Hepatic artery thrombosis (HAT) that follows orthotopic liver transplantation (OLT) is a potentially life threatening complication. Compared to the relatively high incidence reported in the early 80s,1 HAT presently complicates 3–9% of all OLTs.2 HAT when it occurs early post OLT (within 4 weeks) may present with acute graft failure, sepsis or liver abscess, and bile duct complications like biliary leak or stricturing.3 The presentation of delayed HAT is however much more variable, which may present with features of cholangitis, with or without strictures or abscesses, bile leaks, or may remain clinically silent with altered liver function tests.3
It was initially proposed that surgical technique was the most important risk factor for HAT.4 However it is now clear that other factors like graft preservation, ischemia reperfusion injury, immunological factors, coagulation abnormalities, infections, transplantation across ABO blood group, a positive cytotoxic antibody crossmatch, genetic factors, and multiple rejection episodes could also be aetiologically linked to HAT.5
Treatment of early HAT usually requires surgical intervention, with retransplantation being an option.6 Retransplantation is however limited by both organ availability and the patient's condition. Urgent revascularization with thrombectomy and a combination of thrombectomy with revision of anastomosis has been successful in some patients with an early diagnosis.7, 8 A number of patients with delayed HAT are not diagnosed immediately, and surgical intervention is not a viable option in some.3
In this study, we present our experience in the management of HAT following adult liver OLT over a 10-year period. Special emphasis is given to presentation in both early and late HAT, identifiable aetiological factors, complications as a result of HAT, and survival outcome of graft and patient.
HAT, Hepatic artery thrombosis; OLT, orthotopic liver transplantation; CMV, cytomegalovirus.
MATERIALS AND METHODS
Data was derived from a prospectively collected data base from April 1994 to April 2004. Where required, patent records and radiological reports were obtained. There were 1,257 adult OLTs conducted at our centre during this period. Six hundred sixty were males, the median age being 51(range: 16–73) years. Early and late HAT was defined by the occurrence of HAT within or later than 4 weeks following OLT. respectively.3, 9 Additionally, donor demographics, preoperative, and intra-operative details, such as cold and warm ischemia times, and intra-operative transfusion requirements of whole blood and fresh frozen plasma, duration of surgical procedure, type and number or arterial reconstructions, along with type of biliary reconstruction, were compared between patients with and without HAT.
Orthotopic Liver Transplantation
The majority of OLTs were performed with veno-venous bypass using end to end interposition of retrohepatic donor cava to resected retrohepatic recipient cava, until 2001. Following 2001, piggyback veno-venous caval reconstruction without venous bypass became standard practice at our centre. This was usually performed in the presence of a porto-caval shunt following the division of the portal vein. Reperfusion of the graft was performed following completion of the caval and portal venous anastomoses. Arterial anastomosis was usually performed using a Carrel patch at the recipient hepatic artery gastroduodenal artery junction and a branching point on the donor common hepatic artery, with a running suture using prolene 7/0 (Ethicon, Voerderstedt, Germany). Where anastomosis to the native recipient hepatic artery or celiac trunk was not possible, aortic conduits, using a donor iliac artery interposition graft (supraceliac or infrarenal), was used for arterial reconstruction. Biliary reconstruction was performed last with an end to end choledochocholedochostomy or where indicated Roux en-Y choledochojejunostomy. The transfusion policy during the recipient procedure was tailor made to suit the individual recipient and governed by thromboelastography the “near-patient” test of whole blood coagulation.
Routine prophylaxis for prevention of HAT is not standard practice in our unit. However, in a selected subset of recipients, where there was more than one arterial anastomosis, especially with small diameter vessels and when aortic conduits were used for reconstruction, it was our practice to initiate prophylaxis on day one post transplant with aspirin. Post transplant haemoglobin levels were maintained above 8 g/dL and below 10 g/dL. Similarly, platelet counts were maintained around 100,000/μL so as to reduce the risk of arterial thrombosis. These patients also had a Doppler ultrasound scan on the first day following the transplant to assess arterial patency. Doppler ultrasound scan was not however routine for all recipients post liver transplants.
Early HAT when it occurred was diagnosed with Doppler imaging and angiography following derangement of liver functions or acute graft failure. The mode of presentations of late HATs were variable. The majority of late HAT presented with sepsis and/or graft dysfunction. Some presented with late biliary complications and were found to have HAT as did some presenting with associated viral sepsis (Table 1).
Some late HATs had more than one presenting symptom.
Late biliary complications
Associated viral infection
Statistical analysis was performed using SPSS, version 11.0 (Statistical Package for the Social Sciences, Chicago, IL). Categorical variables were compared using chi-square test for association and Mann Whitney U test was used for non parametric data analysis. Survival curves were calculated using Kaplan-Meier method and compared using log-rank tests. Multivariate binary logistic regression was performed to identify independent predictors of HAT, using variables that were significant (P < 0.05) on univariate analysis.
In this series of 1,257 adult OLTs, 61 HATs were observed with an overall incidence of 4.9%. Thirty of these 61 patients were male. Early HAT occurred in 22 (1.8%) patients and late in 39 (3.2%). Median age of recipients in whom HAT occurred early was 48 (range: 19–66) years and 48 (range: 21–65) years in those with late HAT. The median duration for HAT to occur was 13 (range: 0–27) days in the early group and 82 (range: 30–1921) days in the late group. The indications for OLT in those recipients subsequently developing HAT are given in Table 2.
Table 2. Indications for OLT in Those Developing HAT
Indication for OLTx number
% HAT for indication
Abbreviation: AICAH, autoimmune chronic active hepatitis.
Five recipients with cholestatic liver disease, 2 viral hepatitis B/C, and 3 in the acute/sub-acute liver failure group were also regrafts.
Cholestatic liver disease/AICAH
Viral hepatitis B/C
Alcoholic liver disease
Acute/sub-acute liver failure
Hepatocellular carcinoma +Viral hepatitis B/C
Donor cytomegalovirus (CMV) immune status was available on the data base in 691 cases of which 342 were CMV positive. There were 11 late HATs in recipients transplanted with these 691 donor livers. Nine were CMV positive (P = 0.031). Statistical analysis of age at OLT, recipient MELD score and BMI, donor to recipient age ratio, and sex ratio were not different between the groups with and without HAT.
The median duration of the operative procedure for OLT in patients with early HAT was 7 (range: 3–9) hours compared with 5.3 (range: 2.6–15) hours in those recipients who did not develop HAT (P = 0.001). In patients with and without HAT cold ischemia times were 770 (range: 537–993) minutes and 693 (range: 165–1,325) minutes (P = 0.12) and warm ischemia time was 38 (range: 34–69) minutes 45 (range: 20–278) minutes, respectively (P = 0.002).
Analyzing intra-operative transfusion requirements and the incidence of HAT; the incidence of HAT did not reach statistical significance when whole blood transfusions up to 5 units were used intra-operatively (P = 0.091: NS). However the incidence of HAT reached statistical significance when intra-operative whole blood transfusion of 6 or more units were used (P = 0.031). The infusion of 15 or more units of fresh frozen plasma intra-operatively increased the incidence of HAT significantly (P = 0.041). The number of units of platelets transfused the number of units of cryoprecipitate transfused and the use of aprotinin had no significant impact on the incidence of HAT.
The use of split or reduced liver grafts, and a previous history of upper abdominal surgery, did not show a correlation with the incidence of early HAT.
Day 1 and 3 post operative factors were also analyzed. Haematological factors such as haemoglobin levels, platelet count, prothrombin time and INR were not associated with early HAT in our series. The post operative liver function, serum albumin levels and renal function did not correlate with the incidence of early HAT.
Hepatic Arterial Reconstruction
Depending on anatomical variations of donor hepatic artery and the quality of the native recipient hepatic artery, reconstruction required one or more anastomoses or the use of a donor iliac artery interposition graft (aortic conduit). In our series, the number of donor hepatic arteries and the resultant multiplicity of anastomoses for reconstruction had no statistically significant relationship to the incidence of HAT (P= 0.49: NS) (Table 3). This was true for both early and late HAT (P= 0.54).
Table 3. The Incidence of HAT in Relation to the Number of Donor Hepatic Arteries
Number of donor hepatic arteries
P = 0.49: NS.
The number of donor iliac artery interposition grafts in this series was 40, and 6 of these resulted in early HAT. An aortic conduit was used primarily in 10 recipients (0.8%). There were 149 (12%) retransplants in this series and aortic conduits were used in 30 procedures (30/149; 20%). The use of an aortic conduit both in the primary transplant and at retransplantation was found to be significantly associated with early HAT (P = 0.01).
HAT and Type of Biliary Reconstruction
One thousand thirty-three recipients had end to end choledochocholedochostomy biliary reconstructions, of which 41 developed HAT. This was significantly less than the incidence of HAT in those recipients who had Roux-en-Y choledochojejunostomy biliary reconstructions (20/221), (P = 0.01). Of the 22 recipients who developed early HAT in this series, 10 had end to end choledochocholedochostomy and 12 had Roux en Y choledochojejunostomy biliary reconstruction. In the 39 OLT recipients who developed late HAT, 31 had end to end choledochocholedochostomy and 8 Roux-en-Y choledochojejunostomy biliary reconstruction. The incidence of early HAT was therefore found to be significantly higher in those recipients who had Roux-en-Y choledochojejunostomy biliary reconstruction as compared to end to end choledochocholedochostomy (P = 0.008). Primary sclerosing cholangitis was not an independent risk factor for HAT in our series, although Roux-en-Y choledochojejunostomy biliary reconstruction is used more commonly in these patients (P = 0.48: NS). Roux-en-Y biliary reconstruction was however used more commonly in patients receiving more than one graft (97/149) (P = 0.001).
In total there were 149 regrafts in the 1257 transplants during this period. Early HAT occurred in 10 of these regrafts (10/22). Relaparotomy was carried out for varied indications in 28 recipients during the first month post transplant, of which five subsequently developed early HAT (5/22). Both regraft (P = 0.001) and relaparotomy (P = 0.009) were found to be risk factors for early HAT in this series.
A bile leak was associated with 12 of the 61 patients who developed HAT, as compared to 39 of the 1196 recipients who did not develop HAT (P = 0.001). Cholangitis was diagnosed in 5 of the 61 recipients who developed HAT, with 14 cases of cholangitis diagnosed in the 1196 without HAT (P = 0.002). It is however unclear whether these biliary complications were the cause or a consequence of HAT.
Following multivariate analysis of the data using binary stepwise logistic regression, type of biliary anastomosis was the only significant factor found to impact the incidence of HAT. Factors significantly associated with HAT are given in Table 4.
Table 4. Risk Factors Associated With an Increased Incidence of HAT
Factors significantly associated with HAT
CMV (+) donor > CMV (−) recipient
P = 0.031
P = 0.001
P < 0.01
More than 6 units blood transfused
P = 0.031
More than 15 units FFP transfused
P = 0.041
Aortic conduit for HA reconstruction
P = 0.01
Biliary reconstruction R-D > D-D
P = 0.008
P = 0.001
P = 0.002
P = 0.009
Outcome Following HAT
As a consequence of HAT, 36/61 recipients developed graft dysfunction and there were 7 deaths. Overall, 11 were regrafted for graft failure following HAT. There were 7 mortalities following regrafts in these recipients. Of the 22 recipients who developed early HAT, two recipients had angioplasty following angiographic diagnosis of HAT and a further two required relaparotomy and surgical thrombectomy. Twenty-three of the recipients who presented with late HAT developed complications of bile leak (12), cholangitis (4), liver abscess (4), and associated acute pancreatitis (3). Twenty-five of the 61 recipients who developed HAT were alive at the time of this study, four of which were regrafts. The incidence of HAT had a significant (P = 0.001) impact on recipient survival (Fig. 1). There however was no statistical difference in recipient survival between those who developed early vs. late HAT (Fig. 2).
The clinical presentation of HAT varies from a mild elevation of serum amino transferase and bilirubin levels to fulminant hepatic necrosis. This variation is associated in part with the time at which HAT develops. The acute presentation is usually associated with early HAT. The time point that divides early and late is however arbitrary. We considered 30 days as the cut off between early and late HAT.3 Early HAT results in massive injury to hepatocytes and bile duct epithelium, partly due to disruption of collaterals as a consequence of the recipient hepatectomy.9
The overall incidence of HAT in our series was 4.9%. Early HAT however occurred in 22 patients with an incidence of 1.8%. Early HAT is largely related to surgical technique,9 and this was likely to be a major aetiological factor in the 1980s.1 Surgical technique is probably not a major risk factor at our centre, since the number of arterial anastomoses for reconstruction had no bearing on the incidence of HAT in our series. Based on evidence from previous studies, recipients with multiple anastomoses for arterial reconstruction at our centre also receive post OLT aspirin and a day 1 doppler ultrasonographic assessment for patency of the reconstructed hepatic artery.8, 10
Late HAT is generally associated with a milder clinical course and a more insidious on set.11, 12 Possible risk factors for late HAT are less clear than those for early HAT. However positive CMV status in the donor with negative CMV status in the recipient has been repeatedly shown to be linked as a possible risk factor for late HAT.11–13 CMV positive status in the donor was identified as a risk factor in our series too.
The use of an aortic conduit was a significant risk factor for HAT. Although some early studies do not corroborate these results,14 the majority of previous studies indicate a significantly higher risk of HAT in those recipients with aortic conduits used for reconstruction.14–16 Reasons for this higher incidence however is unclear.9 In a recent comparative retrospective analysis of 101 aortic conduit reconstructions and 512 primary hepatic artery reconstructions, the use of an aortic conduit was found to be a predictive factor for early HAT.17 Similar to our findings, this study also showed that retransplantation increased the risk of HAT significantly.
The only donor factor associated with HAT in our series was a seropositive CMV donor liver being transplanted into a seronegative recipient. Disruption of the endothelial lining in these cases may limit the repair of vascular damage and result in continuous activation of the inflammatory and coagulation system.9 Other viral infections such as herpes simplex virus types 1 and 2, human herpes virus types 6 and 7 have also been linked with HAT. In our series however only four cases of recorded herpes infection were noted in those developing HAT. Other donor factors such as low donor/recipient age ratio which has been described previously as a risk factor.9 Donor age greater than 50 years has also been identified as an independent risk factor for HAT.17 Although our data did not collaborate this, other studies have identified increased donor age as a risk factor for HAT.15
It has previously been reported that an increased operation time and prolonged cold ischemia and warm ischemia times increases the risk of early HAT.2, 9 A prolonged cold ischemia time was found to be a risk factor for HAT in our series. However paradoxically, a shorter warm ischemia time was found to be associated with HAT for which a credible explanation eluded us. The use of 6 or more units of transfused blood and 15 or more units of fresh frozen plasma increased the risk of HAT. This has also been previously reported.2, 9 The use of platelet transfusion, cryoprecipitate, and aprotinin did not increase the risk of HAT in this study. Although genetic predisposition of recipients to hypercoagulable states have been implicated in the incidence of early HAT,5, 9 monitoring for conditions like factor V Leiden mutation, antithrombin III, protein C, and protein S deficiency is not routine at our centre. A recent publication has indicated that cigarette smoking in recipients is associated with an increased risk of HAT.18 Comparative data in our series however was not complete in this regard for us to draw conclusions on smoking as a risk factor for HAT.
Split or reduced liver grafts could result in small diameter donor hepatic arteries with a higher possibility of donor-recipient artery diameter missmatch. Large published series have not shown an increased risk of HAT in recipients of these grafts.19, 20 In our series, there was no statistically significant increase in the risk of HAT in recipients of split or reduced liver grafts. Previous upper abdominal surgery could result in increased adhesions and prolonged operation time and although operation time was significantly associated with HAT, previous upper abdominal surgery was not a risk factor for HAT in our series.
Day 1 and 3 post operative factors such as haemoglobin levels, platelet count, prothrombin time, and INR were not associated with increased HAT. Although a high haematocrit and haemoglobin level is associated with increased risk of HAT, it was not a significant factor in our series, probably because it is unit protocol to maintain haemoglobin levels less than 10 g/dL in the post transplant recipient.
The incidence of early HAT was significantly higher in those recipients with Roux-en-Y biliary reconstructions both in univariate and multivariate analysis. Roux-en-Y biliary reconstructions were used more commonly in regrafts (97/149), and the incidence of HAT was significantly higher in recipients receiving regrafts. Roux-en-Y biliary reconstruction is usually used in patients transplanted for end stage liver disease due to primary sclerosing cholangitis. We therefore considered the possibility of primary sclerosing cholangitis being a risk factor for HAT since the condition is also associated with a pro-coagulant state. However primary sclerosing cholangitis was not an independent risk factor for HAT in our series. Bile leak and cholangitis was also found to be common in patients with HAT. It is however unclear whether these biliary complications were the cause or a consequence of HAT.
The incidence of HAT had a significant impact of graft and recipient survival. Retransplantation however does not necessarily need to follow late HAT. Eleven OLT recipients were regrafted in our series. Two recipients had angioplasty following angiographic diagnosis of HAT and a further two had relaparotomy and surgical thrombectomy. However relaparotomy and thrombectomy is not the preferred option for graft salvage if angioplasty is possible. In total there were 14 deaths related to HAT with an overall mortality of 23% in those developing HAT post OLT.
The overall incidence of HAT in this series was 4.9%. The incidence of early HAT was 1.8% and late in 3.2%. HAT has a significant impact on survival, with 23% mortality in those with the complication. Possible aetiological factors for early HAT can be identified in the peri-operative and post-operative period following liver transplantation. In this series the use of a Roux-en-Y biliary reconstruction was a risk factor for early HAT in both univariate and multivariate analysis, although the mechanism for such an increase in incidence is unclear. The increased use of a Roux-en-Y biliary reconstruction in those receiving regrafts may be an association. Positive cytomegalovirus serology was the only donor factor associated with HAT and is the only aetiological factor that has been implicated as a cause of late HAT. The use of aortic conduits for arterial reconstruction is a risk factor that warrants initiation of prophylaxis in the post transplant period.
The authors thank Dr. Allen White of the Statistic Advisory Service, University of Birmingham, Edgbaston, Birmingham UK, for scrutiny of the multivariate analysis of this study and Dr. Emad Esmat, Visiting Surgical Fellow, The Liver Unit. Queen Elizabeth Hospital, University Hospital Birmingham NHS Trust, for his help in referencing this paper.