Stratifying Endoscopic Submucosal Dissection Risks: A Validated Perforation Preoperative Risk Scoring Tool.

Abstract

This study aimed to develop a predictive model for perforation to improve patient safety during Endoscopic Submucosal Dissection (ESD). ESD provides a less invasive approach for colorectal lesion removal but carries perforation risks that can cause considerable morbidity. Data of patients who underwent ESD for colorectal lesions at a tertiary care center between March 2011 and November 2023 were reviewed from a prospectively maintained database. We stratified the subjects into derivation and validation sets using 50-50 allocation. Predictors of perforation identified in univariate analysis (P<0.22) were evaluated for independent association using multivariate analysis. Model efficacy was assessed using the ROC curve. The study involved 1051 patients (515 males and 536 females) with lesions mainly in right colon (638; 61%), left colon (181; 17%), rectum (137; 13%), and transverse colon (98; 9%). Perforation occurred in 108 (10%) patients. Significant (P<0.05) perforation predictors; first 100 procedures (63.3% vs. 36.7%, P<0.001), lesion size >35 mm (59.2% vs. 38.8%, P=0.00369), extended operative time (mean 169 vs. 76.9 min, P<0.001), higher lift degree (P=0.00114), and incomplete resections (18.4% vs. 2.9%, P<0.001). In multivariate analysis, fibrosis (odds ratio [OR]:2.67, P=0.010), the first 100 procedures (OR:14.6, P<0.001), and lesion size >35 mm (OR: 3.50; P<0.001) emerged as significant independent preoperative. ESD perforation risk score validation achieved adequate performance (AUC: 0.7078; accuracy: 80.19%). This study introduces a validated ESD perforation risk model, providing clinicians with reliable tool for patient risk assessment and potentially improving preprocedural strategies.