Characterization of intestinal fibrosis in cats with chronic inflammatory enteropathy

Abstract Background Intestinal fibrosis (IF) is commonly identified on histopathology of intestinal biopsy specimens (IBSp) from cats with chronic inflammatory enteropathy (CIE) however, its clinical relevance is unknown. Objectives Characterize and determine the clinical relevance of IF in cats with CIE. Animals Sixty‐five client‐owned cats diagnosed with CIE after gastrointestinal histopathology from a single referral hospital in the United Kingdom. Methods Medical records were retrospectively searched for cases of CIE on the basis of histopathology of IBSp. The IBSp from eligible cats were re‐reviewed by a single board‐certified veterinary pathologist for inclusion. Masson's trichrome (MT) stain and immunohistochemical labeling using antivimentin and anticollagen I antibodies to identify IF. For each case, various variables at the time of diagnostic investigation were recorded and referring veterinarians were contacted for follow‐up information. Results Mucosal fibrosis was identified in 51% of duodenal and 76% of colonic hematoxylin and eosin (HE)‐stained IBSp. Vimentin labeling and MT staining identified additional cases of IF in 65% and 58% of the duodenal biopsy specimens, respectively. Vimentin labeling detected IF in 79% of the colonic biopsy specimens. Positive vimentin labeling and MT staining of the colonic mucosa were associated with decreased likelihood of attaining clinical remission and increased risk of death because of CIE (P < .05). Conclusions and Clinical Importance Additional stains at initial histopathologic examination of IBSp allow for better identification of IF compared to routine HE staining. Identification of IF in colonic biopsy specimens by vimentin immunolabeling and MT staining may provide prognostic information in cats with CIE.


| INTRODUCTION
Chronic inflammatory enteropathy (CIE) in cats describes a group of idiopathic diseases resulting in gastrointestinal (GI) signs of at least 3 weeks' duration. Definitive diagnosis requires ruling out all known causes of chronic GI signs, including infectious, obstructive, neoplastic and extraintestinal diseases by histopathological examination. 1 The exact etiology of CIE in cats is currently unknown, but its hypothesized etiopathogenesis is extrapolated from inflammatory bowel disease (IBD) in humans, another similar enteropathy. Consequently, CIE in cats is postulated to occur because of genetic defects in the ability to differentiate between commensal vs pathogenic bacteria, aberrant innate immune system responses, and elusive environmental factors in susceptible hosts. [1][2][3] The presence of intestinal mucosal fibrosis is commonly noted on histopathologic examination of intestinal biopsy specimens (IBSp) of cats with CIE. 1,4 Intestinal fibrosis (IF) is characterized by an alteration in collagen metabolism after chronic inflammation, resulting in its excessive and irreversible deposition in the extracellular matrix (ECM). 5 An increase in the resident mesenchymal cell population responsible for the deposition of collagen is the principal mechanism in intestinal fibrogenesis. 6 Mesenchymal cells can be broadly classified into fibroblasts, myofibroblasts, and smooth muscle cells.
Despite IF occurring in cats with CIE, its clinical relevance remains unknown.
In humans with IBD, IF is an expected complication and a distinct cause of patient morbidity and mortality. 5,[7][8][9] In humans with IBD, myofibroblasts have been identified as key cellular mediators of collagen deposition, with collagen types I and III considered the 2 major ECM proteins associated with IF. [10][11][12] Intestinal fibrosis persists in a self-perpetuating manner in the absence of inflammation 13 with evidence also indicating that ECM stiffness itself is capable of perpetuating collagen deposition. 14 Therefore, targeting signaling pathways that lead to upregulation of IF in humans with IBD is a diagnostic and therapeutic focus. 15,16 Currently, identification and quantification of IF in cats with CIE are based solely on the assessment of morphologic features using World Small Animal Veterinary Association (WSAVA) histopathologic scoring of hematoxylin and eosin (HE)-stained tissue sections. However, concern exists among veterinary pathologists that IF may be underreported based on HE staining alone and, additionally, may be masked by edema or missed because of sample orientation. 17 Therefore, use of additional stains, such as Masson's trichrome (MT) or immunolabeling for vimentin and collagen I, may better identify the presence, pattern and pathology of mucosal fibrosis.
Our aim was to determine the frequency of IF in cats with CIE by utilizing additional stains and immunolabeling. We also aimed to correlate the presence of IF with clinical findings, results from diagnostic investigations, and outcome to further characterize its clinical relevance.

| Follow-up information
A minimum of 6 months' follow-up was required for all cats. An exception was cats that died or were euthanized for any cause before 6 months after IBSp retrieval. Follow-up information was gathered by telephone contact or email questionnaire from the referring veterinarian who originally referred the cat. Follow-up information included: whether the cat was in clinical remission, its current treatment regimen, patient outcome (alive or dead and reason for death), and the date of this information. Clinical remission was defined as the absence of GI signs (vomiting, diarrhea, hyporexia, weight loss). Follow-up time was defined as the time from histopathologic diagnosis of CIE to the date of last consultation with the referring veterinarian as stated in the questionnaire.

| Initial histopathologic assessment
For cats that met inclusion criteria, original formalin-fixed paraffin embedded (FFPE) and HE-stained slides of duodenal and colonic biopsy specimens were reviewed again by a single board-certified veterinary pathologist, with a special interest in GI disease, to confirm the diagnosis of CIE and were assigned a WSAVA histopathology score. 20 Absence of intestinal mucosal fibrosis was defined as a narrow band of stroma up to 1 to 2 fibroblasts in width, whereas the presence of intestinal mucosal fibrosis was described as crypts separated by a narrow band of stroma with a width of >2 fibroblasts. 20 Cats were excluded if the HE-stained IBSp were of poor quality, poor tissue orientation or absent. Biopsy specimens suspicious for alimentary neoplasia on examination underwent further immunohistochemical labeling using CD3 and CD20 antibodies.
Intestinal biopsy specimens that were confirmed as alimentary neoplasia on immunohistochemistry (IHC) were excluded.

| Evaluation of immunolabeling of collagen I
Collagen I deposition in the duodenal mucosa was graded by visually assessing its (1) labeling intensity and (2) frequency, over 5 nonoverlapping (Â100) fields, to give a composite collagen score (CCS) which was calculated as a multiple of these 2 values. 21 Regions with the highest immunolabeling intensity were selected.
The CCS, therefore, was graded with scores of 0 to 1: mild collagen labeling, 2 to 4: moderate collagen labeling and ≥6: marked collagen labeling. A photo guide, visually detailing how collagen I labeling intensity was scored, is presented as Figure 3. Because collagen I frequency was graded over multiple fields and tissue sections, only labeling intensity, and not frequency, is represented graphically. Univariable binary logistic regression was performed to assess the association between clinicopathologic variables and the presence of IF using MT stain and immunolabeling for vimentin. Results were presented as odds ratios (OR) and 95% confidence intervals (CI). Because of an extreme categorical problem (zero in contingency table), Kendall's tau-b was employed to assess the correlation between the presence of IF in colonic biopsy specimens and the following variables: history of diarrhea and thickening of muscularis propria layer on ultrasound examination, and these results were reported descriptively.

| Statistical analyses
Kendall's tau-b was used to examine the association between IF and the attainment of clinical remission and outcome of death because of GI disease.
For the 20 duodenal biopsy specimens that underwent IHC for ACA, a Spearman's rank correlation (r s ) was used to evaluate the correlation between CCS and all continuous clinicopathologic variables.
Kendall's tau-c correlation was reported between CCS and ordinal/ binary variables.
Type I error rate for all statistical analyses was set at 0.05.   Full-thickness (FT) biopsy specimens were obtained by laparotomy in 9 cats (14%) and partial-thickness biopsy specimens were obtained by endoscopy in 56 cats (86%). Partial-thickness and FT biopsy specimens of the duodenum were collected in 50 cats (77%) and 7 cats (11%), respectively. In 2 cats (3%), FT ileal and jejunal biopsy specimens also were collected. Partial-thickness biopsy specimens of the colon were collected in 29 cats (47%). Partial-thickness biopsy specimens of both the duodenum and colon were collected in 23 cats (35%).

| Treatment
At discharge from the university referral hospital, 37 cats (57%) were

| Questionnaire response and outcome information
Questionnaire response rate by referring veterinarians was 83% (n = 54) of the 65 cats in the study group. For the 11 cats (17%) for which no follow-up information was obtained, the reason was that the medical records were unavailable (n = 6) or these cats were lost to follow-up (n = 5).
Of the 54 cats for which follow-up information was obtained,

| Identifying IF in duodenal and colonic biopsy specimens with the addition of MT staining and IHC for vimentin
Of the 57 HE-stained duodenal biopsy specimens, WSAVA scoring by the board-certified veterinary pathologist found no evidence of mucosal fibrosis in 28 specimens (49%), mild mucosal fibrosis in F I G U R E 3 Photoguide for the assessment of the intensity of mucosal collagen labeling in duodenal biopsy specimens (Â200) from cats with CIE having undergone IHC using a collagen I antibody. Positive control image depicts collagen labeling of the duodenal submucosa. Black squares indicate areas that appear magnified in the right column.  All results from statistical analyses for the clinicopathologic variables for duodenal and colonic specimens are presented in Tables 3 and 4, respectively. Multivariable analysis was not carried out because of limited predictors identified in the univariable analysis (Table 3) and small sample size (Table 4). All results from statistical analysis for outcome variables are presented in Table 5.

| Composite collagen scoring
Of the 20 cats that had duodenal specimens that underwent IHC for ACA, 11 cats (55%) were given a CCS between 0 and 1 to indicate mildly increased collagen, 7 cats (35%) were given a CCS between 2 and 4 to indicate moderately increased collagen and 2 cats (10%) were given a CCS of ≥6 to indicate markedly increased collagen. The scoring breakdowns are depicted in Table 6.

| Results of association analyses for IHC of collagen I
Kendall's tau-c correlation of CCS indicated that a history of weight loss was strongly correlated with increased CCS (ꞇ 0.630, P < .001).
T A B L E 1 Demographic data of the 65 cats that met final study inclusion criteria.

Number of cats in study 65
Demographic information Spearman's rank correlation indicated that decreased serum albumin concentration (r s À.179, P < .001) was weakly correlated with CCS.
Spearman's rank correlation indicated that decreased serum cobalamin concentration (r s À.740, P = .01) was strongly correlated with CCS. Additionally, increased WSAVA score of the duodenum was moderately correlated with CCS (r s .554, P = .01). All remaining results from statistical analyses are presented in Table 7.

| DISCUSSION
We identified duodenal and colonic mucosal fibrosis as a common finding in cats with CIE. When using HE staining alone, IF was  intestinal segments, so-called "creeping fat" as seen in Crohn's disease, also has been directly linked to increased intestinal wall fibrosis in humans with IBD. 25 No such parallel has been identified in cats, and a recent study evaluating disease associations in 9062 overweight and obese cats found no correlation between increased BW or BCS and GI disease. 26 Additional investigation of these variables, and consideration of the use of percentage weight change as a more discriminating measure, are required to better understand the relationships among BW, BCS, and IF in cats with CIE.
A history of diarrhea was weakly but positively correlated with the presence of IF as identified by MT staining of colonic biopsy specimens. In humans with IBD, diarrhea is thought to lead to a selfperpetuating cycle by instigating a state of chronic inflammation that propagates the development of IF, which in turn further exacerbates diarrhea because of impaired intestinal wall function. 27,28 The severity of diarrhea in humans with IBD is considered an important determinant of clinical disease activity. 29 This conclusion is similar to the Feline Chronic Enteropathy Activity Index, which correlated the presence of frequent diarrhea with increased histologically identified intestinal inflammation. 30 Our study only assessed history of diarrhea as a binary variable (yes vs no). Therefore, determining diarrhea severity and characteristics (i.e., small, mixed, or large bowel) could more specifically correlate diarrhea with the histologic presence of IF. Additionally, it would help to examine the value of obtaining IBSp in all cats that are presented with diarrhea. Future studies also should evaluate the potential effect of IF on colonic motility.
Thickening of the intestinal muscularis propria layer on transabdominal ultrasound examination was correlated with the presence of colonic mucosal fibrosis as detected by IHC labeling for vimentin. In humans with IBD, increased colonic wall thickness, as recognized ultrasonographically, has been positively correlated with histologic inflammation and IF. 31 One study that evaluated ultrasonographicallyidentified SI muscularis propria thickening in cats identified low positive predictive value of 18.9% to 57.1% for histologic disease. 32 Although the aforementioned study did not examine the colon, its results potentially could be extrapolated similarly. However, an additional consideration is that ultrasonographic studies in clinically T A B L E 5 The association between the presence of intestinal mucosal fibrosis in the duodenum or colon of cats with chronic inflammatory enteropathy and outcome. Significant P values are listed in bold.
T A B L E 6 Composite collagen score combinations and number of cats with this scoring combination.

Collagen intensity Â collagen frequency
Composite collagen score Number of cats with this scoring combination Note: A collagen frequency score of 0 referred to a visual estimate of area labeled as <25%.
T A B L E 7 Spearman's rank correlation and Kendall's tau c* correlation coefficient were used to assess the correlation between clinicopathologic variables and composite collagen score on 20 duodenal biopsy specimens of cats with chronic inflammatory enteropathy and increased duodenal mucosal fibrosis. healthy cats have repeatedly identified the colon to be the thinnest segment of the GI tract. 33,34 Therefore, it also could be hypothesized that colonic muscularis propria thickening identified in cats with GI signs and CIE is clinically relevant and should not be disregarded.
Along with our results, these findings should prompt the specific examination of the relationship between colonic ultrasonography and IF, which has not been assessed previously. Additionally, further evaluation of vimentin as a sensitive marker for IF is warranted.
The use of ACA resulted in a moderate correlation between CCS and composite WSAVA duodenal score, which is likely a reflection of the unblinded process by which the 20 cases were selected. This correlation therefore could emphasize how individual WSAVA scoring variables are more reflective of the intricacies between clinicopathologic and histopathologic relationships than the composite score. This conclusion is supported by 1 study in which measuring colonic fibrosis as an individual score resulted in a better correlation with clinical activity in dogs with immunosuppressant-responsive enteropathy. 35 Additionally, because fibrosis occurs as a sequela to inflammation, it is possible that the presence of inflammatory cells and fibrosis are inverse to each other, which may not be accounted for in the composite WSAVA score.
A history of weight loss and decreased serum cobalamin concentration were strongly correlated with the CCS with decreased serum albumin concentration demonstrating a weak correlation. Hypocobalaminemia and hypoalbuminemia previously have been recognized in cats with chronic GI disease, but no dependent relationship has been described between the 2 or reported in relation to IF. [36][37][38] In humans, IBD is characterized by marked GI inflammation, which causes excessive enterocyte turnover and IF. 9,39 This situation subsequently causes loss of the intestinal barrier function and macronutrient and micronutrient deficiency, which could similarly occur in cats with CIE. 40 The relationship identified between hypocobalaminemia and duodenal mucosal collagen I deposition also could represent the diffuse nature of IF throughout the GI tract, given that cobalamin is absorbed in the distal ileum. 36  assessed using image processing programs such as ImageJ. 54 However, we believed that visual estimation by the pathologist would be more representative of how IBSp generally are evaluated in clinical practice, and also allowed for the detection of subtle changes in tissue architecture that could be missed using digital techniques. Biopsy specimens examined were those already collected at the time of diagnosis, not allowing for broader assessment of the GI tract. Consistent with this concern, ileal biopsies were not consistently performed, which could have resulted in missed intestinal neoplasia, considering that SCL is most common in the ileum and jejunum. 1,55 However, evaluation of duodenal and colonic biopsy specimens in our study is representative of how intestinal specimens commonly are obtained in clinical practice. 56 Additionally, despite a study suggesting that incorporation of IHC and PCR for antigen receptor rearrangements increases the ability of SCL identification in IBSp from cats, it also found that ileal specimens seldom changed the diagnosis achieved from duodenal specimens alone. 57

| CONCLUSIONS
Our results identified that IF is a frequent finding in cats with CIE.
Additionally, we found that use of MT staining and immunolabeling for vimentin allowed for better identification of IF in IBSp relative to routine HE staining.
Our study is the first to describe a relationship between IF and outcome in cats with CIE. The presence of colonic mucosal fibrosis, identified using MT staining and IHC for vimentin, was correlated with failure to attain clinical remission and death because of GI disease.
This information is clinically relevant in anticipating the course of CIE in cats as a relapsing-remitting disease. Our results can serve as a stimulus for further research into the complex role of IF in CIE in cats and into IF as a potential treatment target.