Fully covered metal biliary stents confer a higher risk of post-endoscopic retrograde cholangiopancreatography pancreatitis

Abstract

BACKGROUND

Biliary fully-covered self-expanding metal stents (FCSEMS) are increasingly used over plastic or uncovered SEMS due to their long-term patency and removability. However, there is concern that transpapillary placement may lead to post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP).

AIM

To assess the rates of PEP with and without the use of FCSEMS for both benign and malignant indications.

METHODS

We performed a multicenter retrospective cohort study involving three Australian tertiary referral centers. Consecutive adults who underwent ERCP for biliary indications between October 2016 and October 2019 were included. The primary endpoint was the rate of PEP. Secondary endpoints included severity of pancreatitis, other procedure- and stent-related adverse events occurring within 90 days.

RESULTS

A total of 3401 ERCPs were performed (54.2% female, mean age 62.9±18.6 years) with an overall PEP rate of 3.15%. On propensity-score matched analysis, FCSEMS was an independent predictor of PEP (odds ratio = 5.49, 95% confidence interval: 2.10-6.99; P = 0.001). FCSEMS had a higher rate of PEP (7.8%) compared with plastic stents (3.4%; P = 0.0015), and patients who did not receive any stents (2.4%; P = 0.001), but was non-significant when compared with uncovered self-expanding metal stents (3.9%; P = 0.12). The rate of PEP following FCSEMS decreased to 6.0% for malignant indications, and further to 3.9% for biliary obstruction due to pancreatic cancer, but did not reach statistical significance.

CONCLUSION

Biliary FCSEMS are associated with a higher risk of PEP compared to no stents or plastic stents, particularly for benign indications.

Core Tip: Biliary fully covered self-expanding metal stents appear to be associated with a higher risk of post-endoscopic retrograde cholangiopancreatography pancreatitis compared with no stents or plastic stents, regardless of the papilla status (i.e. native vs not native). This is particularly the case when fully covered self-expanding metal stents are compared to plastic stents for benign biliary obstruction. Further prospective studies are needed to validate this finding and guide optimal stent selection for patients.

INTRODUCTION

Endoscopic retrograde cholangiopancreatography (ERCP) is a powerful tool for the management of malignant and benign biliary disease. Post-ERCP pancreatitis (PEP) is the most frequent adverse event arising from ERCP, affecting between 3.5% and 9.7% of all patients undergoing the procedure[1,2]. Reported risk factors for PEP include patient-related factors such as female sex, young age, previous acute pancreatitis, or the absence of chronic pancreatitis; and procedure-related factors such as the duration of cannulation attempts and the passage of wire or contrast into the pancreatic duct[3]. Even with the use of rectal anti-inflammatory therapy and prophylactic pancreatic stenting, PEP remains a common and potentially severe complication. Biliary fully covered self-expanding metal stents (FCSEMS) have become increasingly used for a number of indications, including biliary strictures, bile leaks, and post-sphincterotomy bleeding. Many of these were previously treated using plastic or uncovered SEMS (UCSEMS); however, FCSEMS have been increasingly used due to their long-term patency and removability. On the other hand, there is concern that transpapillary deployment of FCSEMS may lead to PEP due to compression of the pancreatic duct orifice and restriction to the flow of pancreatic juice at the papilla[4,5]. Available literature on this is conflicting, and the risk factors for PEP following FCSEMS placement have not been well elucidated[6-10].

In this retrospective cohort study, we assessed the rates of PEP with and without the use of FCSEMS for both benign and malignant indications, and identified any risk factors that may predict PEP after FCSEMS deployment.

MATERIALS AND METHODS

Study design and patient selection

This was a retrospective cohort study involving three tertiary referral centers in Australia. Consecutive adult patients who underwent ERCP for biliary indications between October 2016 and October 2019 were included in the study. Primary pancreatic indications (such as pancreatic duct strictures or stones) and procedures aborted prior to attempted cannulation were excluded. All ERCPs were performed by experienced interventional endoscopists or advanced fellows under their immediate supervision. Wire-guided biliary cannulation was performed for all intact papillae. Procedural decisions, including endoscopic biliary sphincterotomy, stent choice and use of PEP prophylaxis, were at the discretion of the treating endoscopist. Plastic stents were either 7-French or 10-French caliber with the number placed and the length of the stent decided based on size, etiology and/or location of the stricture. Metal stents were generally 10 mm in diameter and either FCSEMS, UCSEMS, or partially covered SEMS (PCSEMS). The type and length of the metal stent were at the discretion of the treating endoscopist. The study received institutional review board approval (No. ETH14056). All patients provided written informed consent per institutional policy for the procedures.

Data collection and outcomes

Clinical characteristics, procedural information, and any post-procedural adverse events (AEs) were collected from electronic chart review at each center. The primary endpoint was the rate of PEP, according to the Cotton criteria, i.e. new or worsening abdominal pain with an elevated serum lipase or amylase level greater than 3 times the upper limit of normal with or without radiographic evidence of acute pancreatitis[11]. Secondary endpoints included severity of pancreatitis, per Cotton criteria[11], other procedure-related AEs, bleeding, perforation, cholangitis, severity per American Society of Gastrointestinal Endoscopy lexicon[12], and stent-related AEs occurring within 90 days (stent migration or stent occlusion).

Statistical analyses

Continuous variables are presented as the mean and standard deviation. Categorical variables are presented as a number (percentage). To account for baseline differences between patients receiving FCSEMS and those receiving other or no stents, a 1:1 nearest-neighbor propensity score-matched analysis was performed. Matching variables included age, sex, indication (benign vs malignant), papilla status, and use of PEP prophylaxis (rectal indomethacin and pancreatic duct stenting). Logistic regression models were used to identify independent predictors of PEP, adjusting for age, sex, papilla status, prior pancreatitis, sphincterotomy techniques and prophylactic measures. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported, and statistical significance was defined as P < 0.05. Data analyses were performed using Stata version 17 (StataCorp, College Station, TX, United States).

RESULTS

Baseline characteristics

A total of 3401 ERCPs were performed for predominantly benign indications (85.2%). The mean age was 63.5 ± 18.1 years, and patients were mostly female (n = 1843, 54.2%) with a mean American Society of Anesthesiology score of 2.59 ± 0.7. Overall, 2010 cases (59.1%) had an intact papilla, and endoscopic biliary sphincterotomy was performed in the majority of these (n = 1792, 89.2%). In patients with an intact papilla, rectal indomethacin was given in 726 patients (36.1%), and a prophylactic pancreatic duct stent was placed in 287 patients (14.3%). Clinical characteristics were similar among the various stent groups. Although UCSEMS patients were older, all but 1 had a malignant indication, and only 2 patients had prophylactic pancreatic duct stent placement (Table 1). A total of 319 FCSEMS were placed for either benign (n = 186, 58%) or malignant indications (n = 133, 42%). Within the malignant indications, most (n = 77, 57.9%) were for distal biliary obstruction. Plastic stents were used in 984 cases for predominantly benign (n = 835, 85%) indications. UCSEMS were used in 154 cases, almost entirely for malignant indications (n = 153, 99.4%). Of these, 67 cases (43.5%) were for distal biliary obstruction. Four PCSEMS were placed; given the small numbers, these patients were excluded from further analyses.

Table 1.

Baseline characteristics by type of stent, n (%)

	No stent (n = 1979)	Plastic (n = 984)	FCSEMS (n = 319)	UCSEMS (n = 154)
Age, mean ± SD, years	63.5 ± 18.1	63.9 ± 17.2	66.6 ± 15.8	70.7 ± 12.1
Female	1127 (57)	506 (51)	148 (46)	62 (40)
Indication
Benign	1888 (95)	835 (85)	186 (58)	1 (1)
Malignant	91 (5)	149 (15)	133 (42)	153 (99)
Intact papilla	1123 (57)	613 (62)	193 (61)	93 (60)
Sphincterotomy
Precut sphincterotomy	70 (4)	47 (5)	11 (3)	6 (4)
Biliary sphincterotomy	1040 (53)	593 (60)	190 (60)	85 (55)
Sphincteroplasty	114 (6)	81 (8)	20 (6)	5 (3)
PEP prophylaxis
Rectal indomethacin	514 (26.0)	270 (27.4)	77 (24.1)	30 (19.5)
PD stent	152 (7.7)	132 (13.4)	34 (10.7)	2 (1.3)

FCSEMS: Fully covered self-expanding metal stents; PD: Pancreatic duct stent; PEP: Post-endoscopic retrograde cholangiopancreatography pancreatitis; UCSEMS: Uncovered self-expanding metal stents; SD: Standard deviation.

Outcomes

The overall rate of PEP was 3.15%, and the rate of PEP in intact papillae (n = 2008, 59.1%) was 4.23% (vs 1.58% in those with a non-intact papilla) (Table 2). The majority of PEP were classified as mild to moderate severity (87.9%). Univariate analyses revealed that age < 50 years, an intact papilla, performance of a sphincterotomy or sphinteroplasty, use of a pancreatic duct stent or rectal indomethacin, or placement of FCSEMS were significant predictive factors for pancreatitis (Table 3). The remaining variables, including younger age (< 50 years), sex, benign vs malignant indication, and history of chronic pancreatitis or pancreatic cancer, were not significant predictors of pancreatitis. In a propensity score matched cohort (n = 638), the incidence of PEP was significantly higher in the FCSEMS group compared to matched controls (7.8% vs 2.2%) (Tables 4 and 5). FCSEMS placement and sphincteroplasty were independently associated with increased odds of PEP (OR = 5.49, 95%CI: 2.10-6.99; P = 0.001, and OR = 4.20, 95%CI: 1.38-11.44; P = 0.007, respectively) (Table 3). No other covariates were significantly associated with PEP, including female sex, intact papilla or prophylactic interventions such as rectal indomethacin and pancreatic duct stenting.

Table 2.

Adverse events, n (%)

	Overall (n = 3397)	No stent (n = 1979)	Plastic (n = 984)	FCSEMS (n = 319)	UCSEMS (n = 154)
PEP	107 (3.1)	47 (2.4)	33 (3.4)	25 (7.8)	6 (3.9)
Mild	57	30	14	11	3
Moderate	37	12	13	11	2
Severe	13	5	6	3	1
Stent migration	57 (1.7)	NA	29 (2.9)	27 (8.5)	1 (0.1)
Proximal	17		8	8	1
Distal	40		21	19	0
Cholangitis	66 (1.9)	26 (1.3)	18 (1.8)	17 (5.3)	9 (5.8)
Perforation	9 (0.3)	5 (0.3)	2 (0.2)	2 (0.6)	0 (0)
Death	4 (0.1)	2 (0.1)	2 (0.2)	0 (0)	0 (0)

FCSEMS: Fully covered self-expanding metal stents; NA: Not available; PEP: Post-endoscopic retrograde cholangiopancreatography pancreatitis; UCSEMS: Uncovered self-expanding metal stents.

Table 3.

Predictors of post-endoscopic retrograde cholangiopancreatography pancreatitis in all patients

Variable	Odds ratio	95% confidence interval	P value
Age < 50 years	0.43	0.07-1.54	0.263
Female sex	0.89	0.43-1.86	0.757
Intact papilla	1.81	0.86-3.84	0.120
Precut sphincterotomy	2.73	0.57-9.87	0.155
Biliary sphincterotomy	1.30	0.41-4.19	0.661
Sphincteroplasty1	4.20	1.38-11.44	0.007
Rectal indomethacin	1.34	0.55-3.07	0.500
Pancreatic duct stent	1.40	0.38-4.05	0.566
FCSEMS placement1	5.49	2.10-16.99	0.001
UCSEMS placement	3.19	0.63-13.00	0.12
Plastic stent placement	1.45	0.45-3.99	0.49

¹Significant predictor of post-endoscopic retrograde cholangiopancreatography pancreatitis on multivariate analysis.

FCSEMS: Fully covered self-expanding metal stents; UCSEMS: Uncovered self-expanding metal stents.

Table 4.

Comparison of baseline characteristics after propensity score matching, n (%)

	Unmatched cohort			Matched cohort
	FCSEMS (n = 319)	Non-FCSEMS (n = 3082)	SMD	FCSEMS (n = 319)	Non-FCSEMS (n = 319)	SMD
Age, mean ± SD, years	66.6 ± 15.8	63.2 ± 18.3	0.198	66.6 ± 15.8	66.9 ± 15.2	0.021
Female	148 (46)	1682 (55)	0.165	148 (46)	150 (47)	0.031
Benign indication	186 (58)	1800 (58)	0.708	186 (58)	203 (64)	0.109
Biliary sphincterotomy	190 (60)	1710 (56)	0.087	190 (60)	171 (54)	0.120
Precut sphincterotomy	11 (4)	222 (7)	0.027	11 (4)	23 (7)	0.168
Sphincteroplasty	20 (6)	155 (5)	0.004	20 (6)	16 (5)	0.054
Rectal indomethacin	77 (24)	820 (27)	0.047	77 (24)	85 (27)	0.058
PD stent	34 (11)	0 (0)	0.050	34 (11)	0 (0)	0.050

FCSEMS: Fully covered self-expanding metal stents; PD: Pancreatic duct stent; SD: Standard deviation; SMD: Standardized mean difference.

Table 5.

Predictors of post-endoscopic retrograde cholangiopancreatography pancreatitis following fully covered self-expanding metal stent placement (n = 319)

Variable	Odds ratio	95% confidence interval	P value
Age < 50 years	0.49	0.07-1.92	0.369
Female sex	0.67	0.27-1.61	0.368
Benign indication	1.30	0.50-3.52	0.593
Intact papilla	0.35	0.09-1.43	0.143
Precut sphincterotomy1	6.48	1.17-30.32	0.021
Biliary sphincterotomy	1.19	0.31-4.41	0.795
Sphincteroplasty1	3.70	0.99-12.59	0.041
Rectal indomethacin	1.07	0.34-2.93	0.904
Pancreatic duct stent	2.33	0.58-7.71	0.190

¹Significant predictor of post-endoscopic retrograde cholangiopancreatography pancreatitis on multivariate analysis.

In the FCSEMS cohort (n = 319), precut sphincterotomy and sphincteroplasty were independently associated with increased risk of PEP (OR = 6.48, 95%CI: 1.17-30.32; P = 0.021; and OR = 3.70, 95%CI: 0.99-12.59; P = 0.041, respectively) (Table 5). No other variables, including age, sex, papilla status, prior pancreatitis, or prophylactic measures, were significantly associated with PEP in this cohort. FCSEMS had a significantly higher rate of PEP at 7.8% compared with plastic stents (3.4%; P = 0.0015), and patients who did not receive any stents (2.4%; P = 0.001) (Figure 1A and B). There was a non-significant trend to higher rates of PEP compared with UCSEMS (3.9%; P = 0.12) (Figure 1C). Of note, the rate of PEP following FCSEMS decreased to 6.0% when limited to malignant indications, and further to 3.9% if limited to biliary obstruction due to pancreatic cancer (Figure 1D). FCSEMS placement was associated with a significantly higher risk of stent migration compared to plastic (8.5% vs 2.9%; P = 0.017) and UCSEMS (8.5% vs 0.1%; P = 0.0003). FCSEMS was associated with a higher risk of post-ERCP cholangitis (5.3%) compared with plastic (1.8%, P = 0.001) but not UCSEMS (5.8%; P = 0.818). There was no significant difference between stent types with regard to post-ERCP bleeding or perforation. There were no documented episodes of cholecystitis in our cohort. When stratified by benign indications, a total of 186 FCSEMS and 835 plastic stents were placed. There was a higher rate of PEP with FCSEMS placement compared with plastic stents (9.2% vs 3.4%; P = 0.001). In the setting of malignant biliary obstruction, 133 FCSEMS and 153 UCSEMS were placed. FCSEMS had a numerically higher rate of PEP when used for malignant indications compared with UCSEMS, but this did not reach statistical significance (6.0% vs 3.9%; P = 0.427).

Figure 1.

Rate of post-endoscopic retrograde cholangiopancreatography. A: Rate of post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis by biliary stent type; B: Rate of post-ERCP pancreatitis fully covered metal stents vs plastic stents; C: Rate of post-ERCP pancreatitis following fully covered vs uncovered metal stents; D: Rate of post-ERCP pancreatitis following fully covered metal stent according to indication. FCSEMS: Fully covered self-expanding metal stents; UCSEMS: Uncovered self-expanding metal stents.

DISCUSSION

ERCP is an effective first-line therapy for a wide variety of biliopancreatic pathologies. PEP is one of the most common AEs, which can occur in up to 9.7% of cases[1]. Although mild to moderate in the vast majority of cases, PEP can be severe and carries an overall mortality rate of 0.7%[1]. Preventative strategies used to reduce the risk include guidewire cannulation[13], periprocedural rectal non-steroidal anti-inflammatory drugs (NSAIDs)[14], aggressive hydration with lactated Ringer’s solution[15,16], and pancreatic duct stent placement[17,18]. FCSEMS are being increasingly used for indications including benign or malignant biliary strictures, bile leak, and post-sphincterotomy bleeding. There is a concern that transpapillary FCSEMS placement may lead to higher rates of PEP[4,5,19]. The proposed mechanisms for the development of pancreatitis include compression of the pancreatic orifice by radial expansion of the self-expanding metal stent against the opening of the pancreatic duct and obstruction of pancreatic duct drainage from the membrane covering of the FCSEMS. Existing literature regarding the impact of FCSEMS placement on PEP is conflicting, largely limited to malignant indications only[20-22] and generally powered to evaluate other outcomes such as stent patency. Multiple prospective studies, including several randomized controlled trials, comparing FCSEMS to UCSEMS for management of malignant distal biliary obstruction have found no difference in rates of PEP[8,9,23,24]. The largest by Kullman et al[8]with 400 patients, reported rates of PEP of 1.5% and 2.0% (P-value non-significant) following placement of FCSEMS and UCSEMS, respectively. On the other hand, a larger multicenter retrospective trial involving 315 patients with distal malignant biliary obstruction reported significantly higher rates of PEP following FCSEMS when compared with UCSEMS (7.8% vs 1.0%; P = 0.04) but not plastic stents (2.6%; P = 0.13)[25]. This is in contrast to several other studies suggesting that metal stents may be associated with higher rates of PEP[21,26]. A randomized controlled trial in 54 patients with malignant distal biliary obstruction by Gardner et al[21] reported similar rates of PEP following FCSEMS and UCSEMS placement (19% and 18%, respectively), both numerically higher but not statistically significant when compared with plastic stent placement (0%). Similarly, another multicenter randomized study of 70 patients with resectable pancreatic cancer reported a numerically higher rate of PEP following FCSEMS compared with a plastic stent, which did not meet statistical significance (12.5% vs 2.9%; P = 0.19)[27]. Both studies were designed to evaluate stent patency and were not powered to detect differences in PEP[27].

It should be noted that while FCSEMS are often placed for benign and malignant indications (42% benign in our study), UCSEMS are generally placed for malignant indications (99% in our study), resulting in an overall different patient population. In our study, we found a higher rate of PEP with the use of FCSEMS (7.8%) compared to either UCSEMS (3.9%) or plastic stents (3.4%). Notably, although not statistically significant, the rate of PEP following FCSEMS fell to 6.0% if limited to patients with malignant obstruction, and further to 3.9% in patients with pancreatic cancer-related obstruction. This is in keeping with established evidence that performing ERCP in patients with a normal caliber main pancreatic duct or pancreatic ductal obstruction to be at higher risk for PEP[4,6,28] compared to those with pancreatic ductal obstruction by tumour[29].

Our study, analyzing outcomes of over 3000 ERCPs and 300 FCSEMS placements, found the placement of FCSEMS to be an independent predictor for PEP in patients undergoing biliary ERCP. Higher rates of PEP compared to patients who received no stents or plastic stents were seen, which remained numerically higher but not statistically significant when compared with UCSEMS. A notable difference in our study is the inclusion of all biliary indications. Existing comparative literature on the use of FCSEMS for both benign and malignant indications is limited. A recent retrospective study evaluating post-FCSEMS pancreatitis in a similar cohort of 602 patients with benign and malignant biliary disease reported a PEP rate of 9.3%, but did not include comparison to other stent types[28]. Another retrospective cohort study compared 1136 patients with benign and malignant biliary obstruction receiving FCSEMS, UCSEMS or plastic stent placement[7]. Any SEMS placement was a significant risk factor for PEP (OR = 3.67, CI: 1.50-8.97); there was no significant difference when comparing FCSEMS with UCSEMS (9.3% vs 6.1%, P-value not reported).

Established PEP prophylaxis measures, including rectal indomethacin and prophylactic pancreatic duct stenting, were associated with higher rates of PEP in our study, although this did not impact the rates of PEP in FCSEMS. This is likely a reflection of the real-world nature of our data. Prophylactic pancreatic duct stents were only placed for difficult cannulation when wire access into the pancreatic duct had already been established. These patients are therefore at higher risk for PEP. The study period was also during the transition from selective use of rectal NSAIDs to routine use for intact papillae. As such, rectal NSAID therapy was predominantly given for difficult cannulations that are again inherently at higher risk for PEP. A prior study found pancreatic duct stenting and rectal indomethacin administration to be protective against PEP following FCSEMS placement, with risk rising from 9.3% to 18.6% if no prophylactic approaches were adopted[28]. In this study 25.9% of patients received a pancreatic duct stent in comparison to 9.3% of our patients, suggesting differences in practice pattern and/or patient population. A recent study found that routine placement of a pancreatic stent in patients with malignant distal biliary obstruction receiving either FCSEMS or UCSEMS reduced the rate of PEP from 31% to 5%, despite higher rates of pancreatic contrast injection in the stented group[30]. This study was limited by a small sample size (n = 33), being retrospective in nature and using historical controls.

Another potential protective strategy against PEP is an endoscopic biliary sphincterotomy prior to biliary stenting. It has been suggested that an endoscopic biliary sphincterotomy may reduce PEP by separating the pancreatic and biliary orifices and thereby reducing compressive force on the pancreatic orifice. However, two prior randomized controlled trials failed to show a benefit in the prevention of pancreatitis prior to placement of PCSEMS or FCSEMS[31,32]. In our study, biliary sphincterotomy was not shown to be a significant predictor of PEP on multivariate analysis. Almost all patients undergoing biliary sphincterotomy had intact papillae, and all endoscopists routinely performed sphincterotomy prior to stenting unless there was a major contraindication such as uncontrolled coagulopathy. ESGE guidelines suggest against routine endoscopic biliary sphincterotomy before the placement of plastic, uncovered, or partially covered SEMS[33] due to increased bleeding risk; there was insufficient evidence to comment on the impact of this strategy on PEP risk. A specific recommendation regarding sphincterotomy in the setting of FCSEMS placement was not made, acknowledging one small study demonstrating PEP incidence of 50% after biliary FCSEMS without endoscopic sphincterotomy in patients with post-liver transplantation biliary strictures[34].

The strengths of this study include the multicenter design, large procedure numbers and inclusion of consecutive ERCPs performed for biliary indications in a real-world cohort. The study was able to assess patients undergoing all types of biliary stenting to compare between stent types. However, there were several limitations to our study. First, the retrospective design of the study introduced inherent biases associated with retrospective analyses. The retrospective nature of the study also meant that decisions around sphincterotomy, PEP prophylaxis measures (pancreatic stents and topical indomethacin), undertaking biliary sphincterotomy, and choice of stent type were done at the discretion of the endoscopist rather than being controlled in a study protocol. In particular, the choice of FCSEMS or UCSEMS in the setting of malignant biliary obstruction was not controlled. In addition, variables which were not accurately captured on reporting databases (including intra- or extra-papillary stent position and total number of stents) could not able to be captured in this study. These are important variables which should be addressed in future studies. Second, there may have been some underestimation of AE rates as patient data were only available from the electronic medical records at the participating facilities and may not have captured patients who presented to other facilities outside of our center’s catchment. Third, the study centers were high-volume tertiary referral centers with highly experienced endoscopists, and as such, these may not be generalizable across all institutions. Finally, the number of patients in each stent group differed significantly, and the number of PCSEMS was extremely low. This meant no meaningful analysis could be performed on the outcomes of PCSEMS. Further, although propensity-matched analysis was performed to adjust for confounders, the wide CIs seen in the results suggest some uncertainty, likely due to the overall small number of PEP events.

CONCLUSION

This study demonstrated that the use of biliary FCSEMS is associated with a higher risk of PEP compared to patients who received no stents or plastic stents, regardless of the indication or history of previous sphincterotomy. This risk appears to be highest for benign indications. Further prospective studies are needed to validate this finding and guide optimal patient selection to obviate this increased risk.

Data sharing statement

The data that support the findings of this study are not publicly available due to institutional and ethical restrictions related to patient confidentiality.