Classification and endoscopic management of pancreatic fistulas

Abstract

Pancreatic fistula is a common and serious complication that may occur after pancreatic surgery, pancreatitis, trauma, or pancreatic intervention. It can lead to persistent external drainage, pancreatic fluid collections, infection, bleeding, nutritional depletion, repeated interventions, and sepsis. Endoscopic retrograde cholangiopancreatography (ERCP) with transpapillary pancreatic duct stenting is widely used, but its effectiveness depends mainly on ductal continuity, leak location, successful stent placement across the ductal disruption, and the presence of disconnected pancreatic duct syndrome or associated collections. When ductal continuity is lost, the leak cannot be bridged, or mature collections and surgically altered anatomy are present, ERCP alone may be insufficient. In these situations, endoscopic ultrasound (EUS)-guided transmural drainage, EUS-guided pancreatic duct drainage, rendezvous techniques, fistula internalization, and combined approaches may provide alternative or complementary routes for internal drainage. This minireview summarizes current endoscopic strategies for pancreatic fistula and proposes an endoscopy-oriented classification framework to guide treatment selection.

Key Words: Pancreatic fistula; Endoscopic retrograde cholangiopancreatography; Endoscopic ultrasound; Pancreatic duct disruption; Disconnected pancreatic duct syndrome

Core Tip: Pancreatic fistula should be managed according to pancreatic duct anatomy and endoscopic feasibility. Endoscopic retrograde cholangiopancreatography with transpapillary pancreatic duct stenting is useful when ductal continuity is preserved and the ductal disruption can be bridged. When ductal continuity is lost or complex collections and surgically altered anatomy are present, endoscopic ultrasound-guided approaches may provide internal drainage. This minireview proposes an endoscopy-oriented classification framework to guide treatment selection.

INTRODUCTION

Pancreatic fistula is a common and serious complication that may occur after pancreatic surgery, pancreatitis, trauma, or pancreatic intervention. It results from disruption of the pancreatic ductal system and leakage of enzyme-rich pancreatic juice into the peripancreatic space, body cavities, or skin surface. If not properly managed, it may lead to persistent drainage, pancreatic fluid collections, infection, bleeding, repeated interventions, and sepsis[1-3].

Endoscopic therapy has become an important minimally invasive treatment option for selected patients. Endoscopic retrograde cholangiopancreatography (ERCP) with transpapillary pancreatic duct stenting can reduce ductal pressure and promote leak closure when ductal continuity is preserved and the disruption can be crossed by a stent[4-6]. However, ERCP alone may be insufficient in complete duct disruption, disconnected pancreatic duct syndrome (DPDS), mature collections, necrotic cavities, or surgically altered anatomy[2,7-10].

Endoscopic ultrasound (EUS)-guided transmural drainage, EUS-guided pancreatic duct drainage, rendezvous techniques, fistula internalization, and combined approaches have expanded the treatment options for complex pancreatic fistula. Therefore, treatment selection should be guided not only by etiology, fistula output, or clinical severity, but also by ductal anatomy, leak location, collection status, and endoscopic feasibility[3,11]. This minireview summarizes current endoscopic strategies for pancreatic fistula and proposes an endoscopy-oriented classification framework to guide treatment selection according to pancreatic duct anatomy (search strategies are detailed in Supplementary Tables 1 and 2).

CLASSIFICATION OF PANCREATIC FISTULAS

Classification is important in pancreatic fistula because clinical severity and ductal anatomy guide different aspects of management. For postoperative pancreatic fistula (POPF), the 2016 International Study Group of Pancreatic Surgery (ISGPS) classification is based on clinical severity. A biochemical leak is not considered a true POPF. Grade B requires a change in management. Grade C is defined by the need for reoperation, organ failure, or fistula-related death[1]. Although useful for assessing severity, this classification provides limited guidance for endoscopic treatment because it does not reflect pancreatic duct anatomy or procedural feasibility[3,11]. Table 1 summarizes the ISGPS grading of POPF.

Table 1 International Study Group of Pancreatic Surgery grading of postoperative pancreatic fistula.

Grade	Required biochemical criterion	Clinical criteria	Clinical implication
Biochemical leak	Drain fluid amylase activity > 3 times the upper limit of the institutional normal serum amylase value on or after postoperative day 3	No clinically relevant change in management; drains may remain for observation for up to 3 weeks; no POPF-related intervention, organ failure, or death	Not considered clinically relevant POPF; usually requires observation or routine drain care
Grade B POPF	Fulfills the biochemical criterion above	Requires a clinically relevant change in management, such as persistent drainage > 3 weeks, percutaneous or endoscopic drainage, angiographic intervention for bleeding, or infection without organ failure	Clinically relevant POPF requiring active management
Grade C POPF	Fulfills the biochemical criterion above and represents severe clinically relevant POPF	Associated with reoperation, organ failure, or death	Severe POPF requiring urgent multidisciplinary rescue, intensive care, and/or surgery

Management changes and adverse events should be postoperative pancreatic fistula-related. Grade B and grade C are clinically relevant postoperative pancreatic fistulas, whereas biochemical leak is not considered a true pancreatic fistula. POPF: Postoperative pancreatic fistula.

Previous pancreatography-based classifications have shown that main pancreatic duct (MPD) anatomy is a key determinant of treatment selection. Nordback et al[12] classified pancreatic pseudocysts according to MPD anatomy, distinguishing a normal duct, direct communication between the MPD and pseudocyst, and MPD stricture. The classification was intended to use pancreatographic anatomy to help guide treatment selection. Nealon and Walser[13] further demonstrated that MPD anatomy could guide the choice between percutaneous drainage and surgery. Mutignani et al[3] proposed an endoscopy-oriented classification based on leak location and ductal anatomy. These features were used to help select the drainage route, including transpapillary, transmural, or anastomotic drainage. Proença et al[11] also highlighted the importance of distinguishing partial duct disruption from complete disruption or disconnected duct anatomy. Together, these systems indicate that ductal continuity, ductal stricture, leak location, and DPDS are more relevant to endoscopic feasibility than etiology or fistula output alone.

Therefore, anatomical assessment before endoscopic treatment should define the leak site, MPD continuity, upstream duct access, whether the disrupted segment can be crossed by a guidewire and stent, the presence and maturity of pancreatic fluid collections, DPDS, and surgically altered anatomy. Computed tomography (CT) and magnetic resonance cholangiopancreatography (MRCP) provide noninvasive structural evaluation, whereas ERCP and EUS provide ductal assessment and therapeutic access[9-11,14]. The key question is not only whether contrast leakage is present, but whether the anatomy permits duct decompression, stent placement across the disruption, internal drainage, or surgical reconstruction. Figure 1 and Supplementary Table 3 summarize classification systems.

Figure 1 Previous classification systems relevant to pancreatic fistula and endoscopic decision-making. A: Nordback et al[12] classification: Type I, normal main pancreatic duct (MPD) with (IA) or without (IB) collection filling; Type II, MPD opens to the collection; Type III, MPD stricture with (IIIA) or without (IIIB) collection filling; B: Nealon et al[31] classification: Type I, normal MPD; Type II, MPD stricture; Type III, MPD occlusion; Type IV, chronic pancreatitis; Subtype A indicates no communication between the pancreatic duct and the collection, and subtype B indicates communication; C: Mutignani et al[3] classification: Type I, side-branch leak in the pancreatic head, body, or tail; Type II, MPD leak with open or closed proximal stump; Type III, post-pancreatectomy leak after proximal or distal pancreatic resection; D: Proença et al[11] classification: Type I, normal MPD; Type II, MPD stricture; Type III, partial disruption; Type IV, complete disruption with (IV-A) or without (IV-B) contrast extravasation.

TRANSPAPILLARY PANCREATIC DUCT STENTING

The principle of transpapillary pancreatic duct stenting in pancreatic fistula is to reduce pancreatic duct pressure and redirect pancreatic juice into the duodenum[6,15]. When the MPD is continuous or only partially disrupted, a guidewire and stent can often be advanced across the leak site. Placement of the stent across the disruption helps bridge the leak and facilitates fistula closure. Therefore, the success of ERCP depends largely on whether the ductal disruption can be crossed and bridged by a guidewire and stent[16,17].

Successful bridging is associated with better outcomes, although the supporting studies are predominantly retrospective and include mixed etiologies[4,6,17]. Chen et al[17] reported a 3-month composite clinical success rate of 79.2% in 106 patients, with higher success after bridging than non-bridging drainage (94.4% vs 71.4%). This endpoint included improvement in pancreatic fluid collections and should not be equated with confirmed duct healing. In 22 patients with non-POPF, Adachi et al[18] reported higher clinical success for proximal than distal disruptions (89% vs 23%) and poorer outcomes for complete disruption than other types (0% vs 65%). These findings support anatomy-based selection, but differences in etiology, disruption severity, and outcome definitions limit direct comparisons and extrapolation to postoperative stump or anastomotic leaks.

The optimal duration of pancreatic stent placement remains uncertain. Testoni[5] suggested leaving the stent in place for 4-6 weeks. Varadarajulu et al[4] reported stent exchange at 6-8 weeks intervals and a median duration of stenting of 58 days, whereas Chen et al[17] exchanged stents every 3 months until ERCP confirmed leak closure. Reported adverse events include pancreatitis, stent occlusion or migration, and stent-associated duct changes[4,15-17] (key studies on transpapillary pancreatic duct stenting are summarized in Supplementary Table 4).

EUS-GUIDED STRATEGIES

In pancreatitis-related disruption, infected necrosis and symptomatic mature collections require priority management[8,19]. Complete disruption or DPDS often prevents restoration of ductal continuity by ERCP[2,7,10,20,21]. In such cases, EUS-guided approaches may provide alternative routes for internal drainage. EUS-guided transmural drainage is most useful when a mature pancreatic fluid collection or walled-off necrosis is adjacent to the stomach or duodenum. In this setting, the treatment strategy shifts from restoring pancreatic duct continuity to EUS-guided internal drainage. In selected patients with DPDS, EUS-guided pancreatic duct drainage may also provide an alternative route for drainage[9,11]. Lumen-apposing metal stents provide a large-caliber tract and may facilitate drainage or direct endoscopic necrosectomy in walled-off necrosis[8,19,22], whereas double-pigtail plastic stents remain useful for longer-term drainage, especially after metal stent removal or when recurrence risk persists[23,24]. Direct endoscopic necrosectomy should be considered when infected or symptomatic necrotic debris is present. The value of adding transpapillary stenting to transmural drainage remains uncertain. Ni et al[25] and Trevino et al[26] reported favorable outcomes in selected patients with duct disruption, whereas Amin et al[27] and Yang et al[28] found no clear overall benefit in populations dominated by pancreatic pseudocysts.

Other EUS-guided techniques may be considered when conventional ERCP fails or is anatomically impossible. EUS-guided pancreatic duct drainage can provide access in symptomatic patients with obstructed pancreatic ducts when retrograde intervention is unsuccessful or not feasible[9]. EUS-guided rendezvous may be useful when the papilla or anastomosis can ultimately be reached but conventional cannulation fails[9]. EUS-guided fistula internalization has also been reported for persistent pancreaticocutaneous fistula or postoperative leakage when an external tract or collection can be converted into internal enteric drainage[29,30]. Evidence for fistula internalization is limited to small uncontrolled reports, whereas guideline evidence for EUS-guided pancreatic duct drainage mainly concerns symptomatic duct obstruction rather than fistula closure[9,29,30]. These technically demanding procedures should be performed in experienced centers.

Treatment selection should balance ductal anatomy, collection characteristics, procedural risk, and surgical fitness. Transpapillary bridging is favored when feasible, whereas transmural drainage addresses suitable collections. Percutaneous drainage and surgery remain alternatives when endoscopic access or drainage is inadequate[8,19]. Recommendations also differ for disconnected pancreatic remnants: The European Society of Gastrointestinal Endoscopy supports long-term transluminal plastic stenting in DPDS, whereas the American Gastroenterological Association advises distal pancreatectomy for suitable patients with a disconnected left remnant after necrotizing mid-body pancreatitis[8,19]. These context-specific recommendations do not establish universal superiority of either approach. Bleeding, perforation, infection, and stent-related complications must be weighed alongside the risks of repeated intervention (Supplementary Table 5).

NEW CLASSIFICATION FOR ENDOSCOPIC MANAGEMENT OF PANCREATIC FISTULAS

Endoscopic treatment of pancreatic fistula should not be determined only by etiology, fistula output, or clinical severity, because these factors do not directly reflect ductal anatomy or endoscopic feasibility. The key questions are whether the MPD remains continuous, whether the ductal disruption can be crossed by a guidewire and stent, whether the upstream pancreas is disconnected, and whether associated collections or surgically altered anatomy require internal drainage. Based on previous classifications of pancreatic duct anatomy and endoscopic management, we propose the B-P-C-A classification [type B: Branch-duct leak with intact MPD continuity. Type P: Partial MPD disruption, with contrast beyond the disruption. Type C: Complete MPD disruption or disconnected pancreatic duct syndrome, with non-opacification beyond the interruption interpreted alongside CT/MRCP to exclude obstruction and assess the upstream pancreas; C1 indicates a patent proximal duct and C2 indicates a stenotic or closed proximal duct. Type A: Postoperative stump or anastomotic leak, including distal pancreatectomy stump leak (A1) and pancreaticojejunostomy anastomotic leak (A2)] to guide treatment selection according to pancreatic duct anatomy[3,11-13,31] (Figure 2).

Figure 2 Proposed B-P-C-A classification framework for pancreatic fistula. Type B: Branch-duct leak with intact main pancreatic duct (MPD) continuity. Type P: Partial MPD disruption, with contrast beyond the disruption. Type C: Complete MPD disruption or disconnected pancreatic duct syndrome, with non-opacification beyond the interruption interpreted alongside computed tomography/magnetic resonance cholangiopancreatography to exclude obstruction and assess the upstream pancreas; C1 indicates a patent proximal duct and C2 indicates a stenotic or closed proximal duct. Type A: Postoperative stump or anastomotic leak, including distal pancreatectomy stump leak (A1) and pancreaticojejunostomy anastomotic leak (A2).

In type B, a pancreatic branch duct is disrupted, while the MPD remains intact. The anatomical injury is confined to a branch duct, and clinically stable patients with adequate drainage can initially be managed conservatively[3].

In type P, the MPD is partially disrupted, with residual ductal continuity and opacification of the upstream duct beyond the disruption on pancreatography. When endoscopic intervention is indicated, transpapillary stenting is preferred if the disruption can be bridged. A guidewire is advanced across the disruption into the upstream duct, allowing placement of a pancreatic stent to redirect pancreatic juice into the duodenum and promote fistula healing[4,6].

In type C, the MPD is completely interrupted. In type C1, this segment remains patent, and short transpapillary stenting may be considered for decompression in selected patients; however, this does not restore drainage from the disconnected upstream pancreas. In type C2, the proximal duct is stenotic or occluded, making decompression alone unlikely to be effective unless the obstruction can be overcome. The upstream duct assessed in Figure 3 lies beyond the interruption and is distinct from the papilla-side segment used to define C1/C2. Recanalization or bridging is considered only when technically feasible in selected patients; it is not routine treatment for established DPDS. Persistent leakage may require EUS-guided drainage, combined interventions, or surgery, depending on ductal anatomy and associated collections[3,9,17]. Non-opacification of the upstream duct should be interpreted alongside CT or MRCP findings to distinguish complete disruption from obstruction and assess for DPDS[10,11,14].

Figure 3 Clinical and endoscopic management considerations for pancreatic fistula. This decision-oriented schematic translates the B-P-C-A framework into endoscopic management considerations. Type B generally emphasizes conservative management, pressure reduction, and management of mature collections when needed. Type P highlights endoscopic retrograde cholangiopancreatography decompression and stent bridging when duct continuity can be restored. For Type C, “upstream duct” refers to the gland-side duct beyond the interruption, not the papilla-side segment defining C1/C2. “Open/closed” concerns upstream duct access; dilation, recanalization, bridging, and endoscopic ultrasound (EUS) guided rendezvous are conditional options in selected technically feasible cases, not routine treatment for established disconnected pancreatic duct syndrome. Short transpapillary stenting decompresses the papilla-side segment but does not drain the disconnected upstream gland. Persistent non-bridgeable leakage requires assessment for EUS-guided transmural drainage, long-term plastic stents, EUS-guided pancreatic duct drainage, EUS-guided pancreaticocutaneous fistula internalization, hybrid therapy, or surgery. Type A emphasizes postoperative anatomy, associated pancreatic fluid collection, and possible EUS-guided internal drainage or fistula internalization when conventional transpapillary bridging is unlikely. CT: Computed tomography; MRCP: Magnetic resonance cholangiopancreatography; ICU: Intensive care unit; ERCP: Endoscopic retrograde cholangiopancreatography; EUS: Endoscopic ultrasound; PFC: Pancreatic fluid collection; WON: Walled-off necrosis; MPD: Main pancreatic duct; DPDS: Disconnected pancreatic duct syndrome; EST: Endoscopic sphincterotomy; EUS-RV: Endoscopic ultrasound-guided rendezvous.

Type A fistulas arise from a pancreatic resection stump or pancreaticoenteric anastomosis. Type A1 denotes a pancreatic stump leak after distal pancreatectomy, whereas Type A2 denotes a pancreaticojejunostomy anastomotic leak after pancreaticoduodenectomy. The operative anatomy and localization of the leak distinguish these subtypes. Management centers on adequate drainage of postoperative collections and internal drainage of persistent leakage when feasible[3].

The B-P-C-A framework brings these anatomical patterns together to support endoscopic treatment selection, with particular emphasis on ductal continuity, opacification beyond the disruption, and the feasibility of guidewire passage and stent bridging. Its intended contribution is to translate established anatomical findings into a practical treatment pathway that complements clinical severity grading[3,11] (Figure 3).

LIMITATIONS AND FUTURE DIRECTIONS

This review has several limitations. The proposed B-P-C-A framework has not yet been tested in clinical practice, and its feasibility, reproducibility, and clinical utility remain uncertain. Its applicability may also depend on local expertise and access to advanced EUS and combined interventions. Future studies should first apply the framework to consecutive cases and assess interobserver agreement, followed by prospective multicenter validation across different fistula types and practice settings. Comparison with existing classifications is also needed to determine whether B-P-C-A improves treatment selection.

CONCLUSION

Endoscopic management of pancreatic fistula should be guided primarily by pancreatic duct anatomy and the feasibility of restoring internal drainage. When the MPD is continuous and the disruption can be crossed, ERCP with transpapillary stent bridging is preferred. When the disruption cannot be bridged, or when complete duct disruption, DPDS, mature collections, or surgically altered anatomy is present, EUS-guided drainage, EUS-guided pancreatic duct intervention, combined approaches, or surgery should be considered. The proposed B-P-C-A classification classifies pancreatic fistulas according to anatomical features and may help guide treatment selection. External validation is required to confirm its clinical applicability.