The aim of this study was to analyze the effect of human leukocyte antigen (HLA) class I and HLA-DR mismatching, sharing cross-reactive antigen groups (CREGs), and sharing HLA-DR antigens on the outcome after pediatric liver transplantation. Outcome parameters were graft survival, acute rejection, and portal fibrosis. A distinction was made between full-size (FSLTx) and technical-variant liver transplantation (TVLTx). A total of 136 primary transplants were analyzed. The effect of HLA on the outcome parameters was analyzed by adjusted multivariate logistic and Cox regression analysis. HLA mismatches, shared CREGs, and shared HLA-DR antigens affected neither overall graft survival nor survival after FSLTx. Survival after TVLTx was superior in case of 2 mismatches at the HLA-DR locus compared to 0 or 1 mismatch (P = 0.01) and in case of no shared HLA-DR antigen compared to 1 shared HLA-DR antigen (P = 0.004). The incidence of acute rejection was not influenced by HLA. The incidence of portal fibrosis could be analyzed in 62 1-yr biopsies and was higher after TVLTx than FSLTx (P = 0.04). The incidence of portal fibrosis after TVLTx with 0 or 1 mismatch at the HLA-DR locus was 100% compared to 43% with 2 mismatches (P = 0.004). After multivariate analysis, matching for HLA-DR and matching for TVLTx were independent risk factors for portal fibrosis. In conclusion, an overall beneficial effect of HLA matching, sharing CREGs, or sharing HLA-DR antigens was not observed. A negative effect was present for HLA-DR matching and sharing HLA-DR antigens on survival after TVLTx. HLA-DR matching might be associated with portal fibrosis in these grafts. (Liver Transpl 2005;11:1541–1549.)
Prospective matching of human leukocyte antigens (HLA) is not routinely performed before liver transplantation. Early studies have failed to prove a benefit.1, 2 Furthermore, the time span for such a procedure is generally too short due to the constraints of the cold ischemic period. Additionally, successful matching cannot be obtained due to relatively low numbers of recipients and donors. Finally, the progressing deterioration of the clinical condition in some of the recipients does not permit any choice.
However, reconsideration of HLA matching in pediatric liver transplantation is justified. The donor pool for these patients has been expanded by the introduction of technical-variant liver transplantation (TVLTx), which includes transplantation of reduced-size, split, and living-donor grafts. In experimental studies, regeneration of these technical-variant grafts has been associated with enhanced expression of various cellular antigens, including HLA.3 This may evoke increased immune responses and accelerate cellular- and antibody-mediated rejection.3–5
So far, only one study on HLA matching was performed in a study population that exclusively consisted of pediatric liver transplant recipients receiving cadaveric grafts.6 However, this analysis was performed without considering the type of graft. Also, more information on the effect of HLA can be obtained by analyzing common cross-reactive antigen groups (CREGs). Immunization to a single private HLA antigen often resulted in sensitization to many additional HLA antigens. Analysis of these reactions led to the description of a small number of public antigenic epitopes shared by HLA specificities, also called CREGs.7–9 Sharing of these CREGs has been associated with improved graft survival after kidney10 as well as after cardiac transplantation.11
In a previous report we have shown a high incidence (30%) of portal fibrosis at 1 yr after pediatric transplantation.12 Acute rejection was negatively correlated with portal fibrosis, whereas biliary complications, a prolonged cold ischemia time, and a positive cytomegalovirus (CMV) status of the recipient were risk factors for portal fibrosis.
The aim of this study was to analyze the influence of HLA on graft survival, early acute rejection, and the occurrence of portal fibrosis in a pediatric liver transplant population transplanted with cadaveric grafts. The influence of HLA was analyzed at the level of mismatching for HLA class I and HLA-DR antigens, sharing CREGs, and sharing HLA-DR antigens.
Between November 1982 and September 2001, 151 children underwent a liver transplantation in our hospital. To create a homogenous cohort of patients for analysis, only primary transplantations of children transplanted with an ABO-identical or -compatible graft and children with an initial triple-drug (consisting of prednisolone, azathioprine, and cyclosporine) or double-drug (consisting of prednisolone and tacrolimus) immunosuppressive regimen were included. Furthermore, all transplants had or could have had a minimal follow-up of 1 yr (censoring date, August 31, 2002) and should have complete data on HLA. This resulted in a study population of 136 primary transplantations with a median follow-up of 6.0 yr (range = 0-16.6).
Of 136 transplantations, 65 (48%) were full-size liver transplantations (FSLTx) and 71 (52%) TVLTx: 56 (41%) were reduced-size liver grafts (37 either a right or left liver lobe graft and 19 segmental liver grafts) and 15 (11%) split liver grafts (6 either a right or left liver lobe graft and 9 segmental liver grafts). Median follow-up after FSLTx was 6.3 yr (0-16.6) and after TVLTx, 5.3 yr (0-14.5; P = 0.13).
Demographic data of the recipients and variables related to donors and transplantations based on the type of graft are listed in Table 1. Children receiving TVLTx were significantly younger and weighed significantly less than children receiving FSLTx. Inherent to TVLTx, the cold ischemia times were significantly longer in these procedures compared to FSLTx.
Table 1. Demographic Data of the Recipients and Variables Related to Donors and Transplantations
Immunosuppressive therapy in 111 (82%) children consisted of azathioprine (2-2.5 mg/kg/day), prednisolone (starting dose of 1 mg/kg/day, tapered to a maintenance dose of 0.5-1 mg on alternate days), and cyclosporine (dose aimed at trough levels of 200-250 μg/L whole blood during the first 4 weeks and 100-150 μg/L thereafter). During the first 10 days after transplantation, these children received cyclophosphamide (3 mg/kg/day) as well. In the remaining 25 (18%) children azathioprine and cyclosporine were replaced by tacrolimus (dose aimed at trough levels of 10-15 μg/L whole blood during the first 3 months and 5-10 μg/L thereafter).
HLA Typing and Identification of CREGs
The standard complement-dependent National Institutes of Health microlymphocytotoxicity technique13 was used for HLA class I typing, and the 2-color fluorescence method14 for HLA class II. In addition, DNA typing by polymerase chain reaction-sequence specific primers was introduced in 1994 for HLA class II and in 1998 for HLA class I. For HLA class II DNA typing a local and a commercial kit (Dynal AS, Oslo, Norway; GenoVision AS, Vienna, Austria) were used; for HLA class I only commercial kits (Dynal; GenoVision) were employed. DNA typing was routinely performed for all apparent HLA class II homozygotes, and in cases to resolve any problematic serological typing. Assignment of the HLA type was based on the broad HLA-A, -B, and -DR specificities as assigned by the World Health Organization.15
HLA class I antigens were converted to CREGs, representing cross-reactivity among alleles of both HLA-A and HLA-B loci.7–9
Abbreviations: HLA, human leucocyte antigen; CREGs, cross-reactive antigen groups; FSLTx, full-size liver transplantation; TVLTx, technical-variant liver transplantation; MM, mismatches.
The effect of HLA mismatches, shared CREGs, and shared HLA-DR antigens on graft survival, early acute rejection, and portal fibrosis at 1 yr after transplantation was assessed. Assessments were made on all grafts together, as well as on full-size and technical-variant grafts separately.
Liver biopsies were performed routinely at 1 week and at 1 yr after transplantation. Early acute rejection was defined as a biopsy-proven rejection occurring within 3 months after transplantation and graded according to the Banff criteria.16, 17
One-yr biopsies were reviewed anonymously (A.S.H.G.) for the analysis of portal fibrosis and were classified as normal, portal fibrosis12 or as other changes, including reactive changes, acute or chronic rejection, viral hepatitis, vascular changes, and biliary pathology. Portal fibrosis was based on criteria similar to those for chronic hepatitis.18 To analyze the effect of HLA mismatches, shared CREGs, and shared HLA-DR antigens on portal fibrosis in homogeneous groups, these study variables were compared between normal biopsies and biopsies showing portal fibrosis.
The Pearson's Chi-square test or Fisher's exact test was used to compare categorical variables between 2 groups, the Mann-Whitney U test to compare continuous variables between 2 groups, and the 1-way analysis of variance test for comparison of continuous variables between more than 2 groups. The level of significance was set at 0.05.
To analyze the effect of HLA mismatches, shared CREGs, and shared HLA-DR on the outcome, these study variables were categorized into subgroups: HLA class I with 0 to 2 mismatches (n = 33), 3 mismatches (n = 66), and with 4 mismatches (n = 37); HLA-DR with 0 or 1 mismatch (n = 74) and 2 mismatches (n = 62); 0 to 2 shared CREGs (n = 28), 3 or 4 shared (n = 76), and 5 or 6 shared (n = 32); and 0 (n = 81) vs. 1 shared HLA-DR antigen (n = 54). Only 1 transplant had 2 shared HLA-DR antigens and was excluded from the analysis for this reason.
To identify confounding covariates, subgroups based on HLA mismatches, shared CREGs, and shared HLA-DR antigens were compared according to the recipient, donor, and transplantation procedure-related variables as listed in Table 1. Recently, biliary complications and acute rejection proved to be prognostic factors for portal fibrosis besides a positive CMV status of the recipient and the cold ischemia time.12 So additionally, in the analysis of portal fibrosis, these factors were, together with postoperative CMV infection, also compared between the subgroups to identify covariates. Furthermore, vascular complications, posttransplant lymphoproliferative disease, and causes of graft loss and patient death were also analyzed between subgroups to identify possible confounding covariates.
Biliary complications included bile leakage (either along the drain, from the site of the anastomosis, or from the transection surface in TVLTx), necrosis of the extrahepatic bile ducts in case of hepatic artery thrombosis, and biliary strictures (anastomotic or ischemic type biliary lesions). Vascular complications were defined as hepatic artery thrombosis, portal vein thrombosis, or obstruction of the venous outflow tract of the liver due to torsion, compression, or thrombosis of the hepatic veins or the inferior vena cava.
Cox proportional hazards regression models adjusting for confounding covariates were used for obtaining graft survival curves and analyzing the effect of HLA mismatches, shared CREGs, and shared HLA-DR antigens on graft survival. Multivariate logistic regression models adjusting for confounding covariates were used for analyzing the influence of these parameters on acute rejection and portal fibrosis.
Overall graft survival and postoperative complications and graft survival and complications according to the type graft are listed in Table 3. Apart from a significantly higher incidence of biliary complications after TVLTx than after FSLTx, no other differences were observed between these groups.
Table 3. Graft Survival, Causes of Graft Loss, 1-yr Biopsies, and Postoperative Complications
Survival of technical-variant grafts with 0 to 2 HLA class I mismatches was worse than grafts with 3 mismatches (P = 0.06, Table 4), but not different compared to grafts with 4 mismatches (P = 0.75). Survival of technical-variant grafts was superior in case of 2 mismatches at the HLA-DR locus compared to 0 or 1 mismatch (P = 0.01, Fig. 1). Furthermore, graft survival after TVLTx was better in case of no shared HLA-DR antigens compared to 1 shared HLA-DR antigen (P = 0.004, Fig. 2). When analyzing causes of graft loss after TVLTx in the subgroups of HLA-DR mismatches and shared HLA-DR antigens, no statistical differences could be found (data not shown).
Early Acute Rejection
The incidence of overall acute rejection and the incidence of acute rejection after FSLTx and TVLTx were not influenced by HLA mismatches, shared CREGs, or shared HLA-DR antigens (data not shown).
Routinely, a liver biopsy was performed 1 yr after transplantation. Of 136 grafts, 91 (67%) survived at least 1 yr and 45 (33%) were lost in the first year. A total of 86 of the 91 grafts that survived the first year had a biopsy, but 5 did not for unknown reasons.
Of 86 1-yr biopsies available for histopathological analysis, 33 (38%) showed normal histology, 29 (34%) portal fibrosis, and 24 (28%) other changes. The latter group was excluded from the analysis in order to analyze the influence of HLA and CREGs between 2 homogeneous groups.
The incidence of portal fibrosis after FSLTx was 34% (11/32 grafts) and after TVLTx 60% (18/30; P = 0.04; Table 1). There were no statistical differences between grades of portal fibrosis.
The incidence of portal fibrosis in grafts with 0 or 1 mismatch at the HLA-DR locus was 59% (17/29 grafts) compared to 36% (12/33) with 2 mismatches at this locus (P = 0.08; Table 4).
Analyzing HLA-DR in relation to the type of graft, the incidence of portal fibrosis after FSLTx with 0 or 1 mismatch at the HLA-DR locus was 40% (8/20) compared to 25% (3/12) with 2 mismatches at this locus (P = 0.47). After TVLTx these incidences were 100% (9/9) and 43% (9/21; P = 0.004). A separate multivariate logistic regression analysis of the effect of HLA in the different types of grafts on the incidence of portal fibrosis could not be performed because of too small numbers.
Analyzing HLA-DR mismatches in relation with portal fibrosis revealed a difference in the incidence of biliary complications. All 6 grafts with 0 or 1 mismatch at the HLA-DR locus that were complicated by biliary complications developed portal fibrosis, compared to 11 of 23 (48%) of grafts with 0 or 1 mismatch without biliary complications (P = 0.03). For grafts with 2 mismatches at the HLA-DR locus, these incidences were 35 and 43% (P = 0.69).
Considering the results as described above, the type of graft and biliary complications were, besides matching for HLA-DR, analyzed in an adjusted multivariate analysis to determine the relative contributions of these variables for the occurrence of portal fibrosis (Table 5). Matching for HLA-DR (β = 5.66, 95% confidence interval = 1.44-22.23, P = 0.02) as well as technical-variant liver grafts (β = 6.26, 95% confidence interval = 1.49-23.20, P = 0.01) proved to be independent risk factors.
Table 5. Adjusted Logistic Regression Analysis of the Effect of HLA Mismatching, Sharing CREGs, and Sharing HLA-DR Antigens on the Overall Incidence of Portal Fibrosis 1 yr After Transplantation
NOTE. Logistic regression analysis adjusted for confounding factors related to the recipients, donors, transplantation procedures, and postoperative complications.
Abbreviations: HLA, human leucocyte antigen; CREGs, cross-reactive antigen groups; MM, mismatches; CI, confidence interval.
HLA class I
0-2 MM vs. 3 MM
0-2 MM vs. 4 MM
0-1 MM vs. 2 MM
0-2 shared vs. 5-6 shared
3-4 shared vs. 5-6 shared
0 shared vs. 1 shared
In the past, reports on HLA matching in liver transplantation have shown discrepant and confusing results.19, 20 This might be caused by analyzing heterogeneous populations, including either adult or both adult and pediatric recipients.
The importance of a separate analysis of pediatric recipients from adults should be emphasized. HLA genotype plays an important role in both the susceptibility to and prognosis of primary diseases leading to transplantation,21 which are different between children and adults. Furthermore, HLA plays a role in the nature of disease20 and may be of importance in the recurrence of the primary disease in the liver graft.6 Another important contributing argument for a separate analysis is the difference in graft types that are used in pediatric and adult recipients. In most pediatric transplants, technical-variant grafts are used, whereas in adults full-size grafts are used. From experimental studies, enhanced cellular and humoral responses have been reported in regenerating technical-variant grafts, leading to an increased risk for rejection compared to full-size grafts.3–5 Previous studies on HLA matching in pediatric liver transplantation with either cadaveric6 or living related grafts22, 23 focused on graft survival and the occurrence of acute rejection.
The present series analyzed the effect of HLA mismatches, shared CREGs, and shared HLA-DR antigens not only on graft survival and acute rejection, but also on portal fibrosis. This study also made a distinction between the types of graft. Additionally, analyses were corrected for confounding covariates. Analyses were made in a single-center homogeneous cohort of pediatric patients with only primary grafts. As with other single-center reports, results, especially those on portal fibrosis, should be interpreted carefully because of the relatively small number of grafts involved.
Overall graft survival was not influenced by HLA mismatches, shared CREGs, or shared HLA-DR antigens. HLA-DR mismatching improved overall graft survival, but the difference was not statistically significant. However, when considering the type of graft, the beneficial effect of HLA-DR mismatching on graft survival after TVLTx was statistically significant in contrast to survival after FSLTx. Furthermore, graft survival after TVLTx improved in case of no shared HLA-DR antigens, compared to 1 shared HLA-DR antigen. We found no effect of HLA mismatches, shared CREGs, or shared HLA-DR antigens on either the overall incidence of early acute rejection or the incidence of acute rejection after FSLTx and TVLTx. When analyzing portal fibrosis in relation to HLA mismatches, shared CREGs, and shared HLA-DR antigens, matching for HLA-DR was associated with a higher incidence of portal fibrosis after TVLTx.
Francavilla et al.6 reported a beneficial effect of HLA-DR mismatching on graft survival after pediatric transplantation, especially in the long term. Furthermore, Markus et al.24 showed a survival benefit of HLA-DR mismatching partially explained by a lower incidence of graft failures by primary nonfunction. The group from Pittsburgh suggested a dualistic role of HLA, in which HLA compatibility reduces the incidence of rejection and, on the other hand, increases the incidence of other immune-mediated processes leading to graft loss.24 In the present series, this beneficial effect of HLA-DR mismatches on graft survival could not be explained by the analysis of causes of graft loss. This might be due to the relatively low number of patients included in this series and the large diversity of causes of graft loss. A lower incidence of portal fibrosis was found in technical-variant grafts with 2 mismatches at the HLA-DR locus than in grafts matched for HLA-DR. However, this did not attribute to the survival benefit in technical-variant grafts with 2 mismatches at the HLA-DR locus. Previously, we have identified a prolonged cold ischemia time, biliary complications, and a positive CMV status of the recipient at time of transplantation as independent risk factors for the development of portal fibrosis, whereas acute rejection seems to prevent portal fibrosis.12 However, after multivariate regression analysis in this study, matching for HLA-DR and technical-variant liver grafts proved to be independent risk factors for portal fibrosis. Portal fibrosis in TVLTx might be explained by increased immune responses associated with proliferation of parenchymal and nonparenchymal cells during regeneration in technical-variant grafts, as reported in other studies.3–5 Also, aberrant production of cytokines as reported after partial hepatectomies may explain fibrosis.25
Enhanced immune responses have also been associated with CMV disease.26 It is suggested that matching for HLA-DR increases the risk for CMV hepatitis in both primary and secondary CMV infection.27 Another risk for HLA-DR matched grafts is an increased incidence of chronic rejection,27 especially in combination with CMV infection.28 However, from the present study, these theories could not be confirmed and could not be indicated as causes for decreased survival and a higher incidence of portal fibrosis in grafts matched for HLA-DR. Only 4 (3%) grafts were lost because of chronic rejection, and additionally, CMV hepatitis in children is rare in our experience.
In conclusion, an overall beneficial effect of HLA matching, sharing CREGs, or sharing HLA-DR antigens could not be observed in the present series. However, interesting observations were made concerning a negative effect of HLA-DR matching and a negative effect of sharing HLA-DR antigens on survival of technical-variant liver grafts. Furthermore, HLA-DR matching might be associated with a higher incidence of portal fibrosis in these grafts. Hypothetically, the dualistic role of HLA-DR matching combined with increased immune responses after TVLTx could lead to the results shown in this analysis. Prospective studies including larger patient populations are needed to determine the exact role of HLA matching and to confirm the above-mentioned findings.