Vivarelli M, Cucchetti A, La Barba G, Bellusci R, De Vivo A, Nardo B, Cavallari A, Pinna AD. Ischemic Arterial Complications After Liver Transplantation in the AdultMultivariate Analysis of Risk Factors. Arch Surg. 2004;139(10):1069–1074. doi:10.1001/archsurg.139.10.1069
To minimize the incidence of ischemic arterial complications, risk factors should be clearly identified. Knowledge of the predisposing factors for such complications would make possible the institution of strict surveillance protocols that could ensure early detection of complications and so prevent the progression of ischemic damage to graft failure.
Retrospective univariate and multivariate analysis.
Six hundred fifty-three adults who underwent 747 orthotopic liver transplantations.
Main Outcome Measures
We used univariate and multivariate analyses to retrospectively assess the role of possible risk factors for early and late HA thrombosis (HAT) and stenosis (HAS), including etiology of liver disease, donor and recipient sex and age (aged ≤60 vs >60 years), cause of donor death, preservation solution, cold ischemic time, previous orthotopic liver transplantation, HA back-table reconstruction, direct arterial anastomosis vs interpositional conduit, experience of the surgeon, intraoperative transfusion requirements, acute rejection, and cytomegalovirus infection.
We observed 58 ischemic complications, including 26 early HAT, 13 late HAT, and 19 HAS. Independent predictors of early HAT were donor age greater than 60 years and bench reconstruction of anatomical variants of the HA; of late HAT, arterial anastomosis fashioned using an interpositional graft of donor iliac artery (iliac conduit) and donors who died of cerebrovascular accident; and of HAS, previous orthotopic liver transplantation and cytomegalovirus infection.
Predisposing factors for HAT mostly stem from donor and graft features. Use of iliac conduits should be limited, particularly when using old donors. Frequent screening of the arterial flow to the graft with Doppler ultrasonography is advisable in patients at risk.
Complications that affect the hepatic artery (HA) after orthotopic liver transplantation (OLT) can lead to ischemia of the liver graft, which can result in graft morbidity or loss or even patient death. The clinical feature of these complications varies greatly and depends on the type (thrombosis or stenosis) and timing (early or late presentation with respect to the time of transplantation) of the arterial disease and the promptness of the diagnosis.1- 5
Despite continuous improvements of the surgical technique, these complications still represent one of the main causes of failure of liver transplantation, with an incidence ranging from 2.6% to 20% in adult recipients.1- 6
Retransplantation used to be the only possible therapy for these conditions and involved high patient mortality and wasting of already short resources.7 Great advances were recently made in the treatment of HA thrombosis (HAT) or HA stenosis (HAS) that can spare the need for retransplantation through graft salvage procedures8,9; however, revascularization procedures can be successful only when the disease of the HA is diagnosed within hours or a few days of its development, a target that can be achieved only through a strict surveillance program with Doppler ultrasonography.10,11
As it is hard to establish a really effective screening of each patient undergoing transplantation, the knowledge of the risk factors for HAT or HAS is of paramount importance to guide the frequency of the controls on the basis of the presence of these factors.
Despite this issue's clinical relevance, few analyses of the conditions associated with the single possible clinical ischemic events of the liver graft are available in the literature. The available analyses1,2,4- 6,12,13 are largely drawn from experiences of the early days of liver transplantation, and most fail to demonstrate clear risk factors.
The aim of the present study was to retrospectively review a large series of consecutive liver transplantations in adults to thoroughly analyze with univariate and multivariate testing the conditions related to the graft itself, the surgical technique, and some of the postoperative events that may be associated with the development of complications affecting the HA. We analyzed the arterial complications according to their type and the timing of clinical presentation.
From April 9, 1986, to November 8, 2002, 747 consecutive OLTs were performed in 653 adult patients at the Department of Surgery and Transplantation of the University of Bologna, Bologna, Italy. Of these 747 OLTs, we excluded 27 from the analysis of risk factors for ischemic arterial complications. Thirteen of the 27 were performed on patients who died within 24 hours of transplantation (none of these patients had evidence of HAT or HAS). In 3 transplantations, a kinking of the arterial anastomosis was demonstrated, and early HAT, late HAT, and HAS developed in 1 patient each; these complications were considered a clear consequence of a technical failure and excluded from the analysis. In 2 transplantations, arterial anastomosis was performed with the use of a polytef graft (Gore-Tex; W. L. Gore & Associates, Flagstaff, Ariz) or a direct anastomosis on the splenic artery (in this latter case an HAS was observed). As there is not enough literature to establish whether the results of such procedures are similar to those of the standard ones, these unusual ways of reconstruction of the HA were considered additional risk factors for vascular complications. Six transplantations used split-liver grafts from cadaveric donors. As the incidence of HAT is significantly higher in split-liver transplantation, these cases were excluded.14 Finally, we excluded 3 transplantations performed on the same patient who was diagnosed as having paroxysmal nocturnal hemoglobinuria after the loss of 3 grafts due to HAT.
The analysis of risk factors for ischemic arterial complications was therefore performed in 720 adult OLTs with a mean follow-up of 1311 days.
The arterial anastomosis was fashioned with a running 7-0 polypropylene (Prolene suture; Ethicon Inc, a Johnson & Johnson Co, Somerville, NJ) suture between the celiac artery or the common HA of the graft and the recipient, usually at the level of the origin of the gastroduodenal artery, in 612 transplantations (85%). When the HA of the recipient was not suitable for a direct anastomosis, a donor iliac artery graft was used as an interpositional conduit between the recipient infrarenal aorta and the donor celiac or hepatic artery. An iliac conduit was performed in 108 transplantations (15%); all but 2 conduits came from the same donor of the liver graft. Right accessory branches to the graft arising from the superior mesenteric artery were reconstructed on the back table with an end-to-end anastomosis between the proximal stump of the celiac artery and the proximal stump of the superior mesenteric artery; in 4 cases the right accessory artery was anastomosed to one of the branches of the graft's hepatic artery (gastroduodenal artery in 2 cases, splenic and left gastric artery in 1 each). Bench arterial reconstruction was performed in 74 cases (10.3%).
Allografts were preserved with University of Wisconsin solution (Viaspan; Dupont Pharma, Paris, France) in 470 cases (65.2%), Celsior solution (IMTIX-SangStat, Lyon, France) in 215 (29.9%), Euro-Collins solution (Institut Fresenius, Hamburg, Germany) in 27 (3.8%), and histidine tryptophan ketoglutarate solution (Custodiol; Dr Franz Kohler Chemie, Alsbach-Hahnlein, Germany) in 8 (1.1%).
All the patients who experienced arterial complications had received ABO-identical grafts.
Immunosuppressive therapy consisted of cyclosporine or tacrolimus, corticosteroids, and, in some cases, azathioprine sodium or Orthoclone OKT3 (muromonab-CD3; monoclonal antibodies anti-CD3). Mean cold ischemic time was 8.5 hours (SD, 2.3 hours).
Duplex Doppler ultrasonography was performed on posttransplant days 1, 4, and 7 and before discharge; at the monthly follow-up as a routine test of vascular patency; and whenever indicated by clinical or biochemical findings. In the presence of abnormal findings, HA angiography was performed in all cases to confirm the diagnosis.
We defined HAT as the complete occlusion of arterial blood flow to the allograft and classified it as early when it occurred within 1 month after transplantation and as late when it occurred after this point; the diagnosis was based on Doppler ultrasonographic findings and confirmed by means of hepatic angiography.
We suspected HAS in the presence of a resistive index of less than 0.5 and prolonged systolic acceleration time at Doppler investigation.10 All stenoses were confirmed by means of hepatic angiography showing incomplete occlusion of arterial blood flow to the allograft due to the presence of a narrowing of the HA.
Possible risk factors for HAT and HAS included etiology of recipient end-stage liver disease (related to hepatitis B or C virus, or nonviral), previous liver transplantation, donor sex and age (≤60 vs >60 years), cause of donor death (cerebrovascular accident vs other causes), recipient sex and age (≤60 vs >60 years), type of preservation solution, cold ischemic time, HA back-table reconstruction, type of arterial anastomosis (defined as direct when the celiac axis or the HA of the graft was anastomosed to the HA of the recipient and as conduit when an iliac artery harvested from the donor was interposed between the infrarenal aorta of the recipient and the HA of the graft), experience of the surgeon who performed the procedure (considered low when ≤30 previous liver transplantations had been performed), intraoperative transfusion requirement of red blood cells and fresh frozen plasma, acute rejection confirmed by liver biopsy results, and cytomegalovirus (CMV) infection.
We defined CMV infection as IgM serum conversion in the presence of CMV syndrome (fever, leukopenia, and thrombopenia) or by means of virus isolation in tissue specimens. Since 1993, routine screening for CMV infection with monitoring of pp65 antigenemia has been performed.
Until 1998, we treated preemptively with intravenous ganciclovir sodium each patient with a positive finding for antigenemia. Since 1998, following a prospective trial, we treat preemptively only those patients with a positive finding for antigenemia with a level above 50 copies/200 000 polymorphonuclear leukocytes unless specific risk factors (eg, mismatch of the donor/recipient serological CMV status, treatment of rejection with high-dose steroids or with antibodies to lymphocytes) or symptoms suggestive of CMV disease are present. In the presence of risk factors, any positive finding for antigenemia is treated preemptively.15
Univariate analysis was performed for categorical variables with the use of Pearson χ2 testing to identify independent risk factors for early HAT, late HAT, and HAS after liver transplantation. We analyzed continuous variables with 2-tailed unpaired t test. On univariate analysis, P = .05 was considered significant. Variables with P = .05 in the univariate analysis were entered into a forward stepwise logistic regression analysis to estimate the odds ratio (OR) of each artery complication (dependent variables) and the presence or absence of potential prognostic factors (independent variables). The OR was defined as the exp[β-coefficient] with 95% confidence intervals (CIs). Kaplan-Meier estimates were used to calculate graft survival curves. Differences in survival curves were compared using log-rank statistics.
A total of 58 complications (8%) that affected the HA were documented in 720 consecutive transplantations. Early HAT occurred in 26 cases (3.6%); late HAT, in 13 (1.8%); and HAS, in 19 (2.6%). The actuarial survival rates for all patients in whom vascular complication developed were 74%,70%, and 64% at 1, 3, and 5 years, respectively. The 1-, 3-, and 5-year survival rates of grafts targeted by ischemic complications were 32%, 24%, and 19%, respectively (vs 77%, 72%, and 69%, respectively, of grafts without ischemic arterial complications; P<.001).
The time interval between transplantation and the diagnosis of early HAT ranged from 1 to 28 days (mean, 7.4 days; median, 4.5 days). Early HAT represented the most common vascular complication in our series, accounting for 45% of all of the ischemic arterial complications (26 HAT of 58 total ischemic complications). Twenty-six cases of early HAT occurred in 26 patients. Of these, 6 were successfully treated with urgent surgical revascularization (removal of the thrombus with a Fogarty balloon-tip catheter and refashioning of the arterial anastomosis performed the same day of the diagnosis), 16 underwent retransplantation, and 4 died awaiting retransplantation.
Etiology of recipient end-stage liver disease, previous liver transplantation, donor sex, cause of donor death, recipient sex and age, type of preservation solution, cold ischemic time, type of arterial anastomosis, experience of the surgeon who performed the procedure, intraoperative transfusion requirement of red blood cells and fresh frozen plasma, acute rejection, and CMV infection were found not to be associated with early HAT, whereas donor age of greater than 60 years and back-table artery reconstruction were found to be significantly associated with the complication. The incidence of early HAT in 175 transplants using a liver from a donor older than 60 years was significantly higher than that recorded when using a liver from a younger donor (8.6% vs 2.0%; P = .001). The 74 transplants that required back-table artery reconstruction had a higher incidence of early HAT when compared with transplants that did not require any artery reconstruction (10.8% vs 2.8%; P = .001) (Table 1).
At logistic regression, donor age (OR, 4.26 [95% CI, 1.90-9.54]; P = .001) and artery reconstruction (OR, 3.78 [95% CI, 1.56-9.22]; P = .003) proved to be independent predictors of early HAT at logistic regression. Early HAT developed in 5 of 24 recipients (20.8%) who had the simultaneous presence of these 2 factors (OR, 8.46 [95% CI, 2.88-24.82]).
The time interval between transplantation and the diagnosis of late HAT ranged from 50 to 2405 days (mean, 689 days; median, 273 days). In 13 cases, graft failure developed in 10 patients who underwent retransplantation, 1 underwent successful revascularization, and ischemic damage of the common bile duct that developed in 2 was treated with Roux-en-Y bilioenteric anastomosis.
Etiology of recipient liver disease, previous liver transplantation, donor sex and age, recipient sex and age, type of preservation solution, cold ischemic time, HA back-table reconstruction, experience of the surgeon who performed the procedure, and acute rejection were not associated with late HAT.
Late HAT was more common among the recipients of 377 donors who died of cerebrovascular accident (3.2%) when compared with the 345 who died of other causes (0.3%; P = .004). The use of an arterial conduit during transplantation was associated with 6.5% of HAT compared with 1.0% for the modal type of arterial anastomosis (P = .001). Cytomegalovirus infection was significantly associated with an increased incidence of late HAT (4.7% vs 1.3%; P = .02) (Table 1).
After logistic regression, independent predictors of late HAT included only cause of donor death (OR, 6.92 [95% CI, 2.23-21.54]; P = .001) and use of an iliac conduit (OR, 10.69 [95% CI, 1.37-83.56]; P = .02).
The time interval between transplantation and the diagnosis of HAS ranged from 2 to 1117 days (mean, 210 days; median, 112 days). Of 19 cases observed, 2 never showed abnormalities in liver function and had no treatment. Three were sucessfully treated with angiographic balloon dilatation; 6, with angiographic ballooning and stent positioning; and 4, with surgical revascularization. The remaining 4 cases progressed to graft failure and underwent retransplantation.
Etiology of recipient end-stage liver disease, recipient sex and age, type of preservation solution, cold ischemic time, HA back-table reconstruction, type of arterial anastomosis, experience of the surgeon who performed the procedure, and acute rejection were not associated with HAS.
Previous liver transplantation (P = .05), female donor sex (P = .01), donor age greater than 60 years (P = .02), graft harvested from a donor who died of a cerebrovascular accident (P = .005), and CMV infection (P = .006) were associated with the development of HAS at univariate analysis (Table 1). Previous OLT (OR, 3.32 [95% CI, 1.12-9.89]; P = .03) and CMV infection (OR, 3.32 [95% CI, 1.24-8.86]; P = .02) proved to be independent predictors of HAS at the logistic regression.
Thrombosis of the HA is still one of the main causes of graft loss after liver transplantation. Its incidence remains high, even in the more recent reports from highly specialized and experienced transplantation units, which makes it hard to ascribe to difficulties in handling what is now a well-established technique. In the series from the University of Pittsburgh, Pittsburgh, Pa, even in the more recent era of a retrospective analysis of 4000 consecutive transplants, HAT was the indication for nearly one third of the retransplantations performed.16 In the present study, the experience of the surgeon who performed the operation was not linked to a different incidence of complications to the HA. To our knowledge, this evidence has never been demonstrated before. However, to minimize the impact of technical factors on the results, we excluded from the analysis those cases where the surgical reconstruction of the artery could be considered inappropriate (3 patients with evidence of kinking of the HA) or unconventional (use of a polytef graft or a direct anastomosis on the splenic artery).
Few analyses of the risk factors for ischemic graft complications are available in the literature, of which the only multivariate one is from Oh and colleagues.6 In that study, however, a significant number of pediatric recipients were included, which influenced the results of the analysis; for instance, some of the main risk factors identified, such as recipient body weight of less than 15 kg or recipient/donor body weight ratio of greater than 1.25, can hardly be implicated in adult transplantation. As the incidence of HAT in pediatric recipients is much higher than that in adults, we limited the analysis to adult recipients in the present study.11
Our comprehensive retrospective univariate and multivariate analyses of some 16 different variables identified the following independent risk factors: donors older than 60 years and the need for bench reconstruction of anatomical arterial variants for early HAT, use of an iliac conduit to revascularize the graft and grafts from donors who died of a cerebrovascular accident for late HAT, and previous OLT and CMV infection for HAS. Most risk factors for HAT after liver transplantation are therefore linked to the donor or the graft features.
In a recent study, donor age of older than 60 years was shown to be an independent predictor of preservation injury17; this latter occurrence might affect the arterial bed of the graft that is already suboptimal because of the likely presence of atherosclerosis. On the other hand, grafts from donors who died of a cerebrovascular accident are more likely to carry relevant abnormalities of the arteries that are difficult to detect at harvest.18 The use of organs from old donors has, in the experience of our center, allowed a consistent expansion of the number of transplantations. We used 176 grafts harvested from donors older than 60 and to a maximum age of 87 years (median age, 69 years). Although there was no difference in the incidence of primary graft nonfunction or in the 3-year liver function, this study shows that ischemic complications, and in particular early HAT, occur more frequently with these grafts.
Back-table reconstruction of anatomical abnormalities of the HA was an independent prognostic determinant of early HAT in our series; although it has been mentioned by others,2 this is the first time that this evidence has been supported by a statistical analysis.
We observed a nearly 9-fold increase of the risk of early HAT for those patients whose graft was harvested from a donor older than 60 years and required a bench reconstruction of an anatomical variant; we advise routine daily monitoring of the arterial flow in the early postoperative period in the presence of one or both of these risk factors.
The use of an interpositional arterial graft to revascularize the liver was stated to be a risk factor for HAT by Drazan and colleagues,2 although no statistical evidence was reported. A few years later, Muiesan and colleagues19 found no difference in the incidence of HAT between direct anastomosis and arterial conduit. Oh and colleagues6 found that only the presence of an "old" arterial conduit was linked to early HAT after retransplantation. In the present study, the use of the iliac conduit to revascularize the graft was an independent risk factor for late HAT, confirming the results of a recent univariate analysis from Stange et al13; this finding might be explained by the discrepancy in diameter between the iliac artery harvested from the donor, usually consistently wider, and the HA of the graft that is likely to produce turbulence in the blood flow at the site of the anastomosis. Furthermore, DeBakey et al20 demonstrated that the iliac arteries are the most common site of severe atherosclerotic disease; this can be particularly relevant when the iliac arteries are harvested from old donors. We now limit the use of arterial conduits to those cases where the HA of the recipient is truly not suitable for a direct anastomosis owing to the small size of the vessel or to specific alterations of its wall and preferentially use the carotid artery harvested from the donor rather than the iliac artery as an interpositional graft. The caliber of the internal or external carotid is usually more similar to that of the hepatic artery, which allows a more anatomical reconstruction. In the absolute need of an interpositional arterial graft to revascularize livers from old donors, we recommend the use of preserved vascular grafts procured from different and younger donors.21
Liver transplantation involves many events that may affect the vasculature of the graft. It was shown in the past that an imbalance between the procoagulant and anticoagulant factors synthesized by the liver favors a hypercoagulable state in the first few days after liver transplantation.22 Infectious or immunological events that can follow transplantation, namely CMV infection and rejection, can affect the endothelium of the graft arteries and are suspected to be triggers of thrombotic events.6,23
Madalosso and colleagues23 demonstrated an increased incidence of HAT in CMV-seronegative recipients of a graft from CMV-seropositive donors. Although CMV infection usually develops a few weeks after transplantation, in that23 and other24,25 reports HAT surprisingly developed in the very first days after transplantation in the absence of clinical evidence of CMV disease and results of serological screening tests for CMV infection.
More recently, Gunsar and colleagues12 found a higher incidence of late HAT in those recipients who had CMV infection.
The results of our study confirmed the relevance of a clinically documented CMV infection as a risk factor only for the development of HAS; this finding is more consistent with the clinical course of CMV infection, whereas systematic screening of CMV infection with highly sensitive tests such as pp65 antigenemia is performed and preemptive ganciclovir treatment is instituted before clinical evidence of infection.15 This policy may stop the progression of the endothelial damage caused by CMV whose repair may nevertheless result in arterial stenosis. In our series, there was no relationship between severity of CMV infection and development of HAS; CMV infection of those patients who developed HAS had always been treated preemptively with intravenous ganciclovir with no evidence of tissue-invasive CMV disease.
Hepatic artery stenosis usually developed several months after OLT in our patients. In the presence of risk factors for HAS or late HAT (eg, a graft from a donor who died of a cerebrovascular accident, arterial conduit fashioned at transplantation, retransplantation, and CMV infection), a Doppler investigation of the patency of the HA should be part of every outpatient follow-up. If detected promptly, complications affecting the HA are often susceptible to conservative treatment, sparing retransplantation, which remains a major failure in terms of waste of resources and patient mortality.7- 9,11
In a recent study by Nishida and colleagues,11 daily monitoring of the arterial flow with Doppler ultasonography reduced the need for retransplantation to zero in the presence of early HAT in pediatric recipients.
Unfortunately, routine daily Doppler monitoring of the arterial flow to the graft is unlikely to be feasible in all transplantation units, and intense screening with frequent controls cannot be hypothesized for all transplant recipients in the long-term. The knowledge of the predisposing factor is therefore of paramount importance in tailoring the screening program to the patients at risk. Great attention has to be paid to the characteristics of the donor and the graft.
Correspondence: Marco Vivarelli, MD, Dipartimento di Discipline Chirurgiche, Rianimatorie e dei Trapianti, Chirurgia II, Policlinico S. Orsola, Via Massarenti, 940138 Bologna, Italy (firstname.lastname@example.org).
Accepted for publication February 26, 2003.