Imaging characteristics of brain microstructure and cerebral perfusion in Crohn’s disease patients with anxiety: A prospective comparative study

Abstract

BACKGROUND

Anxiety is a common comorbidity in patients with Crohn’s disease (CD). Data on the imaging characteristics of brain microstructure and cerebral perfusion in CD with anxiety are limited.

AIM

To compare the imaging characteristics of brain microstructure and cerebral perfusion among CD patients with or without anxiety and healthy individuals.

METHODS

This prospective comparative study enrolled consecutive patients with active CD and healthy individuals who visited the study hospital between January 2022 and January 2023. Anxiety was measured using the Hospital Anxiety and Depression Scale-Anxiety. The imaging characteristics of brain microstructure and cerebral perfusion were measured by diffusion kurtosis imaging and intravoxel incoherent motion.

RESULTS

A total of 57 participants were enrolled. Among the patients with active CD, 16 had anxiety. Compared with healthy individuals, patients with active CD demonstrated significantly lower radial kurtosis values in the right cerebellar region 6, lower axial kurtosis (AK) values in the right insula, left superior temporal gyrus, and right thalamus, and higher slow and fast apparent diffusion coefficients (ADC_slow and ADC_fast) in the bilateral frontal lobe, bilateral temporal lobe, and bilateral insular lobe (all P < 0.05). Compared with patients with CD without anxiety, patients with CD and anxiety exhibited significantly higher ADC_slow values in the left insular lobe and lower AK values in the right insula and right anterior cuneus (all P < 0.05).

CONCLUSION

There are variations in brain microstructure and perfusion among CD patients with/without anxiety and healthy individuals, suggesting potential use in assessing anxiety-related changes in active CD.

Key Words: Crohn’s disease; Anxiety; Diffusion magnetic resonance imaging; Brain perfusion; Healthy controls; Prospective comparative study

Core Tip: The imaging characteristics of brain microstructure and cerebral perfusion measured by diffusion kurtosis imaging (DKI) and intravoxel incoherent motion (IVIM) vary significantly among Crohn’s disease (CD) patients with or without anxiety and healthy individuals. IVIM and DKI parameters may be used as imaging biomarkers to assess the brain microstructure and perfusion changes associated with anxiety in patients with active CD.

INTRODUCTION

Crohn’s disease (CD) is a chronic inflammatory bowel disease characterized by alternating periods of remission and exacerbation. During exacerbation periods, patients with CD generally report severe abdominal pain, diarrhea, and overall poor quality of life[1]. Patients with CD also demonstrate varying levels of emotional regulation disorders and may experience anxiety related to their illness[2]. The brain-gut axis is a two-way pathway composed of the central nervous system, autonomic nervous system, intestinal nervous system, and hypothalamic-pituitary-adrenal axis[3]. Changes along this pathway are likely responsible for the mental symptoms experienced by patients with CD, such as anxiety[4].

Neuroimaging markers, including markers identified on magnetic resonance imaging (MRI), can be used to evaluate emotional regulation in patients with CD. Previous neuroimaging studies in patients with CD generally used voxel-based morphology and resting-state functional imaging[5,6], revealing differences in brain regions involved in visceral pain and sensation, external environment monitoring, and cognitive processing between patients with CD and healthy individuals[5]. In addition, changes in brain microstructure and perfusion are associated with anxiety[7,8], including anxiety associated with autoimmune inflammatory diseases like lupus[9]. Few studies examined brain microstructure and perfusion in patients with CD[10], and the different methods available can produce different results.

Intravoxel incoherent motion (IVIM) imaging is an extension of diffusion-weighted imaging (DWI) and can be used to assess brain tissue diffusion and perfusion. IVIM is used to assess three quantitative parameters: Slow apparent diffusion coefficient (ADC_slow), fast apparent diffusion coefficient (ADC_fast), and perfusion fraction (f)[11]. A neuroimaging study showed that IVIM was more sensitive than voxel-based morphology when evaluating the microstructure of the brain in patients with Alzheimer’s disease[12]. Diffusion kurtosis imaging (DKI) is another type of MRI sequence that mainly reflects the non-Gaussian diffusion of water molecules in biological tissues. Compared with diffusion tensor imaging (DTI), DKI offers a more subtle and realistic assessment of the microstructural changes[13]. As a noninvasive quantitative MRI technique, DKI can simultaneously assess whole brain microstructure changes in the gray and white matter. This technique has been widely used in studies of bipolar disorder, brain glioma, Alzheimer’s disease, and other conditions[14,15]. Hypoperfusion of specific brain areas has been reported to be associated with anxiety and its severity[16,17]. Since DKI and IVIM can measure brain perfusion, changes in perfusion in IVIM and DKI could be associated with anxiety in patients with CD.

Therefore, this study aimed to investigate the changes in brain microstructure and perfusion associated with anxiety in patients with active CD and healthy individuals.

MATERIALS AND METHODS

Study design and participants

This prospective comparative study enrolled consecutive patients with active CD treated at the study hospital between January 2022 and January 2023. Healthy individuals matched for age, sex, and education were also enrolled during the same period among individuals seeking routine health checkups at the study hospital. The inclusion criteria were: (1) Patients with active celiac disease who first came to the study hospital and were not receiving formal treatment, as defined by the Consensus on Guidelines for International Inflammatory Bowel Disease Research Organizations (2020); (2) A simplified endoscopic score for CD (SES-CD) of > 3; (3) 18 to 55 years of age; (4) Right-handed; and (5) Capable of independently completing the Hospital Anxiety and Depression Scale-Anxiety (HADS-A) scale. The exclusion criteria were: (1) Known neuropsychiatric disorders; (2) Known neurological disease, such as head trauma, intracranial tumor, cerebral infarction, or neurodegenerative disease; (3) History of drug or alcohol abuse; (4) Contraindications to MRI, such as claustrophobia or the presence of dental fixtures or other exogenous objects in the head; or (5) Excessive head movements during MRI scanning. This study was approved by the Ethics Committee of Affiliated Changzhou Second People’s Hospital of Nanjing Medical University (approval number KY026-01). All participants provided written informed consent.

Procedure

General demographic data such as age, sex, height, weight, and body mass index were collected. For patients with CD, laboratory indicators were also collected, including white blood cell count (WBC), hemoglobin level (Hb), platelet count, erythrocyte sedimentation rate (ESR), and C-reactive protein level (CRP).

The HADS is a self-report scale consisting of two subscales, the HADS-A and the HADS-Depression (HADS-D), with a total of 14 entries, including seven for anxiety and seven for depression. As a screening tool, the scale is widely used in clinical practice to identify hospitalized patients with non-psychotic affective disorders[18]. The entries are scored on a four-point scale from 0 to 3, with a possible score of 0-21 for HADS-A and 0-21 for HADS-D. A score of ≥ 8 indicates the presence of anxiety and/or depression. All participants self-completed the HADS-A under the supervision of a trained neurologist before undergoing the study MRI. During the completion of HADS, the surrounding environment was kept quiet, and the participants were in a relaxed state.

The SES-CD strongly correlates with the CD endoscopic severity index, which is widely used nowadays[19,20]. SES-CD mainly assesses four endoscopic items (ulcer size, ulcer surface area, and proportion, proportion of surface area of diseased intestinal segments, and stenosis), with a score of 0 to 3 for each item and a maximum score of 56; the higher the score, the more severe the condition[21]. Scores of 0 to 2 suggest remission, 3 to 6 suggest mild activity, 7 to 15 suggest moderate activity, and scores > 16 suggest severe activity[21]. SES-CD scores were independently determined before MRI by two gastroenterologists with more than 10 years of clinical experience. The intraclass correlation coefficient (ICC) was ≥ 0.75, indicating consistency between the two physicians. The average score was taken as the final SES-CD score of the patient.

MRI examination

MRI examinations were performed using a Philips 3.0-T scanner (Achieva, Philips, Best, The Netherlands) with a standard 8-channel head coil. During the scans, the participants were asked to close their eyes and remain still and awake. Before the scan, the participants were given earplugs to attenuate scanner noise. Foam pads were placed on both sides of the head to limit head movement. A three-dimensional brain volumetric imaging (3D-BRAVO) sequence was used to collect high-resolution anatomical T1-weighted images, using repetition time (TR) of 6.7 millisecond, echo time (TE) of 3.0 millisecond, matrix of 240 × 240, field of view (FOV) of 240 mm × 240 mm, flip angle (FA) of 8°, slice thickness of 1 mm, slice number of 170, and a total scan time of 1 minute 34 seconds. The DKI data were collected using a single-shot echo-planar imaging sequence using TR of 8097 millisecond, TE of 80 millisecond, matrix of 88 × 185, FOV of 220 mm × 220 mm, FA of 90°, b-values of 0, 1000, and 2000 second/mm, and a total scan time, 9 minutes 3 seconds. The IVIM data collection parameters were TR of 1770 millisecond, TE of 61 millisecond, FOV of 220 mm × 220 mm, matrix of 124 × 121, b-values of 0, 20, 40, 80, 100, 150, 300, 500, 800, and 1000 second/mm, and a total scan time of 4 minutes 52 seconds.

Data processing

DKI data were first preprocessed using the FMRIB Software Library (FSL) (https://fsl.fmrib.ox.ac.uk/fsl) and Diffusion Kurtosis Estimator (DKE) (https://www.nitrc.org/projects/dke/) software. First, all DKI original DICOM format files were converted to 4DNIFTI format files. Using the image with b = 0 as a reference, the eddy correct program in FSL was used to correct the image distortion caused by eddy distortion from motion artifacts and high b-values. The DKE software was then used to calculate the parameters of the corrected images, including mean diffusion kurtosis (MK), axial diffusion kurtosis (AK), and radial diffusion kurtosis (RK)[14]. Finally, using the SPM12 software (https://www.fil.ion.ucl.ac.uk/spm/software/) on the Matlab2018b platform (MathWorks, Natick, MA, United States), the T1-weighted anatomical image was registered with the standard Montreal Neurological Institute template, and the b0 image was registered with the standardized T1 spatial image. The image parameters were entered into the DKI parameter graph to obtain the standardized parameter graph. The parameter graph was then smoothed (full width at half maximum, 6 mm), and the smoothed image was used for group comparisons.

The IVIM data were converted from the original DICOM files into 4DNIFTI files. The human brain module on the WFU software (https://www.nitrc.org/projects/wfu_pickatlas/) was applied using Matlab2018b to produce 10 regions of interest (ROIs), including the bilateral frontal lobes, bilateral temporal lobes, bilateral occipital lobes, and bilateral insular lobes. Then, using the IVIM module function software in SPM12[22], the preprocessed images and 10 interest area templates were imported together. The b-values were set to 0, 20, 40, 80, 100, 150, 300, 500, 800, and 1000 second/mm, and the parameters were automatically calculated. The parameter graphs for ADC_slow, ADC_fast, and f were obtained. The parameters of each brain region were imported into the ITK-SNAP software to extract the parameters for comparisons among groups[23]. The IVIM parameter diagram was consistent with the measured parameter values (Figure 1).

Figure 1 Intravoxel incoherent motion diffusion-weighted images from a 37-year-old man with Crohn’s disease and a 35-year-old healthy control man. A: Slow apparent diffusion coefficient (ADC_slow) map from biexponential fitting with 10 b-values. The left frontal ADC_slow values were 0.000779326 in the healthy control (HC) group and 0.000832584 in the Crohn’s disease (CD) patients measured by SPM12. The color scale from blue to red represents the range of ADC_slow values (0.000100-0.00250 × 10^-3 mm²/second); B: Fast apparent diffusion coefficient (ADC_fast) map from biexponential fitting with 10 b-values. The left frontal ADC_fast values were 0.016722 in the HC group and 0.019411 in the CD patients measured by SPM12. The color scale from blue to red represents the range of ADC_fast values (0.0100-0.0300 × 10^-2 mm²/seconds); C: Perfusion fraction (f) map from biexponential fitting with 10 b-values. The left frontal ADC_fast values were 0.117279 in the HC group and 0.115245 in the CD patients measured by SPM12. The color scale from blue to red represents the range of f values (-0.180% to 0.266%). HC: Healthy control; CD: Crohn’s disease.

Statistical analysis

All baseline data and IVIM parameters were analyzed using SPSS 23.0 (IBM, Armonk, NY, United States). The continuous data with a normal distribution were described as means ± SD and analyzed using a one-way analysis of variance among the three groups, with the independent sample t-test for post hoc test [the P values were corrected using the false discovery rate (FDR) method]; otherwise, they were presented as medians (interquartile range) and analyzed using the Kruskal-Wallis H-test. Categorical data were described as n (%) and analyzed using the χ² test or Fisher’s exact test. For the DKI data, the statistical software SPM12 was used to conduct a one-way analysis of variance for the three groups of parameter graphs, and the correction was performed using the FDR method. Using the xjview (https://www.alivelearn.net/xjview/) software in a single T1 template, different brain regions corresponding anatomical information graphs were analyzed, and different brain regions were extracted as the ROIs. The DKI parameters were measured. Pearson correlation analysis was used to study the correlation between the DKI and IVIM parameters, clinical characteristics, and psychological scores. For patients with CD, the receiver operating characteristic (ROC) curve method was used to identify neuroimaging markers that could be used to distinguish between patients with and those without anxiety. Two-sided P values < 0.05 were considered statistically significant.

RESULTS

Demographic data and psychological scores

Fifty-seven participants (33 males; mean age of 34 years) were enrolled, including 37 patients with active CD and 20 healthy individuals matched for age, sex, and education. Among the patients with active CD, 16 had anxiety. The duration of the disease was 2.69 ± 4.59 years, the SES-CD scores were 8.61 ± 6.03, WBC was 7.96 ± 4.14 × 10⁹/L, Hb levels were 113.88 ± 50.87 g/L, ESR was 27.33 ± 18.38 mm/hour, and CRP levels were 10.24 ± 11.09 mg/L. Compared with CD patients without anxiety, CD patients with anxiety had significantly higher SES-CD (13.82 ± 2.73 vs 5.71 ± 1.97, P < 0.001) and WBC (11.39 ± 2.62 vs 6.06 ± 1.48, P < 0.001). The HADS-A scores of active CD patients with anxiety were significantly higher than in patients with CD without anxiety and healthy individuals (both P < 0.001; Table 1).

Table 1 Demographic and clinical characteristics of the participants.

Variable	CD with anxiety (n = 16)	CD without anxiety (n = 21)	Healthy controls (n = 20)	P value
Demographic
Age (years)	37.44 ± 7.89	36.24 ± 9.26	35.75 ± 9.06	0.8461
Sex (n, male/female)	7/9	12/9	10/10	0.7182
Education (years)	15 (12,16)	12 (9,15)	12.00 ± 4.03	0.1713
Height (m)	1.65 ± 0.87	1.72 (1.58,1.75)	1.65 ± 0.09	0.3583
Weight (kg)	59.08 ± 10.92	60.14 ± 12.22	55 (50.25,71.5)	0.8573
Clinical characteristics
HADS-A score	13.88 ± 3.86	2.95 ± 1.26	3.26 ± 1.66	< 0.0011,a
SES-CD	13.82 ± 2.73	5.71 ± 1.97	-	< 0.0014,a
WBC, 109/L	11.39 ± 2.62	6.06 ± 1.48	-	< 0.0014,a
Hemoglobin, g/L	136.69 ± 18.86	131 (122.5,140)	-	0.6423
ESR, mm/hour	29.10 ± 18.21	25.88 ± 17.59		0.5924
CRP (mg/dL)	11.78 ± 5.81	5.9 (5,10.18)		0.0593
Duration of disease (years)	3.36 ± 1.59	2.10 ± 0.57		0.465

¹One-way analysis of variance.

²χ² test.

³Non-parametric test.

⁴Independent sample t-test.

^aP < 0.05.

CD: Crohn’s disease; HADS-A: Hospital Anxiety and Depression Scale – Anxiety dimension; SES-CD: Simplified endoscopic score for Crohn’s disease; WBC: White blood cells; ESR: Erythrocyte sedimentation rate; CRP: C-reactive protein level.

IVIM values

The ADC_slow values of bilateral frontal lobe, bilateral temporal lobe, and bilateral insular lobe in patients with CD were significantly higher than in healthy individuals (all P < 0.05). The ADC_slow values of the left insular lobe in patients with CD and anxiety were significantly higher than in patients with CD without anxiety (0.743 ± 0.050 vs 0.682 ± 0.025, P < 0.05; Table 2 and Figure 2). At the same time, the ADC_fast of the bilateral frontal lobe, bilateral temporal lobe, and bilateral insular lobe in patients with CD were significantly higher than in healthy individuals (all P < 0.05; Table 3 and Figure 3). There were no significant differences in f values among the three groups (Table 4).

Figure 2 Differences in slow apparent diffusion coefficient values between Crohn’s disease patients with anxiety, Crohn’s disease patients without anxiety, and healthy control groups. ADC_slow: Slow apparent diffusion coefficient; CD: Crohn’s disease; HC: Healthy control; NS: Not statistically significant.

Figure 3 Differences in fast apparent diffusion coefficient values between Crohn’s disease patients with anxiety, Crohn’s disease patients without anxiety and healthy control groups. ADC_fast: Fast apparent diffusion coefficient; CD: Crohn’s disease; HC: Healthy control; NS: Not statistically significant.

Table 2 Slow apparent diffusion coefficient values in the participants.

Brain ROI	ADCslow value (10-3 mm2/second)			P value1
	CD with anxiety (n = 16)	CD without anxiety (n = 21)	Healthy controls (n = 20)
Left frontal lobe	0.876 ± 0.071	0.871 ± 0.028	0.820 ± 0.014	< 0.001a
Left parietal lobe	0.819 ± 0.084	0.827 ± 0.010	0.753 ± 0.065	0.128
Left temporal lobe	0.703 ± 0.038	0.716 ± 0.034	0.648 ± 0.028	< 0.001a
Left occipital lobe	0.656 ± 0.044	0.711 ± 0.026	0.645 ± 0.069	0.052
Left insular lobe	0.743 ± 0.050	0.682 ± 0.025	0.656 ± 0.029	< 0.001a
Right frontal lobe	0.872 ± 0.068	0.852 ± 0.017	0.821 ± 0.028	0.001a
Right parietal lobe	0.818 ± 0.084	0.828 ± 0.011	0.754 ± 0.058	0.142
Right temporal lobe	0.664 ± 0.038	0.671 ± 0.023	0.629 ± 0.044	0.001a
Right occipital lobe	0.684 ± 0.054	0.694 ± 0.047	0.676 ± 0.059	0.193
Right insular lobe	0.758 ± 0.061	0.687 ± 0.024	0.658 ± 0.056	< 0.001a

¹One-way analysis of variance.

^aP < 0.05.

ADC_slow: Slow apparent diffusion coefficient; ROI: Region of interest; CD: Crohn’s disease.

Table 3 Fast apparent diffusion coefficient values in the participants.

Brain ROI	ADCfast Value (10-2 mm2/second)			P value1
	CD with anxiety (n = 16)	CD without anxiety (n = 21)	Healthy controls (n = 20)
Left frontal lobe	0.183 ± 0.017	0.171 ± 0.083	0.150 ± 0.025	< 0.001a
Left parietal lobe	0.189 ± 0.031	0.181 ± 0.027	0.216 ± 0.010	0.372
Left temporal lobe	0.186 ± 0.035	0.176 ± 0.085	0.153 ± 0.013	< 0.001a
Left occipital lobe	0.159 ± 0.019	0.169 ± 0.017	0.186 ± 0.033	0.437
Left insular lobe	0.198 ± 0.013	0.181 ± 0.011	0.153 ± 0.024	< 0.001a
Right frontal lobe	0.183 ± 0.017	0.181 ± 0.019	0.164 ± 0.011	0.001a
Right parietal lobe	0.184 ± 0.036	0.193 ± 0.024	0.254 ± 0.017	0.142
Right temporal lobe	0.197 ± 0.033	0.182 ± 0.015	0.164 ± 0.019	< 0.001a
Right occipital lobe	0.159 ± 0.019	0.169 ± 0.017	0.163 ± 0.019	0.437
Right insular lobe	0.122 ± 0.014	0.154 ± 0.019	0.178 ± 0.027	< 0.001a

¹One-way analysis of variance.

^aP < 0.05.

ADC_fast: Fast apparent diffusion coefficient; ROI: Region of interest; CD: Crohn’s disease.

Table 4 Perfusion fraction values in the participants.

Brain ROI	f value			P value1
	CD with anxiety (n = 16)	CD without anxiety (n = 21)	Healthy controls (n = 20)
Left frontal lobe	0.136 ± 0.026	0.140 ± 0.023	0.140 ± 0.017	0.116
Left parietal lobe	0.120 ± 0.021	0.113 ± 0.018	0.124 ± 0.021	0.614
Left temporal lobe	0.113 ± 0.021	0.104 ± 0.013	0.097 ± 0.011	0.190
Left occipital lobe	0.111 ± 0.038	0.101 ± 0.026	0.097 ± 0.021	0.517
Left insular lobe	0.110 ± 0.021	0.106 ± 0.027	0.092 ± 0.017	0.160
Right frontal lobe	0.142 ± 0.033	0.131 ± 0.014	0.122 ± 0.015	0.148
Right parietal lobe	0.113 ± 0.033	0.106 ± 0.019	0.101 ± 0.020	0.548
Right temporal lobe	0.099 ± 0.024	0.101 ± 0.012	0.096 ± 0.034	0.782
Right occipital lobe	0.102 ± 0.036	0.094 ± 0.018	0.094 ± 0.024	0.719
Right insular lobe	0.109 ± 0.037	0.111 ± 0.028	0.088 ± 0.024	0.267

¹One-way analysis of variance.

f: Perfusion fraction; ROI: Region of interest; CD: Crohn’s disease.

DKI values

Compared with the healthy control (HC) group, the AK values of the right insula (P = 0.019), left superior temporal gyrus (P = 0.002), right thalamus (P = 0.001), and right insula (P = 0.023) and the RK values of the right cerebellum region (P = 0.021) were significantly lower in the CD without anxiety group (Table 5 and Figure 4). Compared with the HC group, the AK values in the right insula (P = 0.023), left superior temporal gyrus (P = 0.015), right thalamus (P = 0.031), left middle temporal gyrus (P = 0.006), right inferior temporal gyrus (P = 0.001), and left lingual gyrus (P = 0.007) and the RK values of the right cerebellar region (P = 0.012) and the left hippocampus (P = 0.004) were significantly lower in the CD with anxiety group (Table 5 and Figure 5). Compared with patients with CD without anxiety, the AK values of the right insula (P = 0.002) and right anterior cuneus (P = 0.017) were significantly lower in the patients with CD and anxiety (Table 5 and Figure 6).

Figure 4 Brain regions with decreased diffusion kurtosis imaging parameters in Crohn’s disease patients without anxiety compared to healthy controls. A: Comparison of the axial kurtosis (AK) and radial kurtosis (RK) values among patients without anxiety Crohn’s disease (CD), patients with anxiety CD, and healthy controls (HC); B: AK values of brain regions were significantly lower in patients with CD without anxiety compared with HC; C: RK values of brain regions were significantly lower in patients with CD without anxiety compared with HC. ^aP < 0.05; NS: Not statistically significant; CD: Crohn’s disease; HC: Healthy control; AK: Axial kurtosis.

Figure 5 Brain regions with decreased diffusion kurtosis imaging parameters in Crohn’s disease patients with anxiety compared to healthy controls. A: Comparison of the axial kurtosis (AK) and radial kurtosis (RK) values among patients without anxiety Crohn’s disease (CD), patients with anxiety CD, and healthy controls (HC); B: AK values of brain regions were significantly lower in patients with CD without anxiety compared with HC; C: RK values of brain regions were significantly lower in patients with CD without anxiety compared with HC. ^aP < 0.05; NS: Not statistically significant; CD: Crohn’s disease; HC: Healthy control; AK: Axial kurtosis; RK: Radial kurtosis.

Figure 6 Brain regions with decreased diffusion kurtosis imaging parameters in Crohn’s disease patients with anxiety compared to Crohn’s disease patients without anxiety. A: Comparison of diffusion kurtosis imaging parameters among patients without anxiety Crohn’s disease (CD), patients with anxiety CD, and healthy controls; B: Brain regions with decreased axial kurtosis value in patients without anxiety CD compared with those with anxiety CD. ^aP < 0.05; NS: Not statistically significant; CD: Crohn’s disease; HC: Healthy control; AK: Axial kurtosis; RK: Radial kurtosis.

Table 5 Brain regions with significant differences in diffusion kurtosis imaging parameters between the Crohn’s disease and healthy control groups.

Indexes	Brain regions	AAL	MNI coordinates			P value
			X	Y	Z
AK	Right insula	30	46	2	-4	0.019a
	Left superior temporal	82	64	20	12	0.002a
	Right thalamus	78	0	2	22	0.001a
	Right insula	30	32	0	-8	0.023a
	Left superior temporal	82	56	0	-2	0.015b
	Right thalamus	78	8	28	0	0.031b
	Left medial temporal	85	56	4	-26	0.006b
	Right inferior temporal	90	44	10	44	0.001b
	Left lingual gyrus	47	28	80	-16	0.007b
	Right insula	30	42	0	8	0.002c
	Right anterior cuneate lobe	68	8	52	6	0.017c
RK	Right cerebellar region 6	100	4	32	40	0.021a
	Right cerebellar region 6	100	4	0	4	0.012b
	Left hippocampus	37	14	6	-14	0.004b

^aP < 0.05 for the Crohn’s disease (CD) without anxiety group vs the healthy control group.

^bP < 0.05 for the CD with anxiety group vs the healthy control group.

^cP < 0.05 for the CD with anxiety group vs the CD without anxiety group.

CD: Crohn’s disease; AAL: Automatic anatomical marking; MNI: Montreal Neurological Institute; AK: Axial diffusion kurtosis; RK: Radial diffusion kurtosis.

Correlation between IVIM values and clinical features and psychological scores

The ADC_slow values of the right insular lobe in patients with CD were positively correlated with the HADS-A scores (r = 0.55; P = 0.004), ESR (r = 0.67; P < 0.001), SES-CD scores (r = 0.60; P = 0.002), and duration of disease (r = 0.52; P = 0.008). The ADC_slow values of the left insula were positively correlated with the ESR (r = 0.40; P = 0.047) and disease duration (r = 0.40; P = 0.047). The ADC_slow values of the right frontal lobe were positively correlated with the disease duration (r = 0.41; P = 0.046). The ADC_fast values of the left frontal lobe were positively correlated with the SES-CD scores (r = 0.68; P < 0.001; Figure 7A). No correlations were found between the clinical features or psychological scores and ADC_slow or ADC_fast values in the remaining brain areas (Figure 7B).

Figure 7 The correlation between intravoxel incoherent motion parameters and clinical features and psychological scores in patients with Crohn’s disease. A: Slow apparent diffusion coefficient (ADC_slow) values in various brain regions and clinical features and psychological scores in patients with Crohn’s disease (CD); B: The correlation between fast apparent diffusion coefficient (ADC_fast) values in each brain region and clinical features and psychological scores in patients with CD. The color scale from blue to red represents the r value from -1 to 1. ^aP < 0.05; ESR: Erythrocyte sedimentation rate; HADS-A: Hospital Anxiety and Depression Scale, Anxiety subscale; SES-CD: Simplified endoscopic score for Crohn’s disease; TIME: Disease duration.

Correlation between DKI values and clinical features and psychological scores

The AK values of the right insula in the CD group were negatively correlated with the ESR (r = -0.39; P = 0.034), HADS-A scores (r = -0.70; P < 0.001), SES-CD (r = -0.77; P < 0.001), and duration of disease (r = -0.47; P = 0.009). The AK value of the left superior temporal gyrus in the CD group was negatively correlated with the HADS-A scores (r = -0.51; P = 0.004) and SES-CD (r = -0.49; P = 0.007; Figure 8).

Figure 8 The correlation between diffusion kurtosis imaging parameters and clinical features and psychological scores in patients with Crohn’s disease. A: The correlation between axial kurtosis (AK) values of the right insula and clinical features and psychological scores in patients with Crohn’s disease (CD); B: The correlation between AK values of the left superior temporal gyrus and clinical features and psychological scores in patients with CD. The color scale from blue to red represents the r value from -1 to 1. ^aP < 0.05; ESR: Erythrocyte sedimentation rate; HADS-A: Hospital Anxiety and Depression Scale, Anxiety subscale; SES-CD: Simplified endoscopic score for Crohn’s disease; TIME: Disease duration.

Differentiating between patients with CD with vs without anxiety

The AK values of the right insula and left superior temporal gyrus demonstrated areas under the curve (AUC) of 0.956 and 0.808, respectively, for distinguishing between patients with and without anxiety. The ADC_slow and ADC_fast values of the left frontal lobe, right frontal lobe, left temporal lobe, right temporal lobe, left insular lobe, and right insular lobe demonstrated AUC of 0.806, 0.718, 0.692, 0.736, 0.921, and 0.968, respectively (Figure 9).

Figure 9 Receiver operating characteristic curves for diagnostic efficacy of diffusion kurtosis imaging parameters and intravoxel incoherent motion parameters in patients with Crohn’s disease who had anxiety. A: The areas under the curve (AUC) for axial kurtosis values in the right insula and left superior temporal gyrus were 0.956 and 0.808, respectively; B: The receiver operating characteristic curves for slow apparent diffusion coefficient values and fast apparent diffusion coefficient values of the left frontal lobe, right frontal lobe, left temporal lobe, right temporal lobe, left insular lobe, and right insular lobe demonstrated AUC of 0.806, 0.718, 0.692, 0.736, 0.921, and 0.968 respectively.

DISCUSSION

The results showed that the imaging characteristics of brain microstructure and cerebral perfusion measured by DKI and IVIM vary significantly among CD patients with or without anxiety and healthy individuals. The DKI and IVIM parameters were correlated with psychological scores and clinical indicators in patients with CD. Therefore, the results suggest that the DKI and IVIM parameters had potential applicability as imaging biomarkers to assess brain microstructure and perfusion changes associated with anxiety in patients with active CD.

The ADC_slow value is an important indicator reflecting brain microstructure. The ADC_slow increases when the brain is affected by unpredictable demyelination, axon injury, or primary hyperplasia[24]. At the same time, the DKI parameters AK and RK can also reflect the destruction of the brain microstructure. AK reflects the kurtosis value of water molecules in the direction of maximum diffusion, whereas the RK value reflects the average kurtosis value perpendicular to the direction of the maximum vector and is susceptible to the influence of the myelin sheath. The complementarity of these two values makes it possible to reflect the MR diffusion characteristics of nerve tissue under normal or pathological conditions and jointly quantify the degree of limited diffusion of water molecules, reflecting the complexity of the tissue structure. When the complexity of the tissue structure decreases, the restriction degree of the water molecules is reduced, and the AK and RK values are reduced[25].

In this study, the ADC_fast values of the two CD groups were higher than those of the HC group. As previously shown, ADC_fast values can be used to evaluate tissue perfusion related to microcirculation[26]. Pain in patients with CD can activate the stress system, causing peripheral vascular contraction and cerebrovascular dilation to ensure blood supply to important organs. It may cause increased perfusion in the bilateral frontal lobe, bilateral insular lobe, and bilateral temporal lobe in patients with CD. In addition, these brain regions are important parts of the gut-brain axis. In CD, they receive pain signals from the gut and are maintained in an excited state over an extended period, leading to blood vessel dilation and thereby causing increased perfusion. Increased perfusion in the frontal, insular, and temporal lobes activates these areas, and an over-active state may lead to negative emotions such as anxiety[27]. Similarly, the ADC_fast value of patients with CD and anxiety was higher than in patients with CD without anxiety, which may be caused by the severity of the disease.

This study found that patients with CD, irrespective of anxiety, had impaired brain microstructure in the frontal, temporal, insula, superior temporal gyrus, thalamus, and right cerebellar region 6 compared with healthy individuals, which may be caused by the secretion of intestinal inflammatory factors caused by CD. Normally, the gut prevents intestinal contents from entering the lamina propria and the bloodstream through the gut-vascular barrier (GVB)[28]. Similarly, when in homeostasis, the central nervous system is separated from the peripheral circulation by the blood-brain barrier (BBB). In patients with CD, pro-inflammatory cytokines produced in the gut increase intestinal permeability and impair the GVB. Due to the similarity of the BBB and GVB in composition and structure, after entering circulation, inflammatory factors will also impair the BBB[29], entering the central nervous system and causing neuroinflammation and structural damage.

The frontal lobe plays an important role in emotional behavior and cognitive function[30], while the temporal lobe can combine emotional information with environmental information to trigger adaptive physiological, behavioral, and emotional responses, including the adaptive fear response[31,32]. Therefore, changes in the microstructure of this region may potentially increase anxiety. In addition, damage to the insula, superior temporal gyrus, thalamus, and right cerebellar region 6 Leads to a decline in patients’ analgesic response and the ability to maintain positive emotions and pleasant experiences[33-36] and causes changes in patients’ perception habits of pressure[37], causing anxiety.

This study also found that the brain microstructure of the middle temporal gyrus, inferior temporal gyrus, and lingual gyrus in patients with anxiety was damaged to varying degrees compared with the HC group. Meanwhile, the insular and precuneus structures were damaged in patients with CD and anxiety compared with those without anxiety. Combined with the laboratory measurements and SES-CD scores, the patients with anxiety were generally sicker and had a longer illness than those without anxiety. Pain response and inflammatory factors generated by the disease activate the stress system of the body, in which the hypothalamic-pituitary-adrenal axis is the main response pathway of the stress system and also the main efferent pathway of the gut-brain axis[38]. The hippocampus, as an important part of the limbic system, is the dual-core of the cerebro-intestinal axis and emotional perception[39]. Damage to the hippocampus may lead to the destruction of internal environmental homeostasis and an increase in sympathetic nerve excitability. The continuous excitation of the sympathetic nervous system will not only lead to anxiety but also increase the secretion of inflammatory factors and further aggravate intestinal symptoms, forming a vicious cycle[40]. Intestinal inflammation triggered by local infection, dysbiosis, or food antigens induces the release of pro-inflammatory cytokines [such as interferon-γ, interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α, among others] that can have dramatic extraintestinal consequences if uncontrolled. Under homeostatic conditions, the central nervous system is completely segregated from the peripheral circulation by the BBB, which shares molecular and cellular similarities with the GVB[41-43]. Therefore, it can be hypothesized that due to the similarities between GVB and BBB, the intestinal inflammatory factors that affect the GVB will also affect the BBB after entering peripheral circulation in patients with CD. The increase in intestinal metabolites in patients with CD causes a decrease in pain regulation and emotional processing, leading to an anxious state in patients with CD. Therefore, for the treatment of patients with CD, controlling the intestinal symptoms and improving the anxiety state at the same time may improve the therapeutic effect.

The increase in intestinal metabolites in patients with CD causes a decrease in pain regulation and emotional processing, leading to an anxious state in patients with CD. Therefore, for the treatment of patients with CD, controlling the intestinal symptoms and improving the anxiety state at the same time may improve the therapeutic effect.

No significant differences were seen among the three groups in the MK (DKI parameter) or f (IVIM parameter) values. MK reflects the average value of excess kurtosis in all directions. It can be speculated that because of the overall short duration of illness of the participants, the excess peaks in all directions could still compensate for each other, so there were no significant differences among groups in MK values. A previous study found that f values had low stability in patients with head and neck cancers[44], which may explain the lack of difference in f values among groups in the current study.

In this study, in patients with CD, the ADC_slow values of the right insula were positively correlated with the HADS-A scores. In addition, the AK values in the right insula and the left superior temporal gyrus were negatively correlated with the HADS-A scores. The changes in ADC_slow and AK values reflect damage to the tissue microstructure in patients with CD, in turn affecting the perception and processing of pain, amplifying the unpleasant emotions generated by pain, and leading to anxiety and depression. Moreover, the higher the ADC_slow values or the lower the AK values, the more serious the tissue damage is and the higher the levels of anxiety and depression. This study also showed that the ADC_slow values of the right insular lobe were positively correlated with the ESR and SES-CD, and the ADC_slow values of the left insular lobe were positively correlated with the ESR. The AK values of the right insula were negatively correlated with the ESR and SES-CD, and the AK values of the left superior temporal gyrus were negatively correlated with the SES-CD. SES-CD reflects disease activity in patients with CD[21], and an increase in ESR usually indicates that patients are in the active stage of the disease. In the active stage, patients with CD produce inflammatory mediators that stimulate the increase of TNF-α, nitric oxide synthase expression, and nitrite content through the hypothalamic-pituitary-adrenal axis. It may cause changes in the excitability and behavior of the central nervous system[4] and affect the psychological state of the patients. The ADC_fast value of the left frontal lobe was also positively correlated with SES-CD, which may be due to the expansion of blood vessels caused by inflammatory stimulation, resulting in increased perfusion. Finally, the ADC_slow values of the right insula, left insula, and right frontal lobe were positively correlated with disease duration, and the AK values of the right insula were negatively correlated with disease duration. It may result from long-term stimulation of the relevant brain regions by inflammatory mediators, which leads to increasingly serious damage to the brain microstructure.

At the same time, the AK values of the right insula and left superior temporal gyrus, the ADC_slow values of the right insula, left temporal lobe, and left frontal lobe, and the ADC_fast values of the right insula, left insula, right frontal lobe, and left temporal lobe could differentiate between patients with and without anxiety. In this study, the HADS-A scores were evaluated by a professional neurologist to ensure the objectivity of the score results. However, in actual clinical diagnosis, the HADS-A score is closely related to the emotional state of the patient during the evaluation, and some patients do not cooperate well with the evaluation. In addition, the HADS-A score is subjective to some extent. The results of the HADS-A scale can be affected by many factors (e.g., the noise and lighting environment, the approach and attitude of the neurologist performing the test, etc.)[45]; it is highly subjective and unstable. Furthermore, HADS-A can be limited by the education level of patients (i.e., patients with low education can understand some questions poorly, and patients with high education can answer according to the diagnostic they would wish to have)[45]. Therefore, the IVIM and DKI parameters can be used as supplementary means to evaluate the anxiety state of patients with CD and are more objective and less affected by the emotional fluctuations of patients. DKI and IVIM parameters could provide more objective and stable evaluation methods. Additional studies are necessary to determine the pathophysiological mechanisms responsible for the changes in neuroimaging biomarkers in patients with CD and anxiety and how those biomarkers can be used to predict or manage anxiety. Nevertheless, the present study provides data about the IVIM and DKI parameters to explore the brain microstructure and brain perfusion changes in the anxiety state of patients with CD in order to explore the mechanism of anxiety in CD patients so as to provide neuroimaging evidence for psychological intervention in clinical treatment. IVIM and DKI parameters have certain diagnostic efficacy for the anxiety state of patients with CD, providing theoretical support for the prediction model constructed using IVIM and DKI parameters in future studies.

The study had several limitations. First, all participants were from a single center, limiting the sample size and possibly introducing bias. Second, this study only studied the changes in brain microstructure and perfusion in patients with CD with or without anxiety and in healthy individuals without anxiety. In the future, patients with anxiety alone will be included for component comparison to explain better the bidirectional relationship between CD and anxiety. Third, the small sample size may result in low statistical power. Fourth, it was a cross-sectional study, preventing causality analyses. In the future, we plan to conduct a longitudinal study with a larger sample size to explore the dynamic changes in brain microstructure during CD and to assess the influence of different treatment schemes on brain microstructure. Fifth, the CD activity index was not recorded because of its limited value for study purposes[46,47]. Finally, some examinations were not performed in this study. At present, the gold standard for brain microstructure changes is a pathological examination of a brain specimen, which has obvious issues and unacceptable risks preventing its use in antemortem research and clinical work. Currently, computed tomographic perfusion (CTP) is recognized as a reliable method for evaluating cerebral perfusion[48,49]. However, CTP examination requires injection of contrast agents and radiation, and it lacks sensitivity to detect small and subtle changes, making it unsuitable for this study.

CONCLUSION

In conclusion, imaging characteristics of brain microstructure and cerebral perfusion measured by DKI and IVIM vary significantly among CD patients with or without anxiety and healthy individuals. IVIM and DKI parameters may be used as imaging biomarkers to assess brain microstructure and perfusion changes associated with anxiety in patients with active CD.

ACKNOWLEDGEMENTS

The authors acknowledge the help of Dr. Wen-Jia Liu in analyzing the clinical data.