Bile leaks that may result from cholecystectomy have been described by McKenzie [1] and by Foster and Wayson [2] and are caused by a slipped cystic duct ligature or a leak from an accessory or anomalous bile duct; those resulting from hepatic surgery are caused by a leak from a biliary anastomotic site, dislodgement or removal of an external drainage tube, or damage to a bile duct during surgery or trauma. Bile leaks are most commonly associated with hepatobiliary surgeries or invasive procedures such as open or laparoscopic cholecystectomy, hepatic resection, hepatic transplantation, liver biopsy, and percutaneous transhepatic cholangiography.

Both open and laparoscopic cholecystectomy can be complicated by bile leaks (Figs. 1A, 1B, 1C and 2A, 2B, 2C, 2D) from unrecognized inadvertent damage to the normal bile duct during surgery. Up to 30% of the population may have anomalies of the union of the intrahepatic bile ducts or cystic duct with the common hepatic duct and gallbladder (Fig. 3A, 3B, 3C, 3D) that may predispose them to bile duct injury during surgery [3]. Identifying variations in intrahepatic ductal anatomy during hepatic resection and liver transplantation is also important to avoid bile duct complications. Bile leaks after transplantation may occur at the site of biliary anastomosis or T-tube placement and may necessitate surgical revision of the anastomosis (Fig. 4A, 4B, 4C). Bile leaks also may be caused by nonoperative circumstances such as blunt or penetrating trauma (Figs. 5A, 5B, 5C and 6A, 6B).

Bile leaks from postoperative or blunt abdominal trauma that result in intraperitoneal collections of bile are typically not contaminated by bacteria and usually do not result in severe bile peritonitis [4]. In contrast, intraperitoneal collections after acute cholecystitis are usually infected. Peritoneal culture is positive in 75% of patients, with Escherichia coli being the most frequently isolated organism. As expected, the outcome is worse than for trauma patients.

The incidence of bile leaks after liver transplantation has been reported to be 4.3% [5]; after major hepatic resection, 3–11% [6]; and after laparoscopic cholecystectomy, approximately 0.35–0.47% [7]. Usually several days pass before a leak is diagnosed [4, 5] because symptoms are nonspecific and could derive from other postoperative complications (Fig. 7). Patients frequently undergo some form of cross-sectional imaging such as CT or sonography, and altough findings may be suggestive of bile leak they cannot reliably distinguish bile from other postoperative fluid collections such as pus, serous fluid, or blood (Fig. 2A, 2B, 2C, 2D).

CT is excellent for showing the site and morphology of a fluid collection. In our experience, on CT, bile collections may be loculate, multiloculate (Fig. 8), or diffuse in the peritoneal cavity (Fig. 1A, 1B, 1C). Bile collections are usually close to the site of the leak, but occasionally they may be remote or may even be intrahepatic (Fig. 9). If a fluid collection is identified on CT, further investigation may be undertaken to confirm and treat the collection. If a T-tube is present, a cholangiogram can usually reveal the presence and site of the leak. However, if a T-tube is not present, more invasive examinations such as ERCP, percutaneous cholangiography (Fig. 10A, 10B), or imaging-guided percutaneous drainage can confirm the diagnosis and potentially treat the bile leak.

Hepatobiliary scintigraphy has the advantage of presenting the physiologic course of biliary excretion and showing the actual leakage of bile from the biliary tract (Fig. 5A, 5B, 5C). This technique, however, is limited by poor spatial resolution that makes identification of the site of leak difficult, and CT correlation may be required, especially if the leak is small.

Mangafodipir trisodium–enhanced MR cholangiography is a recently described alternative approach that combines anatomic and functional imaging of the biliary tract [8]. Mangafodipir trisodium is an MR contrast agent comprising a water-soluble chelate complex salt between a paramagnetic manganese ion (II) and the ligand dipyridoxyl diphosphate, a vitamin B₆ analogue. Mangafodipir begins to increase the signal intensity of the liver 1–3 min after IV injection and reaches steady-state enhancement in approximately 5–10 min. Scanning is initiated 5 min after the injection, and volumetric 3D T1-weighted axial and coronal or oblique coronal images are obtained every 10–15 min until a bile leak becomes apparent (Fig. 11A, 11B, 11C). Mangafodipir trisodium–enhanced MR cholangiography combines the advantages of hepatobiliary scintigraphy and multiplanar cross-sectional imaging with the functional evaluation of biliary excretion and excellent spatial resolution and anatomic depiction (Figs. 12A, 12B, 12C and 13A, 13B) of the biliary tree.

Detecting and locating bile leaks is difficult. It is important to differentiate bile leaks from abscesses because bile leaks can often be treated by temporary stenting. Cross-sectional imaging may be a reasonable first step for detecting postoperative complications, but documenting bile leaks often requires techniques that directly visualize biliary excretion from the injured bile ducts. Mangafodipir trisodium–enhanced MR cholangiography can be helpful in such instances and provide accurate anatomic and functional information that allows prompt diagnosis and treatment of bile leaks.

We thank Eric Jablonowski for his help with the illustrations in Figure 3A, 3B, 3C, 3D.

Address correspondence to V. Kapoor.