Optimizing management of pancreaticopleural fistulas.

Aim: To evaluate the management of pancreaticopleural fistulas involving early endoscopic instrumentation of the pancreatic duct.

Methods: Eight patients with a spontaneous pancreaticopleural fistula underwent endoscopic retrograde cholangiopancreatography (ERCP) with an intention to stent the site of a ductal disruption as the primary treatment. Imaging features and management were evaluated retrospectively and compared with outcome.

Results: In one case, the stent bridged the site of a ductal disruption. The fistula in this patient closed within 3 wk. The main pancreatic duct in this case appeared normal, except for a leak located in the body of the pancreas. In another patient, the papilla of Vater could not be found and cannulation of the pancreatic duct failed. This patient underwent surgical treatment. In the remaining 6 cases, it was impossible to insert a stent into the main pancreatic duct properly so as to cover the site of leakage or traverse a stenosis situated downstream to the fistula. The placement of the stent failed because intraductal stones (n = 2) and ductal strictures (n = 2) precluded its passage or the stent was too short to reach the fistula located in the distal part of the pancreas (n = 2). In 3 out of these 6 patients, the pancreaticopleural fistula closed on further medical treatment. In these cases, the main pancreatic duct was normal or only mildly dilated, and there was a leakage at the body/tail of the pancreas. In one of these 3 patients, additional percutaneous drainage of the peripancreatic fluid collections allowed better control of the leakage and facilitated resolution of the fistula. The remaining 3 patients had a tight stenosis of the main pancreatic duct resistible to dilatation and the stent could not be inserted across the stenosis. Subsequent conservative treatment proved unsuccessful in these patients. After a failed therapeutic ERCP, 3 patients in our series developed superinfection of the pleural or peripancreatic fluid collections. Four out of 8 patients in our series required subsequent surgery due to a failed non-operative treatment. Distal pancreatectomy with splenectomy was performed in 3 cases. In one case, only external drainage of the pancreatic pseudocyst was done because of diffuse peripancreatic inflammatory infiltration precluding safe dissection. There were no perioperative mortalities. There was no recurrence of a pancreaticopleural fistula in any of the patients.

Conclusion: Optimal management of pancreaticopleural fistulas requires appropriate patient selection that should be based on the underlying pancreatic duct abnormalities.