Pancreatic leaks: Endo-therapy first?


Dr D Nageshwar Reddy, Asian Institute of Gastroenterology, 6-3-661 Somajiguda, Hyderabad 500 082, India. Email:

Pancreatic leaks or fistulae occur due to disruption of the main pancreatic duct or its side-branches. This may arise as a complication of acute pancreatitis, chronic pancreatitis, pancreatic malignancy, after pancreatic surgery or pancreatic trauma. Extravasation of pancreatic juice from the gland may lead to a localized collection or pseudocyst, ascites or a pleural effusion, depending on the location of ductal rupture and the integrity of the surrounding tissues. An anterior pancreatic leak is often associated with formation of pseudocyst; rupture from it may lead to pancreatic ascites. A posterior ductal disruption causes fluid to ascend retro-peritoneally into the mediastinum through the aortic or esophageal hiatus, manifesting as pleural effusion.1 Occurrence of pleural effusion is more often on the left than the right. External pancreatic fistulae arise from a break in skin integrity; usually as a result of surgical or radiological drains and rarely spontaneously. The corrosive properties of pancreatic juice may lead to the formation ofarterial pseudo-aneurysms or venous thrombosis, particularly in the splenic or portal veins. Hemorrhagic collections, hemosuccus pancreaticus, or segmental portal hypertensive-related bleeding are consequences of these complications.2

Pancreatic fistulae prolong the hospital stay of patients, with associated increase in morbidity, mortality and overall health-related expenditure.3 These are severe complications, and their management usually involves surgeons, gastroenterologists and radiologists.

Management of pancreatic leaks and collections has evolved over years. Traditionally, surgery has been the mainstay of treatment necessitating resection of the gland to a variable extent with associated high morbidity and mortality.4–7 The trend in the last two decades has tilted towards aggressive medical management. This involves suppression of pancreatic secretion by keeping the patient nil orally, using proton pump inhibitors, and somatostatin or its analogs. Repeated paracentesis, with or without parenteral nutrition or enteral nutrition, has frequently been used as the first-line management. Radiologists have refined their technique of draining peripancreatic collections using percutaneous catheters and even internalizing them when required.

In the last one and a half decades, endoscopic interventions have advanced from biliary endoscopic procedures to therapeutic pancreatic endoscopy. Numerous case reports have been published of successful endoscopic management of pancreatic leaks.8–11 Unlike bile leak, which resolves with the elimination of the bile duct–duodenum pressure gradient by biliary sphincterotomy or stenting, decreasing the pancreatic duct–duodenal pressure gradient by pancreatic sphincterotomy or stent placement across the papilla may not, on its own, be sufficient therapy for all patients with a pancreatic leak.12 Hence, there is necessity for a bridging trans-papillary pancreatic stenting in definitive management of communicating leaks.

Pai et al. in the current issue of the Journal have added a series of successful endoscopic treatment of pancreatic leaks presenting with symptomatic ascites or pleural effusion.13 They report successful resolution of pancreatic ascites and effusion in 27 out of 28 patients with various etiologies, including both acute and chronic pancreatitis, by endoscopic placement of a trans-papillary pancreatic stent. After 3–6 weeks of placement, the pancreatic stent was later extracted with no reported recurrence over a median 18-month follow up.

Vardarajulu et al. also reported a large cohort of patients with pancreatic fluid collections managed by endoscopic trans-papillary stent insertion.14 In their retrospective series, pancreatic stenting was successful in 92 of 97 patients. The stent was left in situ for a median of 8 weeks. On the basis of multivariate logistic regression analysis, they concluded that two factors—partially disrupted pancreatic duct and a stent bridging this disruption—correlated with a successful outcome. A complete pancreatic ductal disruption is unlikely to respond to endotherapy, as only six out of 23 such patients had successful resolution of pancreatic fluid collection.14 This issue was not addressed to by Pai et al. in their study.13 Stents bridging the leak correlate with good outcome, while stents placed at the level of the leak or distally (downstream of the leak) often result in failure of endoscopic treatment.14–16

Apart from stent placement in pancreatic leaks, nasopancreatic drainage (NPD) showed good results. Bhasin et al. effectively used NPD in 10 out of 11 patients with pancreatic ascites and pleural effusion.8 NPD provides the convenience of repeated pancreatograms to demonstrate healing of the leak without repeated endoscopic retrograde pancreatograms (ERP). Also, if the NPD became blocked, it could be flushed and aspirated as and when needed. Once the pancreatogram confirmed absence of a leak, the NPD could be removed without the necessity of an additional ERP or stent.8

Demonstration of the site of pancreatic ductal leak is crucial for a successful endotherapy. Secretin magnetic resonance cholangiopancreatography (sMRCP) is currently the best non-invasive method to localize the site of a pancreatic leak.17 It provides a pancreatic ‘ductal roadmap’, upstream and downstream of the leak, presence of strictures and presence of anatomical aberrations like pancreas divisum. MRCP is available at most tertiary care centers. The addition of secretin injection to stimulate pancreatic secretion is essential to demonstrate the exact site of pancreatic leak. Lack of sMRCP compels the endoscopist to judiciously inject contrast at ERP to get an adequate roadmap.

Some authors have reported a liberal use of contrast injection to get a good pancreatogram to tailor their endotherapy.14,18 While trying to demonstrate the site of pancreatic leak with ERP, there is a potential risk of infecting the sterile collections. On the other hand, a potential drawback of an incompletely opacified main pancreatic duct is accidental placement of a stent in a long parallel side-branch, or suboptimal pancreatic stent length selection. In their study, Pai et al. used the ‘middle path’ of injecting small quantities of contrast to visualize the site of the leak in the pancreatic duct, and to clarify the status of the duct downstream to the leak. In one-third of their patients, a leak was not demonstrable. Irrespective of this information (demonstration of a leak), a long pancreatic stent was placed to ‘bridge the entire pancreatic duct’.

How long should we leave a pancreatic stent in situ in patients with pancreatic leaks? Most patients with NPD or stent in situ require more than 2 weeks for resolution. The ideal duration for leaving a pancreatic duct (PD) stent in situ is unknown, but an optimal duration of 4 weeks is reasonable.19 Pai et al. kept the stent for 3–6 weeks and others have retained it for 8 weeks.13 Pancreatography should be performed after removal of the stent or at NPD extraction. If the ductal disruption persists, a new stent should be placed. Ikenberry summarized in his study that stent occlusion is directly proportional to duration of stent placement, and by 9 weeks all the extracted stents were occluded.20

What about tight PD stricture with a leak of contrast upstream to it? These are often best managed by resective pancreatic surgery, unless the pancreatic stricture can be negotiated and dilated, thereby allowing the stent to bridge the leak. The latter is only possible if adequate anatomy is demonstrated with either sMRCP or a liberal filling of contrast at ERP.

Pancreatic ascites or pleural effusion can occur without a demonstrable leak at pancreatogram, as observed in one-third of Pai et al.'s series. This could be due to spontaneous sealing of the leak or a minor leak from a peripheral branch not demonstrated with suboptimal injection. Should we stent such patients or will pancreatic sphincterotomy alone suffice? This can of course only be resolved by a prospective study.

Management of PD disruption remains controversial, as treatment strategies are not uniform. Comparison of the different modalities in a randomized fashion is difficult. Most of the published work available is in the form of case reports or observational studies. Only prospective randomized controlled trials will tell us which form of therapy—medical, endoscopic or surgical—is most effective in these patients.