Abstract
Background/Aim: Crypt branching in inflammatory bowel diseases (IBD) was previously classified into symmetric (CSB) and asymmetric branching (CAB). We aimed to validate the inter-rater agreement of the assessment of crypt branching (also called crypt fission) in patients with Crohn’s disease (CD) and to elucidate its potential diagnostic impact. Patients and Methods: A total of 100 colon biopsies from patients with active CD were analyzed. Two cohorts of patients with ulcerative colitis (UC) and infectious colitis, respectively, served as controls (UC-controls/infectious controls). Three pathologists scored the total number of branching crypts (TBC) as well as the number of CSB and CAB. Results: Inter-rater agreement ranged from moderate to good for TBC and CAB in the study cohort. The study cohort showed significantly higher counts of TBC, CSB, and CAB than the infectious controls (p<0.001 for TBC, p=0.008 for CSB, and p<0.001 for CAB). Cases with CD showed more TBC and CAB compared to cases with UC (p=0.001 each). Conclusion: Inter-rater agreement for crypt branching is reliable in patients with CD. Crypt branching was shown to be an additional histologic feature to distinguish active CD from infectious colitis. Additionally, it could be helpful in the distinction between CD and UC.
In patients with inflammatory bowel disease (IBD), crypt architectural distortion of the colon mucosa is one of the histopathological diagnostic hallmarks against other inflammatory conditions (1). According to literature, architectural distortion is a sign for chronicity in IBD, usually not seen before 15 days of symptoms, but present in more than three thirds of patients after four months of disease (2).
In the colon mucosa of IBD-patients, architectural distortion comprises an irregular or villiform mucosal surface, crypt atrophy, loss of crypt parallelism, crypt irregularity, tortuosity, dilatation, variation in shape and size, and crypt branching (1, 3–9). Crypt branching can normally be seen between mucosal hillocks/at innominate grooves but is an otherwise rare event in the normal mucosa of adults. However, it is present in the majority of IBD-cases, occurring more often in ulcerative colitis (UC) compared to Crohn’s disease (CD) (1, 5, 6, 8, 9). Until recently, little was known about the morphological spectrum of crypt branching (also called crypt fission) in patients with IBD. In 2020, an increased number of symmetric branching crypts (CSB) and, additionally, asymmetric branching crypts (CAB) had been described in colectomy specimens of patients with ulcerative colitis (UC), CSB and CAB being dependent on the cutting plane (10–14). Those findings were reproducible in two additional biopsy cohorts from Swedish and German patients with UC and the assessment of CSB and CAB turned out to be reproducible in a validation-study afterwards (13–15).
Interestingly, the amount of CAB and CSB was shown not only to have a diagnostic but also a prognostic impact. CSB were found more frequently in non-dysplastic colon mucosa of UC-patients diagnosed with low grade intraepithelial neoplasia alio loco (15). However, most studies on this issue were performed in patients with UC and so far little is known about the biological significance of branching crypts in patients with CD.
The aims of this study were therefore: 1) to validate crypt branching regarding its interobserver reliability on a consecutive cohort of 100 patients with CD, 2) to elucidate its diagnostic impact in comparison with the previously published UC-cohort and a cohort of infectious controls.
Patients and Methods
Study cohort. The study cohort included endoscopic colon biopsies from patients with active CD without dysplasia or carcinoma, under surveillance, selected from the electronic archive (DC Pathos database), Institute of Pathology, Klinikum Bayreuth GmbH (DC Systeme, Heiligenhaus, Germany) in a consecutive manner. A minimum of two endoscopic biopsies were taken from the terminal ileum and six different colon-levels for each patient. The slides were reevaluated by CLS and one representative colon slide was chosen. Selection criteria were: CD with active inflammation (neutrophils) in tunica propria epithelium or crypt abscesses. Erosion was accepted but slides with ulceration and/or granulation tissue only were excluded as assessment of crypt branching was not possible. Eight histological step sections (4 μm thick) were cut for each location per slide and stained with Hematoxylin and Eosin (H&E). The cases described in this study were not identical with and do not include the cases from our previously published work (14).
Control cohorts. Cases from our previously published study served as controls (15). Among them were 100 cases from patients with UC, hereinafter called “UC-cohort” and biopsies from 50 patients, diagnosed as infectious colitis (no IBD), hereinafter called “infectious controls”. The slides (study cohort and control cohorts) were randomly pseudo-anonymized by numbers to achieve blinding. All pseudo-anonymized slides were digitalized at the Institute of Pathology, Klinikum Bayreuth, Germany, using a NanoZoomer S360 scanner (Hamamatsu, Herrsching am Ammersee, Germany). The observers were given access to the slides online via a password protected platform using NDPView 2 (Hamamatsu). To test reliability and to validate if acceptable agreement could be consistently reached among pathologists, three pathologists from two different centers in Europe with special interest in gastrointestinal pathology (CR, CLS, MV) evaluated the slides independently and blinded to clinical data and each others, results.
Definition and assessment of crypt branching. To use pathologists’ common terminology crypt branching is hereinafter referred to as TBC, CAB and CSB. Crypt branching was defined as described (11–15). In brief, CSB show back-to-back amalgamating isometrics crypt-rings in horizontal sections or dichotomous branching of a crypt in vertical sections, the two new crypts having the same architecture allowing for mirroring them on a thought axis of symmetry. CAB show ≥2 anisometric crypt-rings in horizontal sections or dichotomous branching crypts in vertical sections with differences in size or shape of the two new crypts or crypt branching with >2 new crypts. Each pathologist recorded on an excel-sheet the TBC, the number of CSB, and the number of CAB.
The results of the study cohort, infectious controls, and UC-cohort were compared. The results of the infectious controls as well as the UC-cohort had been recently published (15). Example histological images for CSB and CAB in the CD-cohort are shown in Figure 1. The ethics commission of Friedrich-Alexander-University Erlangen-Nuremberg approved the study (study number 272_21 Bc).
Histological image of symmetric (asterisks) and asymmetric (arrows) branching crypts in a patient with Crohn’s disease (magnification 209.2×).
Statistics. Statistical analyses were performed using the statistics program SPSS 21 (IBM Corp. Released 2012, IBM SPSS Statistics for Windows, Armonk, NY, USA). Interrater variability was tested by Intraclass Correlation Coefficients (ICCs) with 95% confidence intervals (CI). Strength of agreement was graded as follows: <0.5 poor, 0.5-0.75 moderate, 0.75-0.9 good, and >0.9 excellent. The t-test for paired samples was applied to compare differences between crypt branching values (total, asymmetric, symmetric). Pearson’s Chi Square test was used to compare differences between equally defined variables. Statistical significance was defined as p<0.05.
Results
Number of branching crypts. In the study cohort, the average (mean) number of branching crypts for all three observers (pooled) was 8.79 for the TBC [symmetric and asymmetric; median 7.17, min 0, max 43, standard deviation (SD): 7.922], 1.51 for CSB (median 1.0, min 0, max 10, SD: 1.798) and 7.27 for CAB (median 4.50, min 0, max 33, SD: 6.845), respectively. The differences between the three groups were statistically significant (p-values: CSB vs. CAB: <0.001; CSB vs. TBC: <0.001; CAB vs. TBC: <0.001, respectively).
Interrater agreement. In the 100 cases of the study cohort, interrater agreement for the TBC in fission was ranged between moderate and good [intraclass correlation coefficient (ICC): 0.742, 95%CI=0.662-0.809], moderate for asymmetric branching (ICC: 0.666, 95%CI=0.572-0.749) but poor for symmetric branching (ICC: 0.322, 95%CI=0.197-0.449). However, in pairwise analysis, ICCs differed between different observers from poor to excellent. In sum, agreement was excellent in three, moderate in four, and and poor in two out of 9 measures, resulting in 33.3% with excellent agreement and 77.8% with at least moderate interobserver agreement.
Correlation between the Crohn’s cohort, UC-cohort, and infectious controls. Study cohort versus infectious controls. A significant difference between the CD-cohort and the infectious controls was found for the three observers (pooled) for the number of crypts in fission in all three measures. Cases with CD showed significantly more TBC, CSB and CAB in fission than the infectious controls (p<0.001 for total, p=0.008 for symmetric and p<0.001 for asymmetric).
CD cohort versus UC-cohort. Correlation analysis between the CD-cohort and UC-cohort revealed significant differences for total and asymmetric fission between both cohorts (all observers, pooled, p=0.001 for total and p=0.001 for CAB, respectively). Patients with CD showed significant more total crypts in fission and CAB compared to the UC-cohort.
No differences were found for the number of crypts in symmetric fission between both cohorts (p=0.295). Detailed results between the CD-cohort, UC-cohort and infectious controls are shown in Table I. Figure 2 illustrates graphically the differences between CD, UC, and infectious controls for TBC.
Comparison of the number of crypts in fission between the study cohort, the UC-cohort, and infectious controls.
Boxplot diagram illustrating the total number of crypts in fission for the Crohn’s disease (CD)-cohort, ulcerative colitis (UC)-cohort, and infectious controls.
Discussion
Crypt architectural distortion is known as one of the diagnostic hallmarks in the diagnosis of IBD against other forms of colitis (1, 8). It comprises a histopathological spectrum of changes in the normal crypt epithelium, among them crypt branching (1, 3–9). Crypt branching has recently been better defined by our group and the assessment method had been validated afterwards on a cohort of UC-patients (10, 12–15). Interrater agreement values reached at least moderate agreement in seven out of nine measures with a wide range from poor to excellent in pairwise analysis between different observers. The fact that the interobserver agreement did not reach the same high values than those in the previous UC-cohort can probably be partly explained by the discontinuous nature of CD, with areas of normal mucosa between inflamed areas, making assessment of crypt branching more difficult. Crypt branching (TBC and CAB) turned out to be significantly more often seen in UC compared to infectious colitis (15). This was especially true for the moderately active UC-cases but a trend towards more branching crypts in UC against infection was also seen in remission. This present study shows consistent results in patients with active CD, having more TBC, CAB, and CSB compared to the infectious controls. Interestingly, patients with CD reached even higher amounts of TBC, CAB, and CSB in comparison to patients with UC. In fact, a descending gradient was seen in the amount of crypt branching with highest values for CD and lowest values for the infectious controls. Therefore, we conclude that crypt branching is not only helpful in the differential diagnosis between IBD and other forms of colitis but, additionally, higher amounts of branching crypts could also be of help in the differential diagnosis between CD and UC. Crypt branching had also been shown to have a prognostic impact. We could recently show that CSB is more frequently seen in patients with UC who were already diagnosed with a low grade intraepithelial neoplasia (LGIN) alio loco. Immunohistochemical and molecular studies demonstrated TP53-mutations or copy number changes in non-neoplastic mucosa of IBD-patients, indicating a mucosal point of no return for neoplasia development (15–17). Further studies must show if higher amounts of CSB might be a histomorphological sign of those molecular alterations. Additionally, in a series of 100 cases with LGIN in UC-patients, it was recently shown, that CAB occurs also in dysplastic mucosa, even if it seems to be a rare event (18). However, intraepithelial neoplasia is much rarer in CD compared to UC. A recently published study by our group demonstrated impressively that there is still space for novel discoveries even in the old-school field of morphological-based histopathology (19).
Conclusion
Even if there are still discrepancies concerning interrater agreement and final definitions of crypt branching, our study shows that assessment of crypt branching is reliable also in patients with CD and has a diagnostic impact even superior to that of UC. Hopefully, this study will further fuel the discussion around crypt branching in patients with IBD. As assessment of crypt branching is reliable in UC and CD; it might be an attractive trainable target for advanced intelligence systems in future studies.
Footnotes
Authors’ Contributions
Conception and study design: Carlos A. Rubio, Corinna Lang-Schwarz. Acquisition of data: All Authors. Article drafting: Corinna Lang-Schwarz. Analysis and interpretation of data: Corinna Lang-Schwarz. Critical revision of the article for important intellectual content: All Authors. Final approval of the submitted version: All Authors.
Conflicts of Interest
The Authors declare that they have no conflicts of interest in relation to this study.
- Received February 24, 2022.
- Revision received March 6, 2022.
- Accepted March 8, 2022.
- Copyright © 2022 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
