Hepatic Artery Ligation for Arterial Rupture Following Liver Transplantation: A Reasonable Option


Emmanuel Boleslawski,



Hepatic artery (HA) rupture after liver transplantation is a rare complication with high mortality. This study aimed to review the different managements of HA rupture and their results. From 1997 to 2007, data from six transplant centers were reviewed. Of 2649 recipients, 17 (0.64%) presented with HA rupture 29 days (2–92) after transplantation. Initial management was HA ligation in 10 patients, reanastomosis in three, aorto-hepatic grafting in two and percutaneous arterial embolization in one. One patient died before any treatment could be initiated. Concomitant biliary leak was present in seven patients and could be subsequently treated by percutaneous and/or endoscopic approaches in four patients. Early mortality was not observed in patients with HA ligation and occurred in 83% of patients receiving any other treatment. After a median follow-up of 70 months, 10 patients died (4 after retransplantation), and 7 patients were alive without retransplantation (including 6 with HA ligation). HA ligation was associated with better 3-year survival (80% vs. 14%; p = 0.002). Despite its potential consequences on the biliary tract, HA ligation should be considered as a reasonable option in the initial management for HA rupture after liver transplantation. Unexpectedly, retransplantation was not always necessary after HA ligation in this series.


hepatic artery


liver transplantation


Rupture of the hepatic artery (HA) is a rare but dreadful complication after liver transplantation (LT) and has been recognized as an important cause of graft loss and mortality [1-3]. This complication in the vast majority of patients is due to a pseudoaneurysm of the HA and could be devastating with massive bleeding that often requires immediate surgery. Various publications have reported single cases or small series of pseudoaneurysms following LT [1-11]. These reports had demonstrated the role of infectious microorganisms in the development of pseudoaneurysms. Radiological findings had also been extensively detailed and various techniques for the management of noncomplicated pseudoaneurysms have been proposed, including percutaneous approaches [9]. However, in half of the cases, HA pseudoaneurysm is not recognized before rupture. In these cases, conservative management is not possible and immediate surgery is mandatory [2], but there are no guidelines regarding the ideal surgical procedure either HA ligation or restoration of the graft's arterial flow.

The aim of the present study was to review all cases of HA rupture following LT in six transplant centers and to analyze the results of different managements.

Materials and Methods


From 1997 to 2007, data from six French transplant centers were retrospectively reviewed to identify patients presenting with HA rupture following LT. HA rupture was defined as a severe hemorrhage from the trunk or from a main branch of the HA resulting in disruption of the graft's arterial blood supply. Patients with a single lesion of HA tributaries without interruption of the hepatic arterial axis (i.e. donor splenic artery stump or recipient gastroduodenal artery), and patients with an intrahepatic pseudoaneurysm were excluded.

Data collection and statistical methods

A survey was sent to all centers including demographic data, clinical presentation, mechanism of the rupture, associated pathological findings, therapeutic modalities, mid-term outcome and patient and graft survival. Data were recorded in a specific Microsoft Access® database (Microsoft France, Issy-les-Moulineaux) and statistical analyses were performed using the SPSS® software, version 13 (SPSS Inc., Chicago, IL, USA). Quantitative data were expressed using a median (range). Comparison of continuous variables between groups was made using the Mann–Whitney U-test. Kaplan–Meier actuarial analyses of patient and graft survivals were performed. Differences between Kaplan–Meier survival estimates in patients with and without HA ligation were tested by using the log-rank test. A p-value <0.05 was considered statistically significant.


Patient characteristics

A total of 2649 liver transplants were performed during the study period. Seventeen patients with HA rupture (0.64%) were identified (13 men and 4 women, ages 27–65 years, mean 47.9 years). Indications for LT are detailed in Table 1. The main indications were hepatocellular carcinoma and alcoholic cirrhosis. The different patterns of graft-to-recipient arterial reconstruction are indicated in Table 1 and depicted in Figure 1. The anatomy of the donor HA was standard (common HA arising from the celiac trunk, Figure 1A–C), except in two donors in whom there was a replaced right HA arising from the superior mesenteric artery. In both cases, an arterial anastomosis was performed between the aortic patches of celiac trunk and superior mesenteric artery, as described by Gordon et al. [12] (Figure 1D). In one patient, the anatomy of the donor HA was standard, but on the recipient side, there was a replaced right HA arising from the superior mesenteric artery. In this patient, a redundant arterial supply was obtained by performing a first arterial anastomosis between the recipient's common HA and the donor celiac trunk and a second one between the recipient's replaced right HA and the donor's splenic stump (Figure 1E).

Table 1. Clinical characteristics of liver transplant recipients, circumstances of diagnosis of the arterial rupture and associated pathological findings
Case #Patient (sex, age, disease)Type of arteriala/biliary reconstructionBiliary drainageTime from LT (days)Type of hemorrhageMycotic pseudo-aneurysmBiliary leakIntra-peritoneal abscessOther pathological findingsTreatment of the arterial rupture
  1. aType of arterial reconstruction: see Figure 1.
  2. LT = liver transplantation; GI = gastro-intestinal.
1M, 46, alcoholic cirrhosisA/duct-to-ductNone16HemoperitoneumNoNoYes
2F, 42, alcoholic cirrhosisD/duct-to-ductC-Tube17HemoperitoneumYesNoYes 
3F, 27, metabolic diseaseC/duct-to-ductC-tube21HemoperitoneumYesNoNoAnastomosis revision
4M, 59, otherC/duct-to-ductNone2HemoperitoneumNoNoNoAscitic fluid infectionAnastomosis revision
5M, 44, malignancyA/duct-to-ductNone8HemoperitoneumYesNoNoNecrosis of the gastric antrumAorto-hepatic grafting
6F, 48, cholestatic diseaseB/roux-en-YNone35HemoperitoneumYesNoNoAscitic fluid infectionAorto-hepatic grafting
7M, 46, alcoholic cirrhosisA/duct-to-ductC-tube92HemoperitoneumNoNoNoPercutaneous embolization
8M, 57, malignancyA/duct-to-ductC-tube85GI bleedingNoNoNoGastro-duodenal ulcerHA ligation
9M, 51, malignancyA/duct-to-ductC-tube76HemoperitoneumNoYesYesHA ligation
10M, 46, alcoholic cirrhosisB/duct-to-ductC-tube33HæmobilaYesYesNo 
11M, 29, malignancyD/duct-to-ductC-tube24GI bleedingYesYesNoGastro-duodenal ulcer 
12M, 64, malignancyB/roux-en-YC-tube51GI bleedingYesYesNo 
13F, 48, alcoholic cirrhosisC/duct-to-ductNone38HemoperitoneumNoYesNo 
14M, 59, malignancyA/duct-to-ductNone74GI bleedingNoYesNo 
15M, 50, alcoholic cirrhosisC/duct-to-ductNone29GI bleedingNoYesNo 
16M, 48, malignancyE/duct-to-ductT-tube9HemoperitoneumNoNoNo 
17M, 42, malignancyA/duct-to-ductT-tube24HematomaNoNoYes 
Figure 1.

Different pattern used for graft-to-recipient arterial reconstruction. Ao = aorta; ct = celiac trunk; sa = splenic artery; sma = superior mesenteric artery; cha = common hepatic artery; pha = proper hepatic artery; gda = gastroduodenal artery; rha = replaced right hepatic artery.

Circumstances of diagnosis of the hepatic artery rupture

The median time between transplant and HA rupture was 29 days (range 2–92 days), but the distribution of events was bimodal with 4 late HA ruptures occurring after 2 months. Hemorrhage was always sudden and had various presentations: hemoperitoneum in 10 patients, gastrointestinal bleeding in 5, hematoma in 1 and hemobila in the remaining patient. A mycotic pseudoaneurysm was confirmed by mycological analyses of the arterial wall in 6 patients and suspected in one additional patient because of a positive bile culture, a low raise in candida‘s specific antibodies, a sudden death from unknown cause in the recipient of the heart from the same donor and a death from mycotic pseudoaneurysm rupture in the left kidney recipient. An associated biliary leak was present in seven patients, whereas HA rupture was isolated in four patients (Table 1).

Immediate treatment

Table 2 indicates the different techniques used for the treatment of the HA rupture and for the management of the biliary leak when it was associated. One patient died before any treatment could be performed. One patient had a percutaneous embolization and died after 80 days from severe cholangitis. The remaining 15 patients underwent urgent laparotomy. In five of them, an immediate restoration of the graft arterial blood supply was intended, either by anastomotic revision (three patients) or by aortohepatic grafting (two patients), whereas in 10 patients, a definitive ligation of the HA was performed.

Table 2. Initial management of arterial rupture and associated biliary leak and outcome
Case #Treatment of the arterial ruptureManagement of the biliary leakRe-LTOutcome
  1. LT = liver transplantation; HA = hepatic artery.
1NoDied immediately before any treatment
2Anastomosis revisionNoDied after 20 days from recurrent hemorrhage
3Anastomosis revisionNoDied few hours after laparotomy from hemorrhage
4Anastomosis revision NoAlive with a functional graft
5Aorto-hepatic graftingNoDied after 23 days from hemorrhage
6Aorto-hepatic graftingEmergencyDied after 40 days from sepsis
7Percutaneous embolizationNoDied after 80 days from cholangitis
8HA ligationNoAlive with a functional graft
9HA ligation“Rendez-vous”NoAlive with a functional graft
10HA ligationPercutaneous trans-hepatic drainageNoDied after 27 months from cholangitis
11HA ligationPercutaneous trans-hepatic drainageElectiveRe-LT at 5 years. Died 40 days after Re-LT from recurrent HA rupture
12 Roux-en-Y hepatico-jejunostomyElectiveRe-LT at 3 months and third LT at 8 years. Died 4 months later from sepsis
13 ElectiveRe-LT at 3 months. Died 27 months later from cholangitis
14 T-tubeNoAlive with a functional graft
15 T-tubeNoAlive with a functional graft
16 NoAlive with a functional graft
17 Endoscopic drainageNoAlive with a functional graft

In the seven patients with an associated biliary leak, only three synchronous surgical procedures were performed to treat the biliary damage (i.e. Roux-en-Y hepaticojejunostomy revision in one case and T-tube insertion in two cases). In the remaining four patients, the biliary leak was treated after surgery by percutaneous and/or endoscopic procedures.

Short-term outcome

There were six early deaths, occurring 0–80 days from the HA rupture and 16–172 days from the transplantation. In these patients, the cause of mortality was hemorrhagic relapse in four and sepsis in two. One of these early deaths occurred 12 days after superurgent retransplantation (Table 2). The remaining 11 patients survived beyond 3 months. The median follow-up in the survivors was 70 months (range: 24–98).

Effect of hepatic artery ligation on survival

After excluding one patient that died before any treatment could be initiated, patients with (N = 10) and without HA ligation (N = 6) were compared (Table 3). There were marked differences between the two groups in the clinical presentation of the HA rupture (hemoperitoneum or other) and the presence of an associated biliary leak.

Table 3. Comparison between patients with and without hepatic artery ligation
 With HA ligation (N = 10)Without HA ligationa (N = 6)
  1. aPatients without hepatic artery ligation have had various conservative managements, including percutaneous arterial embolization (one case), arterial anastomosis revision (three cases) or aortohepatic grafting (two cases).
  2. HA = hepatic artery; LT = liver transplantation.
Age49 (29–64)45 (27–48)
Indication for LT  
 Alcoholic cirrhosis33
 Cholestatic diseases01
 Metabolic diseases01
Time from LT to arterial rupture (days)36 (9–85)19 (8–92)
Type of hemorrhage  
Mycotic pseudoaneurysm  
Biliary leak  
Early mortality (<90 days)  

Of the six patients that underwent percutaneous embolization or revascularization, only one survived after 90 days. In this patient, the HA rupture occurred at postoperative day 2 and was due to a technical defect at the HA anastomosis. The anastomosis was revised and this patient is still alive at last follow-up with a functioning graft. All of the remaining five patients died before postoperative day 90.

By contrast, all of the 10 patients with HA ligation survived after postoperative day 90 (Table 3). One patient died after 27 months, in a context of poor clinical status associated with ischemic cholangiopathy. Three patients underwent an elective retransplantation but died 30, 61 and 100 months after the HA rupture. The remaining six patients are alive with a functioning graft without retransplantation (Table 2). Among the 10 patients with HA ligation, the time from transplantation to HA rupture was not significantly different in the six patients that still have a functioning graft, compared to the other four patients (51 vs. 35 days; p = 0.762). However, there were only three long-term survivors in the seven patients with HA ligation performed less than 2 months after transplant, whereas all three patients that experienced HA rupture more than 2 months after transplant and underwent HA ligation are still alive with a functioning graft. One- and 3-year graft survival probabilities according to the Kaplan–Meyer analysis were 14% and 14% respectively in patients without HA ligation versus 80% and 70% respectively in patients with HA ligation (p = 0.003). One- and 3-year overall survival probabilities were 14% and 14%, respectively, in patients without HA ligation versus 100% and 80%, respectively, in patients with HA ligation (p = 0.002) (Figure 2).

Figure 2.

Kaplan–Meier actuarial patient (A) and graft (B) survival plot in patients with and without hepatic artery ligation. This analysis excludes one patient that died before any treatment could be performed. Patients without hepatic artery ligation underwent percutaneous arterial embolization (1 case), arterial anastomosis revision (3 cases) or aortohepatic grafting (2 cases). HA = hepatic artery.


This collective review is the largest series of ruptured posttransplant HA pseudoaneurysms that suggests the efficacy of primary HA ligation on both early and late survival.

Posttransplant HA pseudoaneurysm is a rare event with an overall incidence ranging from 1% to 2% (3, 13, 14). Most of the previous series have reported various clinical presentations of posttransplant HA pseudoaneurysms and their subsequent management [1-3, 5-11, 13-22]. In fact, HA pseudoaneurysms are characterized by great clinical heterogeneity, varying from incidental diagnosis on posttransplant imaging with hemodynamic stability allowing percutaneous stenting or surgical revision to massive bleeding leading to nearly instant death. Only 48 cases of ruptured extra-hepatic pseudo-aneurysm of the HA following LT have been reported in the literature (Table 4). Moreover, it is likely that cases with uncontrollable fatal hemorrhage have been underreported. Although not always detailed, the most frequent clinical presentation is hemoperitoneum [2] (more than half of the patients in the present series), or gastrointestinal bleeding [3, 7, 13, 16, 21], with or without hemobila [5, 8, 19].

Table 4. Reported cases of ruptured extra-hepatic pseudoaneurysms and present series
AuthorsYearndPost-LT day (median, range)Clinical presentation (n)Biliary leak (n)Microbiological cultures (n)Initial management of the hepatic artery rupture (n)eOutcome (n)
  1. aLiving donor liver transplantation.
  2. bSimultaneous liver and kidney transplant.
  3. cPediatric transplantation.
  4. dTwelve patients were excluded from this review. Six of them [3, 13, 34] had an intrahepatic pseudoaneurysm, five patients [2, 3, 28, 34] had a nonruptured pseudoaneurysm, and in the last patient [34], the pseudoaneurysm did not involve the main axis of the hepatic artery, thus preserving the arterial supply of the graft.
  5. eNone: the patient died before any treatment could be performed; ligation: hepatic artery ligation without revascularization; ReVasc: pseudoaneurysm excision or ligation with hepatic artery revascularization; embolization/stenting: percutaneous embolization and/or stenting of the hepatic artery.
  6. fComplicated by hepatic artery thrombosis at postoperative day 2.
  7. gHepatic artery ligation was associated with portocaval shunt in one patient.
  8. LT = liver transplantation; GI = gastrointestinal; NR = not reported; D = dead; ReLT-D = dead after retransplantation; ReLT-A = alive after retransplantation; A = alive without retransplantation.
Madariaga et al. [15]1992755Hemoperitoneum (2)4Negative (2)Ligation (6)D (1)
   (10-70)GI bleeding (5) Enterococcus (2) ReLT-D (1)
      Fungal (2) A (4)
      NR (1)Embolization (1)D (1)
Riedmann et al.[33]1995160Hæmobila1Enterococcus, CandidaReVascfA
Goldman et al.[7]1998160GI bleedingNRNRNoneD
Settmacher et al.[34]2000342Hemoperitoneum (1)NRNRReVasc (2)D (2)
   (21-56)GI bleeding (2)  Ligation (1)A (1)
Marshall et al.[3]2001824Hemoperitoneum (6)5Fungal (4)None (1)D (1)
   (8-75)GI bleeding (2)  Embolization (3)ReLT-D (2)
        ReLT-A (1)
       ReVasc (1)D (1)
       Ligation (3)gD (2)
        ReLT-D (1)
Leelaudomlipi2003624Hemoperitoneum (4)NRFungal (2)ReVasc (6)D (4)
et al.[2]  (12-60)GI bleeding (2) Streptococcus (2) A (2)
Slater et al.[21]a2004160GI bleeding1NegativeStentingA
Sonzogni et al.[26]2004132Hemoperitoneum and GIPseudomonasEmergency ReLTReLT-D
    bleeding Enterococcus  
Alamo et al.[4]2005195GI bleeding and hemobila1NRTotal hepatectomyD before ReLT
Maleux et al.[35]b2005173GI bleedingNREmbolization + StentingA
Fistouris et al.[13]20061138Hemoperitoneum (4)5Fungal Enterococcus (10)None (2)D (2)
   (14-240)GI bleeding (2)  ReVasc (3)D (1)
    Hæmobila (5)   ReLT-D (1)
        A (1)
       Ligation [6]D (2)
        ReLT-A (4)
Jarzembowski et al.[23]a2008130GI bleeding1FungalReVascA
Jones et al.[28]c2008110Hemoperitoneum1Staphylococcus aureusNoneD
Lee et al.[25]a200817GI bleedingNREmbolizationReLT-A
Zhan et al.[30]b200818HemoperitoneumNRFungalReVascD
Soong et al.[29]a2009285GI bleedingA. Baumannii [1]ReVasc [2]D (1)
   (15-155)    A (1)
Golse et al.[24]2010116HemoperitoneumNRReVascA
Present series1729Hemoperitoneum (10)7Candida [7]None [1]D (1)
   (2-92)GI bleeding (5)  Embolization [1]D (1)
    Hemobila (1)  ReVasc [5]D (3)
    Hematoma (1)   ReLT-D (1)
        A (1)
       Ligation [10]D (1)
        ReLT-D (3)
        A (6)
Total-6532Hemoperitoneum (31)26Fungal [20]None [6]D (6)
   (2-240)GI bleeding (26) Bacterial [18]Ligation [26]D (6)
    Hemobila (7)   ReLT-D (5)
    Hematoma (1)   ReLT-A (4)
        A (11)
       Other [33]D (16)
        ReLT-D (5)
        ReLT-A (">2)
        A (10)

Several therapeutic options are possible in cases of acute bleeding, from percutaneous embolization with or without stenting [9, 14, 20, 22] up to surgical procedures, including anastomotic revision [23], aorto-hepatic grafting [2, 24], HA ligation [15], urgent or elective retransplantation [25, 26]. However, there is no consensus on the indications for these procedures and mortality in these circumstances remains very high [2, 3, 13]. Because the diagnosis of HA rupture is usually made at laparotomy, in contrast to less acute presentations in which posttransplant HA pseudoaneurysm is diagnosed by CT-scan or angiography [14], surgical procedures cannot be compared to percutaneous management.

The majority of HA ruptures occur in a context of sepsis. Most bacteriological or mycological examination of arterial wall samples and/or peritoneal tissues are positive for Candida or various bacterial species, mostly Enterococcus or Streptococcus [1-3, 6, 7, 13, 15, 26-30]. Moreover, HA rupture is usually synchronous with biliary leak [3], gastrointestinal perforation [29], intraperitoneal abscess or ascitic fluid infection [3] (13 of 17 patients in the present series). This raises two questions: First, the role of a biliary leak in the mechanism of HA pseudoaneurysm is unclear. A biliary leak was associated in 7 of 17 patients in the present series, 3 of 11 patients in the series of Fistouris et al. [13] and four of seven patients in the review of Madariaga et al. [15]. Other cases of HA rupture associated with biliary leak were also reported [11]. Posttransplant biliary leak complicated by HA rupture is especially reported after bilio-enteric anastomosis, but not always [13]. Second, any attempt to restore the arterial continuity might be compromised by infection of the periarterial environment with a high risk of hemorrhagic recurrence after anastomosis revision or aortohepatic grafting [27].

In the present series, percutaneous embolization or restoration of the arterial blood supply, either by anastomotic revision or aorto-hepatic grafting, was associated with a higher rate of mortality compared to salvage HA ligation. Two circumstances may have explained the poor results of revascularization. First, among the five patients that underwent revascularization, four patients had a mycotic pseudoaneurysm and three of these died from hemorrhagic recurrence (Table 2). Second, as hemoperitoneum was the predominant clinical presentation of HA rupture in patients undergoing revascularization procedures, it is likely that the amount of blood effusion and the hemodynamic instability may have resulted in technical difficulties. In contrast, HA ligation is much less technically demanding in these circumstances and this may partly explain the better short-term outcome. In addition, biliary leak, which were present in most of the patients undergoing HA ligation, while it was never observed in the other patients, likely had prompted the surgeons to perform HA ligation rather than arterial reconstruction. This was probably the right decision since no patient in the ‘HA ligation’ group experienced hemorrhagic recurrence and all of these patients survived after 18 months. Thus, this study confirmed what others have suggested in a previous report [15]: “Primary repair of the HA should not be performed in the presence of rupture or sepsis”. By contrast, HA ligation had achieved long-term survival in several patients (Table 4). HA ligation should not be considered as a unique procedure but as the first key step of a multimodal management in ruptured HA pseudoaneurysm, especially when a mycotic origin is suspected, i.e. if there has been a gastrointestinal wound during the liver procurement, if a systemic candidiasis has been documented prior to the HA rupture or if the rupture occurs several weeks after transplant, with associated lesions such as biliary leak or gastroduodenal perforation. By saving recipients’ lives and controlling hemorrhage, HA ligation makes subsequent management of sepsis and biliary leak possible, using percutaneous and/or endoscopic approaches, which were feasible in more than half of the patients in the present study. Even though HA ligation was usually followed by ischemic cholangiopathy requiring retransplantation, this multimodal management postponed the time of retransplantation, which was performed electively rather than urgently in three patients. Although elective retransplantation had poor results in the present series, it has recently been reemphasized that elective retransplantation was associated with better survival than urgent retransplantation, which was statistically significant [31]. Therefore, putting off the time of retransplantation after resolution of hemorrhage and sepsis might result in better overall outcomes in patients with better general and nutritional status. In addition, it has been unexpectedly observed that some patients will not need retransplantation after HA ligation (Table 4). In the present series, six patients with HA ligation are alive without retransplantation at last follow-up. Why these patients did not develop severe ischemic cholangiopathy is unclear. It is likely that the time of HA rupture has an important role. It has been recently shown that late HA occlusion in patients with HA thrombosis after LT is more likely to be associated with the development of a “neovascularized liver” [32].

The retrospective design of the present study might have led to several biases including underestimation of the incidence of HA pseudoaneurysms. Data on imaging findings before HA rupture are lacking and it was not possible to determine whether HA rupture would have been predicted by a more accurate diagnosis of asymptomatic HA pseudoaneurysms. Also, the role of preoperative chemoembolization could not be investigated here because of the small number of patients that underwent chemoembolization. Finally, due do the difference between groups with and without HA ligation, univariate analysis of survival should be cautiously interpreted. However, this is the largest series focusing on HA rupture, which represents one of the more dreadful complications of LT with challenging questioning for the transplant surgeon. On the basis of the data presented here, we recommend avoiding HA revascularization whenever a mycotic pseudoaneurysm is suspected. In contrast, HA ligation should be considered as a reasonable life-saving option because it prevents from hemorrhagic recurrence in most cases, allowing biliary complications to be managed afterward, including by percutaneous and/or endoscopic approaches. Finally, HA ligation may achieve successful long-term outcome with, or even without retransplantation.


The authors of this manuscript have no conflicts of interests to disclose as described by the American Journal of Transplantation.