Deep mural injury and perforation after colonic endoscopic mucosal resection: a new classification and analysis of risk factors

Objectives Perforation is the most serious complication associated with endoscopic mucosal resection (EMR). We propose a new classification for the appearance and integrity of the muscularis propria (MP) after EMR including various extents of deep mural injury (DMI). Risk factors for these injuries were analysed. Design Endoscopic images and histological specimens of consecutive patients undergoing EMR of colonic laterally spreading lesions ≥20 mm at a large Australian tertiary referral endoscopy unit were retrospectively analysed using our new DMI classification system. DMI was graded according to MP injury (I/II intact MP without/with fibrosis, III target sign, IV/V obvious transmural perforation without/with contamination). Histological specimens were examined for included MP and patient outcomes were recorded. All type III–V DMI signs were clipped if possible, types I and II DMI were clipped at the endoscopists’ discretion. Results EMR was performed in 911 lesions (mean size 37 mm) in 802 patients (male sex 51.4%, mean age 67 years). DMI signs were identified in 83 patients (10.3%). Type III–V DMI was identified in 24 patients (3.0%); clipping was successfully performed in all patients. A clinically significant perforation occurred in two patients (0.2%). Only one of the 59 type I/II cases experienced a delayed perforation. 85.5% of patients with DMI were discharged on the same day, all without sequelae. On multivariable analysis, type III–V DMI was associated with transverse colon location (OR 3.55, p=0.028), en bloc resection (OR 3.84, p=0.005) and high-grade dysplasia or submucosal invasive cancer (OR 2.97, p 0.014). Conclusions In this retrospective analysis, use of the new classification and management with clips appeared to be a safe approach. Advanced DMI types (III–V) occurred in 3.0% of patients and were associated with identifiable risk factors. Further prospective clinical studies should use this new classification. Trial registration number NCT01368289; results.