Correlation of CT enteroclysis with surgical pathology in Crohn's disease

Background: The response to therapy in Crohn's disease (CD) depends on the inflammatory or fibrostenotic nature of the underlying pathological process. Standard diagnostic tests cannot reliably distinguish between these two entities. Although CT enteroclysis (CTE) has shown promise in the evaluation of small bowel disorders, its accuracy for the differentiation of CD phenotypes is unknown. Aims: To determine the accuracy of CTE compared with surgical pathology in patients with CD and to assess the association of CTE variables with inflammatory or fibrostenotic pathological lesions. Methods: CTE studies from patients who underwent resective bowel surgery for CD were reviewed and compared with the pathological specimens using a standardized scoring system. Patients were excluded if they had incomplete studies, nonresective surgeries, or a diagnosis of malignancy. CTE variables, such as mucosal and mural enhancement, wall thickness, engorgement of vasa recta (comb sign), adenopathy, and the presence and severity of strictures were compared with the pathology results using Mantel-Haenszel chi(2), Spearman's rank coefficient, and logistic regression analyses. Results: Of the 54 patients enrolled, 10 were excluded. The remaining patients (61% female, 84% white) underwent 44 surgical interventions generating 47 bowel segments that were included in the analysis. The indications for surgery were: bowel obstruction in 21; perforating disease in 13; and refractory, nonobstructive disease in 15. The accuracy of CTE for inflammatory and fibrostenotic lesions was 76.6% and 78.7% using a four- and three-point grading system, respectively. There was good correlation between CTE and pathology in regards to inflammation (Spearman's r = 0.7, P < 0.0001) and fibrostenosis (Spearman's r = 0.6, P < 0.0001) scores. The pathological inflammation score was significantly associated with the CTE variables mucosal enhancement, wall thickness, comb sign, and adenopathy (Mantel-Haenszel chi(2) P values 0.04, 0.04, < 0.0001, and 0.016, respectively). The pathological fibrostenosis score was significantly associated with the presence and severity of stenosis on CTE (P = 0.001 and 0.007, respectively). By logistic regression analysis, the strongest association was seen with the comb sign (OR 5.52, P < 0.001) for inflammation and the presence of stenosis (OR 5.87, P = 0.006) for fibrostenosis. There was no interaction between the time interval from CTE to surgery and the strength of these associations. Conclusions: CTE may reliably differentiate between inflammatory and fibrostenotic lesions and may have an important role in the management of CD. Specific CTE variables correlate with each of these phenotypes and deserve further investigations in prospective studies.