Multimodal Brain MRI of Deep Gray Matter Changes Associated With Inflammatory Bowel Disease

Abstract

Background

Behavioral symptoms, including mood disorders, substantially impact the quality of life of patients with inflammatory bowel disease (IBD), even when clinical remission is achieved. Here, we used multimodal magnetic resonance imaging (MRI) to determine if IBD is associated with changes in the structure and function of deep gray matter brain regions that regulate and integrate emotional, cognitive, and stress responses.

Methods

Thirty-five patients with ulcerative colitis (UC) or Crohn’s disease (CD) and 32 healthy controls underwent 3 Tesla MRIs to assess volume, neural activity, functional connection strength (connectivity), inflammation, and neurodegeneration of key deep gray matter brain regions (thalamus, caudate, pallidum, putamen, amygdala, hippocampus, and hypothalamus) involved in emotional, cognitive and stress processing. Associations with sex, presence of pain, disease activity, and C-reactive protein (CRP) concentration were examined.

Results

Significantly increased activity and functional connectivity were observed in cognitive and emotional processing brain regions, including parts of the limbic system, basal ganglia, and hypothalamus of IBD patients compared with healthy controls. Inflammatory bowel disease patients exhibited significantly increased volumes of the amygdala and hypothalamus, as well as evidence of neurodegeneration in the putamen and pallidum. Hippocampal neural activity was increased in IBD patients with active disease. The volume of the thalamus was positively correlated with CRP concentration and was increased in females experiencing pain.

Conclusions

Patients with IBD exhibit functional and structural changes in the limbic and striatal systems. These changes may be targets for assessing or predicting the response to therapeutic interventions aimed at improving comorbid emotional and cognitive symptoms.

Key Messages.

• Patients with IBD experience behavioral symptoms that do not improve with IBD treatment.

• We used multimodal brain MRI to demonstrate changes in structure and function of the limbic and striatal systems.

• Brain MRI can help target behavioral symptom treatment and predict response.

Introduction

Inflammatory bowel diseases (IBD), ulcerative colitis (UC) and Crohn’s disease (CD) are chronic inflammatory conditions of the gut characterized by a relapsing and remitting clinical course.¹ Symptoms can be highly variable in patients with IBD, may not always correlate with objective measures of inflammation, and fluctuate over time and with treatment. In addition to gastrointestinal symptoms, IBD patients experience a spectrum of behavioral, cognitive, and emotional symptoms including disorders of mood (anxiety, depression), sleep, fatigue, cognitive dysfunction, and social withdrawal. Rates of mood disorders in IBD patients are double those reported in the general population,² and depression and anxiety are associated with clinical relapse in both UC and CD.³ In contrast to physical symptoms, however, the psychological manifestations of IBD demonstrate only modest improvements over long-term follow-up,⁴ and current therapies that alleviate physical symptoms have little impact on IBD-associated psychological symptoms.⁵

Other chronic inflammatory disorders associated with an increased prevalence of psychological symptoms have demonstrated structural and functional changes within the central nervous system.⁶ In the context of IBD, gut inflammation can signal the brain through a number of pathways and alter central neurotransmission and behavior.⁷ The brain communicates with the intestines through the brain-gut axis, which includes the sympathetic and parasympathetic components of the autonomic nervous system (ANS) and the enteric nervous system.⁸ The ANS regulates most aspects of intestinal function including gut motility, fluid secretion, electrolyte transport, epithelial barrier function, and immune responses.⁸ The central regulation of the ANS is mediated through the central autonomic network (CAN), which includes the amygdala, hypothalamus, insula, hippocampus, and thalamus. In addition, the CAN is a critical component of the stress response, and hyperactivity of the CAN sympathetic component amplifies emotional reactions to stress.⁹ The basal ganglia (caudate, putamen and pallidum) processes signals across different neural networks within cortico-basal ganglia-thalamic circuits, and with various limbic structures (including the hypothalamus, amygdala, thalamus, and hippocampus) important for controlling cognitive and affective functions. It follows that regions of the CAN and the basal ganglia within the CNS could be impacted by inflammatory processes within the gut. If the impact on these brain regions can be objectively determined, then the assessment of these regions in response to therapy could act as an indicator of treatment efficacy or as a predictor of treatment response.

Magnetic resonance imaging (MRI) provides noninvasive means to quantitatively determine the impact of disease on the brain, both structurally and functionally.¹⁰ For example, structural MRI can delineate changes in the volume of brain regions with high accuracy and reliability.^11,12 Quantitative susceptibility mapping (QSM), derived from susceptibility-weighted MRI data, can provide highly reproducible images¹³ that link increased brain iron levels with neuroinflammation (increased susceptibility)¹⁴ or increased calcium deposition with neurodegeneration (decreased susceptibility).¹⁵ Resting-state functional MRI can determine the strength of interregional brain communication (ie,functional connectivity) and how the neural activity levels of brain regions may be altered by disease (using the amplitude of low-frequency fluctuations [ALFF]¹⁶ which has been shown to possess high temporal stability and reproducibility¹⁷). Thus, these quantitative MRI techniques that exhibit high reproducibility and reliability can objectively assess the impact of disease on brain structure and function when visually discernable lesions are not present on MRI. Indeed, quantitative MRI studies have demonstrated that changes in brain structure and/or function of only a few percent are clinically relevant to behavioral symptoms, including depression,^18–21 anxiety,^22,23 and chronic pain.^24–27

Several brain MRI studies of IBD patients have demonstrated structural and/or functional changes within the cortex or midbrain, with some studies suggesting these changes are dependent on the presence of pain or behavioral changes; however, results are highly variable.²⁸ In the majority of these studies, only a single MRI modality was used with simple correlational analyses of subjective clinical/behavioral measures and MRI. However, MRI can provide multiple image types from a single session, allowing for the interrogation of multimodal data from patients to provide a more complete, integrated, and accurate description of the impact of IBD on the brain. Therefore, we employed multimodal MRI to more clearly define the impact of IBD on the brain. Specifically, we hypothesized that multimodal brain MRI of IBD patients would demonstrate altered structure and function of subcortical gray matter structures of the basal ganglia and limbic systems that play key roles in regulating emotion, pain, stress, and cognition.²⁹

Materials and Methods

Participants

This study was approved by the Conjoint Health Research Ethics Board of the University of Calgary (ID# REB14-1376). Adult patients with IBD diagnosed according to standard criteria were recruited from the University of Calgary Inflammatory Bowel Disease Clinic. All participants provided written informed consent. Electronic medical records were reviewed to extract data on patient demographics, IBD diagnosis and phenotype, steroid use, outpatient clinic consult(s) within 90 days of study MRI (for symptoms and patient-reported outcomes), blood concentrations of C-reactive protein (CRP) within 4 months of MRI (0.3 mg/L imputed if the CRP concentration was below this lower limit of quantification), endoscopic assessment (by colonoscopy, sigmoidoscopy, or pouchoscopy), and diagnostic imaging assessments by bowel ultrasound or other modality (MR or CT enterography, CT abdomen, or MR pelvis) within 12 months preceding or 6 months following study MRI. The presence or absence of patient-reported abdominal pain was recorded based on documentation within the medical record; a pain scale was not used to avoid heterogeneity in terms of assessing pain symptoms, given that patients were recruited across multiple clinical settings. A global assessment of inflammatory disease activity was made based on available contemporaneous endoscopy, diagnostic imaging, and CRP data. Patients were classified as either being in remission/mild activity (Mayo endoscopic subscore of 0 or 1, Simple Endoscopic Score for Crohn’s disease <6 for ileocolonic disease or <4 for ileal only disease, or presence of only aphthous ulcers in <3 colonic segments, C-reactive protein <5 mg/L, radiographic imaging showing no active inflammatory changes) or moderate to severe activity by 2 IBD physicians (R.I., C.M.) who were blinded to the MRI findings.

Based on interview, patients were excluded from MRI if they met any of the following criteria: a full scale IQ less than 80 (as assessed using the National Adult Reading Test); an active history of Axis 1 psychiatric illness, including substance misuse other than nicotine; current or previous central nervous system condition (eg, multiple sclerosis, seizure disorder) or uncontrolled medical condition (eg, thyroid disease, diabetes); a history of head injury with skull fracture or loss of consciousness greater than 5 minutes; contraindications to MRI (eg, metallic implants, pregnancy, claustrophobia); current use of psychotropic medications; a family history of bipolar disorder or psychosis; current symptoms of psychosis, mania, or suicidality. These criteria are typical for MRI studies of brain function, given the potential confounding impact on MRI assessments.³⁰

After IBD patient recruitment was complete, MRI data collected using the same imaging parameters were obtained for healthy participants from the Calgary Normative Study (CNS) (Canadian Institutes for Health Research, FDN-143298) to match the IBD group by age and sex.

MRI Acquisition

All subjects underwent MRI acquired using a 3.0 Tesla MR scanner (Discovery MR750; GE Healthcare, Waukesha, WI). An imaging session consisted of (1) a localizer scan to prescribe slices for subsequent imaging sequences; 2) high-spatial resolution anatomical T1-weighted imaging for anatomical registration of the other modalities and for region-of-interest volumetric analysis; (3) resting-state fMRI for ALFF and functional connectivity analysis; and (4) QSM to assess potential neuroinflammation due to iron deposition (increase in susceptibility) or potential calcium deposition related to neurodegeneration (decrease in susceptibility).

Anatomical images were collected using a 3D magnetization-prepared rapid gradient echo sequence (1-mm isotropic voxels). Resting-state fMRI data were acquired with a gradient-recalled echo, echo planar imaging sequence (repetition/echo time = 2000/30 ms, flip angle = 77°, 150 total volumes, 64 × 64 matrix, 3.75-mm isotropic voxels). Participants were asked to keep their eyes open and keep them focused on a fixation cross. Quantitative susceptibility mapping data were collected using an RF-spoiled, flow-compensated 3D gradient echo sequence (repetition/echo time = 29.5/26.3 ms; flip angle = 20°; FOV = 256 × 256 × 132 mm³; voxel size = 1 × 1 × 1 mm³; 8 echoes).

MRI Processing and Analysis

T1-weighted anatomical images were uploaded to BrainGPS (https://braingps.mricloud.org), which segmented the thalamus, caudate, pallidum, putamen, amygdala, hippocampus, and hypothalamus of each hemisphere as regions of interest (ROIs) using the Multiresolution Intrinsic Segmentation Template (MIST). The volume of each structure and the total intracranial volume were recorded. BrainGPS uses a large collection of preparcellated images to achieve high segmentation accuracy over a wide range of anatomical variability and brain diseases,^11,12 accounting for differences across MR scanners and imaging procotols.^11,31 This approach has become the standard for high accuracy, high precision brain region segmentation.

For each ROI, a repeated-measures (left and right hemisphere) analysis of covariance (ANCOVA) was used to examine group differences (IBD vs controls) in volume, including sex as a factor and age and total intracranial volume as covariates, as well as 2-way interactions among factors. All statistical analyses were performed using IBM SPSS Statistics Version 27.0 (IBM Corp. Armonk, NY. Released 2020)

From the acquired QSM data, QSM images were generated using Cerebra-QSM software (Calgary Image Processing and Analysis Centre, University of Calgary, Calgary, AB). These images are then used to create a quantitative susceptibility map. Pixel values were then converted to Z-scores relative to the whole-head average. A repeated-measures (left and right hemisphere) ANCOVA was then used to examine group differences (IBD vs controls) in ROI susceptibility, with sex as a factor, age as a covariate, and 2-way interactions among factors.

Using FSL (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki), resting-state fMRI data first underwent standard preprocessing steps, including brain extraction, slice timing correction, motion correction, 6-mm Gaussian kernel spatial smoothing, and low-pass temporal filtering (<0.01 Hz). Data from participants whose head motion exceeded 3 mm were excluded. Processed fMRI images were registered to their corresponding T1-weighted images. Independent component analysis (ICA) was performed on the fMRI data using the MELODIC tool of FSL, and time-varying fMRI signal components attributable to noise or artifact were identified visually and subsequently removed by regression.³²

Amplitude of low-frequency fluctuations was computed for each image pixel using the CONN toolbox (http://www.conn-toolbox.org).³³ Amplitude of low-frequency fluctuations values for each image pixel were converted to Z-scores relative to the whole-head average. Average values for each ROI were then calculated and used in a repeated-measures (left and right hemisphere) ANCOVA to examine group differences (IBD vs controls) in susceptibility, including sex as a factor and age as a covariate, as well as 2-way interactions among factors.

For each MRI measure, a repeated-measures ANCOVA was then used for IBD patient data alone to determine if patients differed based on global assessment of disease activity, sex, the presence of pain, or any 2-way interactions among these factors and to determine associations with CRP concentration. C-reactive protein was converted to a log scale to generate a normal data distribution.

In addition, for the resting-state fMRI data, the average time courses from ROIs were used to compute the Pearson cross-correlation coefficient for each possible pair of ROIs. This analysis generated an estimate of connectivity between each ROI pair directly. After Fisher Z transformation of the correlation coefficients, a repeated-measures (left and right hemisphere, region 1 and region 2 of the pair) ANCOVA was used to examine group differences (IBD vs controls) in connectivity, including sex as a factor and age as a covariate, as well as 2-way interactions among factors. Repeated-measures ANCOVAs were then used for IBD patient data alone to determine if patients differed based on global assessment of disease activity, sex, the presence of pain, or any 2-way interactions among these factors and to determine associations with CRP concentration.

Results

Thirty-five IBD patients (mean age 34 ± 11 years; 13 female and 22 males; 26 CD and 9 UC) participated in the study. Control MRI data were obtained for 32 healthy participants (mean age 34 ± 11 years; 12 females and 20 males). The number of recruited participants was based on our recent brain MRI study of 20 patients with primary biliary cholangitis (PBC).³⁴ To ensure a statistical power of 0.8 at a type 1 error rate of less than 5%, we required 32 members per group to observe the significant and near-significant changes of the thalamus in that study. A total of 16 patients with IBD reported the presence of pain. Global inflammatory disease activity was considered moderately to severely active in 16 patients (11 of whom endorsed experiencing pain). Only 1 patient was taking steroids at the time of imaging (prednisone 40 mg per day). A summary of patient demographic and clinical characteristics are provided in Table 1. No patients were on prescription narcotics at the time of MRI, and no patients were prescribed opiates up to 30 days prior to MRI.

Table 1.

Characteristics of study population.

Age
mean years ± SD (range)	34 ± 11 years (18–73)
Gender
% female	37%
IBD
CD, n (%)	23 (66%)
UC, n (%)	10 (29%)
CD of J-pouch, n (%)	2 (6%)
Extent of UC by Montreal classification
extent, n (%)	E1: 2 (20%), E2: 5 (50%), E3: 3 (30%)
Phenotype of CD by Montreal classification
age at onset, n (%)	A1: 0 (0%), A2: 15 (65%), A3: 8 (35%)
location, n (%)	L1: 9 (39%), L2: 5 (22%), L3: 9 (39%), L4: 0 (0%)
behavior, n (%)	B1: 13 (57%), B2: 2 (9%), B3: 5 (22%), B2 + B3: 3 (13%)
perianal disease, n (%)	Yes: 4 (17%), No: 19 (83%)
Patient-reported abdominal pain
n (%)	Yes: 16 (46%), No: 19 (54%)
Disease activity
n (%)	Remission/Mild: 19 (54%), Moderate/Severe: 16 (46%)
No. with available contemporaneous objective evidence from endoscopy, imaging, CRP (%)	3 modalities: 22 (63%), 2 modalities: 12 (34%), 1 modality: 1 (3%)
CRP
No. with result within 4 m of study MRI (%)	35 (100%)
median days from study MRI (IQR)	23 days (6–58)
median concentration (IQR, range)	1.9 mg/L (1.0–6.6, 0.5–56.0)
No. with result >5.0 mg/L (%)	14 (40%)
Endoscopic assessment
No. with result within 12 m preceding or 6 m	32 (91%)
following study MRI (%)	(29 colonoscopy, 3 sigmoidoscopy, 1 pouchoscopy)
median days from study MRI (IQR)	41 days (21–94)
Diagnostic imaging assessment
No. with result within 12 m preceding or 6 m	25 (71%)
following study MRI (%)	(15 bowel ultrasound, 5 other modality, 5 both)
median days from study MRI (IQR)	59 days (36–111)

Insufficient results for fecal calprotectin were available to include in analysis. The lower limit of quantification of reported CRP concentration was substituted if the result was reported as below this limit. Both participants with J-pouches were initially diagnosed with UC, with subsequent development of CD of pouch and pre-pouch ileum with peri-pouch fistulization. Other diagnostic imaging modality included MR pelvis for both patients with J-pouch. Abbreviations: CD, Crohn’s disease; CRP, C-reactive protein; IQR, inter-quartile range; m, months; MRI, magnetic resonance imaging; UC, ulcerative colitis; SD, standard deviation.

For all MRI analyses, data were combined across IBD phenotypes, as preliminary analyses revealed no significant differences between UC and CD patients (representative results are shown in Supplemental Figure 1).

Brain Regional ALFF and Susceptibility Changes

Neural activity was significantly elevated in several ROIs in IBD patients as determined by ALFF. Amplitude of low-frequency fluctuation was significantly greater in the hippocampus (F[1,61] = 9.51, P = .003), hypothalamus (F[1,61] = 6.49, P = .013), thalamus (F[1,61] = 23.38, P < .001), insula (F[1,61] = 6.13, P = .016), putamen (F[1,61] = 8.07, P = .006), caudate (F[1,61] = 11.28, P = .001), and pallidum (F[1,61] = 7.49, P = .008) of IBD patients relative to controls (Figure 1). For ALFF of the putamen, there was a trend towards a sex by group interaction (F[1,61] = 2.92; P = .093), and for ALFF of the pallidum, the sex by group interaction was significant (F[1,61] = 4.31; P = .042). As shown in Figure 1, the change in putamen and pallidum ALFF from control to IBD was significantly greater for females but not for males.

Figure 1.

Inflammatory bowel disease patients exhibited significantly greater ALFF of the hippocampus, hypothalamus, thalamus, insula, putamen, caudate, and pallidum relative to controls (CTRL). For the putamen and pallidum, the difference between IBD patients and controls was observed in females only. The image at the lower right shows the anatomical location of the regions of interest on a standardized brain template image, including those that exhibited a significant group difference (in red) and those where the group difference was observed in females only (in yellow) (See online version for color figures).

Susceptibility was significantly lower in the putamen (F[1,61] = 6.64, P = .012) and pallidum (F[1,61] = 5.01, P = .029) of IBD patients relative to controls (Figure 2). Amplitude of low-frequency fluctuation and susceptibility were not associated with active disease or systemic inflammatory response.

Figure 2.

Inflammatory bowel disease patients exhibited significantly reduced susceptibility in the (A) putamen and (B) pallidum, relative to controls (CTRL).

Functional Connectivity

Functional connectivity that differed between IBD patients and controls is summarized in Figure 3. Data from 1 male IBD patient and 1 male control participant were excluded due to excessive head motion. Brain circuits comprising key limbic and striatal structures were “hyperconnected” in IBD patients, with significantly increased functional connectivity for the insula-hypothalamus, insula-hippocampus, insula-putamen, insula-pallidum, putamen-hypothalamus, putamen-pallidum, putamen-caudate, pallidum-hypothalamus, caudate-hippocampus, and caudate-thalamus connections.

Figure 3.

Functional deep gray matter brain region connections (in yellow) whose connectivity was significantly increased in IBD patients relative to controls, including the insula-hypothalamus, insula-hippocampus, insula-putamen, insula-pallidum, putamen-hypothalamus, putamen-pallidum, putamen-caudate, pallidum-hypothalamus, caudate-hippocampus, and caudate-thalamus connections (See online version for color figures).

Changes in the Amygdala and Hypothalamus

Patients with IBD exhibited significantly greater volume of the amygdala (F[1,60] = 5.46, P = .023; Figure 4A). Furthermore, there was a significant effect of sex by disease activity (F[1,24] = 12.10, P = .002), in that the volume of the amygdala was greater in females with moderate/severe disease activity relative to remission/mild activity, but amygdala volume was not associated with disease activity in males (Figure 4B). Overall, ALFF was similar in the amygdala of IBD patients and healthy controls; however, there was a significant group by sex effect (F[1,61] = 7.51, P = .008). As shown in Figure 4C, the change in amygdala ALFF for males from control to IBD was greater than the change in females, eliminating an apparent sex difference in amygdala ALFF in controls.

Figure 4.

A, Inflammatory bowel disease patients exhibited significantly greater volume of the amygdala. B, In IBD patients, the volume of the amygdala increased in patients with mild disease activity (MILD) vs those with moderate/severe activity (MOD) in females (F), whereas males (M) exhibited no change in amygdala volume associated with changes in disease activity. C, The change in amygdala ALFF for males, from control to IBD was greater than the change in females. D, IBD patients exhibited significantly greater volume of the hypothalamus. An increase in (E) hypothalamus-hippocampus and (F) hypothalamus-thalamus connectivity between control and IBD patient groups occurred in females but not males.

The volume of the hypothalamus was significantly greater in IBD patients than in controls (F[1,59] = 17.27, P < .001; Figure 4D). As reported previously and shown in Figure 3, functional connectivity between the hypothalamus and the insula, putamen, and pallidum was significantly greater in IBD patients than in controls. Although there was no group difference in functional connectivity between the hypothalamus and the hippocampus or thalamus, there was a significant sex by group effect, with a decrease in functional connectivity between the hypothalamus and hippocampus (F[1,59] = 16.72, P < .001) and between the hypothalamus and thalamus (F[1,59] = 5.41, P = .023) in females—but not males (Figures 4E and 4F).

Brain Changes Related to Pain and Disease Activity

The increase in volume of the thalamus in IBD patients relative to controls was nearly significant (F[1,60] = 3.90, P = .053). Within IBD patients, the volume of the thalamus exhibited both significant sex by pain (F[1,24] = 6.43, P = .018) and pain by disease activity interactions (F[1,24] = 9.62, P = .005). As shown in Figure 5A, thalamus volume for females experiencing pain was increased relative to females not experiencing pain, but pain was not associated with thalamus volume in male IBD patients. Inflammatory bowel disease patients with moderate/severe disease activity not experiencing pain exhibited a lower thalamic volume compared with patients with low disease activity; patients experiencing pain showed no dependence of thalamic volume on disease activity (Figure 5B).

Figure 5.

A, Thalamus volume for female (F) IBD patients experiencing pain (P) was increased relative to female IBD patients not experiencing pain (NP), but pain was not associated with thalamus volume in male (M) IBD patients. B, IB7D patients with moderate/severe disease (MOD) activity not experiencing pain exhibited a lower thalamic volume compared with patients with mild disease activity (MILD); patients experiencing pain showed no changes with disease activity. C, ALFF of the hippocampus was significantly greater in IBD patients with moderate/severe disease activity relative to mild disease/remission. (D) The volume of the thalamus and (E) functional connectivity between the thalamus and pallidum were significantly correlated with log₁₀ CRP concentration.

Amplitude of low-frequency fluctuation of the hippocampus was significantly greater in IBD patients with moderate/severe disease relative to mild disease/remission (F[1,25] = 11.02; P = .003; Figure 5C).

The volume of the thalamus was also significantly associated with CRP concentration (F[1,24] = 20.88; P < .001; Figure 5D). In addition, there was also a significant effect of CRP concentration on the functional connectivity between the thalamus and pallidum, which is an output relay hub of the basal ganglia, important for modulation of pain responses³⁵ (F[1,24] = 6.86; P = .015; Figure 5E).

Discussion

Our multimodal MRI approach has allowed us to conduct a comprehensive mechanistic study of brain dysfunction associated with IBD. We have identified significant IBD-related brain changes in the subcortical deep gray matter structures that comprise the limbic and striatal systems; we also defined how systemic inflammation, disease activity, pain, and sex impact these changes. The magnitude of our reported findings are consistent with those reported previously for structures within the limbic and striatal system associated with behavioral symptoms in patients with depression,^18–21 anxiety,^22,23 and chronic pain.^24–27 Given the widely accepted roles of the limbic and striatal systems in the regulation of complex human behaviors, including emotional control, mood, stress response, cognition, pain processing, memory, and movement,³⁶ our findings have significant implications for understanding the myriad of behavioral, cognitive, and emotional changes commonly experienced by IBD patients.

The ALFF of fMRI signals reflects the strength or intensity of spontaneous regional brain activity, especially in deep gray matter structures. We found significantly increased ALFF in the thalamus, hippocampus, hypothalamus, and caudate of IBD patients compared with healthy controls. Amplitude of low-frequency fluctuation of the putamen and pallidum, however, was only increased in female IBD patients. Furthermore, we found increased ALFF of the hippocampus in patients with moderate to severe disease activity, relative to those with mild disease activity or remission. The striatum is involved in high-level socio-emotional function and reward processing, and the putamen is a key region within the basal ganglia-thalamic circuit that regulates motivation.³⁷ The hippocampus plays a critical role in memory and in mediating anxiety states in coordination with other limbic system components.³⁸ The thalamus is a relay center for sensory information transmission with widespread connections to cortical/subcortical regions and participates in cognitive and emotional processes.³⁹ Increased ALFF has previously been reported in the striatum and limbic systems of patients with generalized anxiety disorder (GAD), in which emotional dysfunction and hypervigilance is prominent.⁴⁰ In addition, patients suffering with primary insomnia, which manifests by hyperarousal and cognitive and emotional abnormalities, exhibit increased ALFF in the putamen, hippocampus, and insula.⁴¹ In both GAD and primary insomnia patients, increased ALFF in the striatum positively correlates with anxiety scores.⁴⁰ Anxiety and hypervigilance are symptoms commonly reported by IBD patients,⁴² and increased ALFF has been reported previously in CD patients in the insula and hippocampus.⁴³ These findings suggest that disorders associated with experiences of hypervigilence and anxiety, including IBD, may lead to enhanced neural activity in specific brain regions of the striatum and limbic systems.

In our IBD patient cohort, we found increased functional connectivity within and between limbic and basal ganglia structures compared with healthy controls. Previous studies have examined functional connectivity in IBD patients, with decreased functional connectivity between the hippocampus and cortical areas reported in CD patients,⁴³ and increased functional connectivity between the hippocampus and caudate in patients with active UC.⁴⁴ In addition, increased functional connectivity in CD patients between the frontoparietal and salience networks²⁸ and between the frontoparietal and default mode networks have been reported.⁴⁵ Female CD patients demonstrate enhanced functional connectivity compared with healthy controls between the insula and frontoparietal network.²⁸ The insula plays key roles in behavioral arousal and emotional responses, and through neural connections with the hypothalamus it critically regulates the stress response as part of the central autonomic network.⁴⁶ We found increased functional connectivity between the insula and the hypothalamus, hippocampus, and putamen. Similar to our ALFF observations in IBD patients, GAD patients also demonstrate increased functional connectivity between limbic and basal ganglia structures, including hyperconnectivity of the insula. It has been suggested that this hyperconnectivity may predispose GAD patients to perceive neutral or nonconflict stimuli as threatening, leading to abnormal anxiety circuit activation.⁴⁷ Striatal-limbic hyperconnectivity has also been reported in patients with social anxiety disorder, which is characterized by anxiety, fear, self-consciousness and embarrassment due to a feeling of being scrutinized or judged by others.⁴⁸ Moreover, the symptoms experienced by patients with GAD and social anxiety disorder have significant overlap with those commonly reported by IBD patients.⁴⁹ Interestingly, we found increased hypothalamic volume in IBD patients compared with healthy controls—a finding similar to that reported in patients with mood disorders and relevant to IBD patients who commonly experience altered mood.⁵⁰

Our findings of enhanced functional connectivity strength between many key CAN regions in IBD patients compared with healthy controls suggest that IBD alters communication within the CAN, which may in turn have implications for behavioral alterations observed in IBD patients. Interestingly, despite the lack of group differences in functional connectivity between the hypothalamus and hippocampus/thalamus, a significant decrease in functional connectivity strength was found between these regions in females but not in males. The hypothalamus acts as a relay center between limbic regions, the hypothalamic-pituitary-adrenal axis and ANS output neurons to regulate the stress response.⁵¹ Our findings suggest that IBD differentially impacts females and males in hypothalamic regulation of the stress response.

The amygdala is a key part of the limbic system, playing a central role in learning, memory, decision-making, reward, motivation, and regulation of emotional responses (including fear, anxiety).⁵² We found that there was an increase in amygdala volume in IBD patients, findings similar to observations from patients with GAD who also show increased amygdala volumes.⁵³ Moreover, the increase in amygdala volume was observed predominantly in female IBD patients with moderate to severe disease activity, and amygdala ALFF was increased only in male IBD patients. These findings suggest a sex-specific response of the amygdala to IBD disease activity.

Pain is commonly experienced by patients with IBD and forms a key symptom used to evaluate disease activity¹; therefore, we further evaluated associations for patients who reported pain, and those who did not, with MRI changes. We found increased thalamic size, but only in female IBD patients with pain. We also found that in the absence of pain, moderate to severe disease activity is associated with a decrease in thalamic volume. Moreover, we found that thalamic volume and thalamic-pallidum functional connectivity strength positively correlated with CRP levels. The thalamus is a key site for processing and modulating afferent pain impulses and relaying them to the cortex.⁵⁴ The thalamus also relays peripheral interoceptive stimuli reaching the brain from the body, including inflammatory signals.⁵⁵ The reason for the changes in thalamic volume is unknown. In somewhat contrast to our findings, thalamic atrophy has been observed in women with chronic pain syndromes.⁵⁶ What our findings do suggest, however, is that the impact of IBD on the thalamus is multifaceted and that pain and disease processes exhibit independent and sometimes opposing effects.

The implication of decreased tissue susceptibility within the brain is not well understood but appears to reflect increased calcium deposition associated with neurodegeneration.⁵⁷ Basal ganglia calcification is associated with the development of a number of symptoms including altered cognition and mood, as well as tremor/gait disturbances typical of Parkinson’s disease.⁵⁷ We identified significantly decreased susceptibility in 2 basal ganglia regions in IBD patients: the pallidum and putamen. These findings are consistent with the accumulating evidence that IBD appears to increase the risk of Parkinson’s disease and warrants further investigation.⁵⁸

We employed a novel multimodal approach that combines multiple layers of complex MRI data, comparing a broad spectrum of IBD patients with age/sex-matched healthy controls. However, we acknowledge some important limitations. First, due to logistical challenges with scheduling MRI and other investigations, we allowed a time window between MRI assessment and endoscopy/clinical measurements, recognizing that IBD disease activity fluctuates over time. Second, our study was cross-sectional in nature and cannot delineate dynamic changes in MRI, as they correlate with time-varying clinical and endoscopic parameters. Third, the number of patients in our subgroups (male/female, UC/CD, pain/no pain, active/inactive disease) were relatively low. The initial purpose of the present study was to investigate main effects for the entire IBD group by simply considering these subgroups as confounders in our analyses; however, the subgroup results were intriguing and were thus included. Given our findings, studies with larger subgroups are certainly warranted. Finally, assessments of behavioral symptoms and cognitive performance were not performed on our cohort, and so we cannot directly attribute the observed MRI findings to behavior or cognition. Future studies that include neuropsychological assessment are warranted.

In summary, we used multimodal MRI analyses to outline the impact of IBD on deep subcortical gray matter brain regions, including the basal ganglia and limbic system. Using these different techniques, we identified significant alterations within the basal ganglia and limbic systems in ALFF, functional connectivity, volume, and tissue susceptibility. In addition, we outlined the impact of pain and peripheral inflammation in regulating a number of these brain changes in IBD patients, as well as the potential importance of sex. We found hyperconnectivity between key brain regions comprising the central autonomic network. Considering the importance of these brain regions in emotional regulation, central autonomic outflow and the stress response, and the previous description of parallel brain changes found in other conditions that are associated with enhanced anxiety, hypervigilance, and altered pain processing, we postulate that the brain changes we have documented in IBD patients represent relevant striatal-limbic adaptive responses to the chronic inflammatory, emotional, painful, stressful, and unpredictable nature of their disease. An improved understanding of the central networks impacted by IBD may generate novel therapeutic approaches to improving the care of IBD patients and provide MRI measures to assess the efficacy of treatment of behavioral symptoms or to predict treatment response.

Funding

Research was funded by the Canadian Institutes of Health Research (CIHR) Team (Health Challenges in Chronic Inflammation Initiative) THC-135231.