Endoscopic retrograde cholangiopancreatography (ERCP) was first introduced in 1968 by McCune et al. [1], and has evolved over the decades. Currently, it is a valuable, widely used diagnostic and therapeutic tool in hepato-biliary-pancreas diseases. However, ERCP has a relatively high complication rate of nearly 10% and a mortality rate of 0.1 to 1% [2,3]. As therapeutic aspects of ERCP are becoming more important and endoscopists take on increasingly more complex cases, the risk of complication is increasing. Although pancreatitis, cholangitis, and hemorrhage are more frequent ERCP complications [3,4], ERCP-related perforation is one of the most feared, due to its potentially lethal nature.

The reported perforation incidence is 0.3 to 2.0% [3,5-9], and the mortality rate is 12 to 25% [2,3,9-11]. Hazardous as it is, appropriate management of ERCP-related perforation is essential to avoid detrimental outcome. Surgical management was considered the standard management of perforations. Many authors advocated early surgical management for ERCP-related perforations [8,9,12]. Most recent studies, however, indicate that carefully selected patients may recover uneventfully with conservative management alone [10,11,13-17]. Many treatment guidelines are being proposed [10,11,16-18], but a consensus has not been reached. To establish a consensual guideline with appropriate validation, a randomized controlled trial (RCT) is preferable; however, RCTs for ERCP-related perforation are unlikely due to ethical problems. Therefore, detailed analyses of large-scale case series are a better alternative.

This study aimed to identify patients who may be managed conservatively by examining the ERCP perforation treatment results. At the same time, we aimed to establish a simple ERCP perforation management guideline with a management algorithm based on a review of 53 cases.

From January 2000 to December 2010, 8,381 ERCP procedures were done at a tertiary referral center in Seoul, Korea. During that period 53 ERCP-related perforations (0.63%) occurred. The 53 patients were managed according to the management guideline at Seoul National University Hospital. Surgical indications were: 1) remaining problem unsolved by endoscopic procedure that requires immediate correction, such as retained hardware or biliary stone failed to be removed during ERCP; 2) duodenum lateral wall or jejunum injury without possibility of minimal leakage through adhesions or interventions (jejunum was included because of relatively high incidence of altered anatomy due to gastrojejunostomy in Korea); and 3) initial or subsequent symptoms/signs of severe sepsis/peritonitis. Patients with one or more of these indications underwent surgical management, and patients without any of the above were managed conservatively.

Conservative management consisted of antibiotics administration, non-per oral, parenteral nutritional support, and percutaneous/endoscopic drainage where appropriate. Operative management consisted of 9 primary repairs, 4 pyloric exclusions, 3 diverticularizations, 1 common bile duct (CBD) excision, and 1 exploration with irrigation and T-tube insertion.

Injuries were classified into two types: type I were duodenum lateral wall or jejunum injuries and type II were para-Vaterian injuries which includes periampullary injury and bile duct injury.

After acquiring approval from our Institutional Review Board, data on demographic information, ERCP indication and purpose, perforation site, management method, related complication, mortality, hospital stay, and hospital cost of the patients were collected from the patient's electronic medical records. Treatment outcome was assessed by evaluating the hospitalized duration, hospital cost, and related morbidity and mortality rates.

Comparative analysis, according to injury type and management method, was done using chi-square or Fisher's exact tests for categorical variables and Mann-Whitney U tests for continuous variables. Mean values are reported as mean ± standard error of mean. All statistic values with P-value < 0.05 were considered statistically significant. IBM SPSS ver. 19 (IBM Co., Armonk, NY, USA) was used for statistical analysis.

There have been continuous discussions on perforation management, and ERCP-related perforation was traditionally known to require prompt surgical management [8,9,12,19]. However, successful conservative managements have been reported, and currently is being widely accepted in selected patients [10,11,13-18]. A key discussion issue is who can be conservatively managed and who should be promptly operated upon. This is important since many have reported high morbidity and mortality rates in failed conservative management [8,10,14]. As the best outcome in a perforated patient can only be ensured by prompt and appropriate management, an accurate and simple guideline is needed.

Many studies report that around 70% of patients with ERCP-related perforation can be managed conservatively [16,17,20]. This percentage may vary depending on management policy. Following the treatment guideline at our institution, 66.0% of perforated patients were conservatively managed. Excluding mortalities, the conservative management failure rate was only 3.0% (1/33). To evaluate the validity of conservative management, conservative and operative groups were compared. Outcome comparison showed a significantly lower morbidity rate in the conservative group; also, with marginal significance, their hospital stay was shorter and cost was less (Table 1). These results suggest conservative management offers a better outcome than operative management. Therefore, any patient suitable for conservative management should be properly guided as to avoid unnecessary operation that may lead to an inferior outcome.

An important and significant difference between conservative and operative groups was in injury type. Most of the operative group had type I injuries, and nearly all of the type II injuries were managed conservatively. This implies that management method can simply be determined on the basis of injury type. Several researchers have classified perforations according to injury location or mechanism, and have proposed treatment recommendations for each class (Table 3) [10,15,21]. There is general acceptance that duodenum lateral wall perforation requires immediate surgery and that the remaining cases can be managed conservatively, in most cases. Based on our results and on previous opinions, ERCP-related perforations in the current study were classified into two types (rather than 3 or 4 types) to simplify the treatment algorithm.

Our type I was defined as duodenum lateral wall or jejunum injuries. The main reason for including jejunum injury is the high incidence of gastric cancer in Korea. Thus, encountering Billroth II anastomosis (gastrojejunostomy) during ERCP is not rare, especially in large tertiary referral center. In addition to their relatively frequent incidence, complexity due to anatomic alteration increases perforation risk. Eight of 53 cases (15.1%) in this series were patients with Billroth II anastomosis.

Type II represents para-Vaterian injuries defined as periampullary or bile duct injuries. Type II includes sphincterotomy-related and guidewire-related perforations because para-Vaterian injuries are mostly associated with endoscopic sphincterotomy or interventions and guidewire manipulations. We saw little use in classifying periampullary, bile duct, and guidewire-related perforations separately; moreover, nearly all guidewire perforations are periampullary or bile duct injuries. Furthermore, management outcome and treatment strategy do not differ for periampullary and bile duct injuries. These injuries can be managed conservatively and may be further improved by employing immediate endoscopic interventions to divert bile and pancreatic juice from the injury site [16,21,22]. Therefore, we integrated them into a single type for the sake of simplicity and to allow for quick countermeasures. The fact that the outcomes of conservatively managed bile duct injury were equivalent or superior to that of periampullary injury further supports the validity of merging bile duct and periampullary injuries. Although type II injuries can be successfully managed conservatively, an operation should be executed without hesitation in cases with evidence of severe sepsis/peritonitis or unsolved problems requiring correction such as retained biliary stone or hardware. Problems requiring surgical corrections other than the ones mentioned above are better dealt with elective surgery after recovery.

The management outcomes of type I and type II injury differ significantly in terms of hospital stay, cost, and morbidity rate. Table 1 clearly demonstrates superior outcomes for type II over type I injuries. Whether applying conservative management to type I injury would improve type I injury outcome can be questioned. The outcomes of conservative management of type I were not very different from those of operative management in type I injuries. Therefore, conservative management does not seem to guarantee improved outcome in type I injuries.

However, conservative management of type I injury after endoclipping seems promising. With the development of endoscopic techniques and equipment, there are more reports on nonsurgical treatments for small perforations in the upper gastrointestinal tract [23-25]. If endoclipping of small type I perforations at the time of injury is successful, conservative management may be attempted. One type I patient in our series was successfully managed in such a manner, and was discharged 19 days after the incident. Actual full recovery was achieved in 7 days but there was a 12 day delay due to financial disputes. Therefore, in cases where endoscopic clipping is possible, better outcomes can be anticipated, even in type I injuries. This warrants further investigations into the application of endoclipping and the feasibility of conservative management in type I injuries.

Based on our findings, we propose a simple management algorithm which can be readily and easily used (Fig. 3). If type I injury is detected, a prompt operation is recommended. Even in type I injury, conservative management can be attempted if endoclipping is possible at injury occurrence. In type II injury, conservative management may be initiated unless there are unsolved problems requiring surgical correction or any signs of severe sepsis/peritonitis, in which case surgery would be recommended. Defining severe sepsis/peritonitis objectively remains a task to be solved.

An obvious limitation of this field of study is that an RCT is not possible due to major ethical issues; thus, such study inevitably leads to retrospective case reviews. Another limitation is the small number of ERCP-related perforation cases. To our best knowledge, our number of analyzed cases (53 cases) from a single center is second only to a study by Fatima et al. [17] (75 cases). Even so, our sample size is insufficient for subgroup analyses. Rare as such cases are, acquiring enough cases from a single center is inefficient. For a quality conclusion, accumulation of data from multiple centers with same management guidelines or a comprehensive meta-analysis is required.

In summary, conservative management is feasible and produces better outcome than operative management in well-selected patients. Nearly all type II injuries and some type I injuries where endoclipping is possible may be successfully managed conservatively. Therefore, for convenience, injury may be classified simply into type I and type II injuries, and the appropriate management method determined accordingly. By following a simple and easy-to-remember management algorithm, satisfactory treatment outcomes can be expected.