Endoscopic management of major bile leaks complicating hepatic resections for hepatocellular carcinoma

Abstract

Nine patients with postoperative biliary leaks are presented. These were selected from 128 adult patients with hepatocellular carcinoma who underwent hepatic resections at our hospital between March 1990 and July 1995. Types of hepatic resections as defined by Couinaud7 included nonanatomic resection (40), anatomic resection of 3 or less segments (25), left lateral segmentectomy (6), right or left hepatectomy (41), extended hepatectomy (12), and trisegmentectomy (4). Of the 9 patients (7%) who developed bile leaks postoperatively, there were 7 men and 2 women with median age of 49 years (range 35 to 71 years) (Table 1). Types of resections in these patients were right hepatectomy (2), left hepatectomy (2), left extended hepatectomy (2), right extended hepatectomy (1), right hepatectomy with caudate lobe excision (1), and nonanatomic resection (1). One patient who presented with obstructive jaundice from a left lobe tumor invading the common hepatic duct required reconstruction of the common hepatic duct in addition to left hepatectomy. The incidence of bile leaks after extended hepatectomy was significantly higher (4 in 16 vs. 5 in 112, p = 0.01, Fisher’s exact test). The median tumor size was 8 cm (range 3 to 10 cm). Resection margins were free of tumor in all patients. All 9 patients except 1 had cirrhosis confirmed at laparotomy. ERCP was performed at a median of 19 days (range 11 to 80 days) from surgery. The indications for referral were recurrent infected bilomas in 5 patients, high output fistula (greater than 500 mL daily from subhepatic drains) in 2 patients and nonhealing fistula beyond 2 weeks in 2 patients. Radiologic-guided percutaneous aspirations and pigtail catheter drainage (median 3 sessions, range 1 to 5 sessions) were required in patients with infected bilomas prior to ERCP. The volume of aspirates from bilomas varied from 100 mL to 1.5 L. At ERCP in 1 patient with a previous Billroth II gastrectomy, the papilla could not be accessed because of a redundant afferent limb. He was managed conservatively and his bilocutaneous fistula was healed 2 months later. In another patient, deep cannulation of common bile duct was not achieved. Percutaneous transhepatic biliary drainage was not attempted because of a coagulopathy. She subsequently died of sepsis and liver failure. ERCP was successful in 7 patients. Endoscopic sphincterotomy was performed in 2 patients in whom deep cannulation of the bile duct had been difficult. Bile leaks were demonstrated in all patients to be at or proximal to the liver hilum at segmental branches of intrahepatic ducts. In 5 patients without associated strictures, 10F stents were inserted across the sphincter with the proximal ends of the stents near the site of leakage (Figure 1). A nasobiliary drain was initially placed in 1 patient and subsequently changed to a stent following drying of the fistula. After stent insertion, a significant reduction in fistula output was noted in all patients. Fistulas healed in a median of 11 days (range 9 to 47 days). The effect of stenting was best appreciated in patients with high fistula output. The output often fell from 1.5 L to 300 mL on the day after stenting. Two patients had associated strictures with proximal fistulation, one at a reconstructed common hepatic duct and the other from a segment II duct. In the patient with a common hepatic duct stricture, a stent was inserted across an anastomotic stricture proximal to the site of leakage after balloon dilation of the stricture. Distal migration below the stricture occurred twice leading to recollection of bilomas. Longer stents were placed on two occasions. The fistula took 31 days to heal, 18 days from the last stent placement. In the other patient with a tight segmental II duct stricture following right trisegmentectomy, a nasobiliary drain was initially placed. This subsequently became dislodged. At a second ERCP, the stricture was found to be resistant to balloon dilation and a 6.5F stent was inserted across the stricture. Because of the poor bile duct drainage, healing of fistula was slow. The patient was left with a persistent discharge sinus that finally healed after 110 days. Stents were removed in all cases from 6 to 8 weeks after fistula healing. The duration of follow-up in these patients ranged from 2 to 69 months. No median-term complication was noted. Two patients had died of tumor recurrence, one at 2 and the other at 4 months from surgery.