Bile leak is among the most common complications of cholecystectomy. Endoscopic therapy is empiric; a systematic approach to management of bile leak has not been established.

The severity of bile leak was classified by endoscopic retrograde cholangiography into low grade (leak identified only after intrahepatic opacification) or high grade (leak observed before intrahepatic opacification). Therapy was based on this distinction: biliary sphincterotomy alone for low-grade leaks and stent placement for high-grade leaks. The success of this strategy in consecutive patients treated between 1989 and 1999 was reviewed.

A total of 207 patients (127 women, 80 men; median age 57 years) with bile leak were referred for endoscopic management; 134 had undergone laparoscopic, and 72 had open cholecystectomy. Patients presented at a median of 9 days (range 1-50 days) after surgery. Symptoms included pain (56%), jaundice (16%), fever (11%), and abdominal distension (7%). Persistent percutaneous drainage was present in 48%. Endoscopic retrograde cholangiography identified the leak site in 204 patients: cystic duct stump, 159 patients (78%); duct of Luschka, 26 (13%); other, 19 (9%). Of 104 patients with low-grade leaks, 75 had sphincterotomy alone; improvement occurred in 68 patients (91%). Subsequent treatment was required in 7 patients (6 stent, 1 surgery). Stents were placed in the remaining 29/104 patients for the following reasons: biliary stricture (11/29); coagulopathy, precluding sphincterotomy (8/29); severe sepsis (3/29); inadequate drainage after prior sphincterotomy (2/29); and unclear reasons (5/29). Of 100 patients with high-grade leaks, 97 had stent placement. Persistent leakage necessitated another stent insertion in 4 patients. Closure of the leak was documented by endoscopic retrograde cholangiography in all 97 patients. Three patients with leaks not amenable to endoscopic treatment were referred for surgery. Bile-duct stones were identified in 41 patients (28, low-grade group; 13, high-grade group) and were extracted in all cases. Overall, complications occurred in 3 patients (2 pancreatitis, 1 perforation) and were managed conservatively with no mortality.

A simple, practical endoscopic classification system for bile leak after cholecystectomy is proposed. This classification has clinical relevance for selection of optimal endoscopic management.