Brain structural changes in regions within the salience network in patients with functional constipation

Functional constipation (FCon) is one of the common functional gastrointestinal disorders (FGID). Previous studies reported alterations in cortical morphometry as well as changes in white matter (WM) fiber tracts and thalamo-limbic/parietal structural connectivity (SC). However, whether patients with FCon are implicated in changes in gray matter (GM) volume and associated SC remains unclear. Voxel-based morphometry (VBM) was first employed to examine differences in GM volume between 48 patients with FCon and 52 healthy controls (HC). Diffusion tensor imaging (DTI) with probabilistic tractography analysis was then employed to explore alterations in SC of these regions. Results showed abdominal symptoms were positively correlated with anxiety (SAS). Two sample t-test showed patients with FCon had decreased GM volumes in the right middle frontal gyrus (MFG_R), left insula (INS_L), and anterior cingulate cortex (ACC, PFWE < 0.05) which were negatively correlated with abdominal symptoms and difficulty of defecation respectively. Seed-based SC analysis showed patients with FCon had decreased fractional anisotropy of the ACC-right MFG and bilateral INS-MFG tracts. These findings reflect FCon is associated with changes in GM volumes and corresponding SC in brain regions within the salience network.

In recent years, neuroimaging has been gradually employed to assess brain structural and functional changes in patients with FCon (Ma et al. 2021;Liu et al., 2021;Hu et al., 2020;Li et al., 2021;Zhu et al., 2016;Zhang et al., 2020;Jin et al., 2019). Previous resting-state functional magnetic resonance imaging (RS-fMRI) report showed that FCon related brain functional abnormalities were in areas associated with emotional modulation including anterior insula-aINS, hippocampal gyrus-HIPP, orbital frontal cortex-OFC, and dorsal anterior cingulate cortex-dACC (Zhu et al., 2016). Further Granger causality analysis indicated that a stronger effective connectivity from the dACC and HIPP to the aINS and HIPP to induce abnormal emotional processing regulating visceral responses; and weaker effective connectivity from the supplementary motor area (SMA) and precentral (PreCen) to aINS and HIPP, reflecting abnormalities in sensory and behavioral responses (Zhu et al., 2016). Another RS-fMRI report was designed to examine gender-related differences in resting-state brain activity and functional connectivity in FCon patients, and results showed lower basal brain activities (i.e., insula-INS, PreCen, thalamus, and OFC), and decreased INS-OFC connectivity in female FCon patients than in males and its negative correlation with anxiety in women and abdominal distension in men (Jin et al., 2019). One study combining graph theory with RS-fMRI showed that FCon had small-worldness and a low normalized clustering coefficient, indicating reduced brain functional connectivity, which were predominately related to the rostral anterior cingulate cortex (emotional processing), SMA (motor-control), and thalamus (somatic/sensory) (Liu et al., 2021). One newly published study reported FCon showed decreased activity in fractional amplitude of lowfrequency fluctuation in the perigenual anterior cingulate cortex, dorsomedial prefrontal cortex and precuneus and enhanced PreCen-thalamus connectivity and attenuated precuneus-thalamus connectivity (Li et al., 2021). Another published study about alterations of white matter network reported patients with FCon showed increased nodal characteristics in the right middle frontal gyrus (MFG), left superior frontal gyrus, right anterior cingulate and paracingulate and decreased nodal characteristics in the left caudate and left thalamus (Ma et al. 2021). Those aforementioned fMRI studies highlight that FCon is mainly associated with dysfunction of emotional processing implicated in salience network (SN) including ACC and INS.
Growing evidence have showed that brain functional abnormalities might be associated with structural changes (Hao et al., 2013;Hubbard et al., 2016;Wang et al., 2020). Previous MRI study reported cortical morphometric alterations in brain areas involved with emotional processing (MFG, dACC, and OFC), self-referential processing (precuneus/posterior cingulate cortex), and somatic/motorcontrol (SMA) in FCon group . Cortical morphological abnormalities may be related to changes in WM microstructure (Tamnes et al., 2010). Another recent study showed that FCon was related to changes in thalamolimbic/parietal structural connectivity (SC), highlighting the integration of the thalamus in the brain structural network . However, few studies have been performed to examine the alterations in gray matter (GM) volume and associated SC in patients with FCon.
In addition, one structural MRI study on irritable bowel syndrome (IBS), a typical subtype of FGID, displayed significant alterations in GM volumes in the MFG, anterior cingulate cortex (ACC), INS, and HIPP in patients with IBS (Zhao et al., 2018). Diffusion tensor imaging (DTI) also detected abnormalities in SC between INS and ACC in IBS patients (Chen et al., 2011). Female IBS patients had decreased GM volumes in the INS, MFG, ACC, and HIPP (Labus et al., 2014). These previous brain imaging studies on IBS consistently recorded changes in brain regions implicated in emotional processing and modulation, especially in the aINS and ACC, which are key nodes of the SN (Chang et al., 2013;Garcia-Larrea, 2012;Mayer et al., 2015). The SN, which includes INS, ACC, and MFG, is responsible for integrating all kinds of sensory information and directing specific brain regions to respond appropriately (Peters et al., 2016). However, whether FCon is implicated in structural abnormalities in the SN has not been investigated yet.
The current study was designed to explore the brain structural changes in FCon patients. Thus, voxel-based morphometry (VBM) was first employed to explore differences in GM volumes between 48 FCon patients and 52 healthy controls (HC), and regions showing significant differences were then selected as seed regions for probabilistic tractography to feature regional SC as quantified by fractional anisotropy (FA), mean-(MD), axial-(AD), and radial diffusivity (RD). The study aimed to test whether: (a) FCon is associated with decreased GM volumes in brain regions involved with interoceptive and emotional processing, particularly the SN. (b) FCon is associated with alterations in SC of brain regions within the SN.

Participants
The experimental protocol was approved by the Institutional Review Board of Xijing Hospital and registered in the Chinese Clinical Trial Registry Center as: ChiCTR-OOB-15006347 (http:// www. chictr. org. cn). The experiments were conducted in accordance with the Declaration of Helsinki. Fifty-four patients with FCon were recruited from a clinical site at Xijing Hospital affiliated with the Air Force Medical University in Xi'an, China. Fifty-two HC subjects were recruited from the local community. All participants in the experiment were right-handed. They signed an informed consent form as well as performed a full physical examination with history. FCon was diagnosed using Rome IV criteria through a gastroenterologist experienced in FGID diagnosis (Drossman, 2016). FCon subjects with various main bowel habits such as functional defecatory disorders, slow transit constipation, and a combination of the two types were included in the experiment and constipation after childbirth, congenital giant colon/redundant sigmoid colon/ pelvic floor muscle relaxation, psychiatric/medical/neurological disorders demanding instant treatment, and existing medicine taken that might influence the central nervous system were removed. Four females and two males with dental implants were excluded from the experiment. Then, the study included fifty-two HC subjects and forty-eight patients with FCon. All subjects were demanded to finish several self-administered questionnaires, such as difficulty of defecation and Patient Assessment of Constipation Symptoms Questionnaire which consists of abdominal symptoms, rectal symptoms, and stool symptoms (Frank et al., 1999). All subjects finished the ZUNG self-rating anxiety scale (SAS) (Zung, 1971) and ZUNG self-rating depressive scale (SDS) (Zung, 1965) to estimate the severity of their anxiety/depression (Table 1). The study excluded all subjects with anxiety/ depression.

MRI acquisition
MRI scans were implemented between 9 A.M. and 10 A.M. to guarantee uniformity of judgments and to minimize diurnal variation. The experiment of the study was based on a 1.5 T Signa HDXT scanner (GE, Milwaukee, WI, USA). Firstly, the high-resolution structural images of all subjects were obtained by three-dimensional magnetizationprepared fast acquisition gradient-echo sequences with a voxel size of 1 mm 3 as well as an axial fast-spoiled-gradient-echo sequence (matrix size = 256 × 256, TR = 9.1 ms, TE = 3.0 ms, slice thickness = 1 mm as well as 248 slices, and field-of-view = 512 × 512 mm 2 ). Secondly, diffusionweighted (DW) images (single shot spin echo planar imaging sequence) with 1 non-diffusion weighting (b = 0 s/ mm 2 ) and 25 non-collinear diffusion sensitizing directions (b = 1000 s/mm 2 ) were obtained, and the parameters were: matrix size = 128 × 128, TR = 8000 ms, TE = 98.3 ms, fieldof-view = 256 × 256 mm 2 , 30 continuous axial slices, and slice thickness = 5 mm.

Statistical parametric mapping (identification of region of interest)
Voxel-wise analysis on GM brain mappings was implemented using SPM 12. Between HC and FCon groups, they were examined by a two-sample t-test with age/gender regressed out as covariates. Prominent GM volumes were chosen as regions of interest (ROIs) after whole brain family wise error (FWE) correction (P < 0.05) at the minimum cluster size of 30 voxel level. Contralateral brain areas were also identified as ROIs. Five ROIs were selected for fiber tracking using Subcortical Structural probabilistic atlases (Desikan et al., 2006)

DW-MRI tractography analysis
FMRIB's Diffusion Toolbox (FDT), which is embedded in FMRIB Software Library (FSL) version 5.0.9 software package (http:// www. fmrib. ox. ac. uk/ fsl/), was employed to pretreatment DTI information as well as implement  (Smith, 2002) was employed to eliminate non-brain tissue. Eddy-current correction and head motion were conducted by FDT in the FSL. Secondly, Diffusion Tensor Imaging Fit (DTIFIT) in FDT was employed to fit a diffusion tensor model at each voxel of the pretreated eddy current corrected diffusion-weighted data and to obtain the FA, MD, AD, and RD maps. Then, Bayesian Estimation of Diffusion Parameters Obtained using Sampling Technique was used to evaluate the fiber direction distribution at every brain voxel (Hirsiger et al., 2016). Probabilistic fiber tractography was calculated for every participants by probtrackx2 to produce the most likely connectivity distribution between ROI pairs. Tractography parameters were: 5000 samples per seed voxel, 0.2 curvature threshold, subsidiary fiber volume threshold of 0.01, step length of 0.5 mm, and a maximum number of steps 2000. Final-tract mask was obtained by overlapping the i-tract and j-tract, which respectively represent a seed mask and a target mask. The mean value of FA/MD/AD/RD within the final-tract mask connecting each ROI pair was calculated (Hirsiger et al., 2016). Participants who failed tractography/ ROI conversion were removed from the data analysis (resulting in 2 HC subjects and 3 patients with FCon exclusions) (Hirsiger et al., 2016).
Diffusion is usually measured with FA, MD, AD, and RD. FA is a commonly used method to measure anisotropy, which is sensitive to WM microstructure changes (Basser, 1995;Madden et al., 2012). Decreased FA may indicate disordered or damaged WM and fiber structure caused by demyelization or axonal loss. AD and RD are specific to the type of changes: axial and redial (Alexander et al., 2007;Thomason & Thompson, 2011). AD is sensitive to axons and changes in AD can be caused by decreased axonal branching or damage (Budde et al., 2009). RD is sensitive to myelin and demyelination or myelination may lead to changes in RD (Song et al., 2002). MD is the average rate of water diffusion independent of direction, which decreases with heightened myelination (Madden et al., 2009). MD values increased when water diffusion is not restricted by fibers, reflecting WM metabolism at the cellular level (Chanraud et al., 2010).

Statistical analysis
Demographic and clinical variables were analyzed by SPSS Statistics (Version 22, IBM). Chi-square test was employed to evaluate differences in gender distribution. Two sample t-test was employed to examine the discrepancies in behavioral measurements. Diffusion parameters were employed to assess the degree of connectivity for each pair of ROI by regressing out age and gender (Bonferroni correction, P < 0.0005 (0.05/108)).

Association between behaviors and GM/DTI
Partial correlation analysis with age and gender as covariates was conducted to evaluate the correlations among GM volumes, diffusion parameters, and clinical variables. Bonferroni correction was used for multiple comparisons, and significance level was set at P < 0.0028 (0.05/18) between GM volumes/DTI parameters and clinical variables.

Alterations in GM volumes
FCon group showed significant decreases in GM volumes in INS_L, ACC, and MFG_R than HC (P FWE < 0.05, Fig. 1A, B).
GM volume in the INS_L was negatively correlated with abdominal symptoms (r = -0.51, P = 0.0003), and GM volume in the ACC was negatively correlated with difficulty of defecation in the FCon group (r = -0.46, P = 0.0011, Fig.1C).

Alterations in SC
FCon group showed significant decreased FA of the right MFG-ACC, left INS-left MFG, and right INS-right MFG tracts than HC (Bonferroni correction, P < 0.0005, Fig. 2 and Table 2).

Discussion
VBM and DTI with probabilistic tractography analysis were employed to explore differences in brain GM volume and SC between FCon and HC groups. FCon group showed abdominal symptoms were positively correlated with SAS and significant decreases in GM volumes in the INS, ACC, and MFG involved in interoceptive and emotional processing. GM volume in the INS was negatively correlated with abdominal symptoms, and GM volume in the ACC was negatively correlated with difficulty of defecation in the

Alterations in GM volume within SN
Results showed FCon group had decreased GM volumes in the INS, ACC, and MFG, which are key nodes of the SN (Peters et al., 2016). SN is essential for the integration of visceral and sensory information (Peters et al., 2016;Uddin, 2015). Patients with FCon had decreased GM volume in the INS, which is consistent with changes in patients with IBS (Labus et al., 2014;Zhao et al., 2018). The INS is responsible for interoceptive information processing related to the organism, and plays a crucial role in integrating visceral sensation, emotional processing, and high-grade cognition (Craig, 2002;Critchley et al., 2004;Garcia-Larrea, 2012;Mayer et al., 2009;Starr et al., 2009;Uddin et al., 2017). A previous study reported that electrocortical stimulation of the INS caused visceral sensory responses, including abdominal sensation, stomach vibration, and vomiting (Stephani et al., 2011). Reduced GM volume in INS in FCon group might disrupt the internal information perception. Data showed the negative association between abdominal symptoms and GM volume in the INS in FCon group, and abdominal symptoms included pain in your stomach, stomach cramps, and bloating in your stomach, reflecting the role of the INS in sensing visceral pain (Keszthelyi et al., 2012). One RS-fMRI reported that FCon showed abnormal baseline brain activities in aINS and amplitude of low frequency fluctuation (ALFF) in aINS was negatively correlated with difficulty of defecation (Zhu et al., 2016). Another RS-fMRI study reported that FCon showed gender effects on baseline brain activity in the INS, and that ALFF in the INS was positively correlated with abdominal pain (Jin et al., 2019). A previous IBS study also reported that decreased excitatory neurotransmitter grades in aINS were related to abdominal pain (Bednarska et al., 2019). These previous studies indicate that INS is related to interoceptive processing of visceral sensation including abdominal pain and/or distension, and decreased GM volume in the INS might account for its functional abnormality reported in patients with IBS and FCon.
FCon group had also decreased GM volumes in areas related to emotional processing (ACC, MFG). The ACC is primarily concerned with executive-control of internal/external stimuli, affective responses to visceral sensation, and modulation of emotional responses (George et al., 2000). The MFG is the additional cortical node of the SN (Peters et al., 2016), which is critically implicated in visceral sensory information, emotional processing, and cognitive-control Liu et al., 2021). One study reported that FCon displayed reduced cortical thickness in the MFG and dACC, and duration of constipation was negatively correlated with cortical thickness in the ACC . FCon also showed abnormal baseline brain activities in the ACC (Zhu et al., 2016). One MRI study on IBS revealed abnormalities in the ACC during rectal stimulation (Wang et al., 2017). MFG was related to attention control and chronic pain modulation (Lorenz et al., 2003). Data showed that there was no significant association between the MFG and constipation symptoms, which might be related to the small sample size (controlling for age, gender, SAS, and SDS). However, in FCon group, GM volume in ACC had a negative association with difficulty of defecation. These findings suggested that constipation symptoms such as difficulty of defecation might not only be associated with abnormalities in function and cortical morphology, but also with changes in GM volumes in regions involved with emotional processing, highlighting the critical role of the ACC in FCon.

Alterations in SC within SN
Previous studies reported abnormal brain activation in the INS, ACC, MFG, thalamus, OFC, HIPP, and PreCen during visceral stimulation in patients with FCon Jin et al., 2019;Liu et al., 2021;Zhang et al., 2020;Zhu et al., 2016). One recent study using DTI with probabilistic tractography reported FCon was implicated in changes in the SC of thalamo-limbic/parietal circuits ). In the current study, alterations in the SC within SN were assessed by FA/MD/AD/RD, indicating that WM fiber tracts make a valuable contribution towards patients with FCon (Ellingson et al., 2013;Fang et al., 2017).
Results showed significant decreased FA of the ACCright MFG. Previous fMRI studies reported that the ACC and MFG were implicated in emotional regulation processing (Beauregard et al., 2001;Hu et al., 2020;Zhu et al., 2016). One cortical morphometric study reported that FCon displayed reduced cortical thickness in the dACC and MFG . Another study combining DTI and RS-fMRI reported that patients with borderline (Koppen et al., 2015) personality disorder showed abnormal SC and functional connectivity between the ACC and MFG (Lei et al., 2019). Thus, our data indicated that abnormal SC between the ACC and MFG might be associated with dysfunction in emotional regulation processing in patients with FCon.
The current study also showed significant decreased FA of the bilateral INS-MFG tracts. A previous RS-MRI report revealed that FCon patients showed increased ALFF in the aINS (Zhu et al., 2016). Another study reported FCon displayed reduced cortical thickness in the MFG . These findings suggested that abnormal SC between the INS and MFG is associated with emotional processing of visceral sensations.

Conclusion
The study was designed to examine brain GM volume and corresponding SC changes in FCon patients. Results showed abdominal symptoms were positively correlated with anxiety (SAS) and FCon group had lower GM volumes in the ACC, MFG, and INS than HC. GM volume in the INS was negatively correlated with abdominal symptoms, and GM volume in the ACC was negatively correlated with difficulty of defecation. Results also showed significant decreased FA of the ACC-right MFG and bilateral INS-MFG tracts in FCon group. These discoveries reflect that FCon is involved in changes in GM volumes and corresponding SC in brain regions within the salience network.