Search in the Abstract Database

Search Abstracts 2013

* = Presenting author

P286. CT based predictive model for differentiation of Crohn's disease from intestinal tuberculosis

S. Kedia1, R. Sharma2, B. Nagi3, V. Pratap Mouli1, R. Dhingra1, L. Kurrey1, S. Srivastava1, V. Ahuja1, 1All India Institute of Medical Sciences, Gastroenterology, New Delhi, India, 2All India Institute of Medical Sciences, Radiodiagnosis, New Delhi, India, 3Post Graduate Institute of Medical Education and Research, Gastroenterology, Chandigarh, India

Background

Intestinal tuberculosis (ITB) and Crohn's disease (CD) have close clinical, radiological, endoscopic, and histological resemblance. There is lack of literature on computed tomography (CT) in differentiating between CD and ITB.

Methods

From October 2010 to Jan 2012 all patients with ITB and CD, who underwent CT enteroclysis/ CT enterography/ CT abdomen were recruited in the study, and all findings were recorded as per the proforma. The gold standard for diagnosis of CD and ITB were ECCO guidelines and Paustian's criteria respectively. The CT images were evaluated by 2 independent observers who were blinded to the diagnosis.

Results

One hundred eleven patients were included in the final analysis (60 CD, 51 ITB, mean age 36.4±14.2 years vs 34±13.3 years, p = 0.36). Significantly more patients with ITB had short segment involvement (70.6% vs 26.7%, p < 0.001), ileocaecal involvement (49% vs 18.3%, p = 0.001), <3 segments involved (84.3% vs 63.3%, p = 0.002), and lymph nodes >1 cm in size (19.6% vs 1.7%, p < 0.001). Patients with CD had significantly greater long segment involvement (61.7% vs 27.5%, p = 0.006), presence of Comb's sign (40% vs 19.6%, p = 0.02), colonic involvement (35% vs 19.6%, p = 0.072) and ≥3 segments involved (25% vs 19.6%, p = 0.138). Based on above features predictive models were developed which could differentiate ITB and CD with >90% accuracy (Table 1).

Table 1. Accuracy of combination of various features in predicting ITB or CD
 Diagnostic accuracy (%)
Predictive model for Crohn's Disease
Long segment involvement + Comb's sign + ≥3 segments involved90
Long segment involvement + Comb's sign + ≥3 segments involved + Colonic involvement100
Predictive model for Intestinal Tuberculosis
Short segment involvement + ileocaecal involvement + <3 segments involved90.9
Short segment involvement + ileocaecal involvement + <3 segments involved + >1 cm lymph node involvement100

Conclusion

Predictive models based on combination of CT findings to differentiate CD and ITB were developed, and they achieved an accuracy of >90%.