Type I perforations, which are typically 1.0-1.3 cm wide, have been successfully treated during ERCP with hemostatic clips either through the gastroscope(with or without cap assistance)[18,19] or with an endoloop[20], or with over-the-scope clips[21]. Contrast injections should be administered to rule out leakage, as 30%-60% of type I perforations fail endoscopic closure and require surgery. Patients who undergo immediate surgical correction typically remain in the hospital for 12-16 d. Prompt diagnosis is crucial, as evidenced by a study of Miller et al[11] that reported two deaths out of five cases involving type I perforations, resulting from delayed diagnosis in one case, and delayed operation in the other, and six deaths from delayed surgical repair of type II perforations that were initially managed with a conservative treatment strategy[11]. Thus, some surgeons recommend immediate diagnosis and early surgery for ERCP-related perforations[22-24]. In our endoscopic center (within a tertiary care, university-based hospital), 4082 ERCP procedures were performed between January 2009 and June 2013, with a post-ERCP perforation rate of 0.29% (n = 10 type I; n =2 type II perforation cases). All of the patients (65-91 years old) were diagnosed during or immediately upon finishing the ERCP procedures. Eighty-three percent of these cases underwent surgical correction while only 17% received conservative treatment, with no deaths.

Mao et al[25] reported on nine cases of ERCP-related perforation (mostly type II), six of which were managed conservatively with hospital stays ranging from 4 to 75 d. Subcutaneous emphysema was a significant clinical sign found in seven of the nine patients. To aid in the success of conservative treatments, which included antibiotics, endoscopic treatment (when possible), intensive observation and follow-up imaging, nasobiliary and gastrointestinal drainage was recommended to reduce the leakage of digestive juices. However, very high success rates were also reported with conservative treatment (with or without nasobiliary drainage) by Park et al[26] and Vezakis et al[27], using fully covered self-expandable metal stents for closure of perforated sites at the ampullary region.

The majority of type III perforations involve only mild contrast leakage or pneumoperitoneum without contrast leakage, and can be managed conservatively, requiring hospitalization for only 5-6 d[11,24]. Genzlinger et al[28] found that up to 29% of patients who underwent ERCP showed pneumoperitoneum on plain radiography without clinical significance. However, further investigation is recommended in patients who have pneumoperitoneum and clinical signs of infection.

The surgical indications in the studies mentioned here differed from recommendations stated by Stapfer et al[12], which include large extravasation of contrast, fluid collection in the peritoneal or retroperitoneal space on follow-up computed tomography scan, massive subcutaneous emphysema, or retained choledocholithiasis. Our recommendation is to perform salvage surgery in patients with: (1) failure of initial endoscopic closure, especially type I or type II perforations; (2) no improvement in clinical signs of sepsis, or follow-up abdominal signs worsening within 12-24 h after successful endoscopic closure (type I) or conservative management (type II); and (3) retained instruments or choledocholithiasis requiring surgical removal.

The prognosis is generally worse for any type of perforation with a delayed diagnosis, and most patients develop clinical signs of peritonitis and sepsis within 1-5 d after the procedure[26-28]. It is important to note that abdominal signs might be relatively benign within the first few hours, even for type I perforations, however the presentation of peritoneal signs as a late finding is related to a poor clinical outcome[11]. Surgical drainage and correction is recommended in cases of delayed type II diagnosis with clinical sepsis[11,21,26,27]. However, most patients (50%-90%) having type II perforations with minimal leakage or type III perforations will respond to conservative treatment.

Some patients have a higher risk of perforation because of surgically altered anatomy. Such patients should undergo “endoscopic scanning” using an end-view gastroscope before starting ERCP. In cases of failure to pass the end-view scope owing to anastomotic stricture or too much resistance, continuation with the ERCP procedure should first be discussed with the patients and their families. If the endoscopist decides to follow through with the procedure, a duodenoscope should be inserted in an “over-the-guidewire” fashion to minimize duodenal wall injury. When narrowing of the anastomosis site is encountered, balloon dilatation over the guidewire can be helpful prior to duodenoscope insertion. However, this is not recommended for cases with malignant strictures, but rather duodenal stenting with a self-expandable metal stent followed by ERCP in the subsequent weeks is more appropriate.

There are several comments and recommendations concerning the procedure that should be kept in mind: (1) Patients with periampullary diverticula have a higher risk of perforation, especially from sphincterotomy. Therefore, the endoscopist must be very careful during this process. A balloon sphincteroplasty is recommended in such cases, rather than endoscopic sphincterotomy; (2) Insertion of any ERCP instruments into the bile duct should be performed in an “over-the-guidewire” fashion to prevent a “false track.” Furthermore, the endoscopist should perform ERCP as gently as possible to minimize tissue injury; (3) Endoscopic sphincterotomy should be carried out in the suggested direction of 11 to 12 o’clock for the common bile duct and 12 to 2 o’clock for pancreatic sphincterotomy. Over-bowing of the sphincterotome should be avoided to prevent a “zipper cut,” and cutting should not go beyond the second fold above the papilla. Adequate knowledge and intensive observation of ampulla anatomy are essential for all endoscopists who perform ERCP; (4) Needle-knife sphincterotomy should be performed only by experienced endoscopists. Those who are less experienced should recognize their limitations and be responsible enough to request assistance; and (5) Dilatation of the ampulla increases the risk of perforation. Therefore, the endoscopist should consider the length of sphincterotomy or the size of balloon sphincteroplasty suitable for the objective of the maneuver, such as limited or no sphincterotomy for stenting, dilating the ampulla no more than the size of the common bile duct above, and an appropriately sized sphincterotomy depending on the size of the stone.

The endoscopist should observe the patient’s abdominal signs and perform fluoroscopy intermittently during the ERCP procedure. A diagnosis can be immediately made upon recognition of an abnormal guidewire position, extravasation of contrast, or pneumoperitoneum, which can allow for prompt treatment. Furthermore, a fluoroscopic abdominal scan should be performed after any difficult procedure to identify possible complications.

Patients should be encouraged to promptly report any abdominal symptoms and should be watched for abnormal clinical signs after the ERCP procedure. The ward staff should be notified to maintain special observation of patients with difficult cases for early detection of possible complications.