Original Article - (2019) Volume 20, Issue 1
Department of Gastroenterology, Endoscopy Unit, Hospital das Clinicas of University of São Paulo Medical School, Brazil
Received June 25th, 2018 - Accepted January 20th, 2019
Background and Aims Acute pancreatitis is one common and severe complication after endoscopic retrograde cholangiopancreatography. Randomized controlled trials have reported that pancreatic stenting may efficiently prevent post-ERCP pancreatitis. Methods Systematic review was conducted on MEDLINE/PubMed and other databases randomized controlled trials comparing patients undergoing endoscopic retrograde cholangiopancreatography with pancreatic stent versus endoscopic retrograde cholangiopancreatography without stent. Two independent reviewers assessed the eligibility. Primary outcome is to assess the degree of severity of pancreatitis (mild, moderate, and severe) and secondary hyperamylasemia, cholangitis, abdominal pain, duration of endoscopic retrograde cholangiopancreatography and length of hospital stay. Results Twelve randomized controlled trials selected a total of 3.310 patients. Meta-analysis showed that pancreatic stenting reduced pancreatitis risk to PEP (mild PEP: RD 0.06, 95% CI-0.09 - 0.03; moderate PEP: RD 0.03, 95 % CI-0.06 -0.01; severe PEP: (RD 0.02, 95% CI-0.05-0.01); Hyperaylasemia (RD-0.62, 95% CI-0.65-0.59) showed statistically significant difference. Cholangitis (RD 0.03, 95% CI-0.03-0.09), abdominal pain (RD 0.10, 95% CI-0.21-0.01), length of hospital stay after ERCP (RD 1.55, 95% CI-4.39 -1.29), total duration of the procedure (RD 2.97, 95% CI-0.19-6.12), showed no statistical significance. Conclusions Pancreatic stent placement is effective in reducing post-endoscopic retrograde cholangiopancreatography acute pancreatitis incidence in high-risk patients.
Cholangiopancreatography, Endoscopic Retrograde; Pancreatitis
ERCP endoscopic retrograde
The Endoscopic Retrograde Cholangiopancreatography (ERCP) is an advanced endoscopic procedure for diagnosis and treatment of several biliary and pancreatic disorders [1]. The ERCP is usually safe and carries expedited postoperative recovery. However, related complications such as post-ERCP pancreatitis (PEP), gastrointestinal bleeding/perforation, biliary infection (cholangitis), and even death may occur [2].
PEP is the most common complication following ERCP. The estimated related annual costs exceed 150 million dollars in the United States [1]. The incidence ranges from 2% to 10% and most patients present non-severe and selflimited forms of PEP. However, some patients may present severe pancreatitis and demand aggressive medical interventions, particularly those with associated risk factors [3]. Risk factors for PEP entail young age, female gender, previous history of cholangitis or pancreatitis, prior post-ERCP pancreatitis, recurrent pancreatitis, sphincter of Oddi dysfunction, repeated bile/pancreatic duct cannulations, iatrogenic procedural injury, presence of gallstones, periampullary diverticulum, and insufficient pancreatic drainage. In high-risk patients, the incidence of PEP ranges from 20% to 30% and increase length of stay and medical costs [4, 5].
Pancreatic stenting is a prophylactic intervention to avoid PEP, even though its mechanism of action is not clear [6, 7]. Previous meta-analyses [3] have shown that pancreatic stent placement benefits especially highrisk patients. However, some of those studies included both abstracts and full-text articles, and did not assess related complications, which may have led to inaccurate conclusions. Others, enrolled different populations in the same analysis (high and low risk patients) [3, 4]. Once again, there was no evaluation of complications.
Therefore, we performed this systematic review and meta-analysis to elucidate efficacy and safety of pancreatic stenting to prevent PEP in high-risk patients through a strict and clear methodology.
This systematic review was performed according to the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [8] and was registered in the PROSPERO database (CRD42017056261).
Eligibility criteria
Types of studies: Only RCTs were included. There was no language or publication dates restriction. Abstracts and studies enrolling low-risk patients were excluded [9, 10, 11].
Types of participants: Patients older than 18 years who underwent ERCP.
Types of intervention: Pancreatic stenting (intervention) versus no stent (control).
Types of outcomes: The main outcome was the incidence and severity of PEP; secondary outcomes were: hyperamylasemia rate, presence of abdominal pain, incidence of cholangitis, duration of ERCP and length of stay.
Databases
We searched MEDLINE/PubMed, LILACS, Embase, and Cochrane CENTRAL databases from inception to October 2016; then, a manual search was performed using references of the selected studies and previous systematic reviews regarding PEP in high-risk patients. The search strategy for MEDLINE was: (Endoscopic Retrograde Cholangiopancreatography OR ERCP OR Cholangiopancreatography, Endoscopic Retrograde OR Retrograde Cholangiopancreatography Endoscopic) AND (stent*) AND (pancreatitis OR pancreatic).
Selection of studies
Two independent reviewers assessed the studies for eligibility. Discrepancies were resolved by consensus with the others authors.
Risk of bias in studies
We analyzed the risk of bias in the studies by means of a standardized table considering randomization, allocation, blinding, withdrawal, prognostic factors, outcomes, and intention-to-treat analysis. Also, all studies were classified according the JADAD scoring system [12].
Data extraction
The first author extracted data from eligible studies and organized spreadsheets divided in two groups (intervention and control). A second author checked the extracted data.
Statistical Analysis
The analyses were carried out with RevMan 5.3 software. Risk difference (RD) expressed differences between outcomes for dichotomous variables and means difference (MD) for continuous variables using the Cochrane–Mantel–Haenszel test (95% confidence interval). Also, we used inverse variance with fixed-effect (95% confidence interval).
The heterogeneity among studies was assessed with the Higgin`s test (I2). Sensitivity analysis was performed when the heterogeneity was greater than 50%. A funnel plot analysis aimed at identifying the study most likely to carry publication bias (outlier). If an outlier was identified, we excluded it and ran another analysis. If an outlier study was not distinguishable, we considered true heterogeneity and ran analysis using random-effect model to reduce the impact of heterogeneity on outcomes. Forest plots expressed graphically the results from the meta-analysis.
The initial search identified 2.805 articles screened through title and abstract assessment. Fifteen articles were selected. Among them, two studies presented only abstracts [9, 10] while another did not evaluate highrisk factors [11] and thus were excluded from this metaanalysis. Finally, 12 RCTs enrolling 3.310 patients were considered eligible (Figure 1). All trials were published between 1990 and 2016 and randomly assigned 1673 high-risk patients to pancreatic stenting group and 1673 to the no stent group (Table 1).
Figure 1. PRISMA flowchart: flow of information through the systematic review.
Pancreatitis Severity Subgroups
A. Mild pancreatitis: The incidence of mild pancreatitis was higher among no stent group patients: 11% (70/632) vs. 4.8% (30/617). The meta-analysis showed statistical difference in mild PEP incidence between groups (p<0.0001) with RD of -0.06 (95% CI-0.09 -0.03) and low heterogeneity (I2=0%) (Figure 2). The number needed to treat was 2.5.
Figure 2. Forest plots and funnel plot. Forest plot depicting the incidence of post-endoscopic retrograde cholangiopancreatography pancreatitis (PEP) in the analyzed studies according to pancreatitis severity. (a). Mild pancreatitis; (b). Moderate pancreatitis; (c). Severe pancreatitis.
B. Moderate pancreatitis: The incidence of moderate pancreatitis was also higher in the no stent group: 4.8% (28/572) versus 1.4% (8/557). The risk difference was -0.03 (95% CI-0.06 -0.01) favoring the pancreatic stenting group (p<0.0001) in a highly homogenous analysis (I2=0%). The number needed to treat was 5.
C. Severe pancreatitis: Again, the incidence of severe pancreatitis was higher in the no stent group: 2.5% (12/469) versus 0.2% (1/463). The risk difference between groups was -0.02 (95% CI- 0.04 -0.00) with p<0.0001 and I2=47% (Figure 2). The number needed to treat was 9.1
Hyperamylasemia
All articles assessed hyperamylasemia rate. In pancreatic stenting group, 167 out of 710 patients (23.5%) presented increased amylase in comparison with 619 among the 721 patients from the no-stent group. The pancreatic stenting group showed significant lower risk for hyperamylasemia (p<0.0001) (RD -0.62, 95% CI-0.65 -0.59). The number needed to treat was 0.2. However, we detected high heterogeneity in this analysis (I2=100%) but there was no distinguishable outlier in the funnel plot. Therefore, we assumed heterogeneity to be real and ran a random-effect analysis (Figure 3).
Figure 3. Forest plots and funnel plot. Meta-analysis showing the effects of pancreatic stent placement for hyperamylasemia.
Abdominal pain
Our results showed that 47 of the 399 patients in pancreatic stenting group developed abdominal pain compared to 69 of the 338 patients in the no-stent group (11.7% vs. 20.4%). The meta-analysis showed no statistical difference (p<0.08) between groups (RD-0.10, 95% CI-0.21 0.01). Again, we identified high heterogeneity (I2=67%) but no study was distinguishable as an outlier after funnel plot analysis. Therefore, we assumed true heterogeneity and performed a random-effect analysis (Figure 4).
Figure 4. Meta-analysis showing the effects of pancreatic stent placement for abdominal pain.
Cholangitis
Only 2 of the 67 patients (2.9%) in the pancreatic stenting group presented post-ERCP cholangitis versus none among the 74 patients in the no-stent group. The meta-analysis showed no difference (p<0.29) between groups (RD -0.03, 95% CI-0.03 0.09) and low heterogeneity (I2=0%) (Figure 5).
Figure 5. Meta-analysis showing the effects of pancreatic stent placement for cholangitis.
Duration of ERCP
The comparison regarding the total duration of ERCP was highly homogenous (I2=0%) and showed no difference between pancreatic stenting and no-stent group (p<0.07) (RD 2.97, 95% CI−0.19 6.12) (Figure 6).
Figure 6. Meta-analysis showing the effects of pancreatic stent placement for duration of ERCP.
Length of stay
The length of stay was similar between groups (p>0.05) with mean difference of -1.55 (95% CI-4.39 1.29). The meta-analysis showed high heterogeneity (I2=99%) but since only two trial assessed length of stay, a single exclusion would impair the pooled analysis (Figure 7). Therefore, we assumed it as true heterogeneity and performed a random-effect analysis.
Figure 7. Meta-analysis showing the effects of pancreatic stent placement for length of hospital stay.
ERCP is central for the treatment of several biliary and pancreatic diseases. However, the endoscopic manipulation may impair the normal flow of the pancreatic duct and activate intracellular proteolytic enzymes. Also, it may cause bacterial contamination which results in chemical pancreatic injury. All those factors trigger pancreatitis [3, 4].
Pancreatitis is the most common adverse event of ERCP. PEP increases morbidity and mortality significantly, especially in severe cases. Likewise, it carries increased length of stay, which raises associated costs. The reported incidence of PEP ranges from 2% to 30% [4, 12, 13, 14, 15, 16]. This wide variation is due to differences in patient selection and procedure-related risk factors. Risk factors for PEP include female gender, age ˂60 years, previous history of PEP, pancreatic sphincterotomy, sphincter of Oddi dysfunction, difficult cannulation, long duration of procedure, pancreatic duct biopsy and less experienced endoscopist [17, 18].
Several drugs and techniques have been investigated to avoid PEP [18, 19, 20] and the pancreatic stenting is one of the most robust in terms of available evidence. The stent may assure the pancreatic flow in case of papillary edema or sphincter spasms caused by the endoscopic manipulation. Possibly, this is the mechanism of action through which the pancreatic stent effectively prevent PEP [3, 16].
The trials included in our meta-analysis showed that pancreatic stenting reduced the risk of mild and moderate pancreatitis. This data is in accordance with previous meta-analyses [3, 13, 19, 20, 21, 22, 23]. However, our study demonstrated that such intervention does not affect the incidence of severe PEP, even though there was a subtle trend toward risk decrease. Possibly, the tiny severe PEP incidence might explain the lack of statistical difference between groups. Therefore, further and larger studies might change this statement.
In addition, pancreatic stenting reduced the incidence of post-ERCP hiperamylasemia. This result goes according to with from previous meta- analyses [13, 23, 24, 25, 26]. these results support the claim that post-ERCP pancreatic stent prevents pancreatitis.
Regarding abdominal pain, we showed that pancreatic stenting group was similar to no-stent group, and the difference was not statistically significant. It should be noted that abdominal pain is one of the criteria to define acute pancreatitis and, once diagnosed, patients have a longer hospital stay and costs increase proportionally. This result goes according to with from previous meta- analyses [27, 28, 29]. Moreover, the analysis of cholangitis as a post-ERCP, showed no difference between groups. The heterogeneity between the trials was low. This data is in accordance with previous metaanalyses [6, 26, 30, 31].
The mean duration of ERCP was longer in the pancreatic stent group than in the no-stent group; however, there was no statistical significance. Although the insertion of a pancreatic stent reduces PEP incidence, it is important to note that it is a challenging procedure. Difficulties may arise in the identification and cannulation of the pancreatic duct ostium, and the progression of the guidewire and the pancreatic stent itself in the duct lumen may be complex, with increasing local manipulation and duration of the procedure.
The mean length of hospital stay was shorter in the pancreatic stent group than in the no-stent group; however, there was no statistical significance. Because there was no stratification in the length of hospital stay, it was not possible to evaluate or predict costs; therefore, prospective studies focusing on this topic should ideally be performed.
Our study has some limitations. The first we could not evaluate the risk of post-ERCP pancreatitis in patients who failed PD stent placement. Finally various lengths of stents were used, which may result in evaluation bias.
ERCP is an essential procedure for the treatment of several pancreatobiliary disorders. In our study, we demonstrated that the placement of pancreatic stents is important because it prevents post-ERCP pancreatitis (mild and moderate). In addition, no serious complications were reported in individuals in the stent group. Thus, it is assumed that post-ERCP pancreatic stent placement leads to a reduction in morbidity, mortality, and possibly costs to the health system.
Pancreatic stenting is effective in reducing the incidence of mild and moderate post-ERCP acute pancreatitis in high-risk patients.
The authors have no conflicts of interests to declare.
