Because of the introduction of effective highly active antiretroviral therapy (HAART) and the subsequent improvements in mortality and opportunistic infection rates, human immunodeficiency virus (HIV)–positive patients with end-stage liver disease are being offered liver transplantation (LT).1-3 Experience with transplantation in patients infected with HIV is slowly accumulating; to date, more than 200 of these patients have undergone transplantation worldwide.4, 5 For these patients, concerns remain about the risk of hepatitis C virus (HCV) recurrence with accelerated fibrosis, interactions between HAART and immunosuppression, and HAART-associated toxicity.6-10 In addition, the relatively recent recognition of HIV infection as a prothrombotic state with associated vascular changes could have potential implications after transplantation.11, 12 Therefore, we analyzed our own cohort of HIV-positive LT recipients so that we could review the incidence of posttransplant thrombotic complications and evaluate potential risk factors.
Liver transplantation (LT) for human immunodeficiency virus (HIV)–positive recipients with end-stage liver disease has become an accepted practice. However, because these patients are increasingly being recognized as prothrombotic, we reviewed their posttransplant thrombotic complications. Because morphological changes might be responsible in part for this prothrombotic state, we also conducted a histopathological review of explants from HIV-positive patients. Between 1990 and 2010, 24 of 3502 recipients (including 23 adults) were HIV-positive at LT. These patients and their postoperative courses were reviewed with a particular focus on vascular complications, risk factors, and outcomes. Another patient in whom HIV was detected 12 years after LT was also examined. Among the 24 HIV-positive LT recipients (17 males and 22 whole liver grafts; median age = 40 years), 5 developed arterial complications [including 3 cases of hepatic artery thrombosis (HAT), 1 case of generalized arteriopathy (on angiography), and 1 case of endoarteritis (on histological analysis)]. Multiple arterial anastomoses were performed in 8 of the 24 recipients, and HAT occurred twice within this anastomosis group. The outcomes of the 3 patients with HAT included retransplantation, biliary stenting for ischemic cholangiopathy followed by retransplantation, and observation only. In addition, 5 separate venous thrombotic events were detected in the 24 recipients during this period. Moreover, the delayed-HIV recipient developed delayed HAT and subsequently ischemic cholangiopathy and was being assessed for retransplantation at the time of this writing. In conclusion, the prothrombotic state associated with combined HIV and liver disease is a cause of morbidity after LT: 8 of the 24 recipients (33%) in this series suffered vascular thrombotic complications. There is a potential increase in the risk of HAT: the rate for the HIV-positive cohort was higher than the rate for historical HIV-negative controls [12% versus 3.2%, P = 0.016 (Fisher's exact test)]. The minimization of complex arterial reconstruction, coagulopathy screening, and risk-adapted antithrombotic chemoprophylaxis appear to be reasonable precautions. Liver Transpl 18:83–89, 2012. © 2011 AASLD.
PATIENTS AND METHODS
Among 3502 patients undergoing transplantation between March 1990 and March 2010, we identified and reviewed 24 HIV-positive LT recipients, and we paid particular attention to thrombotic complications and associated factors, which included the disease etiology, the transplant procedure, the donor characteristics, the postoperative course, and the long-term outcomes. Furthermore, another patient who was HIV-negative at the time of LT but subsequently developed an HIV infection 12 years later (a delayed-HIV case) was examined separately. Whenever it was possible, the Model for End-Stage Liver Disease (MELD) score was used as a marker of liver disease severity, although some of these patients underwent transplantation in the pre-MELD era. The prerequisites for transplantation in HIV-positive recipients in our unit included evidence of viral control [except for patients presenting with acute liver failure (ALF)], the potential for a response to antiviral therapy, and CD4 cell counts > 100 cells/μL. Opportunistic infections before the initiation of HAART were not considered contraindications.
The immunosuppression included tacrolimus and prednisolone in 22 patients and cyclosporine, prednisolone, and azathioprine in the 2 earliest cases. Prednisolone was started at 20 mg/day; this dosage was reduced by 5 mg every week after the second week, and prednisolone was completely withdrawn at a median of 4 months. No protocol changes were made to the immunosuppression regimen because of the presence of HIV or concurrent HAART. Prophylaxis for hepatitis B virus (HBV) recurrence included the intraoperative use of hepatitis B immunoglobulin (10,000 IU) during the anhepatic phase, 5000 IU of intravenous hepatitis B immunoglobulin daily for the first week, and then hepatitis B immunoglobulin as required to maintain hepatitis B surface antibody titers higher than 100 IU. Patients received cotrimoxazole (480 mg daily) from day 7 to 3 months as prophylaxis against Pneumocystis carinii and toxoplasmosis [with Toxoplasma immunoglobulin G (IgG)–positive donors]. Intravenous ganciclovir (250 mg twice daily) and, more recently, oral valganciclovir (900 mg daily) were used for 3 months as prophylaxis against cytomegalovirus (CMV) in high-risk cases (CMV IgG–positive donor grafts to CMV IgG–negative recipients). Fungal (candidial) prophylaxis after transplantation with oral fluconazole (50 mg once a day except for ALF patients, who received 200 mg once a day) was administered until the steroid dosage was tapered to less than 10 mg/day. Around 1995, routine postoperative antithrombotic prophylaxis at the study center began to include standard subcutaneous heparin (5000 IU) twice a day. Currently, subcutaneous low-molecular-weight heparin (40 mg of enoxaparin) is used once a day regardless of body weight until the recipient is fully mobile; above-knee thromboembolic deterrent stockings are used while the recipient is an inpatient. Moreover, for all patients, an on-table, below-knee, sequential compression pump system (Covidien, Gosport, United Kingdom) is routinely used during orthotopic transplantation.
Despite publications suggesting that vascular morphological changes have the potential to contribute to a prothrombotic state in HIV-positive patients, we are not aware of any reports describing the histopathological changes within the livers of patients undergoing transplantation. Therefore, 2 histopathologists (W.A. and A.Q.) conducted histological reviews of the native livers removed at transplantation (explants) to look for specific histological features suggestive of an obliterative vasculopathy, obstructive venopathy, or vascular injury (eg, venulitis or endoarteritis).
We performed a statistical analysis with Stata 11.1 (StataCorp., College Station, TX) to compare these HIV-positive cases and unpaired, historical HIV-negative LT recipient controls in fields relevant to the aims of this study (Fisher's exact test was used). All potential risk factors were analyzed for correlations with the Mann-Whitney U test for nonparametric, continuous data. Because the use of a complex arterial anastomosis (a potential confounder) appeared to be a particular risk factor in our series, a statistical association was sought between this confounder and our results. In addition, survival estimations were made with a Kaplan-Meier survival analysis, and comparisons between HCV and non-HCV subcohorts were performed with log-rank tests.
Twenty-four of the 3502 LT recipients (0.7%) were infected with HIV at the time of transplantation [23 adults; median age = 40 years (range = 9-59 years)]. All the adult patients received whole liver grafts; the pediatric patient received a reduced left lobe. The average duration of the HIV infection before transplantation was 10 years (range = 0-21 years). Five patients were Afro-Caribbean, and 19 were Caucasian. Twelve patients were coinfected with HCV, and 8 were coinfected with HBV. The liver disease etiology was HBV in 8 patients, HCV in 12 patients, seronegative hepatitis (SNH) in 2 patients, alcohol-related liver disease in 4 patients, and drug-induced ALF in 1 patient. Five of the 24 recipients presented with ALF [HBV (2), SNH (2), and drug-induced ALF (1)]. In 3 of these 5 ALF patients, HIV was detected only at presentation. The median MELD score at the time of transplantation for patients with available results (n = 18) was 13 (range = 5-40).
The pretransplant HIV RNA load was <50 copies/mL in 12 patients and was unknown in 4. In the remaining 8 patients, the median load was 10,316 copies/mL (range = 230-132,690 copies/mL). The HIV RNA load at the time of the arterial event was available for 4 of the 5 patients and was <50 copies/mL in 3 patients and 178 copies/mL in 1 patient (generalized arteriopathy) according to the Roche Cobas AmpliPrep/TaqMan HIV1 assay (the threshold was 40 copies/mL). The HIV viral loads in 3 of the 5 patients who presented with ALF were 25,000, 132,690, and 47,666 copies/mL, and the underlying liver disease was HBV in 2 patients and SNH in 1 patient. The other 2 ALF patients with SNH and drug-induced ALF had viral loads < 50 copies/mL. The median pretransplant CD4 count was 309 cells/μL (range = 17-1121 cells/μL). The values were not significantly different in the ALF cohort. There were 6 patients with counts < 200 cells/μL and 1 patient with a count < 100 cells/μL. Twenty-one of the 24 patients received an antiretroviral treatment (HAART in 12 cases).
Overall, 5 of the 24 HIV-positive recipients developed arterial complications. Three developed hepatic artery thrombosis (HAT), 1 developed generalized arteriopathy (diagnosed by angiography), and another developed endoarteritis (diagnosed by liver biopsy). Among the 2174 HIV-negative recipients undergoing transplantation at our center over the previous 15 years, the overall incidence of HAT was 3.2% [2.7% for adults and 4.8% for children (95% confidence interval = 1.6%-3.2%)]. In comparison, the incidence of HAT was significantly higher (12%) in the current series of HIV-positive recipients [P = 0.016 (Fisher's exact test)].
The timing of the arterial events was variable: at 2, 3, and 8 months for the 3 patients with HAT; at 10 days for the patient with nonthrombotic arteriopathy; and at 20 days for the patient with endoarteritis. The patient with HAT at 2 months developed arterial collaterals and did not require any treatment. The patient with HAT at 3 months required retransplantation at 11 months for ischemic cholangiopathy. The patient with HAT at 8 months was initially managed with biliary stenting but required retransplantation at 13 months for worsening ischemic cholangiopathy. According to a review of the sites of HAT, HAT was noted in all our patients at the origin of the hepatic artery proximal to the anastomosis at the time of the scan. In all these cases, intrahepatic arterial flow was present through arterial collateralization.
The recipient with generalized arteriopathy on angiography was initially found to have poor arterial flow according to a Doppler ultrasound examination (requested to investigate liver dysfunction) 10 days after transplantation. A subsequent liver biopsy sample of his allograft liver showed areas of infarction and preservation injury, even though this was a nonsteatotic allograft from a 32-year-old brain-dead donor with an unremarkable cold ischemia time (CIT) of 10 hours. This patient eventually died because of a combination of renal and liver failure (bilirubin level = 446 μmol/L, aspartate aminotransferase level = 113 IU/L, creatinine level = 423 μmol/L) in the presence of aggressive HCV recurrence 96 days after transplantation. The last arterial complication was found in the 9-year-old pediatric recipient, who had undergone reduced left lobe transplantation (segments 2-4) for seronegative/drug-induced (antiretroviral therapy) ALF. He required retransplantation on day 29 for persistent and severe graft dysfunction, again with a reduced left lobe graft. A histological examination of the first allograft revealed features of hilar endoarteritis despite a patent hepatic artery. Despite the second transplant, the patient died from chronic rejection and severe cholestasis 26 months later.
Five venous thrombotic events were detected among the 24 HIV-positive recipients; they included a nonocclusive left portal vein thrombus detected at the time of the transplant assessment, a pulmonary embolus 7 years after retransplantation, 2 posttransplant nonocclusive portal vein thrombi, and 1 deep vein thrombosis within 3 months of transplantation.
Analysis of Risk Factors
Previously recognized risk factors for HAT, including the donor's age and cause of death, back-bench reconstruction, a long CIT, acute cellular rejection, CMV disease, retransplantation, and the disease etiology, were reviewed (Table 1). All donors except one were less than 60 years of age. The causes of death included cerebrovascular accidents in 2 cases [recipients 1 (HAT) and 3 (generalized arteriopathy)]. Eight recipients had a complex arterial reconstruction (more than a single anastomosis), and 2 of the HAT patients were in this group. Because multiple arterial anastomoses are a recognized risk factor for arterial thrombosis, it could be argued that these findings are simply effects due to the type of arterial reconstruction. However, a statistical comparison found no significant association (P = 0.53) to suggest that HAT is related to complex arterial reconstruction. Beside these factors, because the majority of HAT cases are potentially the result of surgical/technical mishaps, it is important to note that during the course of this study, 7 consultant surgeons performed these independent transplants. Although 2 of the senior surgeons were more involved than the others in these 25 patients, this was merely a reflection of the fact that they were instrumental in developing the service over the last 2 decades, whereas the others joined the team at various stages. There were no other noteworthy differences or changes in the surgical personnel.
|Patient Number||Age (Years)||Race||Arterial Anastomosis||CIT (Hours: Minutes)||HIV RNA (Copies/mL)||CD4 Count (Cells/μL)||CMV||Retroviral Therapy|
|2||35||Caucasian||Multiple||10:19||1093||260||No||Lamivudine, azidothymidine, and indinavir|
Thirteen of the 24 recipients were known to be CMV IgG antibody–positive before transplantation. However, none of the 5 recipients with arterial complications had evidence of CMV viremia at the time of the event. The plasma cell volume (range = 0.28-0.39 l/L) and the hemoglobin level (range = 8.7-12.5 g/dL) were also unremarkable in all 5 patients at the time of the event. The platelet counts were either normal or low in all patients at the time of the arterial event. The median CIT for the 24 patients was 8 hours 22 minutes (range = 4.25-15.20 hours). The longest CIT among the recipients with arterial complications was 11 hours 15 minutes (the others were 9.5, 10.5, 7.5, and 4 hours). There was no correlation between the CD4 counts or the HIV RNA loads and the incidence of HAT among the HIV-positive patients when we used the Mann-Whitney U test for nonparametric, continuous data. Five of the 24 patients experienced acute cellular rejection. One of these patients also suffered HAT, and only 1 received an allograft from a donation after cardiac death donor; this recipient developed HAT at 3 months and required retransplantation.
Mortality and Survival
Eight of the 24 patients died [intracranial bleeding (2), HCV recurrence (3), biliary sepsis or multiorgan failure (2), and fungal infection (1)] at a median follow-up of 29 months (mean = 52.2 months) after transplantation. Fifteen were alive, and 1 had been lost to follow-up at the time of this writing. Three of the 5 patients with arterial events were alive (median survival = 88 months). The median overall 5-year survival rate in our series of HIV-positive recipients was 71%. A Kaplan-Meier survival analysis estimated median 1-, 3,- and 5-year survival rates of 92%, 83%, and 83% in the non-HCV subcohort and rates of 66%, 57%, and 57% in the HCV-coinfected subcohort of our HIV recipients (Fig. 1). HCV recurrence was associated with a poor outcome; the median 5-year survival rate was lower in comparison with the rate for the non-HCV cohort, although the difference did not reach statistical significance [P = 0.20 (log-rank test)].
Slides of 22 explants were available for the 24 recipients. Despite careful examinations, our review of these explants from the HIV-infected patients found only nonspecific pathological features, and they mostly reflected the features of the coexisting disease (eg, viral hepatitis, alcohol liver disease, drug toxicity, or neoplasia). The histology of the 11 patients with vascular complications did not differ significantly from the histology of the others, and there were no particularly notable features exclusive to them. There was no evidence of intrahepatic arterial thrombosis in any of the 22 explants (except in 1 recipient who had undergone transarterial chemoembolization before transplantation), but in the absence of other features of venous outflow block or sclerosis as the dominant histological pattern, hepatic vein thrombi were noted in 4 patients, and portal vein thrombosis was noted in 2 patients. There was no evidence of venulitis, endoarteritis, or portal obstructive venopathy as the dominant cause of liver injury in any of the explants. Despite careful histopathological reviews of all 22 explants available for inspection, no other patients were found to have endoarteritis, so the significance of this finding remains unclear.
The delayed-HIV patient is worth highlighting because her HIV infection was detected only 12 years after LT; this patient then developed HAT next to the anastomosis 6 years after the diagnosis despite an uncomplicated single arterial anastomosis during the initial implantation.
Left intrahepatic portal vein thrombosis was also noted at the time of the diagnosis of HAT. This remained nonprogressive on subsequent imaging, although progressive ischemic cholangiopathy led to secondary cirrhosis, and she was being assessed for retransplantation at the time of this writing. Because she was not HIV-positive at the time of LT, we reviewed this case in isolation (especially in our statistical analysis), although we believe that de novo HAT 18 years after LT is unusual and can potentially be precipitated by the onset of an HIV infection.
It is now well recognized in the nontransplant literature that HIV-infected patients are prothrombotic. Retrospective cohort studies of HIV-infected patients suggest a venous thromboembolism rate of 1% to 2%, which is 10 times that expected among people without HIV.11 Pregnancy with an HIV infection exemplifies the risk; several studies have suggested that the risk of venous thromboembolism is 120-fold higher in comparison with HIV-positive nonpregnant controls.13 A review of the LT literature found 2 recent publications reporting deaths directly attributable to thrombotic complications after LT in HIV-positive patients: the first was due to acute HAT leading to primary nonfunction, and the second was due to massive intrahepatic and extrahepatic portal vein thrombosis.14, 15
Several explanations exist for the increased risk of thrombosis in these patients. Multiple acquired and persistent thrombophilic abnormalities, such as protein C and S deficiencies and increased factor VIII concentrations, are more frequently observed in HIV-infected patients versus the healthy population, and advancing HIV is associated with stepwise increases in these abnormalities.16, 17 Furthermore, a review of the structural and functional alterations in HIV-associated arterial disease has confirmed the presence of several prothrombotic changes, including intimal and smooth muscle hyperplasia associated with perivascular infiltrates, increased adhesion molecules in the vascular endothelium, and nonspecific vasculitis presumably due to viral protein irritation.12 Accordingly, a 20.7% prevalence of peripheral arterial disease (6 times that of age-matched non-HIV adults) was found.
In addition, several other independent factors may work synergistically to increase the risk of thrombosis in these patients; these factors include proatherogenic dyslipidemia, increased levels of high-sensitivity C-reactive protein (an inflammatory marker associated with endothelial dysfunction and promoting thrombosis), and concurrent infections (another risk factor for thrombosis).18 Antiretroviral treatment has been linked to inflammation, atherosclerotic progression, and the destabilization that precedes thrombosis.19
A recent systematic review of posttransplant HAT noted a median incidence of 4.4% among adult recipients.20 In agreement with this and other large series, the incidence of HAT at our center was 2.7% among adult HIV-negative recipients and 4.8% among pediatric HIV-negative recipients. In comparison, the rate of 12% among HIV-positive recipients (even after the exclusion of the delayed-HIV recipient) appears high (P = 0.016). Although these findings need to be confirmed in larger prospective studies, they should raise our awareness of the potential increase in the risk.
The significance of the endoarteritis noted in 1 of these patients is not obvious. The explant revealed mainly intimal inflammation of the hilar hepatic artery and extensive graft necrosis without direct evidence of HAT. The absence of venulitis and the isolation of the inflammatory infiltrate to the arterial intima suggest that these findings unlikely reflect merely acute cellular rejection. A vasculitis-induced prothrombotic state has been proposed as a possible mechanism.12, 21 It is possible that such arteritis could be an initiating factor or alternatively provide a second hit in the progression toward thrombosis. However, such findings are not routinely sought, and their true significance is difficult to ascertain.
According to an analysis of the retrieval data, there was no association between donor age, CIT, or perfusion adequacy and posttransplant thrombotic events. CMV disease has been implicated in the pathogenesis of HAT after solid organ transplantation; however, there again was no evidence for such a role in our series.22, 23 There is substantial evidence that thrombophilic abnormalities in HIV-positive patients increase with disease progression, yet none of the recipients with high HIV viral loads or low CD4 counts developed HAT.17, 18 The higher risk of thrombosis with multiple anastomoses has been recognized previously.24 Therefore, the presence of complex arterial reconstruction in 2 of the patients with HAT might be relevant, and because of the enhanced risk, it seems sensible to refrain from donor-recipient matching that might result in complex arterial reconstruction.
Screening Doppler studies have been reported to be beneficial for the early detection of acute HAT.20, 25 The screening strategy for postoperative HAT in the current study consisted of routine transabdominal Doppler ultrasound of the graft vasculature on days 1 and 5. However, because all the HAT patients in the study cohort presented more than 8 weeks after transplantation, this strategy did not contribute to their detection. Because of the variability of the timing of HAT in the study cohort, it appears unlikely that protocol screening scans would enable earlier detection or indeed reduce morbidity. However, in light of the previous observation about the timing of HAT in the current study, a comparison was made with the published literature. In general, there is a consensus that HAT appearing more than a month after transplantation is considered late rather than early.26 Accordingly, all HAT cases in our study were late, and none of our patients presented with HAT within 4 weeks of the surgery. We believe that this somewhat reduces the argument that the increased incidence was due to multiple arterial anastomoses, which have been shown to be a risk factor mainly for early HAT.26, 27 Moreover, we can speculate that the absence of early HAT suggests that a more insidious process was responsible for thrombosis in this cohort instead of technical errors or factors.
This current retrospective series has limitations. First, a selection bias is possible because these HIV-positive patients received greater medical scrutiny, which led to a higher rate of detecting complications. Second, as might be expected from a retrospective study spanning 2 decades, a procoagulant screen was not part of the routine transplant assessment in all these cases, so this information was not available. Therefore, given our findings, we believe that a hypercoagulable state should be sought in HIV-positive patients at the time of their listing for transplantation. Obviously, some of the several thrombogenic factors in HIV-positive patients (eg, endothelial dysfunction) are difficult to measure. However, increases in the factor VIII serum concentration or activity on an assay, decreases in antithrombin III levels, and decreases in the activity of natural anticoagulants (eg, free or functional protein S and protein C) as well as lupus anticoagulant, von Willebrand factor, and D-dimer levels can be assayed in these patients at the time of listing. Nevertheless, the correct interpretation of the results is potentially difficult: deficiencies related to the underlying end-stage liver disease might confound the findings because the liver is responsible for the synthesis of many of these proteins. Nonetheless, significant findings might allow the early commencement of adequate prophylaxis, regardless of the etiology. Third and finally, despite the posttransplant complications, only 2 of the 24 recipients had evidence of venous thrombosis (portal vein) before transplantation. Still, subcutaneous low-molecular-weight heparin early after transplantation and later antiplatelet therapy for at least the first 12 months appear prudent because this has been shown to effectively reduce HAT after LT.28, 29 If this increased risk is supported by future prospective studies, long-term systemic anticoagulation with warfarin may have a role to play.
In conclusion, the prothrombotic state associated with combined HIV and liver disease is a cause of morbidity after LT: 5 of the 25 recipients (20%) in the present series suffered arterial complications. There is a trend toward an increased risk of HAT: the rate for the HIV-positive cohort was higher than the rate for our own historical HIV-negative controls [12% versus 3.2%, P = 0.016 (Fisher's exact test)]. Larger prospective studies with complete coagulation screens are now required to substantiate these findings. Meanwhile, the minimization of donor-recipient matching (which might result in complex arterial reconstruction), a full coagulopathy screen at the transplant assessment, and a risk-adapted case-by-case consideration of antithrombotic chemoprophylaxis for HIV-positive recipients appear to be reasonable recommendations.