Outcomes of endoscopic treatment of leaks and fistulae after sleeve gastrectomy: results from a large multicenter U.S. cohort

Background Sleeve gastrectomy is the most commonly performed bariatric surgery in the United States. Leaks after sleeve gastrectomy (SGL) occur in 1% to 3% of patients. Endoscopic therapies are increasingly used for treatment of SGLs, but few data exist on their outcomes. Objectives The aim of this study was to assess technical success, leak resolution, and reoperation rates of patients undergoing endoscopic therapy for repair SGLs. Setting Eight high-volume academic endoscopy centers. Methods Patients undergoing endoscopic therapy for SGLs from 2007 to 2017 were identified. Patients were excluded if the index endoscopic therapy for SGL was performed elsewhere or if no follow-up data were available. Leaks were classified as acute (≤7 d of SG), early (1–6 wk), late (7–12 wk), and chronic (>12 wk). Leak resolution was defined as lack of extraluminal air, extravasation on oral contrast radiography, cross-sectional imaging, or resolution of percutaneous drain output. Demographic and procedural data were recorded as rates of additional therapy, adverse events, and surgical revision. Results A total of 85 patients met criteria for analysis (70 women, age 42.6 ± 10.8 yr). A total of 295 endoscopic sessions (median 3, range 1–14) were performed across the cohort. SGLs resolved after index endoscopic therapy in 43 (50.1%) patients. The primary outcome of endoscopic resolution of SGL was observed in 62 patients (72.9%). There were 34 (11.5%) PRAE (the majority occurring with self-expandable metal stents), all but 1 of which were managed endoscopically. Surgical revision was required in 23 (21.7%) patients. On univariate analyses independent variables associated with the need for surgical revision included both acute and chronic SGLs (P = .028), loculated subphrenic collections/abscesses (P = .03), and intraabdominal sepsis (P = .03). On multivariable logistic regression using statistically significant predictors from the univariate analyses, acute SGLs were significantly associated with a need for surgical revision (odds ratio 4.8, 95% confidence interval 1.2–18.9, P = .025). Conclusion Endoscopic therapy for SGLs is associated with good clinical success, avoiding the need for surgical revision in 73% of patients, with an acceptable adverse event profile. Patients with acute or chronic SGLs and those with loculated abscesses or intraabdominal sepsis are more likely to undergo surgical revision. Endoscopic therapy is an appropriate first-line modality for the management of SGLs, especially those not classified as acute or chronic.