Bilomas, infected hepatic fluid collections, are a frequent complication of liver transplantation. We report a case-control cohort study to determine the incidence and microbiologic profile of bilomas and risk factors for biloma formation in 492 patients undergoing liver transplantation from 1994 to 2001. Fifty-seven patients (11.5%) developed one or more bilomas; 95% in the first year post-transplantation. The most common initial infecting pathogens were enterococci (37%), one-half resistant to vancomycin (VRE); coagulase-negative staphylococci (26%); and Candida species (26%). Infection by coagulase-negative staphylococci was strongly associated with the presence of a T-tube (OR 9.60, p = 0.02). In stepwise logistic regression multivariable analyses, hepatic artery thrombosis (OR 90.9, p < 0.0001), hepatic artery stenosis (OR 13.2, p < 0.0001) and Roux-en-Y choledochojejunostomy (OR 5.8, p = 0.03) were independent risk factors for biloma formation; ursodeoxycholic acid use was highly protective (OR 0.1, p = 0.002). Strategies to prevent biloma formation must focus on measures to prevent hepatic artery thrombosis and colonization of liver transplant patients by multiresistant nosocomial pathogens. T-tube drainage post-transplantation bears reassessment. The protective effect of ursodeoxycholic acid found in this study warrants confirmation in a prospective multicenter, randomized trial.
Major advances in transplantation biology, organ procurement and preservation, surgical techniques and immunosuppressive therapy over the past two decades have made liver transplantation a highly effective and cost-beneficial option for the management of end-stage liver disease (1–4), with a 1-year survival exceeding 80% in most centers (5). However, infection, especially involving the biliary tract, remains one of the most challenging complications (6,7). Bilomas, infected intra or parahepatic bilious fluid collections (8), are associated with frequent graft loss and necessity for retransplantation, greatly increased medical costs and significant mortality (9–12).
The epidemiology of biloma formation after transplantation has not been characterized (10,13–15). We report a case-control study, nested within a retrospective cohort, undertaken to determine the incidence and microbiology of post-transplant bilomas at a major transplant center, and risk factors predisposing to this serious complication, which point up potential strategies for prevention.
Sources of data
Using data from a prospectively maintained database, we identified all patients undergoing liver transplantation at the University of Wisconsin between January 1, 1994 and October 31, 2001, who developed one or more post-transplant bilomas. A control group was constructed by randomly selecting two uninfected liver transplant recipients (controls) for each biloma patient, matched by year of transplantation.
Donor and recipient characteristics were extracted from the database, including potential risk factors, operative factors, medications used – particularly immunosuppressive drugs and anti-infectives – biloma location, size and microbiology, hepatic artery patency and concurrent biliary abnormalities, and outcome (resolution with nonsurgical management or following re-transplantation, or death). All cases and control patients were transplanted and followed at our institution; none was lost to follow up.
All liver transplants were performed using the piggyback surgical technique; venovenous bypass was used in only 33 patients. In most patients (72%), a duct-to-duct biliary anastomosis was constructed; percutaneous biliary drainage (T tube) was established at surgery in 82%. In general, T tubes were removed 3–4 months after transplantation. Ceftriaxone was given preoperatively and for 48 h following transplantation; selective digestive decontamination was not employed. The standard immunosuppressive protocol consisted of tacrolimus (Prograf, Fujisawa) or cyclosporine, and prednisone. In cases complicated by renal failure, calcineurin inhibitors were withheld and mycophenolate mofetil (Cellcept, Roche) or basiliximab (Simulect, Novartis) was substituted. Acute cellular graft rejection was treated with intravenous pulses of methylprednisolone, increased doses of calcineurin inhibitors and, when necessary, either antilymphocyte globulin (Thymoglobulin) or OKT3 monoclonal antibody (Orthoclone, Ortho-Biotech) (16). An abdominal sonographic examination was performed routinely on the first postoperative day to assess hepatic artery patency; subsequent examinations were obtained as indicated clinically.
Hepatic artery thrombosis: Lack of arterial flow by Duplex ultrasound, with total occlusion confirmed by angiography in most cases.
Hepatic artery stenosis: Reduced flow, as shown by increased velocities on Duplex ultrasonography, with partial occlusion confirmed angiographically in most cases.
Biloma: An intrahepatic (Figure 1) or parahepatic fluid collection with typical appearance on ultrasonography, computerized tomography or cholangiography. Parahepatic collections were typically associated with extrahepatic bile duct leaks, most often anastomotic leaks. Intrahepatic and parahepatic bilious collections associated with concurrent biliary abnormalities, such as strictures, were included. Four parahepatic collections detected during the first 2 weeks following transplantation, all of which were infected hematomas, were excluded.
The following potential risk factors were examined in univariate analyses:
Recipient: Age at transplantation, gender, race, model for endstage liver disease (MELD) score, Child's Score, etiology of liver disease, previous liver transplantation, high-risk transplantation with respect to cytomegalovirus (CMV) status (CMV+ donor and CMV– recipient), pretransplant comorbid conditions including diabetes mellitus, coronary artery disease, chronic pulmonary disease, deep vein thrombosis, pulmonary embolism, coagulation abnormalities, peripheral vascular disease, renal insufficiency (serum creatinine >1.8 mg/dL), hepatocellular carcinoma, tobacco use and alcohol abuse post-transplantation.
Donor: Age at donation, gender, race, ABO and CMV status.
Surgical and peri-operative factors: Cold ischemia time, duration of surgery, type of biliary anastomosis, hepatic artery reconstruction, and T-tube duration,volume of all intraoperative blood products given.
Postoperative: Biliary tree abnormalities, hepatic artery thrombosis (HAT) and stenosis, HAT management (surgical, angioplasty or other), ursodeoxycholic acid use, anticoagulation (aspirin, clopidrogel, warfarin), other infection, and immunosuppression (maintenance, induction, rejection), number of episodes of acute cellular rejection (total in the first month and first year).
Univariate analysis of potential risk factors for biloma formation was undertaken using Chi-square or Fisher's exact test for categorical variables and Students t-test for continuous variables. Risk factors with a p-value < 0.1 in univariate analyses were entered into a stepwise logistic regression model, and those with p-values < 0.05 were retained. Two-tailed p-values < 0.05 were considered significant. Goodness-of-fit of the final model was tested by the Hosmer-Lemeshow test (17). All statistical analyses were performed with SAS software (Version 8.1, 2000, Cary, NC).
Of 492 patients who underwent primary liver transplantation between January, 1994 and October, 2001, 57 (11%) developed one or more bilomas; 95% occurred in the first year following transplantation. Ninety-eight percent of the entire study population (57 cases, 114 control patients) had undergone orthotopic deceased-donor liver transplantation; only two received a living-related-donor liver transplant.
The median period of follow up after transplantation was 199 weeks (3.8 years), with a minimum follow up of 1 year for patients alive at least 1 year post transplantation. Median follow up after biloma resolution was 143 weeks (2.7 years). No patients were lost to follow up.
A total of 102 bilomas were detected in the 57 liver transplant recipients (Table 1); an average of 1.8 fluid collections per subject. The median time to diagnosis was 9.1 weeks after transplantation, with a range of 2–377 weeks; bilomas were detected within 4 weeks after transplantation in 16 patients (28%), between four weeks and 1 year in 33 (58%), and more than 1 year following transplantation in only eight patients (14%). Sixty-eight bilomas (67%) were intrahepatic and 34 (33%) were parahepatic (subhepatic, perihepatic, in the region of the porta hepatis). The size of collections ranged from 1 to 11 cm, with an average of 5 cm.
Table 1. Characteristics of bilomas in 57 liver transplant patients
Time to diagnosis following transplantation,
weeks, median (range)
Location of biloma
Number of bilomas per patient, mean (range)
Size of bilomas, cm, mean (range)
The initial percutaneous aspirates yielded 87 distinct microbial isolates (Table 2). In six patients the aspirate did not show growth; in each case, however, the patient had received antimicrobial therapy before the diagnostic aspirate, and the gram-stain revealed bacteria. The most common primary infecting pathogens, based on the initial percutaneous aspirate of the biloma, were enterococci (37% of the 57 patients), 48% exhibiting resistance to vancomycin (VRE); coagulase-negative staphylococci (26%); and Candida species (26%). Gram-negative organisms were identified in only 16% of cases, one-half identified as Pseudomonas species. Anaerobic bacteria were recovered from only 5% of patients.
Table 2. Microbial profile of bilomas as found in initial percutaneous aspirate and in later cultures
Initial aspirate no. of patients (%)
Later cultures no. of patients (%)
At some time no. of patients (%)
1Other organisms included Streptococcus viridans, lactobacillus, Proprionibacterium acnes, Staphylococcus aureus, B acteroides fragilis.
Enteric gram-negative bacilli
No growth but gram stain showed microorganisms
Presence of infection elsewhere, no.,%
One or more new superinfecting pathogens (156 isolates) were later cultured from the percutaneous drainage catheter in 95% of cases (Table 2). A much larger proportion of the organisms recovered from later cultures were multidrug resistant, including VRE (44% of patients), Candida spp. (79%) and resistant gram-negative bacteria (35%): Citrobacter freundii, Stenotrophomonas maltophilia and extended-spectrum-beta-lactamase-producing Escherichia coli.
Blood cultures drawn at the time of biloma detection showed concordant bacteremia in nine patients (16%), with the bloodstream isolate matching at least one species isolated from the biloma. Thirteen (23%) patients later had concordant bacteremia. Overall, 15 (26%) patients had secondary bacteremia or candidemia at some time deriving from their biloma.
Isolation of coagulase-negative staphylococci was strongly associated with the presence of a T-tube (OR 9.60, p = 0.02), whereas none of the other species that recovered showed an association with T-tube placement. Coagulase-negative staphylococci were isolated from 15 (38%) of 40 patients with a T-tube and biloma.
A substantial proportion (25%) of cohort subjects had one or more microbiologically documented infections at an extrahepatic site within 2 months preceding biloma detection. Most (nine of 14) were central venous catheter-related bloodstream infections.
The baseline characteristics of patients with bilomas and control patients are compared in Table 3. The mean age was 49 years in both groups, and the majority were men. Although most of the study population was Caucasian, the proportion was significantly lower in the cases (84%) compared with the control patients (94%, p = 0.04). Pre-transplant MELD scores were similar. Etiology of liver disease also was similar, with alcoholic liver disease, hepatitis C or both accounting for more than one-half of all types of liver disease in each group. Diabetes mellitus was more common in patients developing biloma (p = 0.04). No significant differences were seen in the prevalence of other comorbid conditions. Although hepatocellular carcinoma was more common in biloma cases (10%) than control patients (5%), the difference was not statistically significant.
Table 3. Baseline characteristics of patients with bilomas (cases) and control patients
Cases (n = 57)
Controls (n = 114)
MELD, model for endstage liver disease; SD, standard deviation. D+/R–, donor-positive/recipient-negative; D–/R +, donor-negative/recipient-positive; D–/R–, donor-negative/recipient-negative; D+/R +, donor-positive/recipient-positive.
Age at transplant, year, mean + SD
49.2 + 1.4
49.5 + 0.9
Male, no. (%)
MELD score, mean + SD
16.8 + 1.1
17.2 + 0.8
Etiology of liver disease, no (%)
Alcoholic liver disease
Alcoholic liver disease with hepatitis C
Comorbidities, no (%)
Cytomegalovirus status, no (%)
Risk factors for biloma formation were analyzed and the results of the univariate analysis are shown in Table 4. Non-Caucasian race (OR 3.45, p = 0.04), diabetes (OR 2.26, p = 0.04), HAT (OR 17.45, p < 0.0001), hepatic artery stenosis (OR 7.27, p < 0.0001), portal vein thrombosis (OR–6.69, p = 0.003) and a choledochojejunostomy (OR 2.53, p = 0.02), as contrasted with a duct-to-duct anastomosis, were associated with a significantly higher risk of biloma formation; in contrast, ursodeoxycholic acid use was associated with greatly reduced risk (OR 0.19, p = 0.0006).
Table 4. Risk factors for biloma formation by univariate analysis
OR (95% CI)
Hepatic artery thrombosis
Portal vein thrombosis
Cytomegalovirus status of the donor and recipient was not found to be a significant risk factor. Intraoperative and perioperative factors, including cold ischemia time, arterial reconstruction, duration of surgery and T-tube use also were not associated with increased risk, and the frequency of rejection or types of immunosuppressive regimens used did not differ between the biloma and control patients. Aspirin use was not protective and, other than diabetes, there were no differences in preexistent comorbid conditions, such as renal disease, cardio-pulmonary disease or coagulopathy, between the biloma cases and the controls. Smoking before or following transplantation also did not increase risk.
In multivariable analyses (Table 5), HAT (OR 90.9, 6.1–500.0, p < 0.0001) and hepatic artery stenosis (OR 13.1, 95% CI 3.2–52.6, p < 0.0001) were powerful independent risk factors predisposing to biloma formation. A Roux-en-Y choledochojejunostomy anastomosis also appeared to greatly increase risk (OR 5.8, 95% CI 1.2–27.7, p = 0.03). In contrast, use of ursodeoxycholic acid was highly protective (OR 0.1, 95% CI 0.0–0.4, p = 0.002). Other potential risk factors, such as race, diabetes and portal vein thrombosis, did not remain in the multivariable model. The Pearson Chi-square test of the final model was 0.93, indicating a robust model. The model also had good discriminant ability, as shown by an area under the receiver operating characteristic curve of 0.90.
Table 5. Risk factors for biloma formation by multivariable analysis
Adjusted OR 95% (CI)
Goodness-of-fit test by Pearson's Chi-square, p = 0.73; receiver operating characteristic curve c-statistic, 0.90.
Hepatic artery thrombosis
Hepatic artery stenosis
Biloma formation following liver transplantation is a major complication which is associated with substantial morbidity and mortality. While prior studies have found biliary complications occur in 7% to 31% in patients undergoing liver transplantation (18–20), the incidence of biloma has not been well defined. In a retrospective study of 2175 solid-organ transplant patients from 1990 to 2000, Tachopuoulou et al. reported 12 (2.6%) of 459 liver transplant patients with hepatic abscess (10). Their definition of infected fluid collection was limited to parenchymal lesions, while we included both intra and extrahepatic biliary fluid collections. Our analysis shows that 11.5% of all patients undergoing liver transplantation at our center developed one or more infected parahepatic biliary fluid collections. The outcome of parahepatic biloma did not differ significantly from that of the intrahepatic cases (21).
Our analysis shows that the microbiology of postliver transplantation biloma differs strikingly from garden-variety liver abscess (22): anaerobes and enteric gram-negative bacilli such as E. coli were infrequent primary pathogens. In contrast, multiresistant gram-positive organisms (VRE or coagulase-negative staphylococci) were recovered from the initial percutaneous aspirate in 44% of the patients, and the isolation of coagulase-negative staphylococci was strongly associated with the presence of a T-tube, indicating that T-tubes provide an important route for organisms colonizing the skin to gain access to the biliary tree, later infecting vulnerable ischemic areas of liver. The next most frequent infecting pathogens were Candida species (26% of patients) and gram-negative bacilli (16%): half of which were Pseudomonas species.
Analysis of follow-up catheter drainage cultures showed that a much greater proportion of the superinfecting pathogens were multiresistant, including VRE (83% of enterococcal isolates), fluconazole-resistant Candida glabrata (79%), and multiresistant nosocomial gram-negative bacilli, such as pseudomonads (35%). All of these patients had received prolonged and intensive anti-infective therapy in the past and for treatment of their biloma, and antimicrobial pressure is almost certainly responsible for the greatly increased resistance observed (23). However, most of the infecting microorganisms, which are leading nosocomial pathogens in major transplant centers, were probably acquired during hospitalization prior to transplantation. The importance of strategies to prevent nosocomial colonization and infection in vulnerable liver transplant candidates, including the use of barrier isolation precautions, cannot be overemphasized (24).
Understanding the pathogenesis of biloma formation in liver transplantation is essential to devising effective strategies for prevention (Figure 2). Previous studies that attempted to identify risk factors for biloma formation were limited by relatively few cases (13,14), lack of a control group (13,14,25,26) or restricting the study population to patients with HAT (10–12,27), precluding analysis of HAT as a risk factor for biloma formation.
To our knowledge, this is the first study to examine risk factors for biloma formation using a prospectively maintained database and multivariable analysis. The most important risk factor identified was HAT, associated with 90-fold increased risk (Table 5); hepatic artery stenosis, the precursor to HAT, was found to be a commensurately powerful risk factor.
Hepatic artery thrombosis occurred in 1.6–10.9% of cases in other series of adult liver transplantation (28–33) and carries a mortality of 13% to 58% (28,30,32). Several studies have shown that HAT is followed in most cases by ischemic necrosis, which in turn is complicated by secondary infection, vis-à-vis, biloma or abscess formation (28,34). As the transplanted liver is devoid of hepato-portal collaterals and the blood supply to the biliary system is dependent upon hepatic artery flow, bile duct injury from insufficient perfusion presumably leads to leakage of bile from damaged intrahepatic ducts and formation of a collection. If the bile is already colonized, which more often than not is the case, the collection becomes infected at the outset, but if the bile is yet sterile, the collection can later become infected hematogenously (Figure 2).
Risk factors for HAT have been described in previous reports and include operative factors, particularly re-construction of the hepatic artery, or use of an old arterial conduit for the hepatic artery anastomosis; a CMV+ donor–CMV recipient transplant; early rejection; advanced donor age; and female-organ to male-recipient transplantation (12,33–35). Damage to vascular endothelium during surgery may also predispose to HAT by providing a locus for platelet aggregation. Postoperative or other thrombophilic states may further augment this process (36,37). This has led to interest in early postoperative anticoagulation to prevent or treat HAT (30). In an uncontrolled study of 24 pediatric patients undergoing living-related liver transplantation, Hashikura et al. reported that anticoagulation following transplantation prevented HAT in all recipients (38); the regimen consisted of low molecular-weight heparin, antithrombin III concentrates, prostaglandin E, fresh frozen plasma, and a protease inhibitor. Another retrospective study of 529 adult patients, 354 of whom received low-dose (81 mg) aspirin daily beginning immediately after transplantation, failed to show a reduced incidence of HAT in the aspirin treatment group (3.7% vs. 4.0%, p = 0.85; 39). We do not believe the available evidence supports the use of antiplatelet drugs or anticoagulation for the prevention of HAT at this time, although therapeutic modulation of coagulation post-transplantation bears further study.
Biliary injury and associated strictures, in the absence of HAT or stenosis, have also been linked to cold ischemia time, ABO incompatibility, chronic rejection and low-flow states, as well as postoperative hypercoagulability (40–42). Stenting the bile duct with a T-tube after transplantation has been advocated to prevent biliary leaks and strictures and allow access to the biliary tree. However, T-tubes do not reduce the risk of biliary stricture formation or anastomotic leakage (43–46). Moreover, studies have shown that T-tubes increase the risk of biliary colonization and cholangitis (47–50). In our study, use of a T-tube was strongly associated with biloma infection by coagulase-negative staphylococci, most resistant to methicillin and other beta-lactams. T-tube drainage after hepatic transplantation bears critical reassessment.
In the present study a choledochojejunostomy was strongly associated with biloma formation. Prior studies have shown that choledochojejunostomy is associated with a much higher risk of cholangitis following transplantation (51), probably because luminal gut organisms reflux into the biliary tract far more frequently with a gut-to-duct anastomosis than a duct-to-duct anastomosis, where the recipient's ampullary sphincter is intact. Postoperative cholangitis is also known to be far more frequent following choledochojejunostomy in nontransplant hepatobiliary operations (52). Recurrent subclinical cholangitis may be the main mechanism by which sterile intrahepatic bilomas become secondarily infected, even in the absence of HAT.
In our study, ursodeoxycholic acid was associated with a greatly reduced risk of biloma formation (OR 0.1, 95% CI 0.0–0.4). The bile flow-enhancing (choleretic) actions of ursodeoxycholic acid (53,54) may reduce the concentration of hydrophobic bile acids (55) and limit damage to the biliary tree as well as the hepatic parenchyma if there has been ductal injury from ischemia. Ursodeoxycholic acid also has immunomodulatory effects and reduces the expression of MHC type II antigens on the biliary tree, potentially reducing cytopathic T-cell-mediated damage (56). A prospective randomized trial is needed, however, before ursodeoxycholic acid can be recommended routinely for the prevention of biloma after liver transplantation. Three randomized controlled trials in liver transplant patients have assessed the utility of ursodeoxycholic acid for prevention of graft rejection (57–59); only one found a significant reduction in the incidence of rejection with the use of ursodiol (59). Although overall infections occurred less frequently in patients receiving ursodeoxycholic acid in one of the three trials (59), the incidence of biloma was not assessed in any of them.
Our study has several limitations. A retrospective study design usually has the inherent drawback of accurately identifying all risk factors. However, as our database is prospectively maintained, we believe this source of bias has been minimized. It is possible that the incidence of biloma was underestimated in our patients, as only symptomatic patients underwent follow-up imaging to detect hepatic fluid collections. However, in the vast majority of our study population, imaging studies were routinely performed for evaluation of abnormal liver chemistry tests or fever, and we believe that misclassification bias is also unlikely; moreover, most patients with bilomas ultimately become symptomatic or exhibit hepatic enzyme abnormalities, which invariably prompts imaging studies.
In conclusion, infected bilomas are an important complication of liver transplantation. Nosocomial multiresistant gram-positive bacteria and yeasts are the major pathogens implicated, either initially or in later superinfections. Isolation of coagulase-negative staphylococci is associated with T-tube drainage. The most important predisposing factors are HAT or stenosis. While use of ursodeoxycholic acid appeared to be associated with a reduced risk of biloma formation, there is an urgent need for a randomized trial to confirm a protective effect with ursodeoxycholic acid, and to devise more effective strategies to prevent HAT. The routine use of T-tube drainage should be critically reassessed.
We thank Karen Armstrong and Barbara Voss for their help with database management.
This study was presented in part at the American Society of Transplantation Annual Meeting, Washington D.C., 2003, and at the 43rd InterScience Conference on Antimicrobial Agents and Chemotherapy, Chicago, IL, 2003.
Supported by an unrestricted gift for research from the Oscar Rennebohm Foundation of Madison, WI. Abbreviations: MELD, model for endstage liver disease; HAT, hepatic artery thrombosis.