EUS-first versus ERC-first for patients with intermediate probability of bile duct stones: is one strategy superior to the other?

Biliary sphincterotomy at endoscopic retrograde cholangiopancreatography (ERCP) has been used in the treatment of choledocholithiasis for over 25 years [1]. Its principal complications (pancreatitis, bleeding, and perforation) have a total rate of 4 % - 10 %, although this rate appears to be decreasing, probably owing to better understanding of predictors of complications and the increased use of wire-guided cannulation [2] [3]. Although intraoperative open common bile duct exploration was found to be superior to preoperative ERCP in randomized controlled trials (RCTs) that were carried out when open cholecystectomy was the norm, ERCP became more popular in the laparoscopic era, in order to optimize the minimally invasive approach of laparoscopic cholecystectomy [4]. The results of RCTs assessing the optimal approach in the laparoscopic era have yielded disparate results, and have limited generalizability for reasons of patient selection and operator expertise [4]. For patients with a high likelihood of having a common bile duct (CBD) stone, ERCP remains a strategy of choice not only postoperatively, of course, but also preoperatively [4]. The development of highly accurate diagnostic noninvasive technologies for the detection of choledocholithiasis, that do not have an associated risk of pancreatitis, such as endoscopic ultrasound (EUS) [5] and magnetic resonance cholangiopancreatography (MRCP) [6], has offered more adapted alternatives to ERCP as a diagnostic tool for patients with a lower likelihood of having a CBD stone. Appropriate stratification of patients with regard to the probability of choledocholithiasis thus now becomes a critical step for optimal and cost-effective management [4].

In this issue of Endoscopy, Polkowski et al. present the results of a single-center, pragmatic randomized trial comparing an ERC-first with an EUS-first strategy for the management of patients with an intermediate probability for CBD stones, as previously defined [7]. Over a 2-year period, 257 patients were screened and a total of 100 were randomly allocated to either ERC (n = 50) or radial EUS (n = 50). Patients diagnosed with CBD stones at ERC underwent biliary sphincterotomy and stone extraction at the same procedure, whereas those assigned to EUS and with a discovered CBD stone underwent subsequent therapeutic ERC. If any of the initial interventions were unsuccessful, the treating physician could then choose to perform either one of the two interventions under evaluation. The main end point was a composite negative outcome consisting of predefined procedure-related or disease-related complications, including those resulting from false-negative examinations. This methodology closely mirrors that in a previously reported abstract of a randomized trial comparing ERCP with MRCP [8]. Patients were followed with questionnaires for up to 12 months after enrollment. Overall, 28 % and 25 % of patients were found to have CBD stones in the EUS and ERC groups, respectively. The sensitivities of ERC and EUS in the detection of CBD stones, and the number of procedures, did not differ between the groups, but the overall incidence of negative outcomes was higher in the ERCP group (19/50 [40 %] vs. 5/50 [10 %]; 95 % CI of the difference 27 % - 54 %); no procedure-related deaths were reported.

This is an important study that addresses a very practical clinical question and that may not be duplicated. The authors are to be commended for completing such a difficult and thoughtful study. As with all good RCTs, the issues to be raised principally address the generalizability of the results. Although EUS has been shown to be highly accurate for the detection of CBD stones [5], this is the first head-to-head trial of “effectiveness”, that compares its use to that of ERC as the initial investigation following clinical assessment and abdominal ultrasound in this patient population. The advantage of the adopted study design is its increased generalizability or external validity, that closely approximates a “real-life” setting, where patients are not bound by the rigorous rules of a randomized trial, and where the “gold standard” is the patient’s actual clinical outcome. On the other hand, the greater flexibility of the randomized effectiveness trial decreases its internal validity where observed between-group differences might not be attributable only to the interventions under evaluation, since there is greater background variability in the patients’ management after randomization.

Both groups were similar, but the high proportion of females (84 % in the EUS and 92 % in the ERC group), the low incidence of associated gallbladder stones, and the high prevalence of post-cholecystectomy patients limit the generalizability with regard to patient population, and are to some extent, atypical for such a patient population.

As hypothesized by the investigators, an ERC-based approach resulted in a higher rate of adverse events. However, some unusual findings may have introduced some degree of bias which favored the EUS group. The proportion of successful initial ERCPs (36/48; 75 %) was significantly lower than the expected rates of proficiency (greater than 95 %) [3], and perhaps the corresponding endoscopist experience and subsequent need for repeated examinations led to a higher rate of complications than would be expected. This finding may in part have been attributable to when the trial was completed (1999 - 2002), a time period that may have antedated the increased contemporary use of guide wire cannulation [2] [3]. Also, both the overall ERC complication rate (19/48; 40 %) and the attributable risk of complications in the ERC group (ERC risk minus EUS risk; 32 %) were quite elevated considering the indication for the ERC, in the light of published data [2] [3]. Interestingly, had a 5 % - 10 % procedure-related complication rate had been observed in the ERC group, the difference between both groups would not have been statistically significant, all else being equal. Furthermore, the observed difference in complications was attributable to the minimal and mild ERC complications, as there existed no significant differences in the rates of moderate-to-severe complications. This is especially important given the inclusion of abdominal pain by itself as an unvalidated negative outcome whose significance remains unclear, especially taking into account the lack of observer blinding or third-party adjudication. Finally, a cost-effectiveness component to the analysis would have been of interest.

In conclusion, the generalizability of the results of this well-conducted clinical trial are limited because of the patient population, endoscopic expertise, and choice of outcome measures. The readers must decide whether these are representative of their own practice setting or not. If not, these data remain informative, but the final choice of an EUS-first or ERCP-first strategy in patients with an intermediate probability of choledocholithiasis remains dependent on local expertise and availability, as previously suggested [4].

Competing interests: None

References

• 1 Cotton P B. Endoscopic management of bile duct stones; (apples and oranges).  Gut. 1984;  25 587-597
• 2 Freeman ML, Guda NM. Prevention of post-ERCP pancreatitis: a comprehensive review.  Gastrointest Endosc. 2004;  59 845-864
• 3 Freeman ML, Guda N M. Endoscopic biliary and pancreatic sphincterotomy.  Curr Treat Options Gastroenterol. 2005;  8 127-134
• 4 Tse F, Barkun J S, Barkun A N. The elective evaluation of patients with suspected choledocholithiasis undergoing laparoscopic cholecystectomy.  Gastrointest Endosc. 2004;  60 437-448
• 5 Prat F, Amouyal G, Amouyal P. et al . Prospective controlled study of endoscopic ultrasonography and endoscopic retrograde cholangiography in patients with suspected common-bile duct lithiasis.  Lancet. 1996;  347 75-79
• 6 Romagnuolo J, Bardou M, Rahme E. et al . Magnetic resonance cholangiopancreatography: a meta-analysis of test performance in suspected biliary disease.  Ann Intern Med. 2003;  139 547-557
• 7 Barkun A N, Barkun J S, Fried G M. et al . Useful predictors of bile duct stones in patients undergoing laparoscopic cholecystectomy. McGill Gallstone Treatment Group.  Ann Surg. 1994;  220 32-39
• 8 Barkun A, Romagnuolo J, Reinhold C. et al . The effectiveness of ERCP versus MRCP: final results of a randomized clinical trial in patients with an intermediate probability of bile duct obstruction [abstract].  Gastrointest Endosc. 2004;  59 98