P259. Comparative retrospective assessment of prospectively recorded endoscopic and histological findings between CD and GI-TB; the first Eastern European registry data
Y. Erzin1, S.N. Esatoglu2, I. Hatemi1, N. Demir1, G. Dogusoy3, S. Erdamar4, G. Aygun5, S. Sadri2, A.F. Celik1, 1Istanbul University Cerrahpasa Medical Faculty, Gastroenterology, Istanbul, Turkey, 2Istanbul University Cerrahpasa Medical Faculty, Internal Medicine, Istanbul, Turkey, 3Florence Nightingale Hospital, Pathology, Istanbul, Turkey, 4Istanbul University Cerrahpasa Medical Faculty, Pathology, Istanbul, Turkey, 5Istanbul University Cerrahpasa Medical Faculty, Microbiology, Istanbul, Turkey
Background
Diagnosis of gastrointestinal tuberculosis (GITB) and its differentiation from Crohn's disease (CD) may be challenging. Most of the available morphological and histological relatively old data comes from Asian countries where the incidence of TB is high.
Methods
This is the retrospective assessment of the IBD registry cohort including GITB. The aim of this Eastern European study was to revalidate whether endoscopic morphology and/or histological features of inflammed tissue could differentiate CD from GITB as it was claimed in Asian counterpart where IBD is relatively rare. Twenty out of 32 GITB patients whose diagnosis was based on either intestinal tissue culture or tissue TB-PCR positivity, and each with subsequent complete clinical and endoscopic response to treatment, and 40 CD patients who were under the same IBD registry, including the patients before and after each GITB case, were selected. Additionally, both CD and GITB cases has been chosen according to availability of clear endoscopic digital pictures and descriptions leading us to get appropriate morphological informations. Endoscopically, GI inflammation was defined as focal single, focal multiple, segmental or diffuse and the extent of the disease was determined by combination of endoscopic and radiological methods like CT or enteroclysis. Endoscopic and histological features (particularly presence of granuloma) were compared between two patient groups. These patients' endoscopic features were blindly evaluated by 2 gastroenterologists, and kappa statistic was performed to assess interobserver variability.
Results
While all colonic involvements were equally distributed between GITB and CD, cecal involvement significantly indicated GITB (84%/37%, p = 0.001) in contrast to CD with 87%/63% (p = 0.042) ileal involvement.
Caseating or non-caseating granuloma was significantly more common in GI-TB (88%) compared to CD (17%) (=25.186, p = 0.000).
| Endoscopic features | CD (n = 40) | GI-TB (n = 20) | p | kappa |
|---|---|---|---|---|
| Longitudinal ulcers | 23 (57.5%) | 2 (10%) | 0.000 | 0.579 |
| Aphtous ulcers | 17 (42.5%) | 0.001 | 0.512 | |
| Circular ulcer | 3 (7.5%) | 18 (90%) | 0.000 | 0.643 |
| Nodularity | 6 (15%) | 17 (85%) | 0.000 | 0.454 |
| Pseudoplyps | 10 (25%) | 0.023 | 0.386 | |
| Cobblestone apperance | 7 (17.5%) | 0.084 | 0.611 | |
| Stricture | 10 (25%) | 9 (45%) | 0.116 | 0.592 |
| Star shaped | 30 (75%) | 2 (10%) | 0.000 | 0.233 |
| Cecal deformation | 1 (2.5%) | 7 (35%) | 0.001 | 0.932 |
| Distribution pattern and sites of involvement | CD (n = 40) | GI-TB (n = 19) | p |
|---|---|---|---|
| Focal single | 1 (5.3%) | 0.322 | |
| Focal multiple | 18 (45%) | 5 (26.3%) | 0.169 |
| Segmental | 3 (7.5%) | 3 (15.8%) | 0.376 |
| Diffuse | 19 (47.5%) | 10 (52.6%) | 0.733 |
| Mean number of involved segments | 2.4±1.9 | 2.5±1.3 | 0.919 |
Conclusion
The most distinguishing endoscopic features were longitudinal and aphthous ulcers for CD, in contrast to nodularity and circular ulcers for GITB with moderate interobserver variability. None of the endoscopic distribution pattern was significant for neither of the diseases, except highly specific but poorly sensitive CD indicator, cobble stoning. Presence of granuloma (mostly non-caseous) is more suggestive for GI-TB rather than CD.