Endoscopic vacuum therapy (EVT) is an established procedure for treating leaks in the gastrointestinal tract (GIT). This study aimed to investigate overarching factors influencing the success of therapy and the effects of EVT in the upper GIT. Patient characteristics associated with prolonged or complicated courses of EVT should be identified. 

All patients who underwent EVT treatment at a tertiary care centre between 2019 and 2024 were included in this retrospective analysis (n = 35). EVT was performed by placing an open-cell polyurethane sponge either endoluminally or intracavitarily at the site of the leak, typically using a standard gastroscope with an overtube or piggyback technique. Continuous debridement and enhanced secretion drainage were achieved by applying a constant negative pressure of ‑125 mmHg. Sponge changes were carried out every 3–4 days under endoscopic guidance. Commercially available systems were used (Eso-/Endo SPONGE™, B. Braun Melsungen AG, Melsungen, Germany; Suprasorb® CNP wound foam, Lohmann & Rauscher GmbH & Co. KG, Neuwied, Germany; VacStent Medtech AG, Steinhausen, Switzerland).

The defect size was categorised as small (0-9 mm), medium (10-20 mm) or large (> 20 mm) for the leakage and small (≤ 20 mm), medium (21-50 mm) or large (> 50 mm) for the insufficiency cavity.

Therapeutic success was defined as an endoscopically verified complete closure of the leak.

EVT achieved an overall success rate of 83.3% in this cohort, with a mean treatment duration of 33 days and an average of 9 sponge changes per patient.  

A higher Charlson Comorbidity Index (CCI) and ASA score were associated with a poorer therapeutic outcome (p = 0.023 and p = 0.031, respectively). The sponge location significantly influenced the total duration of therapy (p = 0.009), with intraluminal placement being associated with a shorter total duration of EVT. BMI, sex, age, number of sponge changes and intervals of change showed no statistically significant effects on therapeutic success. Patients who had received neoadjuvant chemotherapy tended to have significantly smaller fistula cavities (p = 0.031). Larger insufficiency cavities as well as larger leakages were associated with poorer outcomes (p = 0.0015 and p = 0.006). Kruskal–Wallis analyses demonstrated that medium-sized (p = 0.016) and large cavities (p = 0.001) were significantly associated with longer treatment duration. In contrast, no significant effect was found on the total duration of EVT in this cohort. Complications occurred in 22.9% of cases (n = 12), including therapy-related issues (e.g. suction problems), endoscopic complications and late complications (e.g. stenosis). However, no significant association with treatment duration (p = 0.122) or therapeutic success (p = 0.127) was found.

Our results confirm EVT as an established and clinically relevant treatment option for leaks, with high success rates and a low incidence of serious complications. Compared to surgical methods, EVT is significantly less invasive.

This study also identifies initial risk factors for prolonged treatment courses (fistula cavity, cavity size, high CCI and ASA score) and provides further evidence of EVT effectiveness.