Surgical versus non-operative initial management of post-endoscopic retrograde cholangiopancreatography perforation: a systematic review and meta-analysis

Introduction

Since its introduction in the 1960s, endoscopic retrograde cholangiopancreatography (ERCP) has played both a diagnostic and a therapeutic role. Today, since the establishment of noninvasive diagnostic procedures such as magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasound (EUS), ERCP has evolved to predominantly become a therapeutic tool in hepato-pancreato-biliary diseases [1]. While a safe method, ERCP is associated with considerable complications, such as pancreatitis, bleeding, cholangitis and duodenal perforation [2]. The latter is the most concerning of all complications, with an estimated incidence ranging from 0.09-1.67% and mortality up to 8% [3].

Because of the rarity of this adverse event and the heterogeneity of patients, data in the literature are based on case series and there are no large prospective studies. Thus, initial management of post-ERCP perforations is still controversial, with no consensus among clinicians regarding optimal treatment. Traditionally, surgery was considered a more appropriate method of management for post-ERCP perforation; however, in the past decade a more selective approach has evolved, with both conservative and endoscopic management options being available [4]. There are only limited and isolated data available regarding surgical management, with controversial recommendations on the timing and type of surgical procedures. As a result, management tends to depend on an individual clinician’s preference and expertise.

According to Stapfer et al [5], 4 classes of duodenal perforations, in descending order of severity, based on the mechanism, anatomical location and severity of the injury, may predict the need for surgery. More specifically, lateral or medial wall perforations (type I) caused by the endoscope are often large and remote from the ampulla, with intraperitoneal leakage, and require immediate surgery. In the present study we therefore aimed to perform a systematic review and meta-analysis of all the available literature and data regarding the initial management of perforations following ERCP.

Materials and methods

Results

The literature search resulted in the retrieval of 1725 records (Fig. 1). After the removal of 1080 duplicate records, 645 titles and abstracts were screened. A total of 591 articles were excluded, leaving 54 articles that underwent a full-text review. Most of the studies did not use the Stapfer classification, and as a result only 10 comparative studies [1,5,9-16] regarding the initial management of post-ERCP perforations were eligible for quantitative analysis.

Figure 1 PRISMA flow diagram

The main characteristics of the included studies are presented in Table 1. Nine retrospective studies [1,5,9-10,12-16] and a retrospective review of a prospective database [11] were included in the quantitative analysis. Table 1 also displays demographics information. Six of the included case-series [5,10-11,14-16] were scored as 8/9 and 4 studies [1,9,12-13] as 7/9 in the Case Series Quality Appraisal Checklist [8].

Table 1 Study characteristics

A total of 223 patients with post-ERCP perforations were identified and included in the present evaluation. The incidence of ERCP-related perforations was higher in studies with a small number of total ERCPs. Perforation type was classified according to the Stapfer classification [5]. Of the 58 patients identified with a type I perforation, 39 patients (67.2%) underwent initial surgical management, while 19 (32.8%) had initial NOM. In addition, 131 patients with type II perforation were included in the present study, with 13 (9.9%) undergoing initial surgical management and 118 (90.1%) undergoing initial NOM. There were 22 patients with a type III perforation; only 3 (13.6%) were treated with surgery while the remaining 19 (86.4%) underwent NOM. Finally, 20 patients with a type IV perforation were identified, and all of them (n=20) were treated with initial NOM (Supplementary Table 1).

Supplementary Table 1 Type of perforation (Stapfer) and initial management

Supplementary Table 2 PRISMA 2009 Checklist [6]

Table 2 summarizes the success rates of the initial management of post-ERCP perforations. Specifically, 31 of 39 patients (79.5%) with type I perforation had a successful initial surgical treatment, and 10 of 19 patients (52.6%) with type I perforation had a successful initial NOM. Ten of 13 patients (76.9%) with type II perforation had a successful initial surgical treatment, and 88 of 118 patients (74.6%) with type II perforation had a successful initial NOM. Concerning type III perforations, the success rate of initial management was 100% for both surgery and NOM (3/3 and 19/19, respectively). All patients with type IV perforation (n=20) were treated with NOM, with 100% success.

Table 2 Success of initial management

Data regarding mortality are displayed in Table 3. Regarding type I perforations, the mortality rate was 15.4% (6/39) for the initial surgery group and 31.6% (6/19) for the NOM group. The mortality rate of patients with a type II perforation was 11.1% (2/18) and 8% (9/113) for the initial surgery and NOM groups, respectively. No deaths were reported for either perforation type III or IV, so the mortality rate was zero for both groups.

Table 3 Mortality rate

Quantitative analysis of the primary outcome (success of initial management) was conducted in 3 subgroups relating to the type of perforation (Fig. 2). In type I perforations, the success rate of initial surgical management was higher when compared to the NOM group (79.5% and 52.6%, respectively); however, the difference was not statistically significant (P=0.09). Type II perforations displayed a statistically significant difference between the success rate of initial surgical and NOM management, favoring management with initial surgery (P=0.02). Results for type III perforations could not be evaluated, as the success rates were 100% in both groups and all the patients with type IV perforations were managed with NOM.

Figure 2 Quantitative analysis of the primary outcome (success of initial management)

NOM, non-operative management; CI, confidence interval

Regarding secondary outcomes, there was no statistically significant difference in mortality rates between initial surgical and NOM management for any type of perforation (Supplementary Fig. 1). Similarly, there was no statistically significant difference in the length of hospital stay between the 2 groups for any type of perforation (Supplementary Fig. 2). Visual inspection of the funnel plot of the primary endpoint revealed a symmetrical distribution of the included studies (Supplementary Fig. 3).

Supplementary Figure 1 Quantitative analysis of mortality rates

NOM, non-operative management; CI, confidence interval

Supplementary Figure 2 Quantitative analysis of length of stay

NOM, non-operative management; CI, confidence interval

Supplementary Figure 3 Funnel plot of the primary outcome

SE, standard error; OR, odds ratio

Discussion

During the last decade, because of the wide availability of noninvasive imaging techniques such as MRCP and EUS, the use of the invasive modality of ERCP has been limited to the management and treatment of various pancreatic and biliary pathologies and disorders [1,17]. Even though ERCP is a common procedure nowadays, it still remains a challenging endoscopic procedure, and safe application rates require adequately high expertise, level of training, and experience on the part of the endoscopists and their supporting team. While relatively safe in experienced hands, ERCP is still associated with morbidity and complications, such as pancreatitis, bleeding, and post sphincterotomy perforation, with a reported incidence of 0.08-10% [2-3]. Moreover, regarding the perforations, which are the most dangerous and life-threatening ERCP related complication, medicolegal issues may arise [18], usually related to the provision of information and the acquisition of consent, and to the distinction between not easily avoidable/preventable procedure consequence and medical error.

In the present study, therefore, we aimed to perform a systematic review and meta-analysis of all the available literature and data regarding the initial management of perforations following ERCP, and how this is associated with either positive or negative consequences for the patient. Besides our primary endpoint (success of initial management), mortality and overall length of stay were also reviewed.

In our systematic review, we used probably the most universally used classification system by Stapfer et al [5]. In concordance with the current literature, in our study type I injuries have the smallest rate of NOM. These patients usually need surgical intervention and, as our results show, the success rate of initial management of post-ERCP perforations in type I patients was considerably higher in the group undergoing initial surgical management when compared to the NOM group (79.5% and 52.6%, respectively). Interestingly though, this difference in percentage was not statistically significant.

Type III injuries represent distal bile duct injuries related to wire or basket instrumentation in close proximity to an obstructing entity like a stone, and are often small, while type IV represent pseudo-perforations probably related to the use of compressed air to maintain patency of a lumen, and usually do not require surgical intervention [19,20]. All patients with type III and IV injuries included in this meta-analysis had a favorable outcome, whether treated conservatively or surgically.

Last but not least, type II injuries (peri-Vaterian perforation of the medial wall associated with biliary or pancreatic sphincterotomy or precut papillotomy) vary in severity and pose a clinical dilemma. In these patients, an individualized approach is necessary, based on the severity of the perforation and the patient’s clinical condition. Moreover, discrimination between type II and IV injuries is sometimes only possible intraoperatively, where after no significant perforation is found, a presumed type II injury is upgraded to type IV. This represents a certain limitation of the current classification system. Nonetheless, in type II perforations a statistically significant difference between the success rate of initial surgical and NOM management was noted, in favor of the former, thus dictating the need for an individualized approach to every case.

Regarding secondary outcomes, there was no statistically significant difference in mortality rates between initial surgical and NOM management for any type of perforation. Similarly, there was no significant difference regarding length of hospital stay between the 2 groups in any of the sub-analyses.

Although the present study is the first meta-analysis regarding the initial management of post-ERCP perforations, the results of our review should be interpreted cautiously. The limitations of the present study reflect the limitations posed by the included studies. The vast majority of the studies were retrospective, while one was a retrospective review of a prospective database. No randomized controlled trials were identified in the current literature, which is a great limitation of this study. The small number of the included studies and the heterogeneity of the included data are also a strong limitation.

In conclusion, the current meta-analysis demonstrated the significance of the initial management of patients with post-ERCP perforations. While there is no consensus as to the management of these patients, when a complication such as perforation occurs, an urgent multidisciplinary meeting should be convened to assess the patient’s data and provide a fast and valid treatment plan. Whether a surgical or endoscopic approach is chosen, the patient should immediately be evaluated by an experienced surgeon or endoscopist; if this is not possible they should be urgently referred to a hepatobiliary reference center for treatment.