Circulating and Magnetic Resonance Imaging Biomarkers of Intestinal Fibrosis in Small Bowel Crohn’s Disease

Abstract

Background

We previously identified circulating and MRI biomarkers associated with the surgical management of Crohn’s disease (CD). Here we tested associations between these biomarkers and ileal resection inflammation and collagen content.

Methods

Fifty CD patients undergoing ileal resection were prospectively enrolled at 4 centers. Circulating CD64, extracellular matrix protein 1 (ECM1), GM-CSF autoantibodies (GM-CSF Ab), and fecal calprotectin were measured by ELISA. Ileal 3-dimensional magnetization transfer ratio (3D MTR), modified Look-Locker inversion recovery (MOLLI) T1 relaxation, diffusion-weighted intravoxel incoherent motion (IVIM), and the simplified magnetic resonance index of activity (sMaRIA) were measured by MRI. Ileal resection specimen acute inflammation was graded, and collagen content was measured quantitatively using second harmonic imaging microscopy. Associations between biomarkers and ileal collagen content were tested.

Results

Median (interquartile range [IQR]) age was 19.5 (16-33) years. We observed an inverse relationship between ileal acute inflammation and collagen content (r = −0.39 [95% confidence interval {CI}: −0.61, −0.10], P = .008). Most patients (33 [66%]) received biologics, with no variation in collagen content with treatment exposures. In the univariate analysis, CD64, GM-CSF Ab, fecal calprotectin, and sMaRIA were positively associated with acute inflammation and negatively associated with collagen content (P < .1). The multivariable model for ileal collagen content (R² = 0.31 [95% CI: 0.11, 0.52]) included log CD64 (β = −.27; P = .19), log ECM1 (β = .47; P = .06), log GM-CSF Ab (β = −.15; P = .01), IVIM f (β = .29, P = .10), and IVIM D* (β = 1.69, P = .13).

Conclusions

Clinically available and exploratory circulating and MRI biomarkers are associated with the degree of inflammation versus fibrosis in CD ileal resections. With further validation, these biomarkers may be used to guide medical and surgical decision-making for refractory CD.

Key Messages.

What is already known?

• Despite advances in medical therapy, many patients with ileal Crohn’s disease (CD) develop fibrosis-related complications requiring surgery. Biomarkers to noninvasively assess the degree of inflammation or fibrosis in the ileum are lacking.

What is new here?

• Our study has shown that circulating granulocyte-macrophage colony-stimulating factor autoantibodies, CD64, and ECM1 as well as MRI diffusion-weighted intravoxel incoherent motion measures of tissue perfusion are associated with ileal resection collagen content based on quantitative second harmonic imaging microscopy.

How can this study help patient care?

• Following external validation, these biomarkers may guide clinical practice in terms of therapies targeting inflammation versus fibrosis and might ultimately be used to enrich clinical trials for novel anti-fibrotic agents in CD patients who are more likely to have significant tissue fibrosis at study entry.

Introduction

Recent research has suggested that while inflammation is required for the initiation of fibrosis in Crohn’s disease (CD), intestinal strictures may ultimately progress independently of inflammation.¹ Patients with active inflammation may be offered newer biologics or small molecule therapies, while those with advanced fibrosis would benefit from surgical management. Histologically, fibrostenosis manifests as submucosal fibrosis and/or smooth muscle hyperplasia.² These histologic changes lead to luminal narrowing, stricturing disease with associated luminal dilation, and internal penetrating disease.³ Obstructing and penetrating complications remain common despite the use of advanced medical therapies and are a considerable source of morbidity and healthcare costs.⁴ Moreover, these complications may vary across racial and ethnic groups.⁵ The ability to noninvasively detect the degree of bowel wall fibrosis remains a critical unmet need.

Potential circulating biomarkers of intestinal fibrosis or inflammation include granulocyte-macrophage colony-stimulating factor (GM-CSF) autoantibodies (Ab) and extracellular matrix protein 1 (ECM1), both previously linked to future risk of disease complications, and soluble CD64 (Fcγ receptor I) which has been linked to more severe acute inflammation and a refractory disease course.^6–9 Potential MRI-based biomarkers include bowel wall measurements of magnetization transfer (MT, quantified as MT ratios [MTR]), T1 relaxation mapping, and molecular water diffusion and perfusion (quantified using diffusion-weighted imaging apparent diffusion coefficients [ADC], and intravoxel incoherent motion [IVIM] metrics [namely f, D*, and D, respectively]).^10–13 However, these biomarkers have yet to be systematically compared to comprehensive transmural tissue assessments of bowel wall inflammation and fibrosis. Moreover, while several methods have been used to measure tissue fibrosis in CD, including histologic grading of hematoxylin and eosin (H&E), Masson’s trichrome, or Sirius red staining, recent studies have suggested that quantitative second harmonic imaging microscopy (SHIM) may provide a more reliable, objective approach.^14,15

We recently reported that several circulating and MRI biomarkers differed between pediatric and adult ileal CD patients receiving medical versus surgical management.¹⁶ Patients requiring ileal resection exhibited increased circulating GM-CSF Ab as well as increased MRI ileal 3-dimensional MT ratio (3D MTR) measurements and simplified magnetic resonance index of activity (sMaRIA) scores. The purpose of this study was to evaluate the relationships between these circulating and MRI biomarkers obtained shortly before intestinal resection and ileal fibrosis measured using a standardized, quantitative SHIM measure of tissue collagen. A secondary aim was to compare the relationship of histologic acute inflammation to overall bowel wall collagen content. We found that patients with elevated circulating levels of GM-CSF Ab and soluble CD64, previously linked to acute inflammation, exhibited lower levels of tissue fibrosis, while patients with elevated circulating levels of ECM1 and specific diffusion-weighted IVIM measures of tissue perfusion exhibited higher levels of tissue fibrosis.

Materials and Methods

Ethics Statement

This multicenter, prospective, cross-sectional study was primarily approved by the human research ethics committee at Cincinnati Children’s Hospital Medical Center (2019-0677; initially approved June 20, 2019); the other 3 study sites also approved this research protocol. Oral and written informed consent was obtained from all participants (or a parent/guardian for pediatric participants). For pediatric participants (11-17 years of age), oral and written informed assent was also obtained.

Study Sample

Between December 2019 and August 2022, we enrolled 50 pediatric (8-17 years old) and adult (18-70 years old) CD patients with documented stricturing (B2) or internal penetrating (B3) behavior based on clinical MRI, CT, and/or ileocolonoscopic assessment, who were scheduled to undergo surgical resection of the distal ileum. Prior to surgery, each participant underwent a dedicated research noncontrast MRI examination and blood-based biomarker testing. The median time between the research MRI/measurement of circulating biomarkers and ileal resection was 8 (1, 42) days. At the time of research MRI, 24 (48%), 13 (26%), and 13 (26%) of the participants demonstrated imaging features suggestive of apparent inflammatory (B1), B2, and B3 phenotypes, respectively. However, 45 (90%) showed wall thickening with greater than 50% luminal narrowing on research imaging, demonstrating the persistence of luminal narrowing. A participant flow diagram is presented in Figure S1. Participant exclusion criteria included prior small bowel resection, stricturoplasty, or endoscopic treatment of a small bowel stricture, inability to undergo an MRI examination without sedation, MRI unsafe medical device, and/or known or suspected pregnancy. Please see the Supplementary Methods, Figure 1, and Figure S2 for details of the MRI protocols and biomarker measurements.

Figure 1.

MRI biomarker sequences. Representative images are shown for a 46-year-old with stricturing Crohn’s disease of the terminal ileum. (A) Coronal single-shot fast spin-echo MR image shows ileal wall thickening, luminal narrowing >50%, and dilated upstream bowel, consistent with a stricture (arrows). (B) Similar findings are seen on axial single-shot fast spin-echo fat-saturated MR imaging (arrows). (C) Axial 3D magnetization transfer (MT) MR image without the application of an off-resonance MT radiofrequency saturation pulse shows that the strictured ileum is relatively isointense to fat (arrows). (D) Axial MT MR image with the application of an off-resonance MT radiofrequency saturation pulse applied shows selective loss of signal intensity in the ileal wall (arrows) and skeletal muscle. Bowel wall signal intensity measurements from figure parts C and D are used to calculate an MT ratio (MTR). (E) Axial T1 modified Look-Locker inversion recovery (MOLLI) relaxation parametric map showing the strictured ileum (arrows). (F-I) Examples of axial spin-echo echo planar diffusion-weighted intravoxel incoherent motion (IVIM) images with increasing amounts of diffusion weighting (b-values = 0, 100, 400, and 800 s/mm², respective) through the terminal ileum (arrows). Change in bowel wall signal intensity with increasing b-value is used to calculate IVIM parameters f, D*, and D using a biexponential curve fit.

Our study sample was included in another study by our group that sought to identify MRI and circulating biomarkers associated with the need for surgical management in ileal CD.¹⁶ That study tested differences in the MRI and circulating biomarkers included in the current report between CD patients undergoing ileal resection and CD patients with uncomplicated disease. It included 50 surgical patients also included in the current investigation, as well as 83 patients receiving medical therapy and 42 non-IBD controls that are not included in the current investigation.¹⁶ The other study from our group did not include the ileal resection tissue assessments, including SHIM and histopathology data, which are the focus of the current report. Specifically, the current study builds upon this previous work by systematically testing associations between our novel MRI and circulating biomarkers of interest and measurements of ileal resection inflammation and collagen content (including quantitative SHIM assessments).

Ileal Stricture Tissue Assessments

For each participant, a transmural tissue specimen was obtained from the ileal resection (Figure 2). Specimen location was determined by a pathologist or pathologist assistant, with tissue obtained from the area of the most severe ileal stricturing disease, defined by the most severe luminal narrowing accompanied by the greatest wall thickening. Tissue specimens were embedded in paraffin, and 5-micron-thick sections were prepared for tissue staining as well as SHIM analyses of unstained sections¹⁸. H&E, Masson’s trichrome, and Sirius red tissue staining were performed by the Integrative Morphology shared facility at Cincinnati Children’s Hospital Medical Center (Figures 2 and S3). Further details for pathologist scoring and second harmonic and polarized light microscopy are provided in the Supplementary Methods.

Figure 2.

Inflammatory and fibrotic features of ileal resections. (A) Five-micron-thick sections were prepared from ileal resection strictures and stained with hematoxylin and eosin (H&E) for grading of active and chronic inflammation and smooth muscle hyperplasia, and with Masson’s trichrome and Sirius red with polarized light microscopy for grading of fibrosis. Second harmonic imaging microscopy (SHIM) of unstained sections was used to quantify fibrillar collagen in the submucosa and overall bowel wall. (B) Scatter plots are shown for active inflammation H&E score and fecal calprotectin versus overall ileal collagen content measured by SHIM. Group I: acute inflammation less than the highest quartile with collagen content greater than the median value; Group II: acute inflammation less than the highest quartile with collagen content less than the median value; Group III: acute inflammation greater than the highest quartile with collagen content less than the median value. (C) Representative surgical specimens and ileal tissue section images (H&E, Masson’s trichrome, and SHIM) are shown for a Group I patient with lower acute inflammation and higher submucosal collagen content, and a Group III patient with higher inflammation and lower submucosal collagen content. (D) Heat map for pairwise correlations between ileal histologic features is shown (*P < 0.1). (E) The frequency of B1 inflammatory, B2 stricturing, or B3 internal penetrating disease behavior on the preoperative research MRI based on patient subgroups defined in figure part B is shown. AI = acute inflammation, CI = chronic inflammation, Tri = trichrome, SM = submucosal, MH = smooth muscle hyperplasia, MD = muscle thickness.

Participant Classification

We explored variation in acute inflammation and collagen content between patients using scatter plots of (1) bowel wall acute inflammation histology scores versus bowel wall collagen content measured by SHIM, and (2) fecal calprotectin versus bowel wall collagen content measured by SHIM. Elevated acute inflammation was defined by an ileal bowel wall acute inflammation histology score, or fecal calprotectin, within the highest quartile. Elevated ileal collagen content was defined by a SHIM value within the 2 highest quartiles.

Sample Size

We hypothesized that 1 or more biomarker measurements would correlate with the quantitative SHIM measure of ileal fibrosis. The primary predictor variable for this analysis was the normalized 3D MTR value of the most affected ileal segment. Based upon the recent report from Li et al. in adult CD patients,¹¹ and our study comparing CD patients with medical versus surgical management,¹⁶ we anticipated that the normalized 3D MTR value would increase with increasing ileal collagen content as measured by SHIM. Our sample size of 45 participants (ie, the number of participants with diagnostic SHIM collagen measurements available for analysis) provided 80% power to detect a correlation between normalized 3D MTR and ileal collagen content of 0.4, with an alpha of .05.

Analysis Objectives

Our primary analytic objective was to determine the discriminant power of the normalized ileal 3D MTR measurement to detect the degree of fibrosis in CD ileal resection specimens measured by SHIM. We then planned to assess the additional discriminant power resulting from including the circulating biomarkers (CD64, ECM1, and GM-CSF Ab), and additional exploratory MRI measures (modified Look-Locker inversion recovery [MOLLI] T1 relaxation estimates, IVIM metrics, and sMaRIA scores¹⁷) in a multivariable linear regression model for ileal collagen content. These biomarkers were selected because they were associated with surgical management of CD.¹⁶ We did not include more conventional MRI assessments, such as bowel wall thickness, in our model as they have been previously investigated, and because we were mainly interested in investigating novel MRI and circulating biomarkers. The study design and analytic approach are reported as per the most recent TRIPOD guidelines (Tripod statement [tripod-statement.org]). See Supplementary Methods for the TRIPOD checklist.

Statistical Analysis

Continuous variables were summarized as medians and interquartile ranges (IQRs), while categorical variables were summarized as counts and percentages. Continuous variables were compared using the Mann–Whitney U test, while categorical variables were compared using Fisher’s exact test. Spearman rank-order correlations and heat maps¹⁸ of correlation coefficients were used to evaluate and display the associations between (1) MRI biomarkers, (2) circulating and fecal biomarkers, and (3) tissue assessments of bowel wall acute inflammation, fibrosis, and smooth muscle hyperplasia. Bowel wall collagen content as measured by SHIM was the primary endpoint. Raw P-values are presented without statistical adjustment because of our relatively small sample size and the exploratory nature of our investigation.

The following variables were log-transformed for correlation and regression analyses due to skewed distributions: bowel wall collagen content measured using SHIM, MOLLI T1, ECM1, GM-CSF Ab, CD64, disease duration, ESR, CRP, and fecal calprotectin. The following histologic ordinal outcomes were square root transformed again to make variables more normally distributed: overall bowel wall acute inflammation, chronic inflammation, smooth muscle hyperplasia, and fibrosis. These transformations were performed to help ensure regression normality assumptions were met, including homoscedasticity (ie, residuals having a constant variance).

Multivariable linear regression analysis was used to evaluate the associations between the primary outcome (ileal stricture collagen content measured using SHIM), age, and the prespecified primary predictor normalized 3D MTR (bowel wall 3D MTR normalized to skeletal muscle MTR), and the prespecified secondary predictors CD64, ECM1, GM-CSF Ab, MOLLI T1, IVIM ADC, IVIM f, IVIM D*, IVIM D, and sMaRIA. A stepwise model selection process was employed with the AICc criteria used for final model selection. AICc was used to help protect against model overfitting. The accuracy and precision of the final model fit R² and its associated 95% confidence interval (CI) were obtained by the bootstrapping validation method with 500 replicates. In the model selection process, the initial model was selected based on patients with all complete covariates under consideration. The final model was then re-fit on patients with complete covariates that have been selected above, which is usually slightly more than the initial pool of patients. Since the missing covariates selected were not substantial, and given the exploratory nature of this model building, we did not impute any missing data. To further evaluate the associations between our predictor variables of interest and primary outcome, collagen content measured using SHIM, logistic regression analysis with proportional odds assumption was also performed (Supplementary Results).

All statistical analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC) or R software (https://www.R-project.org). A P-value less than .05 was considered significant for inference testing unless otherwise indicated.

Results

Study Sample

Fifty patients were included in our study, including 17 children and 33 adults (overall median age = 19.5 years; IQR: 16-33 years). Twenty-seven (54%) participants were female, and 40 (80%) were White. Thirty-three (66%) participants were on biologic therapy at the time of research MRI, while 11 (22%) had recent corticosteroid exposure. Biologic therapies included infliximab (n = 13 [26%]), adalimumab (n = 8 [16%]), ustekinumab (n = 10 [20%]), and vedolizumab (n = 2 [4%]). Median disease duration was 943 days (IQR: 362-2428 days). Participant characteristics for the overall study sample are presented in Table 1, including by age group. Pediatric patients had a shorter disease duration, on average, and higher CRP and fecal calprotectin measurements suggesting that at the time of surgery, they were experiencing a higher level of acute inflammation.

Table 1.

Clinical and demographic features.

Variable	Overall (n = 50)	Children (n = 17)	Adults (n = 33)	P-value
Age (years)	19.5 (16.0-33.0)	14.0 (12.0-16.5)	26.0 (19.5-39.5)	<.0001
Sex (F/M, % F)	27/23 (54.0%)	9/8 (52.9%)	18/15 (54.5%)	>.99
Race, self-reported (non-White/White, % non-White)	10/40 (20%)	7/10 (41.2%)	3/30 (10.0%)	.02
Disease location at enrollment (Montreal Classification)	L1–17 L3–33	L1–7 L3–10	L1–10 L3–23	.53
Biologic exposure (Yes/No)	33/17 (66.0%)	14/3 (82.4%)	19/14 (57.6%)	.12
Corticosteroid exposure (Yes/No)	11/39 (22.0%)	5/12 (29.4%)	6/27 (18.2%)	.48
Disease duration (days)	943 (362-2428)	451 (117-909)	1837 (725-3907)	.0003
BMI (kg/m2)	22.9 (19.0-26.1)	20.9 (18.2-24.5)	24.1 (20.4-27.7)	.17
CRP (normalized ULN) (mg/L)	1.0 (0.38-5.1)	1.8 (1.0-7.7)	0.5 (0.38-3.1)	.003
Albumin (g/dL)	4.0 (3.5-4.4)	3.7 (2.9-4.0)	4.3 (3.9-4.7)	.0001
Hematocrit (%)	40.9 (38.3-43.2)	37.4 (34.6-40.9)	41.8 (39.5-44.7)	.0006
Fecal calprotectin (mcg/gm)	362 (130-1718)	1719 (634-4463)	195 (79-1307)	.0003

Abbreviations: BMI = body mass index; CRP = c-reactive protein; ULN = upper limits of normal.

Data are presented as medians and interquartile ranges, or n (%). P-values comparing pediatric and adult values based on Mann–Whitney U test or Fisher’s exact test, as appropriate.

Summary of Ileal Stricture Tissue Assessments and Correlations

Five-micron-thick sections were prepared from ileal strictures, and overall bowel wall and submucosal collagen content were measured using confocal SHIM for regions of interest as shown in Figures 2A and S3. A prior study suggested that a subgroup of CD patients exhibit high levels of both inflammation and fibrosis within the ileal resection.¹⁹ We therefore first tested for associations between inflammation and fibrosis within the ileal specimens. In contrast to this prior study, we observed an inverse relationship between ileal acute inflammation and collagen content (r = −0.39 [95% CI: −0.61, −0.10], P = .008). Next, we created scatter plots for measures of acute inflammation versus collagen content. This revealed 3 subgroups of patients. Group I patients (n = 18) exhibited relatively lower levels of acute ileal inflammation or fecal calprotectin, and increased collagen content, while Group III patients (n = 12) exhibited relatively higher levels of ileal acute inflammation or fecal calprotectin, and lower levels of tissue collagen (Figure 2B and 2C). Group II patients (n = 15) exhibited relatively lower levels of both acute ileal inflammation and fibrosis.

Tissue assessments of histologic acute inflammation, chronic inflammation, fibrosis, and smooth muscle hyperplasia based upon scoring of H&E and trichrome sections as well as measurements of collagen content measured by SHIM and Sirius red staining with PLM are presented in Table 2, including by age group. The highest grades of acute and chronic inflammation were observed in the mucosa, with chronic inflammation also detected in the submucosa. Across the bowel wall layers, the highest level of fibrosis detected by SHIM, trichrome staining, and Sirius red staining with PLM was in the submucosa. The median (IQR) submucosal collagen content measured by SHIM was 31% (21%-46%), while the median (IQR) submucosal collagen content measured by Sirius red PLM was 34% (24%-41%), supporting the reliability of these measures. Measures of inflammation and collagen content did not differ between the pediatric and adult participants. However, adult participants exhibited greater amounts of abnormal smooth muscle.

Table 2.

Summary of ileal resection tissue assessments.

Tissue assessments	Overall (n = 50)	Children (n = 17)	Adults (n = 33)	P-value
Acute inflammation
Entire bowel wall (0-33)	2.8 (0.8-5.3)	3.5 (0.3-5.3)	2.6 (0.9-4.8)	.91
Mucosa (0-15)	2.0 (0.5-4.0)	1.8 (0.3-3.8)	2.1 (0.7-4.3)	.41
Submucosa (0-9)	0 (0-0.8)	0.3 (0-1.0)	0 (0-0.8)	.21
Muscularis propria (0-9)	0 (0-0)	0 (0-0.3)	0 (0-0)	.04
Chronic inflammation
Entire bowel wall (0-30)	11.3 (9.7-13.6)	11.3 (8.6-14.5)	11.8 (9.7-13.4)	.76
Mucosa (0-12)	5.8 (4.6-6.5)	5.3 (4.0-6.0)	5.8 (5.2-6.5)	.21
Submucosa (0-9)	4.0 (2.0-5.4)	4.3 (2.0-5.5)	4.0 (2.0-5.3)	.80
Muscularis propria (0-9)	2.0 (0.4-3.1)	2.3 (1.0-3.5)	2.0 (0.3-3.1)	.50
Fibrosis graded by trichrome
Entire bowel wall (0-9)	2.8 (1.0-3.4)	2.3 (0.8-3.3)	2.8 (1.1-3.8)	.37
Mucosa (0-3)	0 (0-0)	0 (0-0)	0 (0-0.3)	.35
Submucosa (0-3)	1.5 (0.8-2.5)	1.5 (0.8-2.0)	1.5 (0.8-2.6)	.37
Muscularis propria (0-3)	0.3 (0-1.1)	0.5 (0.3-1.0)	0.3 (0-1.3)	.42
Collagen content by SHIM
Entire bowel wall (%)	18 (12-26)	16 (9-22)	19 (13-29)	.23
Submucosa (%)	31 (21-46)	29 (16-50)	31 (21-46)	70
Muscularis propria, inner (%)	13 (8-27)	9 (7-23)	14 (8-29)	.50
Muscularis propria, outer (%)	13 (8-27)	9 (4-17)	4 (3-25)	.31
Collagen content by Sirius red PLM
Submucosa red (%)	18.0 (11.0-21.2)	19.0 (11.8-21.5)	17.0 (11.0-21.4)	.63
Submucosa green (%)	20.0 (12.3-22.0)	18.0 (11.0-22.5)	20.0 (14.0-21.8)	.54
Total (sum of red and green) (%)	34.0 (23.9-41.1)	33.0 (22.2-42.3)	35.0 (27.6-39.8)	.60
Muscle hyperplasia
Entire bowel wall (0-18)	4.3 (2.8-7.3)	3.3 (1.5-4.3)	5.8 (3.6-12.0)	.001
Mucosa (0-6)	1.8 (0.8-2.9)	1.3 (0.3-2.0)	2.1 (0.9-4.1)	.07
Submucosa (0-6)	2.0 (1.1-3.8)	1.5 (0.3-2.5)	2.8 (1.6-5.1)	.008
Muscularis propria (0-6)	0.5 (0-2.0)	0 (0-0.5)	1.0 (0-3.0)	.02
Muscularis propria thickness by confocal microscopy
Muscularis propria (micron)	2084 (1414-2672)	1802 (989-2425)	2159 (1445-2760)	.21
Muscularis propria, inner (micron)	1337 (958-1892)	1304 (721-1748)	1563 (1032-1915)	.36
Muscularis propria, outer (micron)	477 (308-677)	375 (264-623)	536 (328-866)	.06

Abbreviations: PLM = polarized light microscopy; SHIM = second harmonic imaging microscopy.

Data are presented as medians and interquartile ranges. P-values comparing children and adult values based on Mann–Whitney U test. The bold values are statistically significant P-values.

Correlations between different tissue assessments

Correlations between different tissue assessments are presented in Figure 2D and Table S1. As expected, ileal sections with higher overall bowel wall histologic acute inflammation also exhibited higher histologic chronic inflammation (r = 0.53, P = .0008). Similarly, measures of collagen content or muscle hyperplasia correlated with each other. Both overall bowel wall and submucosal collagen content measured by SHIM negatively correlated with overall bowel wall histologic acute inflammation (r = −0.39, P = .008 and r = −0.32, P = .03, respectively), with submucosal collagen content measured by SHIM also trending toward a negative correlation with overall bowel wall chronic inflammation (r = −0.27, P = .08). We did not observe correlations between measures of inflammation or collagen content measured by SHIM with measures of muscle hyperplasia.

Tissue assessments versus disease behavior

While 90% (45/50) of the participants exhibited persistent luminal narrowing on the preoperative research MRI, approximately 50% no longer exhibited pre-stenotic dilation diagnostic of a stricture (B2) or internal penetrating disease (B3) (eg, an abscess or internal fistula), as noted on the prior clinical imaging or endoscopic study required for study entry, and so were re-classified as B1 inflammatory behavior on the preoperative research MRI. This change in MRI appearance could have been due to the escalation of medical therapy between the time of clinical B2/B3 diagnosis and the research MRI or physiology-related variability in the degree of proximal small bowel dilation in participants with B2 disease (eg, a decrease in upstream dilation from 3.1 to 2.9 cm would change an individual’s classification from B2 to B1 disease).

Group III patients with higher levels of acute inflammation exhibited the highest frequency of B3 internal penetrating complications on the preoperative research MRI (n = 12, P = .02), while Group I and II patients with lower acute inflammation and varying levels of collagen content exhibited a higher frequency of B2 stricturing complications (n = 33, P = .01) (Figure 2E).

Summary of Circulating and MRI Biomarkers

We did not observe an association between biologic therapy exposure (median [IQR] collagen content of 18% (11%, 25%) versus 22% (12%, 33%) for exposed versus not exposed) or patient symptoms of pain (r = −0.2, P = .16) or loose stools (r = 0.02; P = .90), and ileal collagen content. Circulating and MRI biomarker measurements are presented in Table S2, including by age group. The circulating biomarkers CD64, GM-CSF Ab, and ECM1 did not differ between children and adults. Likewise, there were no differences between children and adults for any of the MRI biomarkers. The median sMaRIA score was 3.3 (IQR: 2.0-4.2) for the study sample, indicating that most participants had radiologic evidence of bowel wall acute inflammation. MRI biomarkers were measurable as follows: MTR—50/50 participants (100%), MOLLI T1–44/50 participants (88%), and IVIM—48/50 participants (96%).

Univariate Associations With Ileal Collagen Content

Next, we asked if the circulating, fecal, or MRI biomarkers would be associated with ileal inflammation or fibrosis (Table 3). We did not observe an association between the primary predictor variable, the 3D MTR measure, and ileal collagen content. In univariate analysis of the secondary predictor variables (P < .1), the sMaRIA score was positively associated with acute (r = 0.26, P = .09) and chronic (r = 0.38, P = .009) inflammation, and negatively associated with collagen content (r = −0.28, P = .065). Fecal calprotectin was also positively associated with acute (r = 0.31, P = .059) and chronic inflammation (r = 0.43, P = .006), and negatively associated with collagen content (r = −0.36, P = .025). The MOLLI T1 MRI measure was negatively correlated with collagen content (r = −0.28, P = .09), while the normalized 3D MTR measure was positively correlated with the muscularis propria thickness measurement (r = 0.33, P = .03). GM-CSF Ab and soluble CD64 exhibited similar trends as fecal calprotectin for associations with inflammation and fibrosis. GM-CSF Ab was positively associated with acute inflammation (r = 0.27, P = .08), and negatively associated with collagen content (r = −0.33, P = .03). Soluble CD64 was positively associated with acute inflammation (r = 0.36, P = .02) and negatively associated with collagen content (r = −0.24, P = .1). Collectively, these results were consistent with the negative relationship between acute inflammation and collagen content observed in the ileal resections (Figure 2).

Table 3.

Univariate associations between biomarkers and tissue assessments.

	Acute inflammation	Chronic inflammation	Collagen content	Muscularis thickness
sMaRIA	0.255 (−0.041,0.511) 0.091	0.384 (0.102,0.609) 0.009	−0.277 (−0.528,0.018) 0.065	−0.158 (−0.437,0.15) 0.312
Calpro	0.305 (−0.011,0.566) 0.059	0.429 (0.131,0.656) 0.006	−0.359 (−0.606,−0.049) 0.025	−0.142 (−0.445,0.191) 0.402
3D_MTR	−0.233 (−0.493,0.065) 0.124	−0.153 (−0.427,0.147) 0.317	0.119 (−0.181,0.398) 0.437	0.326 (0.028,0.57) 0.033
MOLLI_T1	0.142 (−0.182,0.438) 0.388	−0.008 (−0.322,0.309) 0.963	−0.276 (−0.544,0.043) 0.089	−0.205 (−0.496,0.128) 0.224
IVIM_f	−0.092 (−0.382,0.214) 0.556	−0.22 (−0.488,0.086) 0.156	0.181 (−0.126,0.456) 0.246	−0.086 (−0.384,0.227) 0.592
IVIM_D*	−0.001 (−0.301,0.299) 0.995	0.124 (−0.183,0.409) 0.428	0.115 (−0.192,0.401) 0.464	0.162 (−0.154,0.447) 0.313
IVIM_D	0.016 (−0.285,0.315) 0.917	0.075 (−0.231,0.367) 0.633	0.011 (−0.291,0.31) 0.945	−0.042 (−0.345,0.269) 0.793
IVIM_ADC	−0.058 (−0.352,0.247) 0.712	−0.2 (−0.472,0.107) 0.198	0.092 (−0.214,0.382) 0.558	−0.076 (−0.375,0.237) 0.635
GM_ Ab	0.267 (−0.032,0.523) 0.08	0.174 (−0.13,0.448) 0.259	−0.328 (−0.569,−0.034) 0.03	−0.155 (−0.438,0.157) 0.328
sCD64	0.355 (0.069,0.588) 0.017	0.245 (−0.053,0.502) 0.105	−0.244 (−0.502,0.053) 0.106	0.118 (−0.189,0.404) 0.45
ECM1	−0.227 (−0.488,0.071) 0.133	−0.109 (−0.39,0.191) 0.477	0.205 (−0.094,0.471) 0.176	−0.204 (−0.475,0.102) 0.189

Abbreviations: 3D_MTR = Normalized = bowel wall three-dimensional magnetization transfer ratio normalized to skeletal muscle three-dimensional magnetization transfer ratio; Calpro = fecal calprotectin; ECM1 = extracellular matrix 1; GM_Ab = granulocyte-macrophage colony-stimulating factor autoantibody;IVIM_f = intravoxel incoherent motion perfusion fraction; IVIM_D*=intravoxel incoherent motion pseudodiffusion coefficient; IVIM_D = intravoxel incoherent motion (slow) diffusion coefficient; IVIM_ADC = intravoxel incoherent motion apparent diffusion coefficient; MOLLI_T1 = modified Look-Locker inversion recovery T1 relaxation; sMaRIA = simplified magnetic resonance index of activity; sCD64 = soluble cluster of differentiation 64.

Collagen content was measured by SHIM: second harmonic imaging microscopy; muscularis thickness was measured by confocal microscopy; data are shown as r or rho, 95% confidence interval, and P-value in each cell.

Multivariable Associations With Ileal Collagen Content

Finally, we tested whether these biomarkers could be used to estimate the degree of fibrosis in the ileal resection samples as measured by SHIM. We developed a multivariable linear regression model using stepwise variable selection and AICc criteria (Table 4). As expected from the univariate analysis, the normalized 3D MTR MRI measure did not enter the final model. Five of our secondary predictors of interest were found to be jointly associated with ileal collagen content. These included circulating log GM-CSF Ab (β = −.15, P = .012), log soluble CD64 (β = −0.27, P = .19), and log ECM1 (β = .47, P = .062), and the MRI measures of perfusion IVIM D* (β = 1.69, P = .13), and IVIM f (β = 0.29, P = .10). The R² of the final model was equal to 0.31 (95% CI: 0.11, 0.52, by bootstrapping validation method). A logistic regression analysis across three increasing levels of collagen content confirmed these associations (Supplementary Results).

Table 4.

Multivariable model for predicting log ileal collagen content by second harmonic imaging microscopy (SHIM)^*

Variable	Parameter estimate	Standard error	P-value
logGM-CSF Ab	-0.15	0.058	.01
logECM1	0.47	0.246	.06
logCD64	-0.27	0.204	.19
IVIM f	0.29	0.176	.10
IVIM D*	1.69	1.10	.13

Five variables were selected based on AICc (R² = 0.31 [95% CI: 0.11, 0.52 by bootstrapping validation method]). Non-significant variables were included to improve model goodness of fit based upon AICc criteria.

^*Analysis based on 45 participants with diagnostic SHIM collagen measurements available.

Discussion

Our study is the first prospective investigation to test the relationships between a panel of circulating and MRI biomarkers associated with CD surgical management, and ileal resection fibrosis measured by SHIM.¹⁶ We observed a heterogeneous patient population, including a subset with increased inflammation and reduced collagen as well as others with reduced inflammation over a wide range of collagen content. This is largely consistent with a prior study that defined subgroups of patients based upon relative levels of inflammation and fibrosis, although we did not observe a subgroup with both high inflammation and high collagen content.¹⁹ Our investigation supports the idea that while inflammation is required for the initiation of fibrosis, intestinal fibrosis can progress in the setting of minimal or no acute inflammation.^20,21 In this context, a prior study noted more complete suppression of acute inflammation, relative to expansion of tissue fibrosis, in CD patients receiving anti-TNF therapy relative to patients receiving corticosteroids.²² The multivariable analysis demonstrated that overall bowel wall collagen content measured by SHIM, our primary outcome of interest, was associated with circulating levels of GM-CSF Ab, ECM1, and CD64 as well as the MRI IVIM microcirculation perfusion metrics D* and f.

A recent consensus statement noted that current approaches for assessing tissue collagen content have limitations and that there is currently no gold standard.^13,14 Trichrome staining is subject to batch effects, and it is difficult to localize the Sirius red PLM signal to the appropriate bowel wall layers.¹⁴ We chose to use SHIM as our primary endpoint for measuring tissue collagen content because of its ability to quantify the key fibrillar collagens implicated in CD strictures.¹⁵ Two prior studies have applied SHIM in CD. One study defined differences in collagen between patients with CD and intestinal tuberculosis,²³ while our group utilized SHIM to quantify collagen content relative to preoperative ileal ultrasound shear wave elastography.²⁴ While we found that the degree of overall submucosal collagen content measured by SHIM and Sirius red PLM was comparable, the agreement across trichrome histologic grading, SHIM, and Sirius red PLM for individual patients was less than perfect. Additional studies are needed to further assess the relationships between these different methods for defining tissue collagen content with the goal of arriving at a reference standard.

Our investigation also confirms prior studies demonstrating that in some patients, abnormal smooth muscle may be a dominant histologic feature.^2,24 While we did not confirm an association between the normalized 3D MTR measure and ileal collagen content, we did observe a positive association between the ileal 3D MTR MRI measure and ileal resection muscularis propria thickness measured by confocal microscopy. This may account for the increase in 3D MTR which we recently reported in CD patients requiring surgical therapy.¹⁶ It is conceivable that this abnormal muscle may be a viable target for both future diagnostics and therapeutics, and normalized 3D MTR, if further validated, could serve as a longitudinal biomarker.

Our multivariable model included 3 circulating biomarkers, GM-CSF Ab, soluble CD64, and ECM1, with GM-CSF Ab and soluble CD64 negatively associated with tissue collagen content measured by SHIM and ECM1 positively associated with tissue collagen content. In this regard, GM-CSF Ab has been linked to ileal neutrophil activation, expansion of IFNG-producing type I innate lymphoid cells (ILC1), and clinical relapses with disease complications in CD, and so may play a pathogenic role in controlling the balance between acute inflammation and fibrosis.^6,7,25 Soluble CD64 provides an additional direct noninvasive measure of ileal acute inflammation, while ECM1 provides a direct noninvasive measure of ileal extracellular matrix production. Collectively, these data are consistent with a recent ex vivo ultrasound study demonstrating CD ileal resection tissue stiffness was negatively associated with acute inflammation and positively associated with tissue fibrosis.²⁶ It will be of interest in future studies to test for the role of CDH11+ fibroblasts implicated in CD ileal fibrostenosis in regulating the range of ileal collagen content observed in patients with elevated ECM1 and reduced acute inflammation (group I/II patients), and conversely the role of the IL-1-inflammatory fibroblast–neutrophil axis implicated in refractory IBD in those with elevated GM-CSF Ab, CD64, and acute inflammation, and reduced collagen (group III patients).^27,28

Two diffusion-weighted IVIM parameters related to bowel wall microcirculation were positively associated with bowel wall collagen content in our final multivariable model. While neither IVIM f nor IVIM D* was individually significantly associated with collagen content by SHIM at univariate analysis, further investigation revealed that these variables had moderate interactions with other variables including ECM1, which likely explains how they entered our final multivariable model. The IVIM technique has been increasingly used to assess tissue perfusion without the need for intravenous contrast material.¹³ The IVIM D* measure has emerged as a useful biomarker for cardiac fibrosis.²⁹ Previous studies have demonstrated that the IVIM parameter f is consistently decreased in patients with acute inflammatory CD^30–34 perhaps related to microvascular ischemia and a reduction in microvessel volume,³⁵ with measurements increasing with decreasing acute inflammation. Interestingly, 2 other studies showed a negative correlation between the IVIM parameter f and bowel wall fibrosis, although these relied upon conventional histologic scoring and univariate assessments, illustrating the importance of testing more quantitative measures of tissue collagen content as in the current study.^36,37

Our study has several strengths, including the prospective multicenter design, inclusion of both children and adults with CD, and the use of several complementary biomarkers with redundant reference standards and histopathological features. However, there are also some limitations. First, our study included research MRI examinations from only 2 scanner manufacturers. However, there is no reason to believe our results should not be generalizable to other scanner manufacturers. Second, only a single pathologist scored each participant’s stricture specimen for histologic fibrosis, smooth muscle hyperplasia, and inflammation. Third, our study contained nearly twice as many adults as children, although no differences in either circulating or MRI biomarkers between age groups were observed. Fourth, we did not study the repeatability and reproducibility of our MRI or SHIM measurements. Finally, our multivariable regression results require further external validation due to the potential for model overfitting. The R² of our model is relatively low, possibly due to the inherent variability of our tested biomarkers and/or unknown and unmeasured factors that were not accounted for in our study design. This model should be interpreted as an exploratory model indicating interesting associations and trends that should be verified by future investigations.

We have systematically evaluated relationships between circulating and MRI biomarkers and tissue assessments of fibrosis and inflammation. Circulating levels of GM-SCF Ab, ECM1, and CD64, and the MRI IVIM perfusion parameters f and D*, were associated with ileal collagen content. Lower degrees of histologic acute inflammation at the time of resection may be a consequence of effective anti-inflammatory medical therapies and underscores that inflammation and fibrosis may proceed along separate pathways. Importantly, there were no differences in our novel MRI and blood-based biomarkers between children and adults. Additional prospective studies are needed to further understand how these MRI and blood-based biomarkers perform in clinical practice and impact both outcomes and healthcare costs, as well as potentially guide the enrichment of clinical trials for novel anti-fibrotic agents with patients more likely to have significant tissue fibrosis at entry.

Supplementary Data

Supplementary data is available at Inflammatory Bowel Diseases online.

Funding

This study was supported by The Leona M. and Harry B. Helmsley Charitable Trust (J.R.D. and L.A.D.) and the Integrative Morphology Core shared facility of the National Institutes of Health (NIH)-supported Cincinnati Children’s Hospital Research Foundation Digestive Health Center (P30 DK078392 to L.A.D.). The funders did not play a role in the data analysis or manuscript writing.

Conflicts of Interest

J.R.D.: Unrelated research support from Philips Healthcare, GE HealthCare, Siemens Healthineers, Motilent, and Perspectum. J.A.T.: Unrelated research support from Philips Healthcare. J.G.F.: Unrelated grants to institution from Siemens Healthineers, The Leona M. and Harry B. Helmsley Charitable Trust, and Alimentiv, Inc. Consulting with funds to institution from Genentech, Boehringer Ingelheim, Glaxo Smith Kline, Janssen, Medtronic, Takeda, Alimentiv, and Redx Pharma. D.H.B.: Unrelated research support from Medtronic and Takeda; Consulting: Janssen. A.L.: Nothing to disclose. S.K.: Nothing to disclose. A.L.A.: Nothing to disclose. J.K.-S.: Nothing to disclose. R.W.S.: Unrelated research support from Janssen, Abbvie, Bristol-Myers Squibb. Consultant or advisory board member for AbbVie, Bristol-Myers Squibb, CorEvitas, Eli Lilly, Exact Sciences, Gilead, Janssen, Merck, Pfizer, and Takeda. J.A.: Consultant for Janssen Research & Development. P.M.: Nothing to disclose. B.C.T.: Nothing to disclose. E.L.B.: Nothing to disclose. H.J.: Nothing to disclose. O.L.-N.: Nothing to disclose. M.H.C.: Consultant for Allakos, Arena/Pfizer, AstraZeneca, Calypso Biotech, EsoCap Biotech, GlaxoSmithKline, Receptos/Celgene/BMS, Regeneron Pharmaceuticals, Robarts Clinical Trials Inc./Alimentiv, Inc. and Shire, a Takeda company. S.D.S.: A United States utility patent application was filed for the imaging phantom technology reported in this paper as US 18/196,229 on 11 May 2023. L.F.: Nothing to disclose. L.Q.: Nothing to disclose. A.J.T.: Consultant: Applied Radiology; Author Royalties: Elsevier; Funded travel: Merge. M.K.: Nothing to disclose. C.G.A.: Nothing to disclose. R.A.I.: Nothing to disclose. J.A.D.: Nothing to disclose. L.A.D.: Nothing to disclose.

Ethical Considerations

This multicenter, prospective, cross-sectional study was primarily approved by the human research ethics committee at Cincinnati Children’s Hospital Medical Center (2019-0677; initially approved June 20, 2019); the other three study sites also approved this research protocol. Oral and written informed consent was obtained from all participants (or a parent/guardian for pediatric participants). For pediatric participants (11-17 years of age), oral and written informed assent was also obtained. Research procedures were performed in a manner consistent with the Declaration of Helsinki.

Data Availability

Data generated or analyzed during the study, and the analytic code, are available from the corresponding author by request. The CCHMC IRB protocol is available upon request. This noninterventional, observational cohort study was registered on clinicaltrials.gov (NCT04088773).

Patient Involvement

Patient and public stakeholders were not involved in the design and conduct of the study, or the analysis and reporting of the results.