Qian YB, Liu CL, Lo CM, Fan ST. Risk Factors for Biliary Complications After Liver Transplantation. Arch Surg. 2004;139(10):1101-1105. doi:10.1001/archsurg.139.10.1101
Biliary complications after liver transplantation can be predicted from perioperative factors.
Retrospective analysis of data collected prospectively.
Tertiary referral center.
From October 5, 1991, through June 22, 2002, 230 patients received 241 consecutive orthotopic liver transplants. Patients were divided into those with (group 1) and those without (group 2) biliary complications.
Main Outcome Measures
Postoperative outcomes, biliary leakage, and anastomotic stricture.
The overall biliary complication rate was 20.7%, including bile leakage rate of 7.1% and anastomotic stricture rate of 16.2%. By means of univariate analysis, risk factors associated with biliary complications were preoperative serum bilirubin level (P = .003), international normalized ratio (P = .04), the use of stent or T-tube splinting of the anastomosis (P = .02), and the use of live-donor liver graft (P = .03). Stepwise logistic regression analysis demonstrated that the preoperative serum bilirubin level (relative risk [RR], 1.00), use of stent or T-tube splinting of the anastomosis (RR, 2.10), and use of live-donor liver graft (RR, 2.01) were independent risk factors predicting biliary complications after liver transplantation. Graft survival rates at 1, 3, and 5 years were 89.5%, 84.7%, and 79.7%, respectively, in group 1 and 84.7%, 78.4%, and 75.1%, respectively, in group 2 (P>.05). Patient survival rates at 1, 3, and 5 years were 89.1%, 86.5%, and 86.5%, respectively, in group 1, and 86.1%, 82.8%, and 81.0%, respectively, in group 2 (P>.05).
Preoperative serum bilirubin level and the use of stent or T-tube splinting of the anastomosis and live-donor liver grafts were independent risk factors for biliary complications after liver transplantation. We postulated that high preoperative serum bilirubin level reflected severe liver disease and difficult hemostasis, leading to inadvertent injury to the anastomosis during graft rotation or manipulation for hemostasis. The use of a stent or a T tube predisposes to more complications. Further technical refinement is necessary for biliary reconstruction in live-donor liver transplantation.
Orthotopic liver transplantation is now considered a standard procedure for patients with end-stage chronic liver disease. However, despite recent advances in organ preservation, inmmunosuppression, and surgical techniques and better management of complications, the incidence of biliary tract complications has still been high in recent years, ranging from 22% to 64%.1- 8 The most common biliary complications were bile leakage and anastomosis stricture. In this study, we retrospectively reviewed the results of 241 consecutive orthopotic liver transplantations to determine the incidence of biliary complications and to identify risk factors that might predispose to the development of biliary complications.
From October 5, 1991, through June 22, 2002, 241 orthotopic liver transplantations were performed in 230 consecutive patients at the Department of Surgery, Queen Mary Hospital, the University of Hong Kong, Hong Kong. Cadaveric grafts included full-size liver (n = 97) and right (n = 2), left (n = 3), and segments 2 and 3 (n = 3) liver split from whole liver grafts; the live-donor grafts included right lobe (n = 103), left lobe (n = 10), and segments 2 and 3 (n = 23) grafts. Biliary reconstruction was performed using end-to-end choledochocholedochostomy (n = 116) or hepaticojejunostomy (n = 124). The latter was used when the former was not feasible for anatomical reasons or because of the presence of any underlying hepatobiliary disease. One patient receiving the right lobe live-donor graft underwent separate duct-to-duct reconstruction and hepaticojejunostomy for 2 separate hepatic ductal orifices in the right lobe graft. Choledochocholedochostomy was performed with a T-tube splinting the anastomosis in 47 patients. After March 2000, the T tube was not used, except in 2 patients in whom bile leakage was noted intraoperatively from needle holes of the anastomoses. For hepaticojejunostomy, an external stent was used in 1 patient and a short internal stent was used in 67 patients. A stent was not used in hepaticojejunostomy after March 2000.
Patients were divided into 2 groups. Group 1 consisted of patients with biliary complications after liver transplantation (n = 50), and group 2 of patients without biliary complications after liver transplantation (n = 191). There were 34 male and 16 female patients in group 1, with a median age of 41 years (range, 0.6-61 years), and 129 male and 62 female patients in group 2, with a median age of 44 years (range, 0.33-68 years). Pretransplantation diagnoses of the 2 groups are listed in Table 1. The severity of liver disease was assessed by means of liver biochemistry findings, international normalized ratio, and the Model for End-Stage Liver Disease score.9 Patients with fulminant hepatic failure and children (n = 45) were not included in the Model for End-Stage Liver Disease score calculation.
Biliary complications included bile leakage and anastomotic stricture, which were diagnosed by means of T-tube, percutaneous transhepatic, or endoscopic retrograde cholangiography or computed tomography. The T tube remained in situ for at least 3 months after the operation. Before the T tube was removed, T-tube cholangiography was performed. Antibiotics were given before the cholangiography.
Patients with and without biliary complications were compared with reference to 21 preoperative, operative, and postoperative variables. Preoperative variables included sex; age; serum levels of bilirubin, creatinine, albumin, alanine aminotransferase, and aspartate aminotransferase; international normalized ratio; Model for End-stage Liver Disease scores; and urgency of liver transplantation. Operative variables included cold ischemic time, the type of liver graft (cadaveric or live-donor graft), the use of a T tube or an internal stent for splinting the anastomosis, type of suture materials, type of biliary reconstruction (choledochocholedochostomy or hepaticojejunostomy), and the number of biliary anastomoses. The number of units of blood, fresh frozen plasma, and platelet concentrates were also included in the analysis. Postoperative variables included hepatic artery thrombosis and cytomegalovirus infection. The data were collected prospectively by a single research assistant. We performed univariate analysis using the Mann-Whitney and χ2 tests. Variables with statistical significance at a level of P<.05 on univariate analysis were subjected to stepwise logistic regression analysis to identify the independent predictors of biliary complications. We performed Kaplan-Meier analysis for survival estimation. The log-rank test was used for comparison of survival rates. All statistical analyses were performed using SPSS software (SPSS version 11.0; SPSS Inc, Chicago, Ill). We considered P<.05 to be statistically significant.
The overall biliary complication rate was 20.7%. According to the types of biliary reconstruction, the difference in biliary complication rates of 17.2% in those with choledochocholedochostomy and 24.2% in those with hepaticojejunostomywere not significantly different between the 2 groups. Bile leakage occurred in 17 transplantations (7.1%) on postoperative days 2 to 172 (median, day 16). The reasons for bile leakage included leakage from the anastomosis (n = 11), missed hepatic duct branches (n = 1), leakage around the T tube from the choledochotomy wound (n = 2), and removal of the T tube (n = 3). They were treated with 1 or more of the following procedures: repair of choledochocholedochostomy (n = 2), conversion of choledochocholedochostomy to hepaticojejunostomy (n = 2), repair of hepaticojejunostomy (n = 6), revision of hepaticojejunostomy (n = 2), laparotomy and T-tube reinsertion (n = 1), laparotomy and suturing of the T-tube tract (n = 2), percutaneous transhepatic biliary drainage (n = 2), endoscopic retrograde cholangiopancreatography and stenting (n = 1), percutaneous drainage of biloma (n = 1), or conservative treatment (n = 1). Anastomotic stricture was diagnosed after 39 transplantations (16.2%) on postoperative days 9 to 1508 (median, day 129). They were treated by 1 or more of the following procedures: percutaneous transhepatic biliary drainage and dilation (n = 19), endoscopic dilation (n = 9), laparotomy for dilation of hepaticojejunostomy and removal of stent (n = 3), and revision of hepaticojejunostomy (n = 7) or conversion of choledochocholedochostomy to hepaticojejunostomy (n = 7). Six patients had bile leakage and subsequent anastomotic stricture.
Univariate analysis indicated that significant predictors of biliary complications were preoperative serum bilirubin level (P = .003) and international normalized ratio (P = .04), the use of a stent or a T tube for splinting the anastomosis (P = .02), the number of units of fresh frozen plasma infused (P = .03), and the type of liver graft (P = .03). Age, sex ratio, Model for End-Stage Liver Disease score, serum creatinine level, cold ischemic time, number of bilary anastomoses, hepatic artery thrombosis, and cytomegalovirus infection were not significantly associated with biliary complications (Table 2). Logistic regression analysis demonstrated that the preoperative serum bilirubin level (relative risk [RR], 1.00; 95% confidence interval [CI], 1.00-1.00), use of a stent or a T tube for splinting the anastomosis (RR, 2.10, 95% CI, 1.09-4.16), and use of a live-donor graft (RR, 2.01; 95% CI, 1.02-4.10) were independent risk factors predicting biliary complications after liver transplantation.
All patients were followed up for a median period of 25.1 months (range, 3-122.5 months). Retransplantation was required in 11 cases. In group 1, reasons for retransplantation were graft hepatitis of unknown etiology (n = 1) and portal vein thrombosis secondary to percutaneous transhepatic biliary drainage and dilation for stenosis of hepaticojejunostomy (n = 1). In group 2, the reasons for retransplantation were primary graft nonfunction (n = 1), portal vein thrombosis (n = 1), acute rejection (n = 1), hepatic artery thrombosis (n = 1), recurrent hepatitis B virus infection (n = 2), de novo hepatitis B virus infection (n = 1), and graft hepatitis of unknown etiology (n = 2). Early or late death occurred in 7 patients in group 1 and 32 in group 2. The causes of death of the 2 groups are listed in Table 3. Four patients died as a result of biliary leakage or complications of radiological intervention for biliary stricture.
Graft survival rates at 1, 3, and 5 years were 89.5%, 84.7%, and 79.7%, respectively, in patients with biliary complications, and 84.7%, 78.4%, and 75.1%, respectively, in patients without biliary complications (P>.05). Patient survival rates at 1, 3, and 5 years were 89.1%, 86.5%, and 86.5%, respectively, in patients with biliary complications, and 86.1%, 82.8%, and 81.0%, respectively, in patients without biliary complications (P>.05).
Biliary complications after liver transplantation have been reported to be related to many conditions, including prolonged cold ischemic time, hepatic artery thrombosis, ABO blood type incompatibility, cytomegalovirus infection, recurrence of primary disease, the use of reduced-size liver transplants, and the method of biliary reconstruction.6,10- 12 In this study, we confirmed that the use of live-donor graft is associated with more biliary complications.3,4 In addition, we identified preoperative serum bilirubin level and the use of a stent or a T tube for splinting the anastomosis as important risk factors for biliary complications, whereas the factors identified by previous researchers were not.
End-to-end choledochocholedochostomy and hepaticojejunostomy have been methods of biliary reconstruction after orthotopic liver transplantation. Choledochocholedochostomy is technically easier and has the advantage of low incidence of enteric reflux, whereas hepaticojejunostomy may have such drawbacks. For choledochocholedochostomy, the role of the T tube is controversial.1,13,14 Although the use of a T tube permits monitoring of bile flow and color and allows easy performance of cholangiography, many researchers reported serious complications, including acute cholangitis, biliary obstruction, tube migration, and bile leakage. In this study, we observed a higher cholangitis rate associated with the use of the T tube (data not shown), especially when cholangiography was performed. Use of the T tube also led to bile leakage in 4 patients, in 2 of whom it was caused by the removal of the T tube, presumably owing to lack of fibrous tissue formation around the T tube caused by steroid therapy. A recent prospective randomized trial also showed that T-tube insertion was associated with more complications than was noninsertion.1 We stopped splinting of biliary anastomosis by means of T tube after March 2000, even in recipients of right lobe live-donor liver transplants, and did not observe a higher incidence of biliary complications. Such practice concurs with that of many liver transplant surgeons who have abandoned the use of the T tube in biliary anastomosis since the late 1990s.15,16 However, the T tube and splintage tube are used almost routinely nowadays for duct-to-duct anastomoses in adult-to-adult live-donor liver transplantation.17,18 The logic underlying the resurgence of the practice is not known. Perhaps a prospective randomized trial is needed to demonstrate whether splinting of the duct-to-duct anastomosis is necessary in these patients.
Intuitively, hepaticojejunostomy might be associated with more biliary complications than duct-to-duct anastomosis, but our analysis did not show this to be true. Because the bowel mucosa may be edematous, leading to occlusion of the anastomosis, splinting has been advocated, but the exact value of splinting has never been well defined. External drainage serves a function similar to that of the T tube. The disadvantage is that the patient has to bear with the tube for at least several weeks. As the jejunal loop could be quite edematous at the time of biliary reconstruction, bringing the splinting tube out of the jejunal loop and creating a tunnel for it could be hazardous. We therefore advocated internal drainage by means of a short tube without anchorage, hoping that the tube would migrate into the jejunum subsequently.2 However the tube migrated proximally into the intrahepatic duct or stayed at the site of anastomosis, causing obstruction and acute cholangitis. We have yet to decide whether the tube itself induced more fibrosis or impaired healing, but it is not unlikely that recurrent episodes of acute cholangitis might induce more fibrosis of the ductal wall and lead to late stenosis.
A number of previous studies had found that the preoperative serum bilirubin level was an important determinant of liver transplantation outcome. It predicted patient and graft survivals after liver transplantation14 and correlated with hepatic artery thrombosis risk.19 However, the relationship between preoperative serum bilirubin level and biliary complications after liver transplantation has not been previously studied. We demonstrated that the preoperative serum bilirubin level was an independent predictor of biliary complications after liver transplantation, but the mechanism remains unexplained. In an experimental study, a high level of unconjugated bilirubin might impair fibroblast proliferation.20 Our hypothesis was that patients with a high preoperative serum bilirubin level have poor liver function, and complete hemostasis is difficult to achieve after implantation. Perfect hemostasis before wound closure may require frequent mobilization and rotation of the liver graft for exposure of retroperitoneum and hepatoduodenal ligament and lead to minor or unrecognized disruption of the biliary anastomosis. The significantly higher preoperative international normalized ratio and larger requirement of fresh frozen plasma in the patients with biliary complications support our hypothesis. In this regard, drugs that might impair wound healing, such as steroids, might enhance the minor leakage. Inflammatory reaction and subsequent fibrosis around the anastomosis may then predispose to anastomotic stricture.
Because steroid therapy was an essential component of the immunosuppressive protocol, we advocated the use of a nonabsorbable suture material (6-0 polypropolylene [Prolene; Johnson & Johnson Int, St Stevens-Woluwe, Belgium] with knots outside the lumen) with the hope that it could hold the biliary anastomosis in the early postoperative period. However, on exploration of several patients with anastomotic stenosis, we found nonabsorbable suture materials inside the bile duct lumen, and they were covered with sludge and debris. It is possible that the sutures were inside the bile duct lumen as a result of impaired wound healing or necrosis of the hepatic duct stump. It is not certain whether the nonabsorbable suture materials predispose to impaired wound healing, but once inside the lumen they could aggravate bile duct obstruction and acute cholangitis. For the benefit of the doubt, we have recently changed to the use of an absorbable suture material and, at about the same time, adopted an immunosuppressive protocol almost devoid of steroid therapy.21 Whether such a strategy will reduce biliary complications awaits longer follow-ups.
Patients undergoing live-donor liver transplantation had a significantly higher biliary complication rate than those receiving cadaveric grafts. We reported the possible reasons in a previous report.2 Among all of these possibilities, ischemia of the right hepatic duct stump in the graft is the most likely reason. A previous report from our institution2 has described the strategy to preserve the blood supply of the right hepatic duct stump in the graft and the satisfactory result in hepaticojejunostomy on adoption of modified approaches. With the use of duct-to-duct anastomosis in the right lobe live-donor graft, we realize that blood supply and venous drainage of the recipient common hepatic duct are also important in ensuring satisfactory healing of biliary anastomosis. Unlike cadaveric liver transplantation, live-donor transplantation requires division of the recipient's common hepatic duct high in the liver hilum to avoid tension in the anastomosis, but the blood supply to the common hepatic duct stump becomes precarious, especially when the common hepatic duct is freed from surrounding structures, because the arterial supply of the common hepatic duct is derived mainly from the gastroduodenal artery after division of the duct.22,23 However, there could be many hepatic artery branches running into the common hepatic duct from the right and left hepatic arteries in the hepatoduodenal ligament. Therefore, in the recipient hepatectomy in live-donor cases, we now try to preserve the tissue around the common hepatic duct to preserve such branches. Such an approach may be beneficial in preserving venous drainage as well, but the long-term outcome awaits follow-up studies.
Risk factors identified in the patients, the type of liver graft, and the technical design contributed to biliary complications in liver transplanation. Because the preoperative serum bilirubin level of the patients could not be altered, more technical refinement in hemostasis and liberal infusion of fresh frozen plasma and platelets are required to reduce the complication rate and to improve the result. Splinting of choledochocholedochostomy by means of a T tube or hepaticojejunostomy by means of an internal stent is probably unnecessary. Although biliary complications appear to have no significant effect on the graft and patient survival, such complications should be avoided and an effort must be made to reduce the biliary complication rate, particularly in live-donor liver transplantation, to perfect the operation outcome.
Correspondence: Sheung Tat Fan, MS, MD, PhD, FRCS(Glasg), FRCS(Edin), Department of Surgery, University of Hong Kong Medical Centre, Queen Mary Hospital, 102 Pokfulam Rd, Hong Kong (firstname.lastname@example.org).
Accepted for publication January 2, 2004.
This study was supported by the Distinguished Research Achievement Award of the University of Hong Kong, Hong Kong, and the S. K. Yee Medical Foundation, Hong Kong.