To clarify inconsistencies in the literature we performed a systematic review to identify the incidence, risk factors and outcome of early hepatic artery thrombosis (eHAT) after liver transplantation. We searched studies identified from databases (MEDLINE, EMBASE, Science Citation Index) and references of identified studies. Seventy-one studies out of 999 screened abstracts were eligible for this systematic review. The incidence of eHAT was 4.4% (843/21, 822); in children 8.3% and 2.9% in adults (p < 0.001). Doppler ultrasound screening (DUS) protocols varied from ‘no routine’ to ‘three times a day.’ The median time to detection was at day seven. The overall retransplantation rate was 53.1% and was higher in children (61.9%) than in adults (50%, p < 0.03). The overall mortality rate of patients with eHAT was 33.3% (range: 0–80%). Mortality in adults (34.3%) was higher than in children (25%, p < 0.03). The reported risk factors for eHAT were, cytomegalovirus mismatch (seropositive donor liver in seronegative recipient), retransplantation, arterial conduits, prolonged operation time, low recipient weight, variant arterial anatomy, and low volume transplantation centers. eHAT is associated with significant graft loss and mortality. Uniform definitions of eHAT and uniform treatment modalities are obligatory to confirm these results and to obtain a better understanding of this disastrous complication.
Early hepatic artery thrombosis (eHAT) is a dreaded complication after liver transplantation. Ever since Starzl's first report on human liver transplantations, eHAT has been recognized as an important cause of graft loss and mortality. During the first decennia of liver transplantation, the reported incidence of eHAT was high, especially in children (42% vs. 12% in adults) (1). With improvements of perioperative care, eHAT rates have decreased.
In general, HAT is divided into two categories: eHAT and late HAT (2–5). The natural history of eHAT, or an occlusion of the hepatic artery shortly after liver transplantation, is bile duct necrosis frequently followed by uncontrollable sepsis in the immunocompromised recipient and ultimately death (6). However, interruption of the arterial blood flow is not necessarily associated with bile duct necrosis, as is known from patients with traumatic liver ruptures, in whom ligation of the hepatic artery was performed (7–9). This discrepancy in outcome can be explained by the presence of collaterals and the fact that vascularization of bile ducts is entirely dependent on the hepatic artery. After liver transplantation, these collaterals (mainly derived from the phrenic arteries) are initially absent, but have been demonstrated angiographically as early as two weeks after liver transplantation (1). Collaterals probably prevent a disastrous outcome in case of late HAT (1,10–12).
The real cause of eHAT remains a matter of debate and is in most cases unknown. Apart from surgical (technical) causes (kinking, stenotic anastomosis), several nonsurgical causes have been described (5). Other assumed causes of eHAT are rejection and sluggish flow through the hepatic artery (1,13,14). Initially, symptoms, signs and abnormal laboratory values are absent in eHAT, therefore screening by routine Doppler ultrasound screening (DUS) is of utmost importance (4). If eHAT presents with biliary leakage and sepsis (15), the prognosis is usually dismal resulting in graft loss or even patient death (16). Other vascular complications like thrombosis and/or stenosis of the portal vein or inferior vena cava have been described, but are far less common than eHAT (1,17).
Reports on eHAT are heterogeneous due to a lack of a commonly accepted definition (18–20). For example, timeframes in which eHAT is supposed to occur range from a period of two weeks (21) to 100 days (14) after liver transplantation. Due to the variability of the definition, it is hard to draw conclusions on risk factors, the impact of revascularization attempts and/or patient and graft survival. Yet the true incidence of eHAT is unknown, and estimates vary widely from 0% to 42% (22). Often the same studies are cited in regard to the incidence, but are in our view dated (1,23). Furthermore, reported risk factors are inconsistent, mainly because of a lack of widespread agreement of what constitutes eHAT.
Therefore we have undertaken a literature review with a predetermined definition of eHAT to systematically describe the reported incidence. Further goals were to update and summarize the associated risk factors for eHAT and to assess the various DUS protocols for detection of eHAT together with the different treatment modalities. Finally we evaluated the outcomes of eHAT in terms of patient mortality and graft loss.
Materials and Methods
A systematic search was conducted of The Cochrane Hepato-Biliary Group Controlled Trials Register, the Cochrane Central Register of Controlled Trials in The Cochrane Library (CENTRAL), MEDLINE, Science Citation Index and EMBASE. The search terms were: ((‘Thrombosis’[MeSH]) OR (thromboses) OR (thrombosis)) AND ((‘Hepatic Artery’[MeSH]) OR (hepatic arteries) OR (hepatic artery)) AND ((liver transplantation*) OR (‘Liver Transplantation’[MeSH])). All abstracts were entered in Reference Manager (2005) version 11.0. Updates of the search were automatically obtained from a personal NCBI account. References of the identified studies were also searched in order to identify further studies.
Inclusion and exclusion criteria
All studies published between January 1990 and January 2007, limited to the English language and humans, were considered for this review. We included only studies describing eHAT. Articles published before 1990 were excluded because new developments in organ preservation, perioperative management, immunosuppression and recipient selection have had a great impact on the survival after liver transplantation and eHAT (1,24). In order to analyze the impact of a study period on eHAT we defined two periods: articles reporting on an early period (1982–1996) and a late period (1993–2006). Twenty-four articles were excluded for this analysis, because of an undefined period or an inclusion period outside the range of the above mentioned study periods. Case-reports or studies describing less than 50 transplantations were not considered for this review. Each author independently extracted the variables from the selected reports. We resolved any differences in opinion through discussion. The identified variables derived from the selected publications were entered into a database for subsequent statistical analysis. Key variables included, but were not limited to the following: definition of eHAT and/or day of detection, number of recipients receiving a liver transplantation, the type of population described (adults, children [<18 years] or both), study period, country and transplantation center, DUS protocol, outcome after different treatment modalities, mortality and retransplantation rates, and assessment of risk factors.
Early HAT was defined as a thromboembolic occlusion of the hepatic artery that occurred within 2 months after liver transplantation. The incidence of eHAT is reported as a percentage and represents the number of transplanted livers with eHAT as a fraction of the total number of transplanted livers. Mortality and graft loss were calculated as percentage of patients who died and grafts that were lost as a result of eHAT or its sequelae. Revascularization attempts were defined as any intervention (radiological or surgical) with the aim to restore arterial blood flow of the occluded (thrombosed) hepatic artery. These interventions encompassed thrombectomy, revision of the arterial anastomosis, thrombolytic drug therapy or any combination. A successful revascularization attempt was defined as a recipient who survived with the revascularized graft. Observation was defined as best supportive care without retransplantation and/or intervention aiming at restoration of the hepatic arterial blood flow. Patients on the waiting list for retransplantation were entered into the observation group.
Continuous variables are presented as median with range and categorical variables as number with percentage. Continuous variables were compared with the Mann–Whitney U-test. Categorical variables were compared with the Pearson's chi-square test or Fisher's exact test where appropriate. We used the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL) version 14.0 for the analysis. The level of significance was set at 0.05.
Results of the search strategy
After the systematic search of the different databases, 71 studies were eligible for this review (2–4,12,13,16,17,21,25–87). Figure 1 shows the number of excluded studies with the reasons for the exclusion. Only studies describing retrospective or partially prospective data with or without controls were found. Cross-references of the identified studies did not reveal any additional papers. The definition of eHAT differed between centers, however 92.3% of the included publications defined eHAT as an event that was detected within 1 month after transplantation.
Incidence of early hepatic artery thrombosis
The studies included in this systematic review comprised 21 822 liver transplantations and 843 cases of eHAT. The mean and median incidences of eHAT were 3.9% and 4.4% respectively. The studies reporting on eHAT were subdivided into three categories according to the population described. These categories were adults only, children only and studies describing both adults and children as one group (Table 1). The incidence of eHAT was significantly higher in children (8.3%) than in adults (2.9%, p < 0.001, Odds ratio 3.2, 95% CI 2.8–3.8). The incidence of eHAT reported in the study period between 1982 and 1996 was 6.9% versus 3.8% in the study period between 1993 and 2006 (p < 0.008). No significant difference in incidence of eHAT was noted among the transplantation centers in Europe (3.7%, n = 27 studies), Asia (3.3%, n = 10 studies), North America (4.6%, n = 25 studies) or other continents (4.8%, n = 9 studies). By choosing a cutoff point (median) of 30 transplants per year, we found that the incidence of eHAT was higher in low volume centers (5.8% vs. 3.2%, p < 0.005). There was no difference in incidence of eHAT between studies reporting only on living donor liver transplantation (3.1%, n = 14 studies) versus studies reporting on deceased donor liver transplantation (4.6%, n = 57 studies; p = 0.1). No difference in incidence of eHAT in the living donor liver transplantation (LDLT) subgroup was found in centers using the operation microscope (12,28,42,51,52,71,73,76,77) for the arterial anastomosis (3.1%) versus centers using loupe magnification (2.1%, p = 0.6) (25,37,53,54,74).
Table 1. Publications are categorized in those describing eHAT after liver transplantation in adults only, children only or in adults and children combined
Continuous variables are presented as median (range) and categorical variables as number (percentage).
Almost all centers screened for eHAT in the postoperative period, however only 54 publications described their screening protocol. In only two centers routine DUS was not performed (3,46). The screening protocols for eHAT are highly variable with respect to frequency and interval of screening, and the time period after operation during which screening was performed (Table 2). The peak incidence of eHAT was obtained from reports in which centers performed DUS daily or more frequently, for a period of at least one week after transplantation. The median time to detection of HAT was 6.9 days (range 1–17.5 days postoperative) (12,27,28,33,42,44,48,51,52,54,62,65,71,84). The usual pattern of screening was via routine DUS, while in case of a suspicion of eHAT a (computed tomography)-angiography or laparotomy was performed. All centers performed Doppler ultrasound examinations on clinical grounds and/or in combination with laboratory findings.
Table 2. Doppler ultrasound screening protocols for adults and children separately
Number of transplantation centers
n/a = not applicable, po = postoperative, DUS = Doppler ultrasound screening.
Once a day
Twice po at day 0–2 or 5–7
Day 1, 2, 3, 4, 5 po
Day 1, 2, 3 po
Day 1, 3, 7 po
Day 1, 3, 5 po
At least 1 DUS within 3 days po
Different treatment modalities
In general, there are three different treatment modalities for eHAT: revascularization, retransplantation and observation (Table 1). Revascularization was attempted in 257 out of a total of 510 cases of eHAT, as reported in 47 studies. Two centers did not perform revascularization attempts at all (2,76). In only 32 studies, describing 163 revascularization attempts in 315 cases of eHAT, was the outcome reported according to our definitions (12,16,17,21,25,27,28,33,35,41–43,45–48,53–55,57,59,67,70–73,80,82,84–87). Revascularization was attempted in 75% of the adults and in 54.1% of the children (p = 0.5) in centers that reported the outcome of an attempt. The success rate of a revascularization attempt is somewhat similar among these groups: overall around 56%. A correlation was found between early occurrence of eHAT and a high success rate of a revascularization attempt (r =−0.73, r2= 0.53, p < 0.03), when daily DUS was performed (12,27,28,33,42,48,54,71,84). The success rate of a revascularization attempt was 66.1% when DUS was performed daily or more frequently, versus 44.9% with a less frequent screening protocol (p = 0.17). In adults these success rates were 61.4% versus 44.8% (p = 0.36), and in children 91.7% versus 57.5% (p = 0.13). In 57 of the 163 revascularizations (median 30.3%) a retransplantation was necessary after a revascularization attempt. Retransplantation in general was performed in 260 cases, and described in 43 studies (2,12,16,17,21,25,27,28,30,32,33,35,41–43,45–48,53–56,59–65,67–72,77,80,82,84–87). Six centers did not perform a retransplantation after eHAT (12,28,46,54,77,80). Retransplantation was the treatment of choice in 53.1% of the cases of eHAT, and was more frequently performed in children than in adults (61.9% vs. 50%, p < 0.03). Sixty-four observations were described in 39 studies, while fifteen studies had zero observations. In only 20 studies, describing 49 observations and 266 cases of eHAT (median 21.5%), was the outcome reported (16,17,30,35,40,41,45,48,57,59,62–67,70–72,80). The overall mortality in this group was 55.1%; no significant differences were seen between adults versus children.
Mortality and risk factors
The overall mortality of eHAT was 33.3% with a range from 0% to 80% and was significantly higher in adults than in children (34.3% vs. 25%, p < 0.03) (2,12,16,17,21,27,28,30,35,41,42,45,46,48,51–59,61,62,64,65,67–73,77,80,82,84–87). The mortality rate of eHAT in patients with a LDLT was comparable to that in patients transplanted with deceased donor livers (39% vs. 28.6%, p = 0.47). Risk factors for the development of eHAT were retrieved from those studies in which a significantly contributing factor for the development of eHAT was described. These risk factors are categorized in Tables 3–6. A summary of concordant and discordant risk factors is shown in Table 7.
Table 3. Manuscripts reporting donor factors as a risk factor for early HAT
1Pediatric recipients < 5 years.
2Details of multivariate analysis.
Varotti et al. 2005
Donor age greater than 60 years
Low donor age
p = 0.001
Meneu-Diaz et al. 2004
Lack of ABO compatibility
p = 0.022
Reding et al. 1999
p = 0.02
Heffron et al. 2005
Reduced size grafts
Whole liver grafts
p = 0.045
Rela et al. 1996
Whole liver grafts
Reduced size grafts
p = 0.01
Rela et al. 1996
Low donor age (mean 6.5 years)
Mean donor age 14.2 years
p = 0.04
Sieders et al. 2000
Low donor/recipient age ratio
p < 0.02
Oh et al. 2001
Recipient/donor weight ratio >1.25
p = 0.022
Table 4. Manuscripts reporting perioperative factors as a risk factor for early HAT
The aims of this systematic review were to identify the true incidence of eHAT, to analyse the risk factors for the development of eHAT, to gain insight into the various screening schedules used to detect eHAT and to report on the outcome of the different treatment options available for eHAT. The incidence of eHAT was 4.4%, with a range from 0% to 20%. Although this incidence seems to be rather low, it is extremely relevant because of the clinical consequences. The studies presented in this systematic review uniformly report eHAT as a major cause of graft loss (53.1%) and mortality (33.3%) in the early postoperative period. Additionally, the frequent need for a retransplantation is a burden to the already limited donor pool. Furthermore, the sequelae of eHAT are associated with an exponential increase in medical costs (88,89).
Several risk factors are identified in the published literature. This review confirms the often presumed higher incidence of eHAT in children: the median incidence in children is 8.3% versus 2.9% in adults (p < 0.001). The real cause of this difference is a matter of speculation. However, the most likely explanation is the small size of the vessels with associated technical difficulties of anastomosing (52,90,91). The higher incidence of eHAT associated with low recipient weight supports this view (2,85). Although technical failure is considered to be the cause of HAT in 20% of the cases (92), it is difficult to asses the true cause. Part of the problem may derive from the inability of observational studies to identify and measure all relevant covariates that may influence the outcome; similar studies using a few covariates can sometimes come to opposite conclusions (Tables 3–6). Concordant and discordant risk factors for eHAT are summarised in Table 7. A cytomegalovirus (CMV) recipient/donor mismatch emerged as a concordant risk factor. The higher proportion of seronegative recipients receiving a seropositive graft is especially relevant in the context of the higher incidence of eHAT in children. Seronegative recipients have a significantly higher chance of developing eHAT (2). It is a fact that CMV infection is increasingly prevalent with age; thus, most adult donors will be infected (93,94). This may partly explain the higher incidence of eHAT in (CMV seronegative) children receiving seropositive adult donor livers. Other nonsurgical causes contributing to eHAT have recently been reviewed (5). These authors highlighted the relevance of CMV mismatch in the context of eHAT. Another factor, which may contribute to the higher incidence of eHAT and a CMV mismatch is the fact that a primary CMV infection is associated with a procoagulant response (5). Recently, familial amyloidotic polyneuropathy was found to be an independent risk factor for eHAT (95). This suggests a new nonsurgical cause.
An important risk factor for eHAT is retransplantation. In a multivariate analysis of risk factors for development of eHAT in 1257 transplantations, retransplantation was associated with a six-fold greater risk than primary transplantations (3). Although, in one report this risk factor was only significant in the univariate analysis (p = 0.043) (2). A possible explanation is the smaller number of transplantations described in the publication of Oh et al. (n = 424) as compared to the population of Silva et al. (n = 1257). Primary transplantations were associated with an eHAT incidence of 1.6%, increasing to 4.8% and 12.5% in second and third (re) transplantations respectively (2).
Reconstruction of an anatomical variant or damaged hepatic artery has been mentioned as a risk factor for development of eHAT; it is difficult, however, to identify the real contributing factor because prolonged operation time, the presence of extra anastomoses or inadvertent handling during organ recovering might also increase the risk of thrombosis (33). One of the discordant risk factors for eHAT is the use of conduits for arterial revascularization. Three out of four publications considered the use of a conduit to be a risk factor for development of eHAT. Notably, two of these three (2,3) were the result of a multivariate analysis. In these two publications, eHAT developed in about 15% of patients with arterial conduits versus less than 2% in the group of patients without conduits. Two publications reported the results of univariate analysis of which only one described the conduit as a contributing risk factor. Thus, there are strong arguments to consider an arterial conduit to be a risk factor for eHAT.
It is a general belief that reduced size liver grafts are associated with a lower eHAT incidence than whole liver grafts (96,97). The underlying idea is that reduced size liver grafts are often adult grafts (24), which have relatively larger vessel diameter and thus a technically less difficult anastomosis (98,99). For this review, we found only two publications dealing with the issue of whole versus reduced size grafting (55,64). Unfortunately, these authors found conflicting results in children receiving a liver transplantation: Rela et al. identified whole liver grafts as a risk factor for eHAT, whereas Heffron et al. found reduced size grafts to be associated with an increased risk.
Our study indicated that low volume transplant centers had an increased risk of eHAT. This is in line with publications that report higher complication rates for high-risk surgery in low volume hospitals (89,100). The relation between high volume centers and low eHAT incidence was recently confirmed by a study in adult living donor liver transplantation (101). Moreover, the eHAT incidence decreased over time, with a higher overall incidence before 1996. Several factors may contribute to the decreased incidence of eHAT: gain of experience, more refined techniques in vascular surgery and increasing knowledge on rheology and pro-, and anticoagulant factors.
The incidence of eHAT in LDLT has been reported to be significantly lower in comparison with the non-LDLT (62,102). We could not confirm this in the present systematic review. Microvascular techniques may decrease the incidence of eHAT significantly in LDLT, though there is no clear evidence yet to support this view. In one study, which was not included in this systematic review because of an unclear definition of eHAT, this particular technique reduced the incidence of HAT from 22% to 0% (p < 0.05) in pediatric recipients and significantly increased actual graft survival (103). This study, however, used historical controls, comparing their most recent incidence of HAT with that before 1996. Other centers reported that microsurgical techniques and/or microvascular surgeons are essential in reducing the incidence of eHAT in LDLT (28,42,73,76,90,104–108). Our review found no differences in the incidence of eHAT, when microvascular techniques were compared to other methods of anastomosing in LDLT. The limited number of studies, and as a result, limited number of cases of eHAT according to our definition in the LDLT-subgroup precluded firm conclusions.
Finally, primary sclerosing cholangitis was recently found to be an independent predictor of eHAT; patients transplanted for this indication had a higher risk of eHAT (OR 1.63, 95%CI 1.10–2.40) (109). Accordingly, we advocate postoperative screening and consideration of appropriate prophylactic (anticoagulant) therapy for patients with the identified risk factors for eHAT as summarized in Tables 3–6.
In almost one third of the publications, no DUS method was mentioned. The most remarkable outcome was the lack of consensus for DUS screening, both with respect to the frequency and the timing of surveillance after transplantation.
Little is known about the treatment and outcome of eHAT after liver transplantation. Surprisingly, only about half of the 71 publications included in this systematic review mentioned the outcome after revascularization (32 publications) and/or retransplantation (43 publications). Traditionally, retransplantation was the first choice of therapy in case of eHAT. However, this approach is restricted by a limited donor pool. The disparity between the number of potential recipients and the available donors will continue to grow (110). Urgent revascularization of eHAT as a first option offers the opportunity to prevent retransplantation, but probably only in cases of early detection. A revascularization attempt is performed in approximately half of the cases of eHAT, with a reported success rate of about 50%. Nevertheless, some centers lack a back-up system for urgent retransplantation, making urgent revascularization of paramount importance (12,80). In countries where LDLT is the most common type of liver transplantation, difficulties of finding a suitable donor in an emergency more often lead to mortality of the recipient in case of eHAT (28,42,106).
The overall 1-year graft survival rate after liver transplantation for eHAT is approximately 50%, while other vascular complications have graft survival rates up to 86% (29). The 1-year survival rate after retransplantation for eHAT has been reported as low as 60% (111), 55% (112) or 20%. (113). The overall one-year survival rate after retransplantation in children is lower (64%) than in case of primary transplantations (85%, p = 0.004) (114). Besides the lower survival rate, retransplantation is associated with increased length of hospitalization as well as increased costs (97). Asymptomatic eHAT detected by DUS and treated with early revascularization shows promising results (16,45,84,115). It is, however, important to detect eHAT while the patient is still asymptomatic, because graft survival after revascularization is much better in this group (81.8%) versus 40% in symptomatic patients (16). Nishida et al. report a clear and significant relation to urgent revascularization after eHAT and a better outcome in terms of morbidity and mortality (59). Urgent revascularization has been described by others as a useful treatment option for early HAT (4,16,45,46,84,85,115–117). In conclusion, early detection and thus more frequent screening in the early postoperative period might contribute to a better outcome.
Publication bias is an issue for all systematic reviews. In general, this bias exists when studies report only on positive or substantial differences. It remains possible that studies containing valuable data may have eluded our detection. Language bias may have affected our conclusions, as we limited this systematic review to the English language. A remarkable finding of this review is the significant number of studies of eHAT that were published during the last 18 years: at least 71 studies describing 21 822 liver transplantations. The fact that the total number of studies more than tripled over the past ten years indicates that there is still is uncertainty and discontent with the occurrence and sequelae of eHAT. Interestingly, the overwhelming majority of the papers (687 out of 999) reporting on eHAT had to be eliminated from this review due to a lack of definition of eHAT. Many publications reported both on late and eHAT, which we and others consider to be completely distinct entities. Thus, the major problem is the lack of a commonly accepted definition of eHAT, leading to poor reproducibility. For this review, we arbitrarily defined eHAT as thromboembolic occlusion of the hepatic artery within 2 months after transplantation. Of note, 92.3% of the publications included in this review defined eHAT as an event that was detected within 1 month after transplantation. In order to permit reliable comparisons between studies on eHAT, we advocate guidelines for future studies reporting on this topic (Figure 2).
In conclusion, eHAT still contributes significantly to mortality and graft loss after liver transplantation. The relevance of eHAT is clear: we cannot afford any graft loss due to the increasingly limited donor pool. Early HAT should be reduced to the absolute minimum. As a consequence, the benefits of reducing this dreaded complication are obvious in terms of patient survival and diminishing the associated huge costs of retransplantation. The recommendations for reporting on eHAT, as suggested in this review, will offer the opportunity to more reliably interpret the results of future publications. Additionally, the role of early detection should be evaluated with respect to the possibility of saving grafts with eHAT by performing urgent revascularization. This can, in our opinion, only be guaranteed by using implantable Doppler probes performing continuous flow registration (118–123). A prospective study on the accuracy of continuous flow monitoring after liver transplantation is currently performed in our center to test this hypothesis.
The authors gratefully acknowledge Dr. Diane Black for critically reviewing our manuscript. This study was supported by the Netherlands Organization for Health Research and Development (ZonMW 945-05-013 to KPdJ).
Conflict of Interest Statement
The authors do not have any conflict of interest to report.