Purpose: We aimed to evaluate the diagnostic performance of computed tomography colonography (CTC) in the detection of internal hemorrhoids.
Methods: Three gastroenterologists systematically reported on the presence of internal hemorrhoids in patients with incomplete colonoscopy, for whom they considered a subsequent CTC. For 44 patients with internal hemorrhoids revealed by optical colonoscopy, an age- and gender-matched cohort of 66 patients with normal findings in the rectum was selected. Endoluminal and transaxial CTC views of the rectum were evaluated for the presence of internal hemorrhoids, the anal verge prominence, asymmetry, and cushion-like appearance on a Likert scale by two experienced radiologists and two gastroenterologists.
Results: The sensitivity, specificity, and AUC for identification of internal hemorrhoids were 0.61 (95% CI, 0.53-0.68), 0.69 (95% CI, 0.63-0.75) and 0.66 (95% CI, 0.62-0.70), respectively. The radiologists showed a better specificity, the gastroenterologists a slightly better sensitivity. When only the rating "very likely" was considered as positive, the specificity rose to 0.89 (95% CI, 0.81-0.94) with a sensitivity of 0.50 (95% CI, 0.38-0.62). The interobserver agreement was fair. The best predictor of the presence of hemorrhoids was a prominent anal verge in the supine position (OR=1.789, 95% CI, 1.267-2.525). The difference between supine and prone positions in the evaluated features in patients with internal hemorrhoids was not significant.
Conclusion: CTC has low sensitivity but high specificity in the detection of internal hemorrhoids, if the rater is confident in detecting them. Internal hemorrhoids do not substantially change their shape between prone and supine positions.