Preoperative detection of muscle retraction in colorectal ESD using computed tomography–colonography

BACKGROUND: Muscle retraction (MR) is a major cause of technical difficulty and perforation during colorectal endoscopic sub-mucosal dissection (ESD). However, no established method exists for predicting MR preoperatively. This study investigated whether computed tomography-colonography (CTC) can detect MR before ESD. METHODS: We retrospectively analysed 69 patients who underwent colorectal ESD without muscular invasion (Validation 1). The serosal-side appearance of the tumour on CTC was classified into four types-bulge, matchstick, round depression and sharp depression-and compared with endoscopically observed MR (eMR). To validate the findings, 55 surgically resected colorectal cancers without muscular invasion were analysed for pathological MR (pMR) and correlated with the same CTC classification (Validation 2). Logistic regression analyses were performed to identify predictors of MR. RESULTS: In Validation 1, all eMR-positive lesions (5/5) showed the sharp depression type on CTC, whereas 3 of 64 eMR-negative lesions did (Fisher's exact test, p < 0.0001). In Validation 2, multivariate logistic regression showed that both the sharp depression type (OR 138, p < 0.0001) and severe sub-mucosal fibrosis (OR 4453, p = 0.0079) were independent predictors of pMR. Intero-bserver agreement was almost perfect (κ = 0.93-0.95). CONCLUSIONS: The serosal-surface appearance of colorectal tumours on CTC, especially the sharp depression type, strongly predicts MR. CTC may serve as a simple, non-invasive preoperative tool for identifying MR and selecting optimal treatment strategies, including surgery, before colorectal ESD.