Operator Experience

Increased operator experience also seems to reduce the risk of overall complications slightly, but more importantly, reduces risk of major complications. The role of experience is difficult to quantify because the more experienced centers tend to perform more complicated procedures and thus may have an increased risk for complications. The patients may also be in poorer health or have more difficult anatomy. For example, community gastrointestinal groups often refer patients who have had unsuccessful ERCP cannulations or patients with altered anatomy to tertiary centers. It is intuitively obvious that the number and frequency of ERCPs performed by the endoscopist correlates inversely with rates of complication. However, determining the number of completed procedures that are required to achieve proficiency is difficult because of all the variables involved. The British Society of Gastroenterology has recommended a minimum of 75 ERCPs performed yearly. There is also evidence that the number of ERCP sphincterotomies performed in a week makes a difference in the rate of bleeding complications. Finally, undertaking a dedicated ERCP/EUS (advanced endoscopy) fellowship reduces the risk, but controversy remains and it is still unclear how many procedures need to be performed to gain even the basic competencies. Less than 200 cases does not seem to be adequate, but how many remains a question.

Preprocedure Risk Factors

Patients with SOD carry the highest risk for PEP, ranging from 10% to 40%, and, therefore, it is important to assess this special population of patients carefully (see Table 2 ). SOD is often suspected in younger women with abdominal pain and it seems women, in general, have a higher likelihood of PEP (odds ratio [OR], 2.23). Simply being evaluated for SOD dysfunction can raise the PEP likelihood to 10% to 30%. It was initially thought that biliary and pancreatic sphincter manometry added to this risk factor. Cotton and colleagues retrospectively analyzed 11,497 ERCPs and found biliary and pancreatic manometry to have ORs of 1.16 and 1.43, respectively, neither of which were statistically significant. Other risk factors usually attributed, but not exclusive, to this population, include an age less than 60 to 70 years and normal caliber bile duct. History of alcohol use has been shown to be a significant risk factor (2 to 3 relative risk). Pancreas divisum has shown some increased risk in some studies but the data are not uniform. A more recent retrospective review showed that there is a low PEP rate of 1.2% if dorsal duct cannulation is not attempted, but with dorsal duct cannulation attempts, the risk of PEP is 8.2% and goes up to 10.6% when a minor duct sphincterotomy is performed. A previous history of PEP also puts patients at risk to develop PEP with future ERCPs.

There are also factors that are protective of PEP. Patients with chronic pancreatitis and pancreatic malignancy have lower risk, which is likely because of the atrophy of the upstream pancreatic parenchyma from the chronic obstruction. Because younger age and normal total levels of bilirubin are risk factors for PEP, ages greater than 80 years and increased total bilirubin have been shown to be protective.

Intraprocedure Risk of PEP

Ampullary trauma plays an important role in the pathophysiology of PEP. This trauma can be caused by prolonged cannulation attempts, which reflects the difficulty of cannulation or the experience level of the endoscopist. Initially, it was thought that alternate ways of accessing the duct after difficult cannulation, such as precut sphincterotomy or needle-knife access, increased the risk for PEP. However, there has been growing evidence that early precut access, after only 4 to 10 attempts for cannulation, actually decreases the risk for PEP compared with persistent attempts at cannulation, which can traumatize the papilla further. Two meta-analyses (6 randomized controlled trials) published in 2010 showed relative risks of 0.46 (95% confidence interval [CI]: 0.23–0.92) and 0.47 (95% CI: 0.24–0.91) when early precut access was obtained, which supports the concept of ampullary trauma after multiple attempts. Precut sphincterotomy performed early by an experienced endoscopist before significant trauma results is most likely helpful in preventing PEP. However, precut papillotomy is considered an advanced technique and may be better suited for more experienced endoscopists. It is often difficult to determine how many attempts can be tried until precut is considered. One suggestion would be to consider this after 10 cannulation attempts.

Contrast dye injection into the pancreatic duct has been shown, in multivariate analysis, to be a PEP risk factor, and avoidance or minimization of contrast injection with guidewire cannulation has been shown to decrease PEP. This is somewhat controversial, because there are many trials with mixed results and a most recent prospective comparison trial by Mariani and colleagues showed no difference in rates of PEP. However, a recent meta-analysis evaluated 7 randomized controlled trials and showed a significant difference of 3.2% versus 8.7% PEP with guidewire cannulation compared with contrast-guided biliary cannulation, with a statistically significant relative risk reduction of 0.38 (95% CI, 0.19–0.76). Although guidewire cannulation seems to have a lower risk of PEP than contrast dye injection, another study found a significantly high rate (34.6%) of PEP when pancreatic guidewire was left in place to achieve selective biliary cannulation. It is reasonable to theorize that any pancreatic duct manipulation is, in itself, a risk for PEP, and that if the endoscopist resorts to leaving a guidewire in the pancreatic duct, it was likely a difficult cannulation, further increasing the risk for PEP.

Balloon dilation sphincteroplasty is often used instead of electrocautery sphincterotomy to reduce the risk of postsphincterotomy bleeding. However, this may increase the risk of PEP in certain patient groups; thus, assessing risks before deciding which modality to use is important.

There is some controversy as to whether self-expanding metal stents (SEMS) placed in the biliary tree for malignant biliary obstruction potentially increases the risk of PEP. Theoretically, the greater diameter and radial expansion of the SEMS has the potential to obstruct or distort the pancreatic orifice or common channel, thus increasing the risk of PEP. A recent, large prospective trial showed that SEMS were associated with a 7.3% incidence of PEP and plastic stents were associated with a 1.3% incidence of PEP, which conflicts with previous preliminary studies of safety using SEMS, in which PEP occurred in approximately 1% of the cases. More studies are needed to determine the true risk.

An intraductal ultrasound probe, inserted through the working channel of the duodenoscope, has been used for investigating suspected malignancies and strictures of the biliary duct. However, intraductal ultrasound has been shown to increase the risk of PEP with a hazard ratio of 2.41 ( P = .004). It is likely that the rigid probe causes mechanical irritation of the papilla as it enters the bile duct.

There are additional methods variably used to reduce the risk of PEP. The use of guidewire cannulation instead of contrast dye injection has already been discussed. Another important technique is the use of prophylactic, plastic, pancreatic stents ( Box 1 ). It is theorized that outflow tract obstructions of the pancreatic duct can occur from edema and inflammation during manipulation of the papilla. This obstruction can trigger iatrogenic pancreatitis. By placing a small (3F or 5F) temporary, plastic stent into the pancreatic duct, flow can be maintained ( Figs. 1 and 2 ). Several studies have shown the benefits of pancreatic stents, and 3 meta-analyses have been published on this subject. The most recent, published in 2011, pooled data from 8 randomized controlled trials and showed an OR of 0.22 (95% CI: 0.12–0.38) with an impressive absolute risk reduction of 13.3%, giving the number of patients needed to treat as 8 patients to prevent a single episode of pancreatitis. This risk reduction was true for both mild and moderate PEP. Although this meta-analysis did not show a decrease in frequency of severe PEP, a previous meta-analysis did. With this overwhelming evidence derived from randomized controlled trials, it has become the standard of care to place prophylactic plastic stents in patients at high risk for PEP (see Table 2 for risk of PEP). The effects of diameter, length, and presence of flanges have also been studied. There does not seem to be a difference in flanged versus unflanged stents, or shorter (<3 cm) versus longer (>3 cm) stents, although the studies focusing on these factors are small. Intuitively, it makes sense that smaller unflanged stents may fall out of the pancreatic orifice more quickly, thus limiting their efficacy in mitigating outflow obstructions. In contrast, longer, flanged stents have a higher retention risk and may require a separate procedure to remove. The data continue to be mixed, and it is still unclear what the perfect length of time that a stent should remain in place to balance out the reduction of risk for PEP with the risk of damaging the pancreatic orifice. However, it seems that 3-F or 5-F short plastic stents are just as effective as large-diameter stents. In the authors’ practice, 1-sided pigtail stents to prevent proximal migration are used. Abdominal plain film radiographs should be performed after 2 weeks to assess for spontaneous passage of these stents, as retained stents have the potential for scarring and damage to the pancreatic duct and parenchyma. Because failed attempts of pancreatic duct stent placement have been associated with increased risk of PEP, an intention-to-treat analysis was conducted and it was found that there is a strong trend toward benefit; however, the study was likely not powered enough to reach statistical significance. In this group, the pancreatic duct placement rate of success was 88%. Despite its ability to decrease PEP incidence in high-risk patients, prophylactic pancreatic duct stent placement does not seem, based on a survey of community and tertiary care practices, to have disseminated to the endoscopy community. Although more than 75% of cases were identified as high risk, fewer than 50% of endoscopists reported attempting pancreatic duct stent placement in this setting. Importantly, 21% did not place prophylactic pancreatic duct stents in any circumstance. This results in the debate of requiring that endoscopists who perform ERCPs to be proficient in pancreatic duct stent placement or whether all high-risk ERCP patients should be referred to appropriate care centers with endoscopists who possess the necessary expertise.

Needle knife sphincterotomy of the minor papilla over a prophylactic, plastic, pancreatic stent. Note a plastic, biliary stent coming out of the major papilla in the background ( arrow ).

Examples of a plastic, pancreatic duct stent in the major papilla for PEP Prophylaxis.

Pharmacologic Strategies for the Prevention of Post-ERCP Pancreatitis

Given the frequency of PEP, pharmacologic agents to reduce PEP have been sought. Although some show promise, most agents either had mixed results or fared no better than placebo in rigorous trials. Nonsteroidal anti-inflammatory drugs (NSAIDs) have most recently been in the spotlight for their potential to reduce the frequency of PEP substantially. NSAIDs are potent inhibitors of the phospholipase A2 and other inflammatory mediators. Phospholipase A2 is believed to play a critical role in the initial inflammatory cascade of acute pancreatitis. Early studies of rectal diclofenac showed some efficacy in preventing acute pancreatitis. Most recently, rectal indomethacin has shown promise. Earlier studies of rectal indomethacin looked at treated patients with average or varying risk factors for PEP. In one of the first large studies, there was a trend toward statistical significance (PEP in 3.2% with rectal indomethacin vs 6.8% with placebo; n = 490; P = .06). Earlier studies showed trends that did not meet statistical significance. Another study showed significant hyperamylasemia differences in patients treated with rectal indomethacin versus placebo. The earlier studies used rectal indomethacin either 2 hours before or immediately before ERCP on the premise that it takes 30 to 90 minutes to achieve a therapeutic level. A recent, large, randomized controlled study that received considerable press showed an absolute difference of 7.7% in patients receiving rectal indomethacin versus placebo (9.2%–6.9%) with the number needed to treat as 13. However, the difference from the previous study was the inclusion of only patients considered to be at high risk of PEP, who were given either rectal indomethacin or placebo immediately after ERCP. Importantly, there were no significant increases in bleeding or renal failure in the indomethacin group when compared with the placebo group. Rectal indomethacin (100 mg) seems to be a safe, effective, prophylactic treatment in patients who are at high risk for PEP. It also has the potential to reduce the risk of PEP for average-risk patients, but further investigation is needed. The timing of administration is still questioned, but the authors’ personal perspective is that it is reasonable to administer rectal indomethacin before the procedure in patients with significant preprocedure risk for PEP. For patients who move into the high-risk group only during ERCP (ie, difficult cannulation, multiple pancreatic duct injections, precut sphincterotomy, and so on), rectal indomethacin can be administered immediately afterward. The potential for an indomethacin suppository may be included in the informed consent for the ERCP.

Various other pharmaceutical agents targeting various components of the pathophysiologic process of acute pancreatitis have been studied, including pretreatment to induce papillary dilation with agents such as nifedipine, lidocaine, epinephrine, and glyceryl trinitrate, on the premise that manipulating the pancreatic orifice causes spasms and temporary duct obstruction, leading to acute pancreatitis. The first 3 agents did not show any efficacy, but glyceryl trinitrate (GTN) showed the most promise, although, again, data are mixed. The most recent meta-analysis of 7 randomized controlled studies comprising 1841 patients showed an OR of 0.56 (0.40–0.79; 95% CI) PEP reduction compared with placebo. Adverse events included hypotensive episodes (systolic blood pressure <100 mm Hg) in 50% of the GTN group compared with 5% of the placebo group, but all responded to intravenous colloid infusions. The GTN group also had more headaches. The best route of GTN administration is not completely clear, although a recent meta-analysis suggests the sublingual approach is more effective than the transdermal approach. An intravenous approach has also been used. Interestingly, centers with a higher baseline incidence of PEP had a higher risk reduction with GTN than centers with a lower PEP at baseline. This difference in risk reduction can be viewed in the following 2 ways: if it is assumed that a higher baseline incidence of PEP is because the center undertook more complicated, high-risk procedures, then, much like indomethacin, GTN should perhaps be reserved for the higher risk cases. Alternatively, if it is assumed that the higher incidence of PEP is because of a lack of expertise in these centers, then GTN may be more protective in cases where there is difficulty in cannulating. Another meta-analysis in the same year reviewed 8 randomized controlled trials and showed a risk reduction with an OR of 0.60 (0.39–0.92; 95% CI). However, using sensitivity analysis, in which they removed the trial that did not use standard diagnostic criteria for PEP, the OR range crossed 1.0 with an OR 0.68 (0.41–1.11; 95% CI). Because of mixed results and the question of the ideal route of administration, recommendations about nitroglycerin remain inconsistent.

Protease inhibitors, including Gabexate, Ulinastatin, and Nafamostat mesylate, have been used extensively to manage acute pancreatitis and are part of the standard of care in Japan. Protease inhibitors inactivate the proteolytic enzymes that are thought to play in important role in pancreatic injury. In the most recent meta-analysis in 2011, in which 19 cohort studies were pooled, the overall results showed a modest but significant risk reduction with a pooled risk difference of −0.029 (95% CI, −0.051 to −0.008). This difference yielded nonsignificant results when only high-quality studies in the analysis were included. Most recently, in a study that compared the use of Gabexate, Ulinastatin, and placebo, the Gabexate group had a 3.5% incidence of PEP compared with the Ulinastatin group with 7.0% and placebo group with 7.3%, a statistically significant difference. Thus, Gabexate, but not Ulnistatin, showed efficacy in preventing PEP. Aside from the mixed data and the cost of these agents, another limiting factor is the administration schedule. The infusion of Gabexate and Ulnistatin begins 30 minutes before the procedure and continues for 24 hours. The newer Nafamostat mesylate infusion is started 1 hour before the procedure and is continued for 6 hours after ERCP. Although their use is a common practice in Japan, protease inhibitors have not been adopted in western countries because of these limitations.

Secretin is a gastrointestinal peptide endocrine hormone that primarily stimulates the secretion of bicarbonate-rich fluid from the pancreas. In 1 study, 869 patients were randomly administered 16 ug (8 mL) of synthetic secretin (equivalent to 80 clinical units of biologic activity) or placebo at the start of the ERCP or the start of pancreaticobiliary manometry. The secretin group had an 8.7% incidence of PEP compared with 15.1% in the placebo group ( P = .004). In this study, secretin was highly effective in reducing PEP in patients undergoing biliary sphincterotomy in subgroup analysis. Although this seems promising, further studies are needed.

Somatostatin and octreotide function by inhibiting the exocrine secretion of the pancreas, which theoretically prevents pancreatitis and autodigestion. Meta-analyses of many randomized trials have not shown any efficacy with these modalities. Other agents, such as low-molecular-weight heparin (based on animal models of inhibition of pancreatic damage), allopurinol, and n-acetyl cysteine, have not shown efficacy in preventing PEP.

Prophylactic corticosteroids, in turn, have also been used to decrease the inflammatory response of acute pancreatitis. The most recent meta-analysis showed an OR of 1.13 (95% CI, 0.88–1.46) with subsequent sensitivity and subgroup analysis confirming no statistical difference in risk reduction.