Establishment and validation of a predictive scoring system for titanium clip retention at six months after...

Titanium clips are commonly used to close mucosal defects after gastric lesion endoscopic submucosal dissection (ESD). However, prolonged titanium clip retention beyond the healing period may have potential negative effects on postoperative outcomes. This study developed and validated a scoring model to predict titanium clip retention at six months post-gastric ESD. A multicenter retrospective study was conducted on 1,055 patients who underwent gastric ESD with titanium clips closure for wound management. Patients were grouped into a training cohort (TC, n = 509), an internal validation cohort (IVC, n = 218), and an external validation cohort (EVC, n = 328). Univariate and multivariate logistic regression were applied to identify risk factors for titanium clip retention at the 6-month follow-up. A scoring system was then built by assigning weighted scores based on the regression coefficients of independent predictors. The model's performance was evaluated using the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and predictive values. Among the patients, 132 (12.5%) experienced prolonged titanium clip retention. Multivariate analysis revealed three independent predictors: tumor in the upper third of the stomach (2 points), diabetes (1 point), and the use of ≥ 8 clips (1 point). The scoring model exhibited strong predictive ability, with AUCs of 0.809 in the IVC and 0.855 in the EVC, along with high negative predictive values (0.946 and 0.984, respectively). Risk stratification classified patients into low-risk (0-1 points), intermediate-risk (2-3 points), and high-risk (4 points) groups. The observed retention rates were 3.5%, 14.8%, and 58.8% in the IVC, and 1.6%, 20.8%, and 76.5% in the EVC, respectively. The proposed predictive score, which combines tumor location, diabetic status, and clip number, effectively stratifies the risk of long-term titanium clip retention after ESD. This tool may assist in risk stratification for personalized postoperative surveillance, pending prospective validation.