Predicting stent success in esophageal anastomotic leaks: the clinical significance of the CT-based defect-to-lumen ratio

Endoscopic stent placement is widely used for managing esophageal anastomotic leaks; however, reliable predictors of success remain poorly defined and often rely on absolute defect size. We hypothesised that relative defect geometry within the lumen might better predict mechanical sealing and clinical success. Patients with CT-confirmed esophageal anastomotic leakage after Ivor Lewis esophagectomy or total gastrectomy who underwent fully covered self-expandable metal stent (SEMS) placement were retrospectively analysed. Defect diameter and luminal diameter were measured on sagittal CT reconstructions to calculate the defect-to-lumen (D/L) ratio. Stent success was defined as complete clinical and radiological resolution without further intervention. ROC analysis was applied as a hypothesis-generating tool to explore the discriminatory performance of the D/L ratio within this cohort. Thirteen patients met inclusion criteria; stent therapy was successful in 7 (53.8%). Absolute defect size did not reliably discriminate outcomes. When patients were stratified according to the defect-to-lumen (D/L) ratio, a clear separation in clinical success was observed. No patient with a D/L ratio < 0.21 achieved successful leak closure, whereas 7 of 9 patients (77.8%) with a ratio ≥ 0.21 were successfully treated ( p = 0.01). Failed cases presented significantly earlier (median 4 vs. 9 days, p = 0.005) and were associated with higher mortality (50%). ROC analysis showed supportive discriminatory performance of the D/L ratio (AUC 0.81), consistent with the observed clinical distribution. Relative defect geometry, expressed by the CT-based defect-to-lumen ratio, showed a potential association with stent outcome in this exploratory cohort.