Pancreaticopleural and pancreaticomediastinal fistulas (PPMF), are rare complications of acute and chronic pancreatitis caused by a disrupted pancreatic duct (PD). It involves an abnormal passage through which enzyme-rich pancreatic fluid may leak into the pleural cavity or the mediastinum. Their optimal management remains unclear due to limited clinical data, as prior studies have included only a small number of patients. The aim of this study was to evaluate the outcomes of endoscopic stent therapy in a larger cohort of patients with PPMF secondary to acute or chronic pancreatitis.

A retrospective, single-center case series was performed involving 59 patients with PPMF secondary to acute or chronic pancreatitis. Their treatment was initiated endoscopically in Helsinki University Hospital between 2011-2024. Eleven patients were excluded due to inadequate follow-up information, resulting in a final cohort of 48 patients. All included patients had a PPMF, which was at minimum established through amylase-rich pleural fluid (greater than 200 U/L) or a radiologically identified fistula tract to the pleural cavity and/or to the mediastinum. The entire cohort had fluid in the pleural cavity, mediastinum or both. All data was retrospectively collected from patient records.

Endoscopic retrograde cholangiopancreatography (ERCP) with transpapillary access to the PD and subsequent stent placement was the preferred method for endoscopic therapy (ET). In ERCP, the aim was to identify the fistula tract and place a plastic stent (5Fr-10Fr) beyond the site of ductal disruption. When indicated, PD dilatation was performed using a 4-6 mm dilatation balloon. Successful outcome was determined when pleural fluid cleared, and the fistula was no longer visible through imaging. If ERCP failed, an endoscopic ultrasound (EUS)-guided transmural route was considered for suitable fluid collections.

Among the patients, 33 (69%) had chronic pancreatitis and 15 (31%) had acute pancreatitis. The etiology of acute or chronic pancreatitis was alcohol in 45 (94%) cases. Of the remaining, one had biliary pancreatitis, one had interferon-induced pancreatitis, and one had idiopathic pancreatitis. CT was performed for all patients and demonstrated the fistula tract in 46 (96%) cases, 

Among 30 patients with a fistula in the head or body of the pancreas, the stent was successfully positioned past the disrupted area in 22 cases. All of them had a successful treatment outcome. After the stent bypassed the fistula tract, the patients’ symptoms resolved in a median time of 60 (range 3-156; IQR 45) days. Of the remaining eight cases, ET failed in one case. When the fistula was in the tail of the pancreas, ET succeeded in 14 out of 18 patients (77.8%) but when it was in the head or body of the pancreas ET succeeded in 29 out of 30 patients (96.7%) (p = 0.059). 

ET was ultimately successful in 90% of patients. The median time from successful stent placement to PPMF closure was 2.3 months (range 1-24; IQR 3). Patients underwent a median of three (range 1-8; IQR 2) ERCPs.The median follow-up time was 54 months (range 1-209; IQR 56). There were no fistula recurrences among any of the 48 patients. Mortality was high (27%) but none died due to PPMF. The patients deceased in a median time of 2.3 years (2.0 months - 10.1 years; IQR 5.4 years) after their diagnosis of PPMF.

ET is an effective treatment method for PPMF. The primary goal of ET is to advance the PD stent beyond the fistula tract, as this is associated with treatment success. When the disruption is located in the pancreatic tail, the stent should be positioned as close as possible to the fistula tract. These findings support endoscopic therapy as the first-line treatment for PPMF secondary to acute and chronic pancreatitis. Future prospective studies with larger cohorts are needed to confirm these results.