Factors predictive of persistent fistulas in EUS-directed transgastric ERCP: a multicenter matched case-control study

Abstract

EUS-directed transgastric ERCP (EDGE) is an established method for managing pancreaticobiliary pathology in Roux-en-Y gastric bypass patients, with high rates of technical success and low rates of serious adverse events (AEs). However, widespread adoption of the technique has been limited because of concerns about the development of persistent gastrogastric or jejunogastric fistulas. Gastrogastric and jejunogastric fistulas have been reported in up to 20% of cases in some series, but predictive risk factors and long-term management and outcomes are lacking. Therefore, our aims were to assess factors associated with the development of persistent fistulas and the technical success of endoscopic fistula closure. This is a case-control study involving 9 centers (8 USA, 1 Europe) from February 2015 to September 2021. Cases of persistent fistulas were defined as endoscopic or imaging evidence of fistula more than 8 weeks after lumen-apposing metal stent (LAMS) removal. Control subjects were defined as endoscopic or imaging confirmation of no fistula more than 8 weeks after LAMS removal. AEs were defined and graded according to the American Society for Gastrointestinal Endoscopy lexicon. Twenty-five patients identified to have evidence of a persistent fistula on follow-up surveillance (cases) were matched with 50 patients with no evidence of a persistent fistula on follow-up surveillance (control subjects) based on age and sex. Mean LAMS dwell time was 74.7 ± 106.2 days. After LAMS removal, argon plasma coagulation (APC) ablation of the fistula was performed in 46 patients (61.3%). Primary closure of the fistula was performed in 26.7% of patients (20: endoscopic suturing in 17, endoscopic tacking in 2, and over-the-scope clips+ endoscopic suturing in 1). When comparing cases with control subjects, there was no difference in baseline demographics, fistula site, LAMS size, or primary closure frequency between the 2 groups (P > .05). However, in the persistent fistula group, the mean LAMS dwell time was significantly longer (127 vs 48 days, P= .02) and more patients had≥5% total body weight gain (33.3% vs 10.3%, P= .03). LAMS dwell time was a significant predictor of persistent fistula (odds ratio, 4.5 after >40 days in situ, P= .01). The odds of developing a persistent fistula increased by 9.5% for every 7 days the LAMS was left in situ. In patients with a persistent fistula, endoscopic closure was attempted in 19 (76%) with successful resolution in 14 (73.7%). Longer LAMS dwell time was found to be associated with a higher risk of persistent fistulas in EDGE patients. APC or primary closure of the fistula on LAMS removal was not found to be protective against developing a persistent fistula, which, if present, can be effectively managed through endoscopic closure in most cases.