Clinical characteristics and overall management outcome
Over the 11-year period, 8,381 ERCP procedures were undertaken and 53 (0.63%) ERCP-related perforations occurred. Among the 53 patients, 22 (41.5%) were male and 31 (58.5%) were female. Mean age was 66.7 ± 1.6 years (range, 21 to 91 years).
The most frequent diagnosis was biliary stones (n = 30), followed by periampullary cancer (n = 6) and Klatskin tumor (n = 4). There were 3 cases each of gallbladder cancer, advanced gastric cancer, and benign biliary stricture. Other diagnoses included one case each of intraductal papillary mucinous neoplasm, biliary cystadenocarcinoma, hepatic cyst, and duodenal carcinoid tumor.
Diagnostic ERCP was performed in 8 patients (15.1%), and therapeutic ERCP in 45 (84.9%). Diagnostic procedures were either biopsy or cholangiography, and therapeutic procedures included endoscopic sphincterotomy, endoscopic naso-biliary drainage, stent insertion, stone removal, and endoscopic mucosal resection.
Type I injuries occurred in 39.6% (21 patients) of the cases. Type I patients included 6 with jejunal insult in Billroth II anatomy. Type II injuries occurred in 60.4% (32 patients). Of the type II, 24 patients received periampullary injury, mostly related to endoscopic sphincterotomy. The remaining 8 type II patients had bile duct injuries: 6 with CBD injuries mostly as result of balloon dilatation or stone removal, 1 with guidewire-related left intrahepatic duct stump injury (that patient previously received hepatic surgery), and 1 with a guidewire-related injury to a choledochoduodenal fistula.
Most of the injuries (39 cases, 73.6%) were immediately detected during the ERCP procedure, but 14 cases (26.4%) had delayed diagnoses. The mean delay time until detection was 28.4 ± 5.7 hours (range, 1 to 55.2 hours). Of the delayed patients, 12 experienced severe postprocedural abdominal pain, with or without fever, which warranted radiologic examinations. Perforation was confirmed by computed tomography (CT) in 11 patients and by plain abdominal radiography in 1 patient (). The remaining 2 cases were incidentally detected during follow-up CT and plain abdominal radiography, each.
Fig. 1 Radiologic findings of endoscopic retrograde cholangiopancreatography perforations are demonstrated. (A) Computed tomography shows pneumoperitoneum, pneumoretroperitoneum, and fluid collection adjacent to duodenum. (B) Massive bilateral subdiaphragmatic (more ...)
The mean hospital stay was 23.7 ± 2.7 days (range, 0 to 86 days) and the mean hospital cost was 13.7 ± 2.1 thousand United States Dollars (USD) (range, 2.3 to 80.4 thousand USD). The morbidity and mortality rates were 34.0% (18/53) and 5.7% (3/53), respectively.
Clinical outcome comparison of type I and type II injuries
The demographics and procedure/injury related data for the 21 type I and 32 type II patients did not show significant differences; however, management method was significantly different. Type I injuries were mainly managed operatively (17/21, 81.0%) because a type I injury without possible adhesion or intervention was the operative indication at Seoul National University Hospital. Nearly all type II injuries were managed conservatively (31/32, 96.9%). The operative indication for the surgically managed type II patient was a large retained, impacted CBD stone with stricture distally. The stone could not be removed due to the stricture, even after balloon dilatation. Free air was detected after the procedure, and the patient was referred for surgery.
Comparative analyses of management outcomes demonstrated superior results in type II patients over type I (). Hospital stay was significantly shorter (19.3 ± 2.8 days vs. 30.6 ± 5.0 days, P = 0.010) and hospital cost was significantly less (9.5 ± 1.1 thousand USD vs. 20.1 ± 4.6 thousand USD, P = 0.028) for type II patients compared to type I patients. The morbidity rate was also significantly lower in type II patients (18.8% vs. 57.1%, P = .004). There was no difference in mortality.
Comparison of results between according to injury types and management methods
Within the type II group, analysis comparing 24 periampullary injuries and 8 bile duct injuries was done. Demographic and procedure/injury related data were similar in both groups. Treatment outcomes were better in bile duct injury patients with shorter hospital stay (11.1 ± 2.4 days vs. 22.0 ± 3.4 days, P = 0.150), less cost (7.0 ± 2.4 thousand USD vs. 10.4 ± 3.4 thousand USD, P = 0.372), lower morbidity (12.5% vs. 20.8%), and null mortality (0.0% vs. 4.2%). However, none of these differences were significant.
Comparison of clinical outcome according to management type
Thirty-five patients received conservative management and 18 patients received operative management. No significant differences were found demographically or with respect to procedure and to detection delay between the management groups. However, between the groups, injury type and mean delayed detection time were significantly different (). In the conservative group, 88.6% (31/35) were type II injuries whereas in the operative group 94.4% (17/18) were type I (P < 0.001). Moreover, the mean delayed detection time was longer in the conservative group (36.0 ± 6.8 hours vs. 14.8 ± 7.7 hours, P = 0.031).
With regard to management outcomes, the mean hospital stay was shorter (20.6 ± 2.6 vs. days 29.8 ± 5.8 days, P = 0.092) and the mean hospital cost was lower (10.6 ± 1.3 thousand USD vs. 19.9 ± 5.3 thousand USD, P = 0.095) in the conservative group than in the operative group (). The conservative group suffered significantly less morbidity (22.9% vs. 55.6%, P = 0.017). The mortality rates were similar: 5.7% (1/18) and 5.6% (2/35) in the conservative and operative groups, respectively. Marginal significance was observed in hospital stay and hospital cost.
According to the surgical indication approach of our institution, type I injuries need surgery. However, 4 of 21 type I patients were not managed in accord with that guideline. Of those, there was 1 mortality case and 1 failed and 2 successful conservative management cases (). One of the successful cases involved endoscopic clipping (). The reason for the conservative management attempts in mainly Billroth II anatomy patients was the possibility of perforation sites localized with adhesions from previous surgery. There were not any specific indications in predicting the possibility of spontaneous sealing, and this prediction was based on the clinician's decision.
Summary of conservative management in four type I injury cases
Perforation site endoclipping is presented. The patient was conservatively managed and discharged uneventfully. (A) The perforation site at posterior wall of duodenal bulb is identified. (B) Perforation site is successfully sealed with endoclips.
The outcomes of conservatively managed type I injuries were similar with those of the operatively managed type I injuries in terms of hospital stay (28.5 ± 6.9 days vs. 31.1 ± 6.0 days, P = 0.658), cost (18.1 ± 6.6 thousand USD vs. 20.6 ± 5.5 thousand USD, P = 0.929), morbidity (50.0% vs. 58.8%, P = 1.000), and mortality (25.0% vs. 5.9%, P = 0.352).
Additionally, although statistical significance was not demonstrated, conservatively managed type I injuries had longer hospital stay (28.5 ± 6.9 days vs. 19.6 ± 2.8 days, P = 0.153), higher cost (18.1 ± 6.6 thousand USD vs. 9.6 ± 1.2 thousand USD, P = 0.265), higher morbidity rate (50.0% vs. 19.4%, P = 0.218), and higher mortality rate (25.0% vs. 3.2%, P = 0.218) than type II injuries.
Morbidities and mortalities
Of 53 patients, 18 (34.0%) experienced complication(s) during management. There were 31 complications: 17 in the surgical group and 14 in the conservative group. In the surgical group, surgery-related complications such as wound complications (n = 5) and repair site leakage (n = 3) were common. In conservative group, intra-abdominal fluid collection requiring percutaneous drainage (n = 5) and sepsis (n = 3) were most frequent.
There were 3 mortalities. One was a 75-year-old male who underwent ERCP for evaluation and stent insertion of CBD stricture. The patient had underlying Child-Pugh class C liver cirrhosis and Billroth II anatomy. Microperforation distal to the ampulla of Vater (type I injury) was detected during ERCP. The patient's condition aggravated, even after percutaneous drainage. Operation was not possible due to severe liver cirrhosis. Sepsis occurred and he expired after 17 days.
A 65-year-old female with extrahepatic duct obstruction, due to gallbladder cancer with liver metastasis, underwent ERCP for stent replacement. Perforation of the duodenum first portion was immediately detected, and primary repair with palliative cholecystectomy was done. The patient recovered relatively well, but expired from the terminal state of gallbladder cancer 66 days postoperatively.
The third mortality was an 85 year-old male with Klatskin tumor. ERCP was done for stent insertion and a type II injury was discovered 2 days later. The patient was recovering well with conservative management, but expired of cardiovascular accident 23 days later.