Author’s Accepted Manuscript Altered Frontal Inter-Hemispheric Resting State Functional Connectivity Is Associated With Bulimic Symptoms Among Restrained Eaters Shuaiyu Chen, Debo dong, Todd Jackson, Yanhua Su, Hong Chen www.elsevier.com/locate/neuropsychologia

PII: DOI: Reference:

S0028-3932(15)30083-X http://dx.doi.org/10.1016/j.neuropsychologia.2015.06.036 NSY5650

To appear in: Neuropsychologia Received date: 8 February 2015 Revised date: 22 June 2015 Accepted date: 27 June 2015 Cite this article as: Shuaiyu Chen, Debo dong, Todd Jackson, Yanhua Su and Hong Chen, Altered Frontal Inter-Hemispheric Resting State Functional Connectivity Is Associated With Bulimic Symptoms Among Restrained Eaters, Neuropsychologia, http://dx.doi.org/10.1016/j.neuropsychologia.2015.06.036 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Resting State Connectivity and Bulimic Symptoms 1

Altered Frontal Inter-hemispheric Resting State Functional Connectivity is Associated with Bulimic Symptoms among Restrained Eaters Shuaiyu Chen*, Debo Dong*, Todd Jackson, Yanhua Su, Hong Chen Key Laboratory of Cognition and Personality (Ministry of Education), Southwest University, Chongqing 400715, China. School of Psychology, Southwest University, Chongqing 400715, China Corresponding Authors: Hong Chen, Prof. Key Laboratory of Cognition and Personality (Ministry of Education), Southwest University, Chongqing 400715, China. E-mail: [email protected] Tel.: + 86 23 6825 7975 Fax: +86 23 6836 3625 *

Note: Shuaiyu Chen and Debo Dong contributed equally to this work.

E-mail addresses: Shuaiyu Chen: [email protected] Debo Dong: [email protected] Hong Chen: [email protected] Todd Jackson: [email protected] Yanhua Su:[email protected]

Resting State Connectivity and Bulimic Symptoms 2

Abstract Theory and research have indicated that restrained eating (RE) increases risk for binge-eating and eating disorder symptoms. According to the goal conflict model, such risk may result from disrupted hedonic-feeding control and its interaction with reward-driven eating. However, RE-related alterations in functional interactions among associated underlying brain regions, especially between the cerebral hemispheres, have rarely been examined directly. Therefore, we investigated inter-hemispheric resting-state functional connectivity (RSFC) among female restrained eaters (REs) (n=23) and unrestrained eaters (UREs=24) following food deprivation as well as its relation to overall bulimia nervosa (BN) symptoms using voxel-mirrored homotopic connectivity (VMHC). Seed-based RSFC associated with areas exhibiting significant VMHC differences were also assessed. Compared to UREs, REs showed reduced VMHC in the dorsal-lateral prefrontal cortex (DLPFC), an area involved in inhibiting hedonic overeating. REs also displayed decreased RSFC between the right DLPFC and regions associated with reward estimation – the ventromedial prefrontal cortex (VMPFC) and posterior cingulate cortex (PCC). Finally, bulimic tendencies had a negative correlation with VMHC in the DLPFC and a positive correlation with functional connectivity (DLPFC and VMPFC) among REs but not UREs. Findings suggested that reduced inter-hemispheric functional connectivity in appetite inhibition regions and altered functional connectivity in reward related regions may help to explain why some REs fail to control hedonically-motivated feeding and experience higher associated levels of BN symptomatology.

Key words: restrained eating; resting-state fMRI; inhibitory control; reward; bulimia nervosa symptoms; VMHC.

Resting State Connectivity and Bulimic Symptoms 3

1. INTRODUCTION Restrained eaters (REs) pay excessive attention to their body weight and attempt to lose or maintain weight by stringently restricting caloric intake (Herman & Mack, 1975; Lindroos et al., 1997). In obesogenic environments, an increasing number of people use restrained eating (RE) as a weight loss strategy (Andreyeva et al., 2010), even though few ultimately maintain weight loss successfully. Furthermore, hedonic eating urges increase among REs exposed to food cues and often result in over-eating, binge-eating and inappropriate compensatory behavior. (Fedoroff et al., 2003; Harris et al., 2009). Some theory and research (Fairburn & Brownell, 2005; Polivy & Herman, 1985), has indicated that chronic hunger among REs results in an over-reliance on cognitive resources to control eating and increases risk for future onset of binge-eating within settings that have abundant external food cues and readily available food (Lowe, 2002; Stice & Agras, 1998). Hence, RE is a potentially important influence on the development of binge eating disorder (BED) and bulimia nervosa (BN). To date, however, few researchers have examined specific neural mechanisms related to RE. For example, neuroimaging studies have not directly evaluated neural correlates of RE and overall BN symptom levels; doing so may clarify why RE contributes to binge-eating and inappropriate compensatory behaviors that can arise upon encountering tempting foods (Allison, 2009; Stice, 2002; Stroebe et al., 2013). Although an increasing number of people engage in chronical dieting (Andreyeva et al., 2010), REs have higher rates of disinhibited eating and are vulnerable to uncontrolled binge-eating in food-rich environments (Van Strien et al., 2005). According to a recent goal conflict model of eating (Stroebe et al., 2013), REs often fail to lose weight because caloric restriction corresponds to increased hedonic-eating reward motivation and reduced capacity to control weight. Supporting these assumptions, REs shown inhibitory control deficits and high impulsivity levels, characteristics that have been implicated in unhealthy eating (Jasinska et al., 2012; van Koningsbruggen et al., 2013). Furthermore, reduced spontaneous activity in the dorsal-lateral prefrontal cortex (DLPFC) corresponds to lowered inhibitory control among REs (Dong et al., 2014). Finally, REs may experience more pleasure from feeding because they are generally more sensitive to reinforcing stimuli (Ahern et al., 2010).

Resting State Connectivity and Bulimic Symptoms 4 In relation to dietary failures, the hedonic-inhibitory model of feeding posits that overconsumption of palatable food results from the disrupted balance between appetitive motivation mediated by the mesolimbic reward system and active inhibitory control mediated by the DLPFC, whereby appetitive motivation may over-ride inhibitory control (Appelhans, 2009). REs also display hyper-responsivity in reward-related regions (Burger & Stice, 2011; Coletta et al., 2009; Alice et al., 2014), especially after dietary violations (Demos et al., 2011). On this basis, high food reward sensitivity and low inhibitory control may interact as influences on disinhibited overeating and weight regain among REs (Appelhans, 2009; Appelhans et al., 2011; Nederkoorn et al., 2010). Despite evidence that RE is related to hyperactivity of reward and motivation circuitry and reduced inhibitory control, research assessing functional coordination between inhibitory control and reward systems among REs is sparse. In addition, substance abuse and non-substance addiction are closely associated with cognitive control deficits accompanied by altered prefrontal cortex (PFC) function (Noel et al., 2013; Volkow et al., 2008). For example, hypo-activation of frontal control regions has been found among adolescent girls with obesity (Batterink et al., 2010), in line with evidence that groups with drug addiction or obesity have altered circuits reflecting disrupted inhibitory control (Volkow et al., 2012). Some researchers have hypothesized frontal asymmetry might be a latent mechanism underlying cognitive control deficits (Kelly et al., 2011; Wang et al., 2013). For example, Silva et al. (2002) found that REs show more right-sided prefrontal asymmetry than UREs do (Silva et al., 2002). Nonetheless, it is still unclear which specific regions are related to such asymmetry and whether prefrontal asymmetry is linked to altered functional interactions, between reward and inhibitory control regions. Furthermore, neuroimaging studies have not been undertaken to assess relations between prefrontal asymmetry and BN tendencies among REs, despite robust behavioral evidence for links between RE and BN symptomatology. Instead, task-based fMRI studies have been the main thrust of RE neuroimaging studies (Burger & Stice, 2011; Coletta et al., 2009; Schur et al., 2012; van der Laan et al., 2014). Given that intrinsic resting-state activity may consume 95% of the brain’s energy (Raichle & Mintun, 2006), resting-state fMRI (RS-fMRI) is a promising avenue for elucidating patterns of

Resting State Connectivity and Bulimic Symptoms 5 coherent spontaneous fMRI signal fluctuations (Biswal et al., 1995). Recently, neuroscientists have proposed voxel-mirrored homotopic connectivity (VMHC) as an effective way to examine inter-hemispheric functional interactions (Anderson et al., 2011; Wang et al., 2014; Zuo et al., 2010). VMHC involves the assessment of correlations of the time series for each voxel in one hemisphere with its mirrored counterpart in the other hemisphere and provides a reliable approach to exploring prefrontal cognitive control circuitry in REs (Kelly et al., 2011; Wang et al., 2013; Zuo et al., 2010). By extension, VMHC may be a useful in examining the relations between prefrontal asymmetry and BN tendencies of REs. Self-imposed food deprivation (i.e.,reduced energy intake) is one of the most common strategies for REs to prevent weight gain or achieve weight reduction (Jeffery et al., 2000). Unfortunately, prolonged dieting increases susceptibility to external food cues and disinhibited eating (Mills & Palandra, 2008; Westenhoefer et al., 1994). Food deprivation has a negative impact on cognitive processing of food cues (Benau et al., 2014). For example, skipping breakfast increases activity in reward value and motivational regions of the brain (Goldstone et al., 2009; Leidy et al., 2011) and predicts poor long-term success among caloric restrictors (Dansinger et al., 2007). Arguably, given that skipping meals (e.g., breakfast) is as common weight control strategy among REs, assessing spontaneous brain activity such conditions may be an ecologically-valid approach to studying the issue in this group. Based on this overview and research implicating the DLPFC in eating control and successful dieting (DelParigi, 2010; DelParigi et al., 2007; Schonberg et al., 2014), we hypothesized that VMHC might indicate REs have relatively impaired inter-hemispheric functional interactions in the DLPFC. Furthermore, we reasoned that interactions between altered inhibitory control (DLPFC) and reward valued-computational regions (e.g., VMPFC) can be examined via resting-state functional connectivity (RSFC) (Grabenhorst & Rolls, 2011). Hence, based on goal conflict and hedonic-inhibitory models of eating, REs were expected to show comparatively reduced functional connectivity between the DLPFC and reward-based value regions (e.g., VMPFC). Finally, in light of links between RE and BN symptomatology, we hypothesized that areas having lower inter-hemispheric functional connectivity and altered functional connectivity, especially within inhibitory control regions

Resting State Connectivity and Bulimic Symptoms 6 (i.e., DLPFC) would correspond to overall BN symptom elevations among REs relative to UREs.

2. Methods and Materials 2.1 Participants Participants were 47 undergraduate women from Southwest University (SWU), Chongqing, who comprised URE (n = 24) and RE (n = 23) subgroups. Participants were selected from Restraint Scale (RS) scores in a prescreening procedure administered to 107 volunteers two weeks before the imaging study. Only women were recruited because men and women differ in how and why they gain and lose their weight (Holm-Denoma et al., 2008), and average weight women are much more likely to diet than are average weight men (van Strien et al., 2014). RE and URE group membership was based on RS scores above 15 or below 12, respectively, following other published accounts (Demos et al., 2011; Dong et al., 2014). Open-ended queries assessed exclusion criteria including current neurological disease (i.e., central and peripheral nervous system diseases such as epilepsy, migraine and other headache disorders, multiple sclerosis or brain trauma), and a formal eating disorder diagnosis or a history of such concerns. Obese participants (BMI ≥ 30 kg/m2) were also excluded because of potential neuro-anatomical differences based on BMI (Gunstad et al., 2008). Smokers, alcohol users and women taking medication known to affect fMRI signals were also excluded.

2.2 Procedure The research was approved by the SWU human research ethics committee. All participants provided written, informed consent prior to participation. Eligible women were instructed to fast overnight following their evening meal and to refrain from eating and drinking any liquids except water before their scan. All scans were conducted between 9:30-11:30. Upon arrival, fasting status and hunger were confirmed by self-reports. During the scan, participants were instructed to keep their eyes closed, not to think of anything in particular and not to fall asleep. Participants completed a demographics questionnaire following the scan, as well as measures of BN symptoms and mood. This sequence was followed because RS-fMRI signals can be modulated by antecedent events (Albert et al., 2009). After finishing all tasks, each volunteer received 60 Yuan

Resting State Connectivity and Bulimic Symptoms 7 (equivalent to about $10 USD) for participation.

2.3 Measures Demographics items included age, height, weight, smoking status, alcohol using, current medications, and phase of menstrual cycle. Dieting and weight history items included “What is the most you have ever weighed since reaching your current height?” and “In your lifetime, how many times have you lost 5 kilograms or more?” as well as queries about having been diagnosed with a clinical eating disorder (Coletta et al., 2009). Participants also completed self-report measures of affect, restrained eating, and bulimic symptomatology that had been back-translated and validated for use in Chinese samples. Hunger ratings. Participants were asked to rate the current feelings of hunger on 100-mm visual analogue scale (VAS), ranging from 0 (‘not at all hungry’) to 100 (‘very hungry’). Body Mass Index. BMI (kg/m2) was calculated from objective measures of height and weight. Specifically, after the removal of shoes and coats, height was measured to the nearest millimeter using a stadiometer and weight was assessed to nearest 0.1 kg using a digital scale. Restraint Scale (RS; Herman & Polivy, 1980). The ten-item Revised RS measures degree of eating restraint and reflects chronic concerns with weight and food intake restriction as a means of weight control. RS items were rated on four- (0 to 3) and 5-point (0 to 4) frequency scales with higher ratings reflecting more RE (Polivy & Herman, 1999). In Chinese college student samples, support has been found for the original RS factor structure, internal consistency and convergent validity (Kong et al., 2013). In this study, alpha coefficients were α = 0.81 for REs and α = 0.78 for UREs. Positive and Negative Affect Scale - Chinese (Jackson & Chen, 2008). The Positive Affect

and Negative Affect Schedule (PANAS) assesses 20 affective states (e.g., distressed, excited) experienced in the last month from 1 = never or little of the time to 4 = most of the time. The original PANAS factor structure (Watson et al., 1988) was replicated in Chinese samples (Jackson & Chen, 2008), except that “Alert” loaded with the 10 original NA items, not Positive Affect. Studies of Chinese males and females reported internal consistency coefficients above  = .80 and high levels of convergent validity with related constructs (e.g., Jackson & Chen, 2008, 2010, 2014). For REs, alpha coefficients were α = 0.92 for PA and, α = 0.91 for NA. Among UREs, alphas were α = 0.86 and α = 0.92 for PA

Resting State Connectivity and Bulimic Symptoms 8 and NA, respectively. Eating Disorder Inventory-Bulimia (EDI-B; Garner et al., 1983). This 7 item scale assess BN symptoms including binge eating episodes accompanied by a perceived loss of control and inappropriate compensatory behaviors (i.e., laxatives, diet pills, excessive exercise) between 1

(never) and 6 (always), with higher scores reflecting more overall BN symptomatology. Previous research with Chinese samples replicated the original factor structure and reported the scale has adequate reliability and construct validity (Lee et al., 1997). In this study, alphas were α = 0.82 for REs and α = 0.84 for UREs.

2.4 MRI data acquisition RS-fMRI data were acquired with a 3.0 T Siemens Tim Trio whole-body MRI system (Siemens Medical Solutions, Erlangen, Germany). Foam pads were used to reduce head movements and scanner noise. To optimize functional sensitivity in the VMPFC, a key region of interest, images were acquired using an oblique orientation of 30° to the anterior commissure–posterior commissure line (Hutcherson et al., 2012). In addition, we used an eight-channel phased array coil which yields a 40% increase in VMPFC signals over a standard head coil. Scans were performed with an echo-planar imaging (EPI) sequence using the following parameters: repetition time = 2000 ms, echo time = 30 ms, flip angle = 90°, field of view = 192 × 192 mm2, acquisition matrix = 64 × 64, in-plane resolution = 3×3 mm2, 32 interleaved 3 mm-thick slices, inter-slice skip = 0.99 mm. For each participant, 242 EPI functional volumes were collected. Subsequently, high-resolution T1-weighted anatomical images were obtained with the application of a magnetization-prepared rapid gradient echo (MPRAGE) sequence (repetition time (TR) = 1900 ms; echo time (TE) = 2.52 ms; inversion time (TI) = 900 ms; flip angle = 9 degrees; resolution matrix = 256 × 256; slices = 176; thickness = 1.0 mm; voxel size = 1 × 1 × 1 mm).

2.5 Data preprocessing Data preprocessing was conducted using the Data Processing Assistant for Resting-State fMRI V2.0 (DPARSF, http://www.restfmri.net/forum/DPARSF) (Yan & Zang, 2010). Following other published accounts (Tian et al., 2012), preprocessing included discarding the first ten volumes, slice timing, head motion correction, spatial normalization in Montreal Neurological Institute (MNI) space using the transformation parameters estimated using

Resting State Connectivity and Bulimic Symptoms 9 unified segmentation algorithm and a re-sampled voxel size of 3×3×3 mm3, smoothing with a Gaussian kernel of 6 mm at full-width at half-maximum (FWHM), linear trend removal, and filtering (e.g., 0.01–0.08 Hz). Finally, 6 head motion parameters, white matter, CSF and whole brain signals were regressed out before VMHC and FC computation.

2.6 Voxel-Mirrored Homotopic Connectivity To account for geometric differences between hemispheres, we refined the registration from individual anatomic to Montreal Neurological Institute (MNI) 152 template space using a group-specific symmetrical template. All 47 registered structural images were averaged to create a mean image, which was then averaged with its left–right mirror to generate a group-specific symmetrical template. Nonlinear registration to this symmetrical template was performed for each participant, and the resultant transformation was applied to each participant’s preprocessed functional data based on Tian, et al. (2012). VMHC computation was performed using DPARSF software. For each woman, homotopic RSFC was computed as the Fisher Z-transformed Pearson correlation between the time series of every pair of symmetrical inter-hemispheric voxels. Resultant values constituted the VMHC and were used for analyses. To test for regional group differences in VMHC, individual-level VHMC maps were entered into a group-level voxel-wise t test. The T-map was set at a corrected significance level of p < 0.05 (combined individual voxel height threshold of p < 0.005 and a cluster size > 1215 mm3 (45 voxels)), using AlphaSim in REST software (REST, www.restfmri.net) (Song et al., 2011). The following parameters were used: individual voxel p-value = 0.001, 10,000 simulations, 2-sided, FWHM estimated by 3dFWHMx, with a unilateral hemispheric brain mask. Age, BMI, fasting time, hunger level and menstrual cycle phase (Table 1) were treated as covariates given that each may induce potentially subtle effects on image data.

2.7 Structural Image Analysis To exclude the influence of structural damage to VMHC measures, possible abnormalities in gray-matter volume (GMV) of regions showing altered VMHC were assessed. Voxel-based morphometry (VBM) was performed to compute GMV of each participant using DPARSF software. Briefly, individual T1 images were co-registered to mean functional images after head-motion correction. Transformed images were segmented into gray matter, white matter,

Resting State Connectivity and Bulimic Symptoms 10 and cerebrospinal fluid and were then normalized to the MNI space using a unified segmentation algorithm. Next registered gray-matter images were modulated and smoothed with a 10-mm Gaussian kernel. We extracted mean GMV values in regions showing RE versus URE group differences in VMHC.

2.8 Seed-Based RSFC RSFC associated with areas exhibiting significantly different VMHC between groups was assessed by computing whole-brain voxel-wise correlations associated with mean time series derived separately for two regions of interest (ROIs), comprising all voxels within the left and right seed DLPFC exhibiting lower VMHC for REs than UREs. Fisher Z–transformed correlation maps were entered into a group-level voxel-wise t-test analysis. The T-map was set at a corrected significance level of p < 0.05 (combined individual voxel height threshold of p < 0.005 and a cluster size > 972 mm3 (36 voxels)), using AlphaSim in REST software (parameters were: individual voxel p-value = 0.01, 10,000 simulations, 2 sided, FWHM estimated by 3dFWHMx, with a whole brain mask including 70,831 re-sampled voxels). Covariates were identical to those used in the VMHC section.

2.9 Brain–Behavior Relationships Correlations between mean VMHC, RSFC of identified regions and EDI-B scores were computed within each group to evaluate relations between differences in spontaneous brain activity and bulimic tendencies. First, we explored relations between EDI-B scores and mean prefrontal inter-hemispheric RSFC within a 4-mm-radius sphere centered on the peak of the VMHC group difference, after regressing out covariates above from both RSFC data and EDI-B scores. Second, we examined cluster-wise association between EDI-B and mean RSFC within a 4-mm-radius sphere centered on the peak of FC group differences.

3. Results 3.1 Sample Characteristics REs and UREs did not differ on age, BMI, fasting time, menstrual cycle phase, or affect levels (Table 1). However, supporting the integrity of the samples, REs had significantly higher RS and EDI-B scores than UREs did.

3.2 Voxel-Mirrored Homotopic Connectivity RE group differences in VMHC, controlling for age, BMI, fasting time, menstrual cycle

Resting State Connectivity and Bulimic Symptoms 11 phase and affect were revealed, as shown in Fig. 1 and Table 2. Specifically, REs showed reduced VMHC in the DLPFC compared to UREs. However, no significant GMV changes were found among REs in regions showing altered VMHC (t = 1.14, df = 45, p = 0.26). Given controversy regarding treatment of the global signal as a covariate (e.g., Murphy et al., 2009), we recalculated the VMHC without controlling for the global signal. Previously significant VMHC differences in the DLPFC remained, although there was minimal change in the peak point (t = 62, df = 45, p < 0.005, Peak = (±18, 51, 24)).

3.3 Seed-Based RSFC Whole-brain RSFC associated with two ROIs (i.e., the right and left hemisphere DLPFC shown in Figure 1) having significant VMHC differences were also assessed. Consistent with VMHC results, REs exhibited lower RSFC between the right DLPFC seed and left DLPFC, VMPFC, and posterior cingulate cortex (PCC) as reported in Fig. 2 and Table 2. Conversely, group differences in the left DLPFC RSFC were not significant based on AlphaSim multiple correlations (with a combined individual voxel height threshold of p < 0.005 and a cluster size > 972 mm3). However, when the analysis was re-run without covarying the global signal RSFC results for each ROI did not attain statistical significance.

3.4 Brain–Behavior Relationships Correlations between mean VMHC, RSFC of identified regions, and EDI-B scores were computed within each group to evaluate relations between spontaneous brain activity differences and BN symptoms. Among the REs, a significant negative correlation between EDI-B scores and mean VMHC was found within a spherical 4-mm-radius ROI centered on the VMHC group difference peak (reported in Fig. 2, Table 2, r = -0.583, n = 23, p = 0.018); his relation suggested that weaker prefrontal inter-hemispheric RSFC was associated with higher BN symptom levels (Fig. 3). In contrast, corresponding correlations above were not significant among UREs (r = 0.053, n = 24, p= 0.80). Fig. 3 indicates that EDI-B scores also had a significant positive association (r = 0.658, n = 23, p = 0.006) with RSFC between the right DLPFC and left VMPFC (4-mm-radius, ROI) in the RE group but not the URE group (r = 0.276, n = 24, p = 0.202). To test group differences in associations between EDI-B scores and brain activity further,

Resting State Connectivity and Bulimic Symptoms 12 multiple regression analyses with “group”, “EDI-B scores” and “group ╳EDI-B scores” as predictors were performed using “Interaction!” (http://www.danielsoper.com/Interaction/). Figure4 illustrates the significant interaction effect for the correlation between EDI-B scores and VMHC in the DLPFC [F (43) =11.08, P 50, FDR corrected). (C) Seed-Based RSFC map of significant differences between REs and UREs (p< 0.05, corrected). Letters L and R correspond to left and right sides of brain, respectively. RSFC: resting-state function connectivity.

Figure 3. Scatterplots showed the correlation of VMHC and Seed-Based RSFC values with bulimic tendencies among REs group. The left plot illustrates a significant negative correlation between VMHC values in the DLPFC and the binge eating tendencies (EDI-B scores). The right plot illustrates a significant positive correlation of FC between the right DLPFC and the left VMPFC with bulimic tendencies (EDI-B scores). FC: functional connectivity.

Figure 4. Moderating effect of group on relations between EDI-B scores and brain activity.

Altered frontal inter-hemispheric resting state functional connectivity is associated with bulimic symptoms among restrained eaters.

Theory and research have indicated that restrained eating (RE) increases risk for binge-eating and eating disorder symptoms. According to the goal con...
952KB Sizes 0 Downloads 11 Views