoms: a longitudinal analysis. Eur. Eat. 20. Moreno-Navarrete, J. M. et al. Neuroinflammation in obesity: circulating Disord. Rev. 27, 34–48 (2019). lipopolysaccharide-binding protein associates with brain structure and cog- 46. Da Cunha, C. et al. Toward sophisticated basal ganglia neuromodulation: nitive performance. Int J. Obes. 41, 1627–1635 (2017). review on basal ganglia deep brain stimulation. Neurosci. Biobehav. Rev. 58, 21. Alarcón, G., Ray, S. & Nagel, B. J. Lower working memory performance in 186–210 (2015). overweight and obese adolescents is mediated by white matter micro- 47. Heller, A. S. et al. Reduced capacity to sustain positive emotion in major structure. J. Int. Neuropsychol. Soc. 22, 281–292 (2016). depression reflects diminished maintenance of fronto-striatal brain activation. 22. García-García, I. et al. Functional network centrality in obesity: a resting-state Proc. Natl Acad. Sci. USA 106, 22445–22450 (2009). and task fMRI study. Psychiatry Res.—Neuroimaging 233, 331–338 (2015). 48. Doré, B. P. et al. Finding positive meaning in negative experiences engages 23. Preuss, H., Pinnow, M., Schnicker, K. & Legenbauer, T. Improving inhibitory ventral striatal and ventromedial prefrontal regions associated with reward control abilities (impulse)—a promising approach to treat impulsive eating? valuation. J. Cogn. Neurosci. 29, 235–244 (2017). Eur. Eat. Disord. Rev. 25, 533–543 (2017). 49. Weygandt, M. et al. The role of neural impulse control mechanisms for dietary 24. Sheehan, D. V. et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): success in obesity. Neuroimage 83, 669–678 (2013). the development and validation of a structured diagnostic psychiatric inter- 50. Balodis, I. M. et al. Divergent neural substrates of inhibitory control in binge view for DSM-IV and ICD-10. J. Clin. Psychiatry 59, 22–33 (1998). eating disorder relative to other manifestations of obesity. Obesity 21, 367–377 25. Wolz, I. et al. Emotion regulation in disordered eating: psychometric properties (2013). of the difficulties in emotion regulation scale among spanish adults and its 51. Olson, I. R., McCoy, D., Klobusicky, E. & Ross, L. A. Social cognition and the interrelations with personality and clinical severity. Front. Psychol. 6, 907 (2015). anterior temporal lobes: a review and theoretical framework. Soc. Cogn. Affect. 26. Browning, L. M. et al. Measuring abdominal adipose tissue: comparison of Neurosci. 8, 123–133 (2013). simpler methods with MRI. Obes. Facts 4, 9–15 (2011). 52. Bracht, T., Linden, D. & Keedwell, P. A review of white matter microstructure 27. Finner, H. Stepwise multiple test procedures and control of directional errors. alterations of pathways of the reward circuit in depression. J. Affect. Disord. Ann. Stat. 27, 274–289 (1999). 187, 45–53 (2015). 28. Kelley, K. & Preacher, K. J. On effect size. Psychol. Methods 17, 137–152 (2012). 53. Olson, I. R., Heide, R. J. V. Der, Alm, K. H. & Vyas, G. Development of the 29. Rosnow, R. L. & Rosenthal, R. Computing contrasts, effect sizes, and counter- uncinate fasciculus: Implications for theory and developmental disorders. Dev. nulls on other people’s published data: general procedures for research Cogn. Neurosci. 14, 50–61 (2015). consumers. Psychol. Methods 1, 331–340 (1996). 54. Giel, K. E. et al. Decreased emotional perception in obesity. Eur. Eat. Disord. Rev. 30. Phan, K. L. et al. Neural substrates for voluntary suppression of negative affect: 24, 341–346 (2016). a functional magnetic resonance imaging study. Biol. Psychiatry 57, 210–219 55. Görlach, M. G., Kohlmann, S., Shedden-Mora, M., Rief, W. & Westermann, S. (2005). Expressive suppression of emotions and overeating in individuals with over- 31. Albein-Urios, N. et al. Re-appraisal of negative emotions in cocaine depen- weight and obesity. Eur. Eat. Disord. Rev. 24, 377–382 (2016). dence: dysfunctional corticolimbic activation and connectivity. Addict. Biol. 19, 56. Shackman, A. J., Maxwell, J. S., McMenamin, B. W., Greischar, L. L. & Davidson, R. 415–426 (2014). J. Stress potentiates early and attenuates late stages of visual processing. J. 32. Morawetz, C., Bode, S., Derntl, B. & Heekeren, H. R. The effect of strategies, goals Neurosci. 31, 1156–1161 (2011). and stimulus material on the neural mechanisms of emotion regulation: a 57. Sabatinelli, D. et al. Emotional perception: meta-analyses of face and natural meta-analysis of fMRI studies. Neurosci. Biobehav. Rev. 72, 111–128 (2017). scene processing. NeuroImage 54, 2524–2533 (2011). 33. Song, X. W. et al. REST: a Toolkit for resting-state functional magnetic reso- 58. Sinke, C. B. A., van den Stock, J., Goebel, R. & de Gelder, B. The constructive nance imaging data processing. PLoS ONE 6, e25031 (2011). nature of affective vision: seeing fearful scenes activates extrastriate body area. 34. Pierpaoli, C. & Walker, L. TORTOISE: an integrated software package for pro- PLoS ONE. 7, e38118 (2012). cessing of diffusion MRI data. Process Diffus. 51, 2010 (2010). 59. Buschschulte, A. et al. Reward- and attention-related biasing of sensory 35. Taylor, P. A., Chen, G., Cox, R. W. & Saad, Z. S. Open environment for multi- selection in visual cortex. J. Cogn. Neurosci. 26, 1049–1065 (2014). modal interactive connectivity visualization and analysis. Brain Connect. 6, 60. Pourtois, G., Schettino, A. & Vuilleumier, P. Brain mechanisms for emotional 109–121 (2016). influences on perception and attention: What is magic and what is not. Biol. 36. Warbrick, T., Rosenberg, J. & Shah, N. J. The relationship between BOLD fMRI Psychol. 92, 492–512 (2013). response and the underlying white matter as measured by fractional aniso- 61. Lyu, Z., Zheng, P. & Jackson, T. Attention disengagement difficulties among tropy (FA): a systematic review. NeuroImage 153, 369–381 (2017). average weight women who binge eat. Eur. Eat. Disord. Rev. 24, 286–293 37. Taylor, P. A. & Saad, Z. S. FATCAT: (an efficient) functional and tractographic (2016). connectivity analysis toolbox. Brain Connect. 3, 523–535 (2013). 62. Bang, L., Rø, Ø. & Endestad, T. Threat-detection and attentional bias to threat in 38. Von Der Heide, R. J., Skipper, L. M., Klobusicky, E. & Olson, I. R. Dissecting the women recovered from anorexia nervosa: neural alterations in extrastriate and uncinate fasciculus: disorders, controversies and a hypothesis. Brain 136, medial prefrontal cortices. Eur. Eat. Disord. Rev. 25, 80–88 (2017). 1692–1707 (2013). 63. Contreras-Rodríguez, O. et al. Visceral adiposity and insular networks: asso- 39. Hiser, J. & Koenigs, M. The multifaceted role of the ventromedial prefrontal ciations with food craving. Int. J. Obes. 43, 503 (2019). cortex in emotion, decision making, social cognition, and psychopathology. 64. Graziano, P. A., Calkins, S. D. & Keane, S. P. Toddler self-regulation skills predict Biol. Psychiatry 83, 638–647 (2018). risk for pediatric obesity. Int. J. Obes. 34, 633–641 (2010). 40. Paret, C. et al. Monitoring and control of amygdala neurofeedback involves 65. Nolen-Hoeksema, S. Emotion regulation and psychopathology: the role of distributed information processing in the human brain. Hum. Brain Mapp. 39, gender. Annu. Rev. Clin. Psychol. 8, 161–187 (2012). 3018–3031 (2018). 66. Val-Laillet, D. et al. Neuroimaging and neuromodulation approaches to study 41. Wilcox, C. E., Pommy, J. M. & Adinoff, B. Neural circuitry of impaired emotion eating behavior and prevent and treat eating disorders and obesity. Neuro- regulation in substance use disorders. Am. J. Psychiatry 173, 344–361 (2016). Image Clin. 8, 1–31 (2015).