Back to Annual Meeting Program

Background: Hepaticojejunostomy (HJ) leaks after pancreaticoduodenectomy (PD) are poorly characterized in the literature, in contrast to more commonly encountered complications such as pancreaticojejunostomy (PJ) leaks.
Methods: We reviewed 650 consecutive PDs performed at our institution between 2005 and 2011 and categorized patients according to whether or not they experienced an HJ leak. Leaks were identified on either transhepatic-cholangiography or an abdominal drain contrast study. Preoperative variables were analyzed to identify risk factors for an HJ leak. The clinical presentation, morbidity, and treatment plan were examined in detail.
Results: An HJ leak was identified in 14 patients (2.2%), whereas 87 patients (13.5%) in the cohort developed a PJ leak. Univariate analysis demonstrated that low preoperative albumin was the only pre- or intraoperative factor found to be associated with increased risk of HJ leaks (3.5 vs. 4.0 mg/dL no leak; p=0.001). Six of 14 patients (43%) had a preoperatively placed endostent in the common bile duct. Patients typically presented on the 6th postoperative day (range: 1 to 14 days), and in all cases a diagnosis was made prior to hospital discharge. Presenting signs and symptoms included leukocytosis (86%, median 15.3, range 6.6 to 26.1), increased abdominal pain (64%), fever (43%), failure to tolerate a diet (36%), abdominal distension (21%), and bilious drainage from the abdominal drain (21%). Thirteen of 14 patients were managed with a percutaneous intervention. Seven patients were managed with a percutaneous transhepatic biliary drain and six patients required manipulation of an intraoperatively placed surgical drain. No patients required surgical intervention. In addition to the HJ leak, patients also frequently developed a wound infection (71%), PJ leak (43%), and sepsis (29%). The median length of stay was 18 days (range: 16 to 55), as compared to 8 days in patients without an HJ leak (p=0.000). Readmission rates were 26% in the HJ leak group and 15% in the total cohort (p=NS). There was a single 90-day mortality in the HJ group (7%) as compared to 17 (2.7%) in the entire cohort (p=0.356).
Conclusions: HJ leaks are rare complications after PD and can result in substantial morbidity with increased length of hospital stay. However, early recognition with effective drainage typically results in a full recovery, without the need for surgical intervention when skilled interventional services are available. Low volume leaks are managed with effective abdominal drainage, while larger leaks may require placement of a transhepatic biliary drainage catheter.