Long-term outcome of human leukocyte antigen mismatching in liver transplantation: Results of the national institute of diabetes and digestive and kidney diseases liver transplantation database

Authors


  • This work was prepared for the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database.

  • Potential conflict of interest: Nothing to report.

Abstract

A perfect or nearly perfect human leukocyte antigen (HLA) match has been associated with better immediate and long-term survival of diseased donor kidney transplants. However, the effect of HLA matching for hepatic allografts remains poorly defined. Using data from the National Institutes of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database, we investigated the association between HLA mismatches and hepatic allograft survival, disease recurrence, and immunosuppression interactions. A, B, and DR loci were used to calculate total mismatch scores of 0 (no mismatches in any loci) to 6 (mismatches in all loci). Seven hundred ninety-nine adults (male, 55%; female, 45%) underwent 883 liver transplants. The 10-year graft survival according to total mismatch score was as follows: 0-2, 60%; 3-4, 54%; and 5-6, 57%. There was a negative effect of mismatching at the A locus on patient survival, with shorter survival for patients with 1 or 2 mismatches compared with 0 mismatches [P = 0.05, hazard ratio (HR) = 1.6]. Patients on tacrolimus with 1 or 2 mismatches at B or DR loci appeared to have increased rates of patient and graft survival compared to patients with 0 mismatches, with the appearance of a protective effect of tacrolimus (HR = 0.67). The effect of HLA mismatching was more pronounced on certain disease recurrences. DR-locus mismatch increased recurrence of autoimmune hepatitis (P = 0.01, HR = 4.2) and primary biliary cirrhosis (P = 0.04, HR = 2). Mismatch in the A locus was associated with more recurrence of hepatitis C virus (P = 0.01, HR = 1.6) and primary sclerosing cholangitis (P = 0.03, HR = 2.9). Conclusion: Mismatching at the A locus decreases patient survival in liver transplant recipients, and mismatching at the DR and A loci affects recurrence of autoimmune liver diseases and hepatitis C, respectively. (HEPATOLOGY 2008.)

In contrast to the situation in kidney transplantation, in which perfect or nearly perfect human leukocyte antigen (HLA) matching between the donor and transplant recipient clearly improves short-term and long-term survival, in liver transplantation, HLA matching appears to improve outcome in some cases, worsen it in others, or have no effect at all. Conflicting results may be due not only to differences in HLA typing methods1 but also to differences in the outcome endpoint being assessed, site location of any mismatches, underlying disease, immunosuppression regime, and hidden bias due to inadvertent matching for untyped histocompatibility antigens.2

Five-year patient survival rates after first liver transplantations now range from approximately 70% to 90%, depending on the underlying disease and age of the recipient.3–11 Although current practice does not consider HLA matching in liver transplantation, cellular in vitro studies have shown a significant role for HLA in clinically relevant phenomena.12 Thus, a more sophisticated approach to HLA matching may further improve long-term outcome. The advent of living-donor liver transplantation should help to overcome time constraints that have made donor-recipient HLA compatibility difficult to achieve, thereby increasing the feasibility of HLA matching. Moreover, as patients survive longer, disease recurrence is becoming the primary cause of morbidity.13 Improvements in HLA matching, along with adjustment of immunosuppression, may decrease the probability or delay the recurrence of disease, at least in certain subgroups.

We used the Liver Transplantation Database to examine the relationships between the degree of HLA mismatching and patient survival, hepatic allograft survival, rejection, and recurrence of disease. This database, established by the National Institutes of Diabetes and Digestive and Kidney Diseases in 1990, prospectively collects serum, tissue, and clinical and demographic information from patients being evaluated for liver transplantation who subsequently undergo the procedure.

Abbreviations

AIH, autoimmune hepatitis; ALD, alcoholic liver disease; ALD-C, alcoholic liver disease with hepatitis C; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCV, hepatitis C virus; HLA, human leukocyte antigen; HR, hazard ratio; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis.

Patients and Methods

Study Patients

All patients were enrolled in the National Institutes of Diabetes and Digestive and Kidney Diseases Liver Transplant Database and treated at one of three medical centers: the Mayo Clinic, Rochester, MN; the University of Nebraska, Omaha, Nebraska; or the University of California at San Francisco, San Francisco, CA. The study was coordinated at the University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA. The institutional review boards of each of the centers approved the study protocol.

A total of 916 patients undergoing liver transplantation at the three contributing centers were enrolled between April 15, 1990 and June 30, 1994. Clinical follow-up data were collected through June 30, 1995, and vital status was determined through November 2002. To be eligible, recipients had to be adults (≥16 years), have received a single organ transplant that survived at least 1 day post-transplantation, and provide informed consent. Only primary grafts were used in this analysis. Therefore, the final cohort consisted of 799 recipients who underwent 883 liver transplants.

Immunosuppression

Induction of immunosuppression consisted of cyclosporine, prednisone, and azathioprine at the Mayo Clinic; cyclosporine and prednisone at the University of Nebraska; and antilymphocyte globulin followed by cyclosporine, prednisone, and azathioprine at the University of California at San Francisco. Because all centers were participants in the FK506 Primary Immunosuppression Trial,14 a subgroup at each center received a tacrolimus-based regimen.

HLA Typing

HLA-A and HLA-B typing was performed by standard lymphocytotoxicity testing. HLA-DR typing was performed by cytotoxicity assays. Each patient was assigned an overall (total) score depending on the total number of mismatches at the three loci. Patients were then categorized as having 0-2, 3-4, or 5-6 total mismatches. For analyses of individual loci, patients were categorized as having 0, 1, or 2 mismatches. Analyses of the individual loci were also repeated with the categories of 0-1 and 2 mismatches.

Underlying Disease

Diseases were designated as “recurrent” if positive findings were reported on pathology, clinical, reason-for-hospitalization, or autopsy forms. In this analysis, time to recurrence was defined as the time to first definite histological recurrence. Those recipients without available data were censored at the last available biopsy date. All liver biopsies were reviewed by a single pathologist (A. Jake Demetris, M.D.) prospectively. Criteria for diagnosing the recurrence of individual diseases follow.

Primary Biliary Cirrhosis

  • 1Pretransplant diagnosis including the presence of antimitochondrial antibodies.
  • 2Posttransplant characteristics showing histopathologic changes on biopsy including “florid” duct lesions or nonsuppurative cholangitis and an absence of adverse drug reactions, biliary tract obstruction or stricturing, or deep fungal or mycobacterial infection that would produce similar histopathologic changes.

Primary Sclerosing Cholangitis

  • 1Pretransplant diagnosis.
  • 2Posttransplant cholangiographic evidence of intrahepatic biliary tract obstruction or stricturing typical of primary sclerosing cholangitis, without an apparent cause, and typical histopathologic changes not consistent with acute or chronic rejection or with other well-recognized patterns of unrelated liver disease in recipients in whom other causes of biliary strictures have been reasonably excluded.

Autoimmune Hepatitis

  • 1Pretransplant diagnosis including an antinuclear antibody titer ≥ 1:80; antibodies against liver-kidney microsomal antigen or anti-smooth muscle antibody at a similar titer; hepatitis C virus (HCV) and all other causes ruled out; and an increased gamma globulin level of >2 times the upper limits of normal.
  • 2Antinuclear antibody–positive, anti-smooth muscle antibody–positive, or liver-kidney microsomal antibody–positive with titer > 1:80 post-transplant; HCV-RNA–negative by polymerase chain reaction; and hepatitis B surface antigen (HBsAg)–negative in serum, with biochemical evidence or histopathologic changes typical of chronic hepatitis but not consistent with acute or chronic rejection or other well-recognized patterns of non–chronic hepatitis–related liver injury.

Neoplasm

  • 1Pretransplant diagnosis of neoplasm.
  • 2Posttransplant histopathologic evidence of a neoplastic disorder.

Hepatitis B virus (HBV)

  • 1HBsAg-positive before transplantation.
  • 2HBsAg-positive or HBV-DNA–positive post-transplant with the presence of biochemical or histologic evidence.

HCV.

The outcome of HCV after liver transplantation through an analysis of the Liver Transplantation Database has been previously reported.15 The pretransplant diagnosis of HCV was based on the following criteria:

  • 1Anti-HCV–positive or HCV-RNA–positive by polymerase chain reaction before transplantation. Patients who originally had “ indeterminate” or “cryptogenic” diagnoses and did not meet these criteria but had some pretransplant indication of HCV and later had histologic evidence of HCV were reclassified into the HCV group (n = 10).
  • 2Post-transplant diagnosis of recurrent HCV was based on the following: HCV-RNA–positive by reverse-transcriptase polymerase chain reaction and evidence of graft dysfunction with the presence of biochemical or histologic evidence post-transplant.

Recurrent HCV on histology was diagnosed according to prospectively validated criteria16 when either of the following conditions16 was met:

  • 1Portal inflammation with inflammatory bile duct damage involving >50% of the bile ducts.
  • 2Mononuclear perivenular inflammation involving >50% of the terminal hepatic venules, associated with hepatocyte necrosis and/or dropout.

Also, rejection should have been excluded. The criteria used to diagnose rejection were the following: the presence of an adequate biopsy sample. A biopsy was considered adequate when it contained six or more portal tracts and four or more terminal hepatic venules. Early and late chronic rejection was diagnosed with the Banff criteria.17 The criteria for mild acute rejection require more extensive tissue injury than that listed for the Banff criteria in allografts not otherwise affected by a coexistent disease. Hence, recurrent HCV was diagnosed when lobular or interface necroinflammatory activity was more prevalent and prominent than bile duct inflammation and damage.16

Nonalcoholic Steatohepatitis

  • 1No pretransplant diagnosis of alcoholic liver disease.
  • 2Fatty changes with or without Mallory's hyaline and parenchymal polymorphonuclear infiltrates and necrosis on liver biopsy post-transplant with no evidence of alcohol use. Patients with jejunoileal bypass or prolonged total parenteral nutrition were excluded.

Alcoholic Liver Disease

  • 1Pretransplant diagnosis, presence of fatty changes with or without Mallory's hyaline and parenchymal polymorphonuclear infiltrates, and necrosis on liver biopsy in a patient who has returned to drinking post-transplant. Patients with diabetes, jejunoileal bypass, or prolonged total parenteral nutrition were excluded.
  • 2Admission or recidivism and other objective evidence of significant alcohol intake (blood alcohol levels and desialylated transferrin levels) post-transplant.

Hemochromatosis

  • 1Pretransplant diagnosis of hemochromatosis (presence of HFE gene or hepatic iron index > 1.9).
  • 2Posttransplant increased iron staining defined as more than “mild” on liver biopsy.

Statistical Analysis

Demographic data are presented as measures of a central tendency (mean, median, standard deviation, and range). For analyses of individual loci, patients were categorized as having 0, 1, or 2 mismatches, 0-1 versus 2 mismatches, or 0 versus 1-2 mismatches.

Rates of patient and graft survival were calculated with the Kaplan-Meier method for the total study population and for prospectively defined patient subgroups with HCV, primary sclerosing cholangitis, primary biliary cirrhosis, alcoholic liver disease associated with HCV, autoimmune hepatitis, and HBV. Separate Kaplan-Meier analyses were also performed for patients who received cyclosporine-based immunosuppression versus tacrolimus-based immunosuppression. All groups were compared with the log-rank test.

Predictors of disease recurrence (probable and definite), death, and graft failure were analyzed with Cox proportional hazard modeling. Time-dependent covariates were used where appropriate. Logistic regression was used to analyze predictors of early rejection (<42 days). All regression modeling was performed first by construction of univariate models. Those variables found to be significant at the 0.10 level were then incorporated into multivariate models. Forward and backward selection modeling was also used. In addition to the mismatch scores, basic demographic characteristics such as recipient and donor age and gender and baseline immunosuppression were included in the modeling. P values < 0.05 were considered significant. All analyses were performed with SAS 8.2 (SAS Institute, Cary, NC).

Results

Overall Study Population

Patient Characteristics.

The mean (± standard deviation) follow-up period was 7.4 (±3.9) years (median, 8.9 years; range, 0.003-12.6 years). Of 799 transplant recipients in the study population, 357 (45%) were women and 442 (55%) were men, with a median age of 49.7 years (range, 16.1-77.5 years) and a median Model for End-Stage Liver Disease score of 15.3 (range, 0.64-44.9). Donors were 316 women and 482 men (median age, 35.5 years; range, 2.2-73.3 years). Of the recipients, 637 (79.7%) were Caucasian, 33 (4.1%) were African American, 76 (9.5%) were Hispanic, and 53 (6.6%) did not belong to any of these categories. A total of 693 recipients (86.7%) received a cyclosporine-based immunosuppressive regimen, and 92 recipients (11.5%) received the tacrolimus-based regimen. Data were missing for 14 patients (1.8%). Table 1 lists the underlying diseases that led to transplantation, and Table 2 shows the HLA subtypes of the recipients.

Table 1. Underlying Disease Leading to Liver Transplantation
Pretreatment Diagnosisn%
  • Liver transplant recipients were from 3 centers enrolled in the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database.

  • *

    Metabolic liver disease included hemochromatosis, Wilson's disease, and alpha-1 antitrypsin deficiency.

  • Other diagnoses included cryptogenic (n = 6), other cholangiolitic hepatitis (n = 5), biliary atresia (n = 1), other pediatric disease (n = 3), and miscellaneous or indeterminate chronic liver disease (n = 93).

Hepatitis C14217.8
Primary sclerosing cholangitis12415.5
Alcoholic liver disease10212.8
Primary biliary cirrhosis9912.4
Alcoholic liver disease with hepatitis C516.4
Acute liver failure475.9
Autoimmune hepatitis465.8
Hepatitis B364.5
Metabolic liver disease*202.5
Budd-Chiari syndrome131.6
Hepatic malignancy111.4
Other10813.5
 799100
Table 2. Number of HLA Mismatches in Liver Transplant Recipients
Site and Number of MismatchesnPercentage of Total Study Population
  1. Liver transplant recipients were from 3 centers enrolled in the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database.

  2. Abbreviation: HLA, human leukocyte antigen.

HLA-A  
 0607.5
 135644.6
 233541.9
 Number typed75194.0
HLA-B  
 0374.6
 120125.2
 251164.0
 Number typed74993.7
HLA-DR  
 0556.9
 133441.8
 232440.6
 Number typed71389.2
HLA-A+B+DR  
 0-2445.5
 3-428836.0
 5-636846.1
 Number typed70087.6

HLA Mismatching.

The number of mismatches did not predict patient or graft survival in Kaplan-Meier analyses: 10-year patient survival was 67% for 0-2 mismatches, 59% for 3-4 mismatches, and 62% for 5-6 mismatches (P = 0.74), and 10-year graft survival was 60% for 0-2 mismatches, 54% for 3-4 mismatches, and 57% for 5-6 mismatches (P = 0.63). The number of mismatches was also not a predictor of death in multivariate proportional hazard regression analyses. However, higher total mismatch scores were significantly associated with higher rates of early acute cellular rejection (odds ratio = 1.2, P = 0.03). In Kaplan-Meier analyses of individual loci, HLA-B and HLA-DR mismatching had no effect on patient and graft survival, but mismatching at the A locus showed a trend of decreasing patient survival (P = 0.05, hazard ratio = 1.6) and showed a tendency to decrease graft survival (P = 0.10; Fig. 1).

Figure 1.

Long-term effect of the number of human leukocyte antigen mismatches at the A locus on the outcome of liver transplantation in the unselected Liver Transplantation Database population with respect to patient and graft survival.

Immunosuppressive Treatment

Patients treated with tacrolimus had fewer graft failures than those treated with cyclosporine (hazard ratio = 0.68, P = 0.04). The two groups also differed in the effects of mismatching. Although mismatching had no effect in cyclosporine-treated patients, higher total mismatch scores (>4) significantly increased patient survival (P = 0.02) and graft survival (P = 0.02) in tacrolimus-treated patients (Fig. 2). This effect in the tacrolimus group may have been due to possible beneficial effects of DR-locus mismatching on patient (P = 0.06) and graft (P = 0.06) survival. B-locus mismatching also tended to have a positive but nonsignificant effect. In contrast, as in our overall population, mismatching at the A locus appeared to decrease patient and graft survival in both treatment groups, but these differences were not significant.

Figure 2.

Effect on patient and graft survival of the total number of human leukocyte antigen mismatches at the A, B, or DR loci in liver transplant recipients treated with cyclosporine or tacrolimus.

Disease Subgroups

Recurrence rates differed among diseases, with the highest rates seen in HBV (Fig. 3).

Figure 3.

Prevalence of recurrence of primary disease at 1, 3, and 5 years after liver transplantation. Abbreviations: AIH, autoimmune hepatitis; ALD, alcoholic liver disease; ALD-C, alcoholic liver disease with hepatitis C; HBV, hepatitis B virus; HCV, hepatitis C virus; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis.

In autoimmune hepatitis, disease recurrence was significantly increased with high total mismatch scores (hazard ratio = 2.4, 95% confidence interval = 1.2-4.7, P = 0.01) and with mismatching at the DR locus (hazard ratio = 4.2, 95% confidence interval = 1.4-12.5, P = 0.01). However, high mismatch scores appeared to have positive although nonsignificant effects on graft survival at all loci (Fig. 4).

Figure 4.

Long-term effect of human leukocyte antigen mismatching on graft survival in liver transplant recipients with autoimmune hepatitis.

The risk of recurrence of primary sclerosing cholangitis was significantly increased in patients with mismatches at the A locus (hazard ratio = 3.0, 95% confidence interval = 1.2-7.4, P = 0.02) but significantly decreased in patients with mismatches at the DR locus (hazard ratio = 0.34, 95% confidence interval = 0.16-0.76, P = 0.01). Mismatching had no consistent effects on graft or patient survival.

In primary biliary cirrhosis, HLA-DR mismatching significantly increased the chance of a diagnosis of possible disease recurrence (hazard ratio = 2.0, P = 0.04). Higher total mismatch scores and mismatching at the DR locus tended to decrease patient survival rates, but the differences were not significant, and the same did not hold true for graft survival.

In patients with hepatitis C, mismatching at the A locus significantly increased the rates of recurrence of disease (possible and definite, hazard ratio = 1.6, 95% confidence interval = 1.1-2.4, P = 0.01). No significant effects of mismatching were seen in patients with alcoholic liver disease associated with HCV, although high total mismatch scores and mismatching at the A and B loci tended to have a detrimental effect on graft and patient survival in such patients. However, in patients with alcoholic liver disease alone, 2 mismatches at the B locus significantly raised rates of patient (P = 0.03) and graft (P = 0.02) survival in comparison with 0 mismatches or 1 mismatch (Fig. 5).

Figure 5.

Long-term effect of the total number of human leukocyte antigen mismatches and mismatches at the individual loci on graft survival in liver transplant recipients with alcoholic liver disease.

No significant effects were seen in hepatitis B or other liver diseases, although patient and graft survival were lower, albeit nonsignificantly, with a higher total score and A-locus mismatching in patients with hepatitis B.

Discussion

A review of the literature on HLA mismatching in unselected liver transplant recipients1, 5, 6, 8, 9, 18–37 shows that most studies found no effects (Table 3). The results in our unselected population of 799 adult liver transplant recipients were mostly consistent with such findings. This study was designed to follow participants prospectively, rather than retrospectively. Furthermore, this study utilized carefully defined protocols to determine various outcomes including HLA mismatching in liver transplantation.

Table 3. Studies Reporting Effects of HLA Mismatching on the Long-Term Outcome of Liver Transplantation in Unselected Populations
Endpoint RateLocus of HLA Mismatch
A+B+DRABDR
  1. Numbers outside parentheses refer to the reference list; the sample size (n) is given for a particular reference only when it is first cited in the table.

  2. Abbreviation: HLA, human leukocyte antigen.

Graft survival    
 Increased17 (n = 149), 21 (n = 500), 24 (n = 365)17, 18 (n = unknown)18, 21
 Decreased22 (n = 446), 25 (n = 321)Present study, 15 (n = 205), 16 (n = 466)16, 229 (n = 501), 23 (n = 1500), 26 (n = 208), 27 (n = 52)
 No effectPresent study, 1 (n = 924), 5 (n = 814), 8 (n = 138), 23, 26, 28 (n = 194)1, 8, 21, 25, 26, 27, 28Present study, 1, 17, 21, 25, 26, 27, 28Present study, 1, 8, 16, 17, 25, 28
Patient survival    
 Increased8, 29 (n = 78)29
 Decreased30 (n = 800)Present study, 159, 30
 No effectPresent study, 8, 31 (n = 80)1, 8Present study, 1Present study, 1, 8
Rejection    
 Decreased819, 20 (n = 20)
 IncreasedPresent study, 1, 24, 32 (n = 60)6 (n = 61), 15, 2215, 2221, 22, 30, 33 (n = 762)
 No effect22, 25, 28, 31Present study, 19, 20, 28Present study, 6, 19, 20, 28Present study, 6, 28, 34 (n = 121)

There was some variation in the treatment strategy (for example, different immunosuppressive regimens) among the different sites. On the other hand, this registry includes only three centers, and this provides for a consistent collection of data. It has been demonstrated that in, for example, renal transplantation a large number of patients had to be studied to demonstrate an effect of HLA matching on graft survival. However, such well-characterized data for a large number of patients are not available in liver transplantation. There are a number of studies published thus far examining the effects of HLA on liver transplantation, which are summarized in Table 3. This table also lists the sample sizes in these studies and summarizes the outcome of HLA mismatch on patient and graft survival and rejection rates. These studies had sample sizes varying from 20 to 1500. It appears that the studies with larger sample sizes showed a more definitive effect of HLA.

We have used HLA mismatching by serology instead of using HLA typing of class I and class II allele mismatches by DNA-based polymerase chain reaction methodology. Only the serological method was available when the study was conducted. Nevertheless, HLA mismatching information detected by serology is meaningful. For example, HLA class I antigen mismatches that are serologically detected are more likely to be associated with increased incidence of posttransplant complications, as shown in some organ transplantation. For example, in unrelated bone marrow transplants, HLA class I antigen mismatches that are serologically detectable confer an enhanced risk of graft failure after hematopoietic-cell transplantation. Allele mismatches that are detectable only at the DNA level are less immunogenic than those that are serologically detectable and therefore are associated with a lower risk of graft failure after hematopoietic-cell transplantation. HLA class I antigen mismatches that are serologically detectable confer an enhanced risk of graft failure after hematopoietic-cell transplantation.38 Whether this is applicable to liver transplantation is not known. Furthermore, in an overwhelming majority of studies looking at the effect of HLA in liver transplantation, HLA status was determined by serotyping.

Mismatching at the HLA-A locus did significantly decrease overall patient survival in our population, and this was consistent with the results of one study.18 Reports on the effect of HLA mismatching on rejection have been mixed, but our finding of higher rejection rates with high total mismatch scores was consistent with three other studies.1, 27, 35

The type of initial calcineurin inhibitor used for immunosuppressive treatment may influence the effects of HLA mismatching. In agreement with Neumann et al.,1 we found no effect of HLA mismatching on acute rejection in cyclosporine-treated patients. However, in tacrolimus-treated patients, we found that beneficial effects of mismatching increased graft and patient survival at the B and DR loci, whereas Neumann et al. found an adverse effect of increased acute rejection. Differences between the cyclosporine-treated and tacrolimus-treated groups in our study must be approached with caution because the treatment was not randomly assigned, and our tacrolimus group was comparatively small. It is currently well accepted that tacrolimus-based immunosuppression leads to better graft survival39; hence, the results are not surprising. These results suggest that tacrolimus-based immunosuppression may lead to better outcomes in patients with worse HLA mismatching.

Because of the increasing importance of disease recurrence as the primary cause of posttransplant morbidity,13 we examined the effects of HLA mismatching in various disease subgroups. HLA antigens are associated with underlying liver diseases and in fact may influence recurrence of the original disease after transplantation. In addition, recurrent viral infections may affect the recipient's immune status and may interact with HLA compatibility in a complex way. Furthermore, other immunological parameters such as the T- or B-cell crossmatches or the degree of panel reactive antibodies were not available. In addition to HLA mismatching, these parameters may also affect the outcome.

Markus et al.24 hypothesized a dual effect, particularly in patients at risk for developing recurrent autoimmune diseases or infection. According to this hypothesis, HLA mismatching would increase transplant rejection yet also act beneficially by decreasing the effects of other immunological mechanisms associated with disease recurrence that lead to allograft dysfunction. Neumann et al.1 observed a beneficial effect of mismatching at the A locus on graft survival in autoimmune patients that was consistent with this idea. Doran et al.25 concluded that HLA mismatching at all loci is beneficial in transplant recipients with autoimmune liver disease and detrimental in recipients with other causes of liver disease. Our findings of possible beneficial effects of mismatching on graft survival in patients with autoimmune hepatitis may support this hypothesis. However, Opelz26 found that DR mismatching tended to have a detrimental effect on graft survival in patients with autoimmune liver disease.

Neumann et al.1 also reported a beneficial effect of mismatching in patients with primary sclerosing cholangitis. However, in our patients with primary sclerosing cholangitis, mismatching had a significantly detrimental effect characterized by a higher rate of disease recurrence.

We found no specific studies of the effects of HLA mismatching in patients with primary biliary cirrhosis. In our patients, mismatching at the DR locus adversely affected recurrence of disease and possibly patient survival.

The outcome of HCV after liver transplantation from an analysis of the Liver Transplantation Database has been previously reported.15 Our observation that HLA-A mismatching increased the risk of recurrence of hepatitis C has not been seen by others.7 Furthermore, neither Gane et al.40 nor Langrehr et al.41 found an effect of HLA mismatching on graft loss or survival in patients with hepatitis C. However, Langrehr et al. did find that fibrosis progression was significantly faster in patients with 0-5 HLA mismatches compared to those with mismatches at all 6 loci.

To our knowledge, the effects of HLA mismatching have not been studied specifically in patients with alcoholic liver disease. Our data show that mismatching at the B locus may be beneficial in patients with alcoholic liver disease alone. Recidivism by itself should not have an association with HLA typing. An explanation for this finding may be that the histological injury resulting from recidivism to alcohol use may be less likely if there is mismatching at the B locus. In addition, in patients with hepatitis C with the presence or history of a second toxic insult such as alcohol, it does not affect the recurrence of hepatitis C. Such negative effects by alcohol disease or the recurrence of hepatitis C were previously suggested by Testa et al.42 in Liver Transplantation.

Although we did not see significant effects in patients with hepatitis B, the detrimental tendencies that we observed were consistent with the findings of Neumann et al.1, 4 These authors also found that mismatching at the A and B loci increased rates of recurrence of hepatitis B. We could not examine this question because insufficient data were available. The recurrence rate of HBV was higher than expected. This may be due to the fact that these patients were treated in a time period when more effective antivirals were not available.

This analysis of long-term data from the National Institutes of Diabetes and Digestive Kidney Diseases Liver Transplant Database illustrates the complexity of the effects of HLA mismatching on the outcome of liver transplantation. In this study, some of the subset groups were small. Because of the sample size, our power to detect a change may be low. Patient survival, hepatic allograft survival, number of episodes of acute rejection, and recurrence of disease may be influenced by the locus of mismatch, type of immunosuppressive treatment, and underlying disease. Immunosuppression may need to be tailored individually for patients with autoimmune liver disease or hepatitis C on the basis of their HLA mismatch profile. Therefore, we suggest that the HLA donor-recipient profile may have a role in tailoring immunosuppression. However, prospective multicenter studies are needed to evaluate this idea.

Although it is not feasible with deceased donor liver transplants, consideration of donor-recipient HLA matching may become useful in the setting of living-donor liver transplantation. This may be a consideration in living-donor liver transplantation if multiple donors are available for a particular recipient. We are not suggesting that an HLA profile be used primarily to select the donor. However, this may be a consideration if multiple donors are available. Even here the utility may be minimal as we rarely have the opportunity of choosing between multiple likely donors with the most favorable HLA matching results.

In conclusion, the HLA donor-recipient profile may have a role in deciding the immunosuppression regime. We could postulate that patients with autoimmune hepatitis and primary biliary cirrhosis and DR-locus mismatch may need increased immunosuppression. On the other hand, patients who have undergone liver transplantation and have HCV-associated liver disease and A-locus mismatch may benefit from lower immunosuppression. Our findings do suggest looking at this issue further. This can be clarified further by the performance of randomized controlled studies looking at the effects of immunosuppression on recurrence of both autoimmune and viral diseases, such as increased immunosuppression in autoimmune diseases and primary biliary cirrhosis with a mismatch at the DR locus and lower immunosuppression in viral hepatitis C with a mismatch at the A locus.

Acknowledgements

This article is dedicated to Dr. Detre who was instrumental in the establishment of the NIDDK LTD. Members of the National Institute of Diabetes and Digestive and Kidney Diseases Liver Transplantation Database who contributed to this study included the following: Katherine M. Detre, M.D., Dr.P.H. (principal investigator), A. Jake Demetris, M.D. (co-investigator), Steven H. Belle, Ph.D. (co-investigator), Yuling L. Wei, M.S. (co-investigator), Kris Ruppert, Dr.P.H. (statistician), Tatiana Ledneva, M.S. (data manager), Carol Penovich (data manager), and Heather Eng, B.A. (data analyst), from the Department of Epidemiology, University of Pittsburgh Graduate School of Public Health, Pittsburgh, PA; Russell H. Wiesner, M.D. (principal investigator), Michael R. Charlton, M.D. (co-investigator), Ruud A.F. Krom, M.D. (co-investigator), and Lori Schwerman, R.N. (research nurse coordinator), from the Departments of Medicine and Surgery, Mayo Clinic Foundation, Rochester, MN; Daniel Schafer, M.D. (principal investigator), Byers W. Shaw, Jr., M.D. (co-investigator), Rowen K. Zetterman, M.D. (co-investigator), and Elizabeth Strudthoff (research nurse coordinator), from the Departments of Medicine and Surgery, University of Nebraska Medical Center, Omaha, NE; John Roberts, M.D. (principal investigator), Nathan Bass, M.D., Ph.D. (co-investigator), and Bev Nikolai, C.C.R.C. (research coordinator), from the Departments of Medicine and Surgery, University of California, San Francisco, CA; and James Everhart, M.D., M.P.H. (program director), and Jay H. Hoofnagle, M.D. (division director), from the Division of Digestive Diseases and Nutrition, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD. The authors also acknowledge Kathryn M. Nelson, Stephen M. Myles, Ph.D., and Maura Rynn for a thorough review of the manuscript and the Mayo Clinic Arizona Academic Support Office for its superb editorial assistance.

Ancillary