Introduction

Inflammatory bowel diseases (IBD) are a group of chronic, non-specific intestinal diseases that mostly comprise Crohn’s disease and ulcerative colitis1. They typically occur alternately in episodes and remissions and are difficult to cure2,3. Patients with IBD may have a variety of negative emotional behaviors, primarily characterized by heightened physical discomfort, excessive concern with the possibility of disease recurrence, and apprehension over exacerbating their condition4, and experiences of anxiety and depression5. Lower-order pain-related psychological constructs (such as pain catastrophizing, pain-related anxiety, and pain-related fear) and higher-order personality traits (such as trait anxiety and depression) have been demonstrated to have different impacts on pain in healthy humans6. Despite the close association between negative feelings and repeated symptoms, few studies have considered the different impacts of pain-related negative emotions and trait negative emotions in patients with IBD.

The brain-gut axis is commonly utilized to describe the intricate interactions among neuroendocrine pathways, the peripheral, central, and autonomic neural systems, and the gastrointestinal tract7. These interconnected pathways might be bidirectional8. For example, there have been strong connections between gastrointestinal disorders and pain-related fear9, pain-related anxiety10, and pain catastrophizing11. Negative emotions, a higher-order trait associated with negative feelings, cognitions, and self-concept, are also linked to the intensity of symptoms in some chronically gastrointestinal conditions12,13. Neuroimaging techniques provide a valuable tool for investigating the maladaptive plastic changes of the human brain and their association with psychological factors in gastrointestinal disorders. For example, insights from resting-state functional MR imaging have observed abnormal activation of the anterior midcingulate cortex and altered functional connectivity strength in the right inferior frontal gyrus, associated with anxiety symptoms in patients with irritable bowel syndrome (IBS)14 and functional dyspepsia (FD)15, respectively. Additionally, several studies have also reported links between abnormal brain activity in the frontal middle regions, disrupted amygdala-thalamus functional connectivity, and elevated levels of anxiety16 and depression17 in patients with IBD. However, it is worth noting that there has been limited research examining the distinct impacts and associations of pain-related negative emotion and trait negative emotion on the brain in patients with IBD.

The current analysis of the association between the brain function, pain, and negative emotion is being conducted frequently using mass univariate analysis. Mass univariate analysis assumes correlations between different behavioral indexes and brain function, and ignoring the interrelationships and independent effects of behavioral indexes. Partial least squares correlation (PLSC), as a class of late variable (LV) algorithms to model associations between brain measurements and two or more behavioral measurements18, is a better suit for the interrelationships and independent effects of pain-related and trait negative emotions in patients with IBD19,20.

In the current study, we hypothesized that prolonged nociceptive afferent input from the gut to the brain could modify the brain’s resting function. Additionally, we anticipated that altered brain function would be associated with pain-related negative emotions, trait anxiety, and depression in distinct ways. To address these hypotheses, multiple aspects of blood oxygen level-dependent (BOLD) signals, including the amplitudes of low-frequency fluctuation (ALFF), regional homogeneity (ReHo), and degree centrality (DC), were used to identify the resting brain activity in healthy controls and patients with IBD, respectively. Several measurements for negative emotion were assessed, including pain-related fear, pain-related anxiety, pain catastrophizing, pain sensitivity, trait anxiety, and depression. PLS correlation was employed to assess the relationship among abnormal brain activities, pain-related negative emotions, and trait anxiety and depression.

Results

Clinical, demographic characteristics, and psychological results

A total of ninety-eight right-handed IBD individuals and forty-six healthy individuals were enrolled in the study. Sixteen IBD individuals and four healthy individuals were excluded for some reasons. The flow of participants enrollment was stated in Fig. 1. Descriptive statistics of clinical and demographic characteristics were summarized in Table 1. No significant differences between the IBD and HCs groups were observed in terms of age, gender, and years of education. Patients with IBD had a significantly lower level of GIQLI than HCs. In terms of neuropsychological tests (Table 2), patients with IBD had a significantly higher BDI score and STAI-T score than HC, but no significant difference between the two groups was found in FPQ, PASS, PCS, PSQ, and PVAQ. A heatmap (Fig. 2) was generated using the “heatmap” package in GraphPad Prism software (version 9.5) to provide a clearer representation of the gastrointestinal symptoms and psychological assessment of all the patients with IBD.

Fig. 1
figure 1

The flow diagram of study enrollment. HC healthy control, IBD inflammatory bowel diseases, MRI magnetic resonance imaging.

Full size image
Table 1 Demographic information and clinical characteristics.
Full size table
Table 2 Psychological assessment.
Full size table
Fig. 2
figure 2

The heatmap of gastrointestinal symptoms and psychological assessment of IBD patients. BDI beck depression inventory, FPQ fear of pain questionnaire, GIQLI gastrointestinal quality of life index, IBS-SSS irritable bowel syndrome severity scoring system, PASS Pain Anxiety Symptoms Scale, PCS pain catastrophizing scale, PSQ pain sensitivity questionnaire, PVAQ pain vigilance and awareness questionnaire, STAI-T Trait version of State-Trait Anxiety Inventory.

Full size image

Differences in ReHo, ALFF, and DC values

The IBD group exhibited significantly higher values of ALFF in the right anterior cingulate cortex (ACC) (MNI coordinates, x-y-z: 9, 39, 20) and lower values of ALFF in the left postcentral gyrus (MNI coordinates, x-y-z: -55, -9, 26) than the HCs group (FDR-corrected, p < 0.01, two-sample t test; Fig. 3-a). The IBD group showed significantly higher values of DC in the bilateral ACC (MNI coordinates, x-y-z: -5, 37, 1; 10, 39, 20) and lower values of DC in the left secondary visual cortex (MNI coordinates, x-y-z: -44, -68, 5) compared to the HCs group (FDR-corrected, p < 0.01, two-sample t test; Fig. 3-b). No significant statistical differences were observed in the ReHo between patients with IBD and HCs.

Fig. 3
figure 3

Significant differences in the amplitudes of low-frequency fluctuation (ALFF) and degree centrality (DC) values between the inflammatory bowel diseases (IBD) and healthy control (HC) groups. (a) Compared to the HC groups, patients with IBD exhibited higher values of ALFF in the right anterior cingulate cortex (ACC) and lower values ALFF in the left postcentral gyrus; (b) Compared to the HCs groups, patients with IBD showed higher values of DC in the bilateral ACC and lower values in the left secondary visual cortex. A two-sample t test was conducted to examine the whole-brain differences between groups in ALFF and DC maps. P < 0.05 was considered significant. False discovery rate (FDR) was used to correct for multiple comparisons.

Full size image

Multivariate PLSC analysis

Multivariate PLS correlation analysis was used to determine the association of gastrointestinal symptoms, pain-related negative emotion, and trait negative emotion with the values of ALFF and DC. Two LVs were significant using the 5,000-permutation bootstrap test. These were LV1 and LV2, which made up 59.57% and 14.43% of the cross-block covariance, respectively, and showed a significant contribution to the covariance. The dominant LV1 exhibited positive brain salience in the left postcentral gyrus, and negative brain salience in the left ACC (Fig. 4). The corresponding gastrointestinal symptoms and psychological assessment revealed more emphasized correlations of LV1 to scores of GLIGI and pain-related negative emotion. The brain scores of the left postcentral gyrus were positively correlated with GLIGI (r = 0.246, p = 0.015, Pearson correlation) (Fig. 4). The brain scores of the ACC were positively correlated with PASS (r = 0.270, p = 0.007, Pearson correlation), PCS (r = 0.287, p = 0.004, Pearson correlation), and PVAQ (r = 0.301, p = 0.003, Pearson correlation) (Fig. 4). The dominant LV2 exhibited negative brain salience in the left postcentral gyrus, was associated with the levels of GLIGI, IBS-SSS, and BDI (Fig. 5).

Fig. 4
figure 4

PLS correlation analysis results of LV 1. (a) Bar graph representing Pearson’s correlation coefficients between the behavioral variables and brain scores of LV 1 in the IBD group. (b) The dominant LV1 exhibited positive brain salience in the left postcentral gyrus, and negative brain salience in the left ACC. (c) The brain scores of the left postcentral gyrus were positively correlated with GLIGI, and the ACC were positively correlated with PASS, PCS, and PVAQ. We employed a BSR threshold of ± 3, which corresponds to a p-value of around 0.01. Pearson correlation was conducted. P < 0.05 was considered significant.

Full size image
Fig. 5
figure 5

PLS correlation analysis results of LV 2. (a) Bar graph representing Pearson’s correlation coefficients between the behavioral variables and brain scores of LV 2 in the IBD group. (b) The dominant LV2 exhibited negative brain salience in the left postcentral gyrus. We employed a BSR threshold of ± 3, which corresponds to a p-value of around 0.01. P < 0.05 was considered significant.

Full size image

Discussion

In the present study, we conducted resting-state fMRI to investigate the potential brain functional alterations between healthy controls and patients with IBD. Further, we employed PLS correlation analysis to explore the associations between functional alterations, pain-related negative emotion, and trait negative emotion in patients with IBD. The following results were mainly observed: Firstly, the patients with IBD displayed higher values of ALFF in the right ACC (MNI coordinates, x-y-z: 9, 39, 20), lower values of ALFF in the left postcentral gyrus (MNI coordinates, x-y-z: -55, -9, 26), higher values of DC in the bilateral ACC (MNI coordinates, x-y-z: -5, 37, 1; 10, 39, 20), and lower values of DC in the left secondary visual cortex (MNI coordinates, x-y-z: -44, -68, 5) compared to the HCs group. Secondly, PLS correlation analysis revealed ascending negative brain salience in the ACC was consistent with increasing levels of PASS, PCS, and PVAQ; the higher values of positive brain salience were associated with the levels of GLIGI in dominant LV1, and negative brain salience in the left postcentral gyrus was associated with the levels of GLIGI, IBS-SSS, and BDI in dominant LV2.

The study revealed that patients with IBD exhibited distinct changes in brain regions such as the ACC and postcentral gyrus. These regions are crucial components of the networks responsible for visceral sensory information and the emotions associated with pain. The ACC is a part of the limbic system, which is a region in the front part of the brain that is located above the corpus callosum21. It serves as a central hub that relays different input signals, analyzing requirements from other areas in order to direct adaptive responses21. Studies have found that patients with IBD have an aberrant brain metabolism in the ACC22. A recent animal study revealed the ACC interprets signals from visceral nociceptive afferents and exhibits long-term potentiation, which impacts excitatory synapses and is implicated in neural plasticity23. Previous human neuroimaging studies have shown that increased spontaneous brain activity in the ACC in patients with IBD22,24. The results of our study displayed higher values of ALFF and DC in patients with IBD during a resting state. These findings are in line with previous research and suggest that patients with long-term chronic inflammation had abnormal functional activities in the brain.

The postcentral gyrus, which is a key element of the sensorimotor network and includes the primary somatosensory (S1) cortex, plays a crucial role in the nociceptive pathway by receiving and processing afferent signals to encode the ___location and intensity of nociceptive stimuli25. Both animal model studies and human neuroimaging investigations have shown that the S1 may play a role in regulating nociceptive stimuli. For example, anatomical evidence from monkeys showed that nociceptive neurons in the thalamus connected to the S1 cortex26. Abnormal spontaneous brain activity in the postcentral gyrus has been observed in other recurrent experiences in gastrointestinal disorders, including IBS27. For patients with IBD, Bao et al. and Huang et al. reported a decreased ReHo value in the postcentral gyrus compared to the HCs17,28. These findings were consistent with our results and the abnormal activity of postcentral gyrus may partially explain the nociceptive symptoms in patients with IBD.

Based on PLS correlation, we found that the brain scores of the ACC were positively correlated with the scores of PASS (r = 0.270, p = 0.007, Pearson correlation), PCS (r = 0.287, p = 0.004, Pearson correlation), and PVAQ (r = 0.301, p = 0.003, Pearson correlation), thus suggesting that the dysfunction of ACC may be closely related to pain-related negative emotion. As we know, one of the most common symptoms of IBD was pain. Empirical evidence has shown a strong correlation between pain-related negative emotion and the experience of pain29,30. Traditionally, the ACC has been associated with negative emotions and has mostly been connected to the experience of pain31. It seemed that the ACC is involved in rational cognitive and pain-related processes, including empathy and emotion, particularly pain-related negative emotion32. Animal studies have demonstrated the presence of nociceptive-specific neurons in the ACC, and these neurons are responsible for encoding the integration of nociception as well as the anticipation of pain that leads to avoidance and the reward behaviors that come from cutaneous electric stimulation33,34,35. Furthermore, Cao et al. and Yan et al. used colorectal distension in conjunction with the conditioned place avoidance (CPA) model to discover that an ACC lesion or the inhibition of the ACC’s cholecystokinin or N-Methyl-D-aspartic acid receptors eliminated the CPA score’s measure of aversive visceral pain memory but had no effect on acute visceral pain behaviors36,37. Clinical studies have shown that cingulotomy effectively decreases pain-related negative emotion while not affecting the patient’s capacity to perceive the intensity and ___location of painful stimuli38,39. Previously, a PET study showed a relationship between the level of perceived unpleasantness and the activity in the ACC while selectively adjusting the unpleasantness of painful stimulation without affecting the perceived stimulus intensity using hypnotic suggestions40. These findings were consistent with our results and suggested that the ACC has been involved in pain-related negative emotion in patients with IBD. This relationship indicated that exposure to pain-related negative emotion in the central nervous system may result in brain functional alterations in the ACC.

Following projection in the orthogonal subspace from the dominant LV1, the LV2 exhibited negative brain salience in the left postcentral gyrus, which was associated with the levels of GLIGI, IBS-SSS, and BDI. The postcentral gyrus, as a key hub of the sensorimotor network, plays a crucial role in the nociceptive pathway25. Recent structural and functional MRI revealed abnormal cortical thickness/gray matter volume and altered brain activity in the postcentral gyrus in gastrointestinal disorders27,41. Compared to the HCs, patients with IBD showed lower activity in the postcentral gyrus17,28. Additionally, the postcentral gyrus is also involved in the stages of emotional processing, including identifying emotional meaning in stimuli, generating emotional states, and regulating emotion42. Recent neuroimaging studies have demonstrated that adolescents with major depressive disorder (MDD) have decreased surface area in the postcentral gyrus43. Furthermore, compared to women with premenstrual dysphoric disorder alone, women with comorbid bipolar and premenstrual dysphoric disorders showed decreased functional connectivity between the left postcentral gyrus and left hippocampus44. Notably, we found that negative brain salience in the left postcentral gyrus was linked to both the severity of gastrointestinal symptoms and depression. This may suggest that the postcentral gyrus plays a role in controlling gastrointestinal symptoms and mental states like depression in people with IBD.

There are several limitations in this study. First, we could not exclude the influence of drugs on assessment, such as conventional immunosuppressants and anti-TNF antibody, which may affect the rs-fMRI patterns. Second, in this study, our main focus was on gastrointestinal symptoms, emotional function, and the connections between these data and fMRI data. We did not differentiate between the stages of disease activity. In future studies, we will include patients with different stages of disease activity. Third, FDR correction is relatively liberal and might result in a higher false positive rate. Fourth, we did a cross-sectional study solely on patients with IBD. In the future, longitudinal data should be gathered to study if there will be an increase in negative emotions and their correlation with brain alterations. Finally, the HCs data did not include blood analysis.

Altogether, we found different changes in ALFF and DC values between patients with IBD with and HCs. More importantly, we found that brain salience in the ACC was linked to pain-related negative emotion, and brain salience in the left postcentral gyrus was associated with the higher-order trait depression. These findings can enhance our comprehension of how pain-related negative emotion and trait negative emotion affect the brains of patients with IBD in distinct ways.

Methods

Participants

All research procedures were approved by the Institutional Review Board of the Chongqing General Hospital (Approval No. KY S2023-019-01) and were conducted in accordance with the Declaration of Helsinki. All participants provided written, informed consent after receiving a comprehensive explanation of the experimental procedures.

All patients were screened and diagnosed by an IBD specialist at the Department of Gastroenterology, Chongqing General Hospital. All patients underwent systemic and gastrointestinal examinations, including a colonoscopy and pathology biopsy. We conducted both laboratory tests and the colonoscopy within 2 weeks prior to the MRI. Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) measurements were obtained from all patients. Additionally, we collected data on previous intestinal surgery, biologic therapy, and possible maintenance medications.

The inclusion criteria for the IBD patient group were as follows: (1) age of 18 to 55 years; (2) IBD diagnosed at least 1 year prior to enrollment; (3) an education period ≥ 6 years; and (4) right-handed. The exclusion criteria for the IBD patient group were as follows: (1) severe disease, based on physicians’ examination and ESR level greater than 30 mm/h; (2) current use or use of corticosteroids over the prior 6 months; (3) existence of a neurological disease or psychiatric disorder; (4) immediate plans for pregnancy or a positive pregnancy test; and (5) any contra-indications to MRI.

Right-handed healthy controls (HCs) were recruited with advertisements from Chongqing General Hospital. None of the individuals in the HC group had taken any medication, had gastrointestinal or pain-related illnesses, or exhibited abnormal findings during colonoscopy examination. They were subjected to identical screening protocols as the IBD patients, following the same criteria for inclusion and exclusion, with the exception of the IBD-specific examinations.

Clinical features and psychological assessment

The IBS symptom severity system (IBS-SSS)45 and the Gastrointestinal Quality of Life Index (GIQLI)46 were used to evaluate clinical features, gastrointestinal function, and life quality. The Fear of Pain Questionnaire (FPQ)47, Pain Anxiety Symptoms Scale-20 (PASS-20)48, the Pain Catastrophizing Scale (PCS)49, the Pain Sensitivity Questionnaire (PSQ)50, and the Pain Vigilance and Awareness Questionnaire (PVAQ)51 were conducted to assess pain-related negative emotion and cognition. The Trait version of State-Trait Anxiety Inventory (STAI_T)52 and the Beck Depression Inventory (BDI)53 were employed to evaluate trait negative emotion.

The specifics of these questionnaires were delineated as follows. The IBS-SSS scale is employed to assess the intensity of abdominal symptoms in patients, encompassing five dimensions: intensity of abdominal pain, length of abdominal pain, bloating, satisfaction with bowel movements, and level of interference with daily life. The score range spans from 0 to 500 points45. The GIQL Index employs a total of 36 items distributed over 5 dimensions: gastrointestinal symptoms (consisting of 19 items), emotional dimension (comprising of 5 items), physical dimension (including 7 items), social dimension (comprising of 4 items), and therapeutic impacts (consisting of 1 item)46. The FPQ is a 30-item questionnaire with a 4-point Likert scale that is used to evaluate situational fears of painful stimuli as a trait-like phenomena47. The PASS-20 is a questionnaire consisting of 20 items, rated on a 6-point Likert scale. It is designed to assess underlying, non-specific anxiety connected to pain experienced in daily life48. The PCS is a 13-item self-report questionnaire that assesses catastrophizing in the context of current or expected pain. Catastrophizing is measured by the PCS as a complex construct with three subscales: rumination, magnification, and helplessness49. The PSQ requires participants to envision 14 painful situations and 3 non-painful situations. Participants are required to assess the level of pain they would experience, ranging from 0 (indicating no pain) to 10 (representing the most intense suffering imaginable)50. The PVAQ comprises 16 items, and participants are instructed to reflect on their behaviors during the past 2 weeks. They are then asked to rate the frequency of each item on a 6-point scale ranging from 0 (indicating never) to 5 (indicating always), indicating the extent to which each item accurately describes their behavior51. The STAI-T is a subscale of the State-Trait Anxiety Inventory, which is a 20-item scale used to assess the degree of one’s dispositional anxiety52. The BDI is a 21-item scale utilized to assess an individual’s inherent attitudes and symptoms of depression experienced over the past week53.

Imaging acquisition

We collected MRI data using a 20-channel phase array head coil on a Siemens SKYRA 3.0T magnetic resonance scanner (Siemens, Germany). A high-resolution T1-weighted structural image was collected using a T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) sequence with the following parameters: slices = 144, slice thickness = 1.0 mm, repetition time (TR) = 2250 ms, echo time (TE) = 2.6 ms, FA = 8◦, field of view (FOV) = 256 × 256 mm, voxel size = 1 × 1 × 1 mm; data matrix = 256 × 256.

Resting-state functional images were obtained using an echo-planar-imaging sequence with the following parameters: slices = 33, slice thickness = 3.0 mm, TR = 2,000 ms, TE = 30 ms, flip angle = 68°, FOV = 210 × 210 mm, data matrix = 64 × 64, slice thickness = 3 mm, measurements = 220.

Data preprocessing

The functional data underwent preprocessing using SPM8 (http://www.fil.ion.ucl.ac.uk/spm) and the Data Processing Assistant for Resting-State fMRI toolbox (DPARSF, http://www.restfmri.net/forum/DPARSF). We excluded the first 10 volumes of functional images initially. Next, we adjusted the remaining images for temporal discrepancies between head movement (using 24 motion-related regression) and slices (employing a least-squares method)54. Furthermore, we normalized the realigned images to the MNI template. Subsequently, we converted the normalized functional pictures to Z values and smoothed them using a Gaussian kernel with a full width at half maximum (FWHM) of 8 mm. Meanwhile, we regressed cerebrospinal fluid (CSF) and white matter signal. Finally, we applied a bandpass filter to keep low-frequency fluctuations, restricting the frequency range to 0.01 and 0.1 Hz.

ReHo, ALFF, and DC analyses

The regional homogeneity (ReHo), amplitudes of low-frequency fluctuation (ALFF),, and degree centrality (DC) maps were computed using DPARSF55. ReHo refers to the degree of synchronization of fluctuations in BOLD signals among neighboring voxels within a specific brain region. Alterations in ReHo values indicate possible disturbances in the synchronization and coordination of spontaneous neuronal processes within that particular region of the brain56. For the ReHo map, bandpass filtering (0.01–0.08 Hz) was applied to the normalized images. The quantification of ReHo values involved computing the Kendall’s coefficient of concordance value on a voxel-by-voxel basis between a given voxel and its neighbors56. To minimize the impact of individual variability, the ReHo value of each voxel was divided by the ReHo map’s global average value. Lastly, a Gaussian kernel with an 8-mm FWHM was used for spatial smoothing.

ALFF is a measurement that detects the amplitude of the blood oxygen level-dependent (BOLD) signal in relation to the baseline. It can be used to indicate the level of spontaneous activity at each voxel57. For the ALFF map, the time course for each given voxel was decomposed into the frequency ___domain with the fast Fourier transform (FFT). The ALFF value was calculated as the average square root of the power spectrum following the application of the FFT across the frequency range of 0.01–0.08 Hz for each voxel.

Degree centrality (DC), a graph-based network organization assessment, represents the number of instantaneous functional connections between a region and the rest of the brain within the brain’s total connectivity matrix. As consequence, it can determine how much a node effects the overall brain and integrate information from functionally distinct brain regions. For the DC map, a voxel-based whole-brain correlation analysis was performed on the preprocessed rsfMRI data. Within the gray matter (GM) mask, the temporal progression of each voxel was associated with the temporal progression of every other voxel. Consequently, a matrix of Pearson’s correlation coefficients between any pair of voxels was obtained, with the size of the matrix being n × n, where n is the number of voxels in the GM mask. By applying a threshold of r > 0.25 to each correlation, we converted Pearson’s correlation data into Fisher Z-scores, which follow a normal distribution. Subsequently, we created the functional network of the entire brain. Only Pearson’s correlation coefficients with positive values were included in the current investigation. The DC for a certain voxel was calculated by summing the important connections at the individual level. The DC map provides information about the number of functional connections for a specific voxel in the voxel-based graphs at the individual level. This map is commonly used to describe the properties of nodes in large-scale brain intrinsic connectivity networks58.

PLS correlation analysis

To explore the inter-group differences in brain function and its relationship with pain-related negative emotion and trait negative emotion, PLS correlation analysis was employed. In our analysis, we incorporated variables for gastrointestinal symptoms (IBS-SSS and GIQLI scores), pain-related negative emotion (FPQ, PASS-20, PCS, PSQ, and PVAQ scores), and trait negative emotion (STAI_T and BDI scores) in the behavior matrix Y. The brain functional values of each participant are represented in matrix X.

The cross-covariance between the two matrices was determined after normalizing the columns of X and Y as z-scores to avoid any variance differences that could potentially disrupt the PLS modeling process. Subsequently, the cross-covariance matrix was subjected to singular value decomposition. This allows for the determination of a group of independent LVs that demonstrate the highest level of correlation between brain measures and behavioral measures59. Each LV is linked to three key components: (1) a singular value (represented by the diagonal elements of S) that reflects the strength of the brain-behavior correlation; (2) a vector of left singular values (U), also known as behavior salience, which represents the contribution of each behavior variable to the brain-behavior correlation; and (3) a vector of right singular values (V), also known as brain salience, which represents the contribution of each voxel to the brain-behavior correlation. The brain scores XV were computed as a metric to quantify the resemblance between individual brain data and prominent brain patterns. Brain scores with high absolute values imply a significant contribution to the pattern, whereas scores closer to zero suggest a less significant contribution60. The quantity of latent variables (LVs) is equivalent to the quantity of behavioral measures incorporated in the analysis.

Additionally, we calculated the association between the brain scores and behavioral factors in IBD group. This was done to offer an additional way of visualizing the data that was recorded in the behavior saliences.

Statistical analysis

The demographic characteristics, including age and level of education, were compared between the two groups using independent-samples t-tests. Prior to conducting statistical analysis, Shapiro-Wilk tests were employed to assess the normality of the data (i.e., gastrointestinal symptoms, pain-related negative emotion variables, and trait negative emotion variables). Results showed that the scores of GIQLI, PASS, PCS, PAQ, and BDI were skewed distributed (all p < 0.05), and the values of FPQ, PVAQ, and STAI were normally distributed (all p > 0.05). Consequently, nonparametric Mann-Whitney U-tests were conducted to examine variables with skewed distributions. Parametric independent-samples t-tests were employed to examine the variables that followed a normal distribution. The gender ratio difference was examined using the chi-square test.

A two-sample t test was conducted to examine the whole-brain differences between groups in ALFF, ReHo, and DC maps. An FDR correction was utilized to account for multiple comparisons. A significance threshold of p < 0.05 was deemed statistically significant.

In order to assess the statistical significance of each LV in the PLS correlation analysis, we generated a null distribution of the singular values. This was done by performing 5,000 permutations, which involved randomly rearranging the rows of the brain matrix while keeping the behavior matrix identical. Any LV with a p-value less than 0.05 was included in order to ensure that the LV could be applied to the entire study population. The 5,000 bootstrap samples were generated to assess the dependability of the impact on each voxel. For each bootstrap sample, the brain and behavior saliences were recalculated, yielding bootstrap ratios (BSRs) that represent the original saliences divided by the bootstrap standard errors. We employed a BSR threshold of ± 3, which corresponds to a p-value of around 0.01. This entailed the random selection (with replacement) of both the brain and behavioral matrices.