Needle-knife sphincterotomy and post-ERCP pancreatitis: time to lower the threshold for the needle?

Successful biliary cannulation is a prerequisite to biliary therapy during ERCP. However, biliary cannulation may be unsuccessful in up to 20% of cases using conventional techniques.1 Precut papillotomy can be used to obtain biliary access in this situation. This technique usually refers to needle-knife sphincterotomy. In this free-hand needle-knife technique, the cut is made starting at the orifice and extending cephalad for a variable distance. However, there are other precut papillotomy techniques such as the use of a papillotome to perform papillary roof incision or “fistulotomy,” which starts above the orifice; transpancreatic sphincterotomy (intentional seating of the tip of a standard traction papillotome into the pancreatic duct and cutting through the septum in the direction of the bile duct); or an intramural incision (a false tract created with a guidewire to place a papillotome through the intramural portion of the papilla and to unroof the biliary orifice). Several large, prospective studies2, 3, 4, 5, 6 and a meta-analysis7 have identified precut papillotomy as an independent risk factor for post-ERCP pancreatitis. However, there are other large studies that have not identified this technique as a risk factor.8, 9 The current standard of practice is that precut papillotomy is used as a last-resort measure when multiple attempts at conventional cannulation have failed. Thus, the hypothesis is that the resulting greater number of cannulation attempts at the papilla and multiple inadvertent pancreatic injections may be the dominant causative factor in the association between precut papillotomy and post-ERCP pancreatitis, rather than the precut papillotomy per se. The study by Bailey et al10 in this issue of Gastrointestinal Endoscopy has attempted to evaluate the relationship between needle-knife sphincterotomy (using the precut papillotomy technique) and post-ERCP pancreatitis.

In their study, Bailey et al10 analyzed prospectively collected data from two randomized trials of ERCP techniques, with post-ERCP pancreatitis as the primary endpoint. Needle-knife sphincterotomy was performed in 94 of 732 patients (12.8%) and was successful in achieving bile duct access in 80 of 94 (85%) patients. There were no perforations or major bleeding events. The overall frequency of post-ERCP pancreatitis with needle-knife sphincterotomy was significantly higher at 14.9% (14 of 94 patients) compared with 6.5% (48 of 734 patients) without. In multivariate analysis, independent predictors of post-ERCP pancreatitis were female sex, suspected sphincter of Oddi dysfunction, partial pancreatic drainage, 10 to 14 attempts at papilla, and >15 attempts at papilla. The incidence of post-ERCP pancreatitis increased with increasing number of attempts at the papilla; 11.5% with 10 to 14 attempts and 15% with >15 attempts. Needle-knife sphincterotomy was not found to be an independent predictor of pancreatitis, leading the authors to conclude that the number of attempts at the papilla is an independent risk factor for pancreatitis, rather than needle-knife sphincterotomy.

The authors suggest that instead of repeated, multiple, unsuccessful attempts at cannulation, which is associated with post-ERCP pancreatitis, needle-knife sphincterotomy should be considered earlier when it appears that multiple cannulation attempts are likely to fail. They suggest that there is a watershed moment in selective biliary cannulation when several important variables converge.10 At this point, successive attempts at biliary cannulation have not succeeded, and the risk of post-ERCP pancreatitis is escalating. The complete failure of conventional techniques of cannulation is imminent, and the use of needle-knife sphincterotomy becomes significantly more likely. Earlier use of needle-knife sphincterotomy (ie, lowering the threshold) for cases of difficult cannulation may ameliorate the occurrence of post-ERCP pancreatitis in this difficult cannulation subgroup.

The evidence presented by Bailey et al10 is compelling, but were there any caveats? This study was not randomized for needle-knife sphincterotomy intervention. The only prospective, randomized trial, to date, compared needle-knife sphincterotomy versus persistence with a non-wire guided, single-lumen papillotome.11 The investigators of that study, who were experienced endoscopists from Toronto, Canada, defined a difficult biliary cannulation as failed cannulation after 12 minutes. There were 32 patients randomized to the needle-knife arm and 30 to the persistence arm. From the authors' description of their needle-knife technique, it appears at first glance that the standard needle-knife technique was used, although it is unclear whether a fistulotomy was performed instead. Primary success rates and complication rates were similar in the needle-knife and persistence arms (75% and 4%, respectively for the precut arm vs 73% and 9% for the persistence arm). There was no post-ERCP pancreatitis in the needle-knife group, although there was one case of pancreatitis in the persistence arm. Thus, the conclusion from this study is that there was no significantly increased risk of complications with multiple cannulations compared with needle-knife sphincterotomy, which is different from that of Bailey et al.10 However, the incidence of pancreatitis with multiple cannulations appears to be higher than with needle-knife sphincterotomy, and, therefore, perhaps was not statistically significant because of the small numbers.

As mentioned earlier, there are many different techniques for precut papillotomy. The authors in this study10 used a conventional needle-knife sphincterotomy technique in which the cut was commenced from the top of the papillary orifice and extended upward in 2- to 3-mm increments, by using blended current. The goal was to completely divide the majority of the papillary mound in a controlled, stepwise fashion with a single pass, thus unroofing the biliary orifice. A pancreatic stent was placed when the duct was repeatedly instrumented. An alternative needle-knife technique is a fistulotomy, in which the cut is started above the orifice. A randomized trial of fistulotomy versus conventional needle-knife sphincterotomy found a lower rate of pancreatitis for fistulotomy (0% vs 8%) but similar overall complication rates and success at initial cannulation.12 Therefore, it is difficult to generalize the conclusion of the Bailey et al study, which used a conventional needle-knife sphincterotomy technique, to other precut techniques. It is not known whether the other techniques have comparable success and complication rates. Even in the same technique group, not all patients had pancreatic stents placed (<25% in this study). A higher pancreatic stent usage may reduce the incidence of pancreatitis even further.

The authors acknowledge that their study was performed by endoscopists experienced in high-volume biliary endoscopy, and this might have mitigated against needle-knife sphincterotomy complications. A multicenter study from Italy6 supports this observation. That study found that low-volume centers (<200 procedures per year) were associated with an increased risk for overall major complications. A multicenter study from the United States found that trainee involvement was a significant risk factor for post-ERCP pancreatitis (odds ratio 1.5).9 These data suggest that procedures done by experienced, high-volume endoscopists are likely to be associated with a lower risk of complications. However, at present, there is no universally accepted definition of what constitutes an experienced endoscopist or what numbers are required to be a high-volume endoscopist.

There are other alternatives to precut papillotomy for obtaining biliary access in cases of difficult biliary cannulation. These include (1) placement of a pancreatic guidewire (or stent) to assist biliary cannulation and (2) less commonly used, but previously described, novel devices such as endoscopic scissors, an endoscopic dissection technique using a cotton swab, and papillectomy for duct access.13 Placement of a guidewire or stent into the main pancreatic duct can facilitate cannulation of the bile duct by serving a number of functions, including opening a stenotic papillary orifice, stabilizing the papilla, lifting the papilla toward the working channel, straightening the pancreatic duct and the common channel, and draining the pancreatic duct, potentially to minimize repeated injections into the pancreatic duct.13 A pancreatic stent can be placed to reduce the risk of post-ERCP pancreatitis. There are no comparative data among these techniques, and even if there were, they may not be generalizable beyond the endoscopists and centers performing these techniques.

Should the threshold for performing needle-knife sphincterotomy be lowered? First of all, it should be considered only in cases of difficult biliary cannulation. Second, it should be performed only by experienced endoscopists. We think that placement of a pancreatic guidewire or stent to aid biliary cannulation should be attempted first when the pancreatic duct has been repeatedly cannulated. When the pancreatic duct has not been cannulated, then there is a case to be made for earlier consideration of needle-knife sphincterotomy, rather than repeated attempts at biliary cannulation, but, again, only in experienced hands. The definitive answer in the difficult cannulation scenario as to whether or not needle-knife sphincterotomy should be performed earlier requires larger, prospective, randomized studies comparing this technique to persistent attempts at cannulation.