Endoscopic management of bile duct leaks: Current strategies and controversies

Abstract

Bile duct leak (BDL) is a common complication of hepatobiliary surgery, including cholecystectomy, hepatic resection, and liver transplantation. Timely recognition and effective management are essential to prevent serious complications such as peritonitis, intra-abdominal abscesses, and sepsis. Endoscopic retrograde cholangiopancreatography (ERCP) has become the first-line diagnostic and therapeutic approach. Despite the high overall success rate, the optimal treatment strategies remain a subject of ongoing debate. This review summarizes the existing classification systems and current endoscopic management of BDLs, including drainage strategies, optimal timing of ERCP and stent removal, and approaches for complex cases.

Key Words: Endoscopic retrograde cholangiopancreatography; Bile duct leaks; Endoscopic biliary drainage; Biliary stent; Biliary fistula

Core Tip: Bile duct leak is a common but challenging complication after hepatobiliary surgery. This review summarizes current classification systems and endoscopic treatment strategies, including drainage strategies, optimal timing of endoscopic retrograde cholangiopancreatography and stent removal, and management of complex cases. The review emphasizes individualized treatment approaches based on leak characteristics.

INTRODUCTION

Bile duct leaks (BDLs) most commonly occur as complications following hepatobiliary surgeries such as cholecystectomy, hepatic resection, or liver transplantation[1,2]. Early recognition and prompt management are critical, as delayed management may lead to life-threatening complications such as electrolyte disturbances, peritonitis, intra-abdominal abscesses, and, in severe cases, mortality rates approaching 40%-50%[3]. ERCP is the first-line diagnostic and therapeutic approach for BDLs, with diagnostic sensitivity exceeding 98% and therapeutic success rates ranging from 71.4% to 100%[4-7]. Treatment typically involves endoscopic sphincterotomy (EST), biliary drainage [plastic stents or nasobiliary drainage (NBD)], or a combination of both, aiming to reduce the bile duct-duodenum pressure gradient to promote bile flow and facilitate leak closure[8-10]. However, endoscopic management of BDLs remains challenging due to the lack of consensus on classification, drainage strategies, optimal timing of stents placement and removal, and strategies for complex cases[11,12].

This review aims to provide a comprehensive overview of current classification systems and endoscopic treatment strategies, including drainage strategies, optimal timing of ERCP and stent removal, and management of complex cases.

CLASSIFICATION OF BDLS

Classification of BDLs is critical for assessing injury severity, guiding individualized treatment, and improving clinical outcomes. The Bismuth classification, first proposed in 1982, was originally designed for surgical planning by describing strictures relative to the hepatic duct confluence, but is limited to bile duct injuries with strictures and has limited value for endoscopic management[13,14].

To address these limitations, the Strasberg classification was introduced in 1995, offering a more comprehensive classification that includes a wider range of injuries and is now widely used in clinical practice[15]. Types E1-E5 align with Bismuth types I-V, while type A injuries (cystic or Luschka duct) are the most common (up to 85%) and generally respond well to endoscopic treatment[1] (Figure 1).

Figure 1 Strasberg and Bismuth classification of bile duct injuries. Type A: Bile leak from a cystic duct stump or Luschka’s duct; Type B: Occlusion of the right posterior sectoral duct; Type C: Bile leak from a transected right posterior sectoral duct; Type D: Bile leak from the main bile duct involving < 50% of its circumference; Type E1: Transection of the main bile duct with a stricture located > 2 cm from the hepatic hilum; Type E2: Stricture located < 2 cm from the hilum; Type E3: Hilar stricture with preserved communication between the right and left hepatic ducts; Type E4: Hilar stricture with complete separation of the right and left hepatic ducts; Type E5: Combined stricture involving the main bile duct and the right posterior sectoral duct. Types E1-E5 correspond to Bismuth types I-V.

In 2004, Sandha et al[16] proposed an ERCP-based classification that stratifies BDLs into low-grade (leak visible only after complete intrahepatic opacification) and high-grade (extravasation evident before full opacification). High-grade leaks are associated with poorer outcomes and may require additional drainage or surgery. This system has been validated in several studies and aids in predicting the success of endoscopic treatment[7,17].

The International Study Group of Liver Surgery (ISGLS) classifies BDLs into grades A-C based on clinical severity: Grade A requires no intervention, Grade B necessitates endoscopic treatment, and Grade C requires surgical repair[18]. While the system standardizes the definition of each grade, its clinical utility remains limited, as it has been criticized for being overly simplistic[19,20].

Several other classifications, including the Amsterdam[21], Stewart-Way[22], Hannover[23], McMahon[24], Neuhaus[25], Siewert[26], Csencdes[27] classifications, have been proposed, but none have achieved widespread clinical use. The Strasberg, ISGLS and Sandha classifications remain the most practical for guiding endoscopic management. Table 1 summarizes these classifications and their clinical implications.

Table 1 Bile duct injury classification.

Classification, year	Study objective	Type/grade	Practical implication
Bismuth[13], 1982	To classify post-cholecystectomy biliary strictures for surgical planning and prognosis	Type I: Injury > 2 cm below confluence; Type II: < 2 cm; Type III: Hilar with no confluence; Type IV: Complete separation of right and left ducts; Type V: Aberrant right sectoral duct ± CHD injury	Designed for strictures, not leaks. Offers limited guidance for endoscopic management of BDLs such as cystic duct or Luschka leaks
Strasberg[15], 1995	To classify bile duct injuries after cholecystectomy for treatment planning	Type A: Cystic duct/Luschka leak; B: Occluded right posterior duct; C: Leaking posterior duct; D: Lateral injury (< 50%); E1-E5: Strictures (Bismuth I-V).	Most widely adopted. Types A, C, D may be endoscopically managed; B and E often require surgery. Useful for guiding endoscopic decisions
Amsterdam[61], 1996	To guide endoscopic management of bile duct injuries	Type A: Cystic duct/peripheral leak; B: Major duct leak; C: Stricture; D: Complete transection	Types A and B can be managed endoscopically; C and D generally need surgical repair. Aids decision-making for endoscopic vs surgical management
Stewart-Way[22], 2004	To evaluate the mechanism and impact of RHAI in laparoscopic bile duct injury	Class I: CBD mistaken for cystic duct, recognized pre-transection; Class II: CHD injured by clip/cautery; Class III: CBD transected due to misidentification (most common); Class IV: RHD injured during dissection	Highlights the high RHAI incidence in severe injuries, especially Class III (35%) and IV (64%). Important for surgical planning; limited relevance for endoscopic strategies
Hannover[23], 2007	To guide surgical strategies for bile duct and vascular injuries	Type A: Peripheral leaks (A1: Cystic duct; A2: Gallbladder bed); Type B: Strictures without injury (B1-B2); Type C: Tangential injuries (C1-C4); Type D: Complete transections (D1-D4); Type E: Strictures (E1-E4); Vascular injury suffixes: D, s, p, com, c, pv	Provides comprehensive anatomical and vascular classification, aiding surgical decision-making. Limited utility for endoscopic management
McMahon[24], 1995	To classify bile duct injury severity and guide surgical repair	Minor: Laceration < 25% or cystic-CBD tear; Major: Laceration > 25%, CBD/CHD transection, postoperative stricture	Minor injuries often amenable to T-tube or suture repair; major injuries typically require hepaticojejunostomy. Endoscopy plays a limited role
Neuhaus[25], 2000	To guide surgical/endoscopic management of post-cholecystectomy injuries	Type A (Peripheral leaks): A1, cystic duct leak; A2, gallbladder bed leak. Type B (Occlusion): B1, incomplete (e.g., clip); B2, complete. Type C (Lateral CBD injury): C1, lesion < 5 mm; C2, > 5 mm. Type D (Transection): D1, without tissue defect; D2, with tissue defect. Type E (Stricture): E1, < 5 mm; E2, > 5 mm; E3, confluence; E4, right hepatic/segmental duct	Type A: Sphincterotomy ± stent; percutaneous drainage if needed. Type B1: Endoscopic dilation + stent; B2: Surgery (clip removal) + long-term stenting. Type C1: Sphincterotomy or stent; C2: Surgery + stent ≥ 12 months. Type D: Surgical reconstruction (e.g., hepaticojejunostomy). Type E1: Stenting ≥ 12 months; E2-E4: Surgical resection + hepaticojejunostomy; extended hepatectomy if ischemic cholangiopathy
Siewert[26], 1994	To stratify bile duct injuries for surgical planning	Type I: Immediate biliary fistula; Type II: Late stricture; Type IIIa/IIIb: Tangential lesion ± vascular injury; Type IVa/IVb: Duct disruption ± vascular injury	Type I may be endoscopically treated; Types II-IV typically require surgical reconstruction, especially with vascular involvement
Csencdes[27], 2001	To guide surgical/endoscopic treatment of bile duct injuries	Type I: Small tear of hepatic duct or right hepatic branch. Type II: Injury at cysticocholedochal junction (e.g., from traction, catheter, electrocautery, or close transection). Type III: Partial or complete CBD section. Type IV: Resection of > 10 mm of CBD	Type I/II: May be managed with endoscopic stenting; Type III/IV: Typically require surgical repair (e.g., hepaticojejunostomy)

CBD: Common bile duct; CHD: Common hepatic duct; RHD: Right hepatic duct; RHAI: Right hepatic artery injury; Com: Common hepatic artery involvement; c: Coeliac trunk injury; d: Right hepatic artery injury; p: Portal vein injury; pv: Portal vein branch injury; s: Left hepatic artery injury.

ENDOSCOPIC DRAINAGE STRATEGIES

ERCP is the first-line treatment for most BDLs. However, several aspects of endoscopic management remain controversial, including the choice between sphincterotomy, stent placement, or both; the optimal stent diameter and type; whether the stent should bridge the leak; and the predictors of ERCP outcomes. This section reviews current strategies and controversies in endoscopic management (summarized in Tables 2 and 3).

Table 2 Comparison of treatment strategies.

Ref.	Study design (n)	Key findings	Conclusion	Quality1
EST vs stents
Dolay et al[62], 2010	Single-center, prospective (27)	Stenting led to faster leak resolution than EST (4.5 ± 2.0 days vs 6.5 ± 3.4 days); 2 failures in EST group	Stenting > EST	High
Abbas et al[33], 2019	Multicenter, retrospective (1028)	Stent alone (96%) and stent + EST (97%) had higher success than EST alone (89%)	Stenting ± EST > EST alone	Medium
Rainio et al[32], 2018	Single-center, retrospective (71)	Comparable leak closure time and healing rates between EST and EST + stent	EST and EST + stent offer similar efficacy in type A leaks	Medium
Kaffes et al[63], 2005	Single-center, retrospective (100)	Success: Stent and EST + stent (100%) vs EST alone (78%, P = 0.001)	Stenting ± EST > EST alone	Low
Haidar et al[29], 2020	Single-center, retrospective (100)	Stenting (± EST) had higher success than EST (95.3% vs 72.7%, P < 0.05)	Stenting ± EST > EST alone	Low
Sachdev et al[12], 2012	Single-center, retrospective (65)	EST + stent in 52 (80%), stent alone in 6 (9%), EST alone in 5 (8%), NBD in 2 (3%); all achieved clinical success	EST + stent may be optimal	Low
Sendino et al[64], 2018	Dual-center, retrospective (65)	Stent ± EST (n = 47) led to higher resolution than EST alone (94% vs 58%, P < 0.01). Fewer required percutaneous (4% vs 12%) or surgical intervention (6% vs 42%, P < 0.001)	Stenting ± EST > EST alone	Medium
Chandra et al[31], 2019	Single-center, retrospective (58)	Comparable initial success between EST and EST + stent (92% vs 90%, P = 0.85); stent group had slower resolution (P = 0.02) and more reinterventions (P < 0.01)	EST with or without stent is effective; stenting may delay resolution	Low
Flumignan et al[65], 2021	Single-center, retrospective (31)	EST + stent (n = 22) and EST alone (n = 9) both reduced drainage volume and time to cessation; 2 failures (group unspecified)	No significant difference between EST and EST + stent	Low
NBD vs stent
Raza et al[38], 2019	Systematic review	Stent efficacy: 82.4%; NBD efficacy: 87.2%	Comparable efficacy between NBD and stent	-

¹Quality was assessed using the Newcastle-Ottawa Scale[66] for non-randomized studies: High quality (score ≥ 7), medium quality (score 5-6), low quality (score ≤ 4). Randomized controlled trial was evaluated with Cochrane risk of bias[67].

EST: Endoscopic sphincterotomy; NBD: Nasobiliary drainage; FCSEMS: Fully covered self-expandable metal stent; MPS: Multiple plastic stents; ERCP: Endoscopic retrograde cholangiopancreatography.

Table 3 Factors of endoscopic success in bile duct leaks.

Ref.	Study design (n)	Key findings	Conclusion	Quality1
Leak-Bridging vs short stent
Obata et al[37], 2025	Single-center, retrospective (122)	Bridging stents (P < 0.001), percutaneous drainage (P = 0.0025), and leak severity (P = 0.015) were independent predictors of endoscopic success	Bridging stents across the leak are key to clinical success	Medium
Schaible et al[36], 2017	Single-center, retrospective (35)	Bridging stents achieved 100% success (13/13) vs only 52.6% (10/19) for non-bridging stents	Bridging stents > non-bridging stents	Low
Quintini et al[42], 2024	Dual-center, retrospective (65)	Success rate higher with bridging vs. non-bridging stents (91% vs 53%, P = 0.005)	Bridging stents more effective	Low
Stent diameter
Katsinelos et al[40], 2008	RCT (63)	Success: 93.5% (7 Fr) vs 96.9% (10 Fr)	Stent diameter did not impact outcome	High
Vlaemynck et al[28], 2019	Meta (331)	Success: 95.4% (<10 Fr) vs 97.8% (≥10 Fr).	Stent size did not affect efficacy	High
Predictors
Yabe et al[17], 2017	Single-center, retrospective (58)	Success: 88% (low-grade) vs 59% (high-grade)	High-grade leak predicts failure	Medium
Quintini et al[42], 2024	Dual-center, retrospective (65)	Success: 67% (main duct) vs 90%-100% (others); bridging stents superior (91% vs 53%)	Leak location and bridging predict success	Low
Schaible et al[36], 2017	Single-center, retrospective (35)	Success: 64% (peripheral) vs 92% (central); not significant (P = 0.059); best with bridging at hepatic ducts	Peripheral leaks respond poorly	Low
Tewani et al[1], 2013	Single-center, retrospective (223)	ERCP is more effective for cystic/Luschka leaks (P = 0.028)	Leak site and stenting predict success	Medium
Chen et al[7], 2024	Multicenter, retrospective (106)	Positive: Bridging and cystic duct; Negative: SIRS, high-grade leaks	Location, severity, bridging stent, and systemic status are key predictors	Medium

¹Quality was assessed using the Newcastle-Ottawa Scale[66] for non-randomized studies: High quality (score ≥ 7), medium quality (score 5-6), low quality (score ≤ 4). Randomized controlled trial was evaluated with Cochrane risk of bias[67].

BDLs: Bile duct leaks; ERCP: Endoscopic retrograde cholangiopancreatography; RCT: Randomized controlled trial; SIRS: Systemic inflammatory response syndrome; Fr: French.

Endoscopic sphincterotomy vs stent

Endoscopic management of BDLs typically involves EST, stent placement, or a combination of both[28]. The principle is to reduce the pressure gradient between the bile duct and the duodenum, thereby promoting bile flow into the duodenum and facilitating closure of the leak site[29,30]. However, the optimal strategy remains controversial, particularly regarding whether EST alone is sufficient.

Chandra et al[31] reviewed 58 cases and reported slower resolution (P = 0.02) and higher reintervention rates (P < 0.01) in the stent group. The authors suggested that stents may impair epithelial healing by promoting bacterial colonization and reducing ductal contractility. Rainio et al[32] retrospectively analyzed 71 cases of Amsterdam type A leaks and found no significant difference in clinical success between EST alone and EST plus stent placement (90% vs 90.4%, P > 0.05).

In contrast, a large nationwide study by Abbas et al[33] involving 1028 patients showed better outcomes with stent placement—either alone or combined with EST—compared to EST alone. Similarly, a meta-analysis by Vlaemynck et al[28] demonstrated that combined EST with transpapillary stent placement achieved the highest success rate (98.3%)[34]. Therefore, EST alone may be sufficient for low-grade leaks, but is inadequate in more severe cases.

Leak-bridging vs non-bridging stent

The necessity of bridging the leak during endoscopic biliary drainage remains a subject of debate. Haidar et al[29] emphasized that the primary goal is to reduce the pressure gradient between the bile duct and the duodenum. As the proximal bile duct typically exceeds the stent diameter, bridging the leak may not confer additional benefit. Furthermore, bile is thought to predominantly flow along the stent's outer surface rather than through its lumen[35]. However, a meta-analysis including 11 studies demonstrated that leak-bridging stent was associated with higher clinical success rates[28]. Technically, bridging the leak can be challenging, particularly in cases involving intrahepatic leaks or concurrent biliary strictures. The reported technical success rate of leak-bridging drainage in intrahepatic leaks is only 64%[36].

Therefore, leak-bridging drainage should be prioritized when technically feasible. If bridging is not achievable, non-bridging drainage remains a valid and effective alternative[37].

Stent vs NBD

Both plastic stent drainage and NBD are commonly used during ERCP for the management of BDLs. A systematic review of 34 studies demonstrated comparable success rates in post-liver transplantation BDLs (plastic stents: 82.4% vs NBD: 87.2%)[38]. NBD provides real-time monitoring of bile output and enables cholangiography to confirm BDL healing without the need for repeat ERCP. However, its use is limited by poor patient tolerance, risk of dislodgement, electrolyte loss, and increased nursing demands. Additionally, the average duration of NBD is relatively short (2-11 days; mean 4.7 ± 0.9), compared to biliary stents (14-53 days; mean 29.1 ± 4.4)[39]. Therefore, NBD may be less appropriate for high-output leaks. Plastic stents are preferred for their longer drainage time and better patient tolerance.

Diameter of stents

In a randomized trial of 63 post-cholecystectomy patients, Katsinelos et al[40] found no significant difference in success rates between 7 Fr and 10 Fr stents (93.5% vs 96.9%, P > 0.05). Similarly, a meta-analysis reported comparable outcomes for stents ≥ 10 Fr and < 10 Fr (97.8% vs 95.4%)[28]. Plastic stent diameter may not significantly impact outcomes, unless it exceeds the bile duct lumen and directly seals the leak. However, to minimize the risk of stent occlusion, 10 Fr stents are generally preferred[41].

Predictors of endoscopic outcomes

The location of BDLs is an important factor influencing endoscopic treatment success. Leaks from the cystic duct stump or ducts of Luschka after cholecystectomy are generally associated with high success rates[1,7]. In contrast, a retrospective study reported that leaks from the main bile duct (common hepatic or common bile duct) had a lower success rate compared to those from other sites (67% vs 90%-100%)[42]. This reduced efficacy is likely due to higher bile flow in the main ducts, which may hinder leak closure.

Intrahepatic leaks have also been associated with lower success rates, possibly due to their deeper anatomical location and the technical difficulty of bridging the leak with a stent[36].

Leak severity is another important predictor of endoscopic failure. High-grade leaks, defined as contrast extravasation before complete intrahepatic opacification, are associated with poorer outcomes[16]. In addition, systemic inflammatory response syndrome indicates more severe BDLs and systemic inflammation, and is associated with endoscopic failure[7].

TIMING OF ERCP AND STENT REMOVAL

Optimal timing of ERCP and stent removal is essential for effective management of BDLs. Urgent ERCP is not necessary in all cases, and optimal stent retention time varies by stent type and leak severity (Supplementary Table 1).

Optimal timing of ERCP

Several studies have shown that the timing of ERCP, whether performed within 1 day, 2-3 days, or after 3 days, does not significantly affect clinical outcomes[33,43,44]. In a retrospective study of 593 hepatectomies, 34 cases of bile leaks were reported, with a 76.5% success rate for conservative management[45]. These findings suggest that urgent ERCP may not be necessary in all cases and could impose unnecessary procedural burden. When abdominal drainage is in place, ERCP can be performed electively[46,47].

According to the ISGLS definition, most Grade A BDLs resolve within one week with percutaneous drainage. Persistent output beyond one week or signs of infection are classified as Grade B and should be managed with endoscopic intervention[18]. Viganò et al[45] identified a drain output > 100 mL on postoperative day 10 as an independent predictor of conservative treatment failure. According to the 2020 World Society of Emergency Surgery guidelines[48], ERCP is recommended for minor leaks if symptoms persist or worsen under conservative treatment. For major leaks identified between 72 hours and 3 weeks postoperatively, ongoing inflammation may hinder surgical repair, and ERCP can be considered during this period. Chen et al[49] reported that performing ERCP within 3 weeks was associated with a higher clinical success rate and a lower incidence of biliary stricture.

Therefore, in patients with abdominal drainage in place, conservative management may be appropriate during the first 1-3 weeks. If the bile output does not decrease or symptoms persist, ERCP should be performed within this period to prevent stricture formation.

Stent removal

The duration of stent retention should be individualized based on stent type and leak severity. For plastic stents, a retention period of 4-8 weeks is generally recommended to prevent stent occlusion[5,34,40]. A comparative study reported a mean healing time of 29.1 ± 4.4 days (range, 14-53 days) with plastic stents vs 4.7 ± 0.9 days (range, 2-11 days) with NBD[39]. For fully covered self-expandable metal stents (FCSEMS), optimal stent retention time remains unclear. In one study of 17 patients with refractory postcholecystectomy leaks after failed sphincterotomy and plastic stenting, the median stent duration was 16 days (range, 7-28 days)[6]. Another study reported a median leak closure time of 3.5 days[5]. Overall, FCSEMS should be retained within 3-4 weeks to minimize risks of strictures, pancreatitis, and stent migration[6,41,50,51].

MANAGEMENT OF COMPLEX CASES

Management of complex BDLs, including refractory leaks and bilomas, often requires additional endoscopic or surgical interventions. This section summarizes strategies for cases that fail to respond to standard treatment (Supplementary Table 2).

Refractory leaks

Endoscopic sphincterotomy with placement of a plastic stent is effective in most BDLs, but may be insufficient for high-grade or high-output leaks. Multiple plastic stents (MPS) provide a cost-effective salvage option, while FCSEMS, directly seal the leak, have demonstrated higher efficacy in refractory cases[5,6,41,50,52,53].

In a multicenter study of 178 patients who underwent sphincterotomy plus a single 10 Fr stent, 16 (9%) experienced initial treatment failure and were thus classified as refractory leaks. Among these, 62.5% responded to MPS, while the remainder required FCSEMS[5]. Predictors of MPS failure included the use of fewer than three stents, total diameter < 20 Fr, and high-grade leaks[5]. In a prospective study of 40 patients with refractory postcholecystectomy leaks, FCSEMS achieved superior efficacy compared to MPS (100% vs 65%, P = 0.004)[54].

Despite its effectiveness, FCSEMS is associated with a risk of biliary stricture, particularly in liver transplant cases, and is currently not recommended in this setting[55]. In one case series, 13 patients achieved leak resolution with FCSEMS, with 2 cases developing strictures[51]. In addition, FCSEMS placement across the sphincter of Oddi raises concerns about pancreatitis and stent migration. The use of a dumbbell-shaped FCSEMS may reduce these risks[54]. Persistent leaks unresponsive to endoscopic treatment may ultimately require surgical intervention, such as hepaticojejunostomy[56,57].

Treatment strategies for biloma

Bilomas are typically secondary to BDLs. While small or asymptomatic collections may resolve spontaneously with conservative management, symptomatic or large bilomas (> 5 cm) usually require drainage. Compared with percutaneous drainage, EUS-guided transmural drainage (EUS-TD) offers physiological internal drainage, fewer catheter-related infections, and better access to deep locations such as the caudate lobe.

Both EUS-TD and transpapillary/transfistulary (TP/TF) drainage are effective minimally invasive options[58]. In a retrospective cohort of 30 patients with complex bilomas, clinical success rates were 75% for EUS-TD and 67% for TP/TF drainage[59]. Another study comparing ERCP and EUS-TD in 47 patients reported similar clinical success (87% vs 84%) and technical success (94% vs 100%), while EUS-TD demonstrated shorter procedure time (16.9 minutes vs 26.6 minutes, P = 0.009) and hospital stay (22 days vs 46 days, P = 0.038)[60].

CONCLUSION

ERCP is the first-line approach for the diagnosis and management of BDLs. This review summarizes current classifications, endoscopic strategies, optimal timing of ERCP and stent removal, and the management of complex cases. Figure 2 summarizes the proposed treatment algorithm for BDL based on current evidence. However, most current evidence is based on retrospective studies and case series. Well-designed prospective studies are needed to establish standardized, evidence-based management protocols for BDLs.

Figure 2 Suggested treatment algorithm for bile duct leak. BDL: Bile duct leak; ERCP: Endoscopic retrograde cholangiopancreatography; EST: Endoscopic sphincterotomy; MPS: Multiple plastic stents; FCSEMS: Fully covered self-expandable metal stents.

ACKNOWLEDGEMENTS

We would like to thank Guan-Jun Zhang, Ya-Wen Liang, Yu-Ming Han for their valuable contributions to literature search. Their assistance was essential to the successful completion of this study.