Potential conflict of interest: Nothing to report.
Patients infected with HIV-1 who are heterozygous at HLA class I loci present greater variety of antigenic peptides to CD8+ cytotoxic T lymphocytes, slowing progression to AIDS. A similar broad immune response in chronic hepatitis C (CHC) infection could result in greater hepatic injury. Although specific HLA class II alleles may influence outcome in CHC patients, the role of HLA class I heterogeneity is generally less clearly defined. Our aims were to determine whether HLA class I allelic diversity is associated with disease severity and progression of fibrosis in CHC. The study population consisted of 670 adults with CHC, including 155 with advanced cirrhosis, and 237 non–HCV-infected controls. Serological testing for HLA class I antigens was performed via microlymphocytotoxicity assay. Peptide expression was defined as heterozygous (i.e., a different allele at each locus) or homozygous. Fibrosis staging was determined using METAVIR classification. Heterozygosity at the B locus (fibrosis progression rate [FPR] 0.08 vs. 0.06 units/yr; P = .04) and homozygosity at the A locus (FPR 0.10 vs. 0.08 units/yr; P = .04) predicted a higher median FPR. Age at infection, genotype, and duration of infection were also predictors of FPR. A higher proportion of patients with stage F2-F4 expressed HLA-B18 compared with controls (OR 2.2, 95% CI 1.17-4.23; P = .02). These differences were not observed in patients with advanced cirrhosis. HLA zygosity at 1, 2, or 3 alleles was not associated with fibrosis stage, liver inflammation, or treatment outcome. In conclusion, HLA class I allelic diversity has a minor influence on FPRs and disease severity in CHC. (Hepatology 2006,43:241–249.)
Hepatitis C virus (HCV) is a small enveloped RNA virus that is a common blood-borne infection affecting close to 3 million people in the United States and 170 million individuals worldwide.1, 2 The majority of these patients develop chronic infection, which is characterized by varying degrees of inflammation and hepatic fibrosis. A proportion of patients (<20%) will develop progressive liver damage with subsequent cirrhosis, end-stage liver disease, and hepatocellular carcinoma over 20 to 40 years.3
The immunopathogenic mechanisms responsible for viral persistence and varying degrees of hepatic injury in chronic hepatitis C (CHC) are poorly understood. Cellular immune responses probably play an important role in this regard. CD4+ T helper cells recognize HCV-derived peptides in the HLA class II binding groove of antigen-presenting cells such as dendritic cells, macrophages, and B cells. CD8+ cytotoxic T cell lymphocytes (CTL) recognize intracellularly processed HCV proteins presented on HLA class I molecules on virus-infected cells. The HLA class I and II loci contain a highly polymorphic set of genes that appear to have evolved over time through selection pressures, thus allowing the host to resist a variety of pathogens.4 In comparison with homozygous individuals, persons with overdominant selection or heterozygote advantage at HLA loci may be able to present a greater variety of antigenic peptides to T cells, thus resulting in a broader immune response. Such HLA allelic diversity may protect against severe forms of Plasmodium falciparum malaria5 and persistent hepatitis B infection in West Africa.6 An effective CTL response is vital for the control of HIV-1 infection,7 and maximal heterozygosity at HLA class I alleles appears to slow progression to AIDS and AIDS-related death.8
Although HCV may not cause direct cytopathic injury to infected hepatocytes, the host immune response is important in this regard, resulting in hepatic inflammation, CTL-mediated lysis of hepatocytes, and fibrogenesis. Certainly a weak or shortened immune T helper cell response may lead to persistent infection and the development of chronic hepatitis,9 whereas a strong and polyclonal CTL response is associated with lower levels of viremia.10 Numerous reports have failed to show an association between individual HLA class I antigens and the course of disease in HCV infection.11–14 In contrast, several studies have suggested that specific HLA class II alleles may influence outcome parameters in chronic HCV infection such as disease susceptibility, viral clearance, or disease severity.15–21 However, many of these studies were small and reported inconsistent findings. Despite the propensity of HCV to present a diverse range of epitopes for T cell recognition, heterozygous advantage has not been addressed or demonstrated in these studies.
The aims of this study were to evaluate the role of HLA class I allelic diversity in determining disease progression and severity in a large sample of patients with CHC. In particular, we hypothesized that patients with either overall or specific heterozygosity at these class I alleles would potentially have more rapid disease progression.
HCV, hepatitis C virus; CHC, chronic hepatitis C; CTL, cytotoxic T cell lymphocyte; FPR, fibrosis progression rate; ALT, alanine aminotransferase.
Patients and Methods
Adult patients with previously untreated CHC infection were assessed at 2 centers from 1992 to 2001: Scripps Clinic and Research Foundation (La Jolla, CA) and King's College Hospital (London, UK). Eligible patients with a single chronologically identifiable risk factor were identified from a biorepository database at each center and were included in the case–control study. Each patient required confirmation of CHC on a pretreatment liver biopsy and serological evidence of HCV RNA via reverse-transcriptase polymerase chain reaction assay. Patients were excluded following clinical assessment, serological testing, and histological evaluation if they had evidence of hepatitis B or HIV infection, daily alcohol consumption of more than 50 g/d, or other causes of chronic liver disease. At the initial assessment, patients were asked to complete a detailed questionnaire that collected demographic information, including sex, ethnic background, year of infection, mode of infection (e.g., first year of injecting drug use), and alcohol history (both current and past intake). Past alcohol intake was estimated from patient questionnaires and medical records and was categorically graded as less than 10 g/d, 10-50 g/d, or more than 50 g/d. Most of these patients had participated in other clinical therapeutic trials and thus had an alcohol intake of less than 10 g/d over the preceding 12 months. Other data collected on host and viral characteristics included genotype, HCV RNA levels, serum aminotransferases (aspartate and alanine aminotransferase), and liver histology. Blood samples were collected from patients at or near the time of liver biopsy. Peripheral blood mononuclear cells were isolated within 4 hours of collection and stored at −70°C until analysis. HLA class I antigen typing was performed as described in the “HLA Class I Typing” section below. HLA class I typing was also performed on a subset of 155 patients with advanced cirrhosis listed for transplantation at the two centers. The control population was determined from a donor registry that included HLA data on 237 consecutive deceased organ donors in San Diego County. The study protocol was approved by an institutional review board at both centers, and all patients provided written informed consent.
Liver biopsies obtained at the initial assessment were evaluated by a single pathologist at each center, blinded to clinical and laboratory information, using the METAVIR scoring system.22 Biopsies were deemed adequate if they contained at least 5 portal tracts. The METAVIR scoring system stages fibrosis on a linear scale of 0 to 4: F0, no fibrosis; F1, portal fibrosis without septa; F2, few septa; F3, numerous septa without cirrhosis; F4, cirrhosis. Necroinflammatory activity was graded according to the modified Knodell histological activity index (representing the sum of portal, periportal, and lobular inflammation and Knodell fibrosis stage). Fibrosis progression rate (FPR; fibrosis units per year) was calculated as the ratio between the fibrosis stage (in METAVIR units) and the estimated duration of infection (in years). For example, if a person who received a blood transfusion 15 years ago and a subsequent liver biopsy 10 years later reveals stage 2 fibrosis, that person has a presumed FPR of 0.2 units/yr.
HCV RNA Detection and Genotyping.
Serum HCV RNA was assessed using either multicycle reverse-transcriptase polymerase chain reaction (SuperQuant, National Genetics Institute, Los Angeles, CA; lower limit of detection 100 copies/mL)23 or Cobas Amplicor version 2.0 (Roche Molecular Systems, Pleasanton, CA; lower limit of detection 600 copies/mL). HCV genotyping was performed using a line-probe assay (Inno-LiPA; Innogenetics NV, Zwijnaarde, Belgium) as previously described.24
Isolation of Peripheral Blood Mononuclear Cells.
Peripheral blood mononuclear cells from study patients were isolated from 10 mL of heparinized blood collected before treatment. Cells were separated within 4 hours on Ficoll-Histopaque density gradient (Sigma, St Louis, MO), washed 3 times in Hank's balanced salt solution (Life Technologies, Grand Island, NY), and cryopreserved at −70°C.
HLA Class I Typing.
Serological testing for HLA class I antigens (HLA-A, HLA-B, and HLA-C) in all patients was performed with a standard complement-dependent micro-lympho-cytotoxicity assay.25 This detection method uses well-characterized HLA antisera from multiparous females and monoclonal sources that are placed into individual wells on commercial 144-well class I typing trays (Biotest AG, Dreieich, Germany) organized as a panel to identify a complete HLA type for all 3 loci. In the presence of exogenous complement, HLA antibodies are cytotoxic to lymphocytes expressing the corresponding antigen. After further incubation, cell death was determined by Acridine Orange/Ethidium Bromide (FluoroQuench; One Lambda Inc., Canoga Park, Ca.) vital stain exclusion. The pattern of reactivity is then interpretable as the HLA type of the subject. Patients with different alleles at each locus were considered heterozygous; those with the same allele at each locus were considered homozygous.
For this case–control study, summary statistics are reported as the mean ± SD or median and range for continuous variables, and as median or proportions and ranges for categorical data. The Wilcoxon rank sum test was used to compare differences in the FPR between the heterozygotes and the homozygotes at each of the three HLA loci. The Kruskal-Wallis test was used to assess differences in FPRs among the groups with varying degrees of heterozygosity at the three HLA class I loci. Omnibus Fisher exact tests were used to compare allele frequencies at the HLA loci among controls, patients with CHC, and patients with advanced cirrhosis. Pairwise exact tests were used to identify individual group differences. Kaplan-Meier curves and the log rank test were used to depict and assess any differences in time to progression to fibrosis stage 4 between the heterozygous and homozygous patients without transplants. In this regard, baseline (time 0) was taken as the estimated date of infection, and patients who had not progressed to fibrosis stage 4 are treated as censored at the time of last observation. Multivariable logistic and linear regression were used to assess the significance of various potential predictors on progression of fibrosis —namely, METAVIR fibrosis scores (dichotomized as F0-F1 vs. F2-F4 for logistic regression) and fibrosis units per year (linear regression). Our modeling strategy was to undertake all subset regressions with nominal P values of .10 or less for all variables included in the models reported. We report the logistic regression model with minimal deviance and the linear regression with maximal adjusted R2. All tests and confidence intervals are two-sided.
The demographic and baseline characteristics of the main CHC study population are summarized in Table 1. Of the 515 patients with CHC evaluated in this study, 302 (59%) had injecting drug use and 158 (31%) had blood transfusion as a risk factor for HCV infection. These patients were predominantly Caucasian males with a mean age and duration of infection of 43.4 (range 16 to 70) and 20.2 (range 1 to 56) years, respectively. More than half of the patients had mild to moderate fibrosis (stage 1-2, 297/515; 57.7%) with a mean (± SD) rate of fibrosis progression of 0.11 ± 0.20 METAVIR units/yr.
Table 1. Baseline Characteristics of Patients With Chronic Hepatitis C
F0-4 (n = 515)
F0-1 (n = 270)
F2-4 (n = 245)
NOTE. Numbers (%) and chi-square tests summarize the categorical variables; means ± SDs and Mann-Whitney tests summarize the continuous variables.
Abbreviations: NA, not available; ALT, alanine aminotransferase; ULN, upper limit of normal; HAI, histological activity index.
Hispanic (50%), Asian (14%), Native American (8%), African American (6%), other (12%).
Patients with a homozygous pattern of expression at HLA-A tended to have slightly higher FPRs compared with patients that were heterozygous at this locus (median 0.10 vs. 0.08 METAVIR units/yr; Wilcoxon, P = .04). However, at the HLA-B locus, median FPRs were somewhat higher amongst heterozygotes compared with homozygotes (median 0.08 vs. 0.06 METAVIR units/yr; Wilcoxon, P = .04). FPRs did not differ significantly between heterozygotes and homozygotes at the HLA-C locus (Fig. 1).
There were no significant differences in the median FPRs for patients that were heterozygous at all 3 loci (n = 240; median 0.08 METAVIR units/yr) compared with those that were homozygous at either 1 (n = 198; median 0.08 METAVIR units/yr), 2 (n = 53; median 0.08 METAVIR units/yr), or all 3 loci (n = 15; median 0.05 METAVIR units/yr) (Kruskal-Wallis test, P = .65).
Allelic Frequency and Disease Severity.
We compared homozygote and heterozygote allelic frequencies at HLA-A, HLA-B, and HLA-C between the 237 normal controls and the patients without transplants but with no or minimal fibrosis (stages F0-F1) or with moderate to severe disease (stages F2-F4) (Table 2). There were no statistically significant differences in allelic frequencies at HLA-A, HLA-B, or HLA-C between the controls and patients with either stage F0-F1 or F2-F4 fibrosis. Likewise, there were no differences in allelic frequencies between patients with stage F0-F1 and F2-F4. In general, individuals were more likely to be heterogeneous at HLA-A or HLA-B than at HLA-C. Analysis of individual allele frequency indicated that although a higher proportion of patients with stage F2-F4 fibrosis expressed HLA-B18 compared with controls (OR 2.2; 95% CI 1.17-4.23; P = .02), there were no such differences in this, or other alleles, between patients with minimal and moderate to severe fibrosis (Table 3).
Table 2. Associations Between Frequencies of Allelic Zygosity and Severity of Disease
In a Kaplan-Meier survival model, log rank tests revealed no statistically significant differences between zygosity patterns at HLA-A, HLA-B, or HLA-C and time to progression to cirrhosis (F4) (Fig. 2).
Predictors of Fibrosis Progression Rate.
Multivariable logistic and linear regressions were used to assess the significance of potential predictors of FPR; in particular, we were interested in determining whether heterozygosity at the class I alleles provided independent information relating to FPRs given other known risk factors and potential predictors. We evaluated 2 fibrosis outcome measures: (1) METAVIR fibrosis stage, dichotomized as F0-F1 versus F2-F4 for logistic regression, and (2) FPR in METAVIR units per year for linear regression. The independent predictors included in these multivariable regressions were: age at infection (<40 vs. ≥ 40 yr), duration of infection, sex (male or female), race (Caucasian vs. non-Caucasian), alcohol consumption (<10 g/d, 10-50 g/d, >50 g/day), HCV genotype (1 vs. non-1), log10 HCV RNA, ALT levels, ALT ratio (ALT/upper limit of normal), risk factors (intravenous drug use, needle/tattoos, blood transfusion, other), HLA-A zygosity (homozygosity vs. heterozygosity), HLA-B zygosity, and HLA-C zygosity.
Neither METAVIR fibrosis stage nor FPR could be precisely modeled in our multivariate regression analyses. Three variables (ALT ratio, duration of infection, and race) were independent predictors of METAVIR fibrosis stage (deviance = 696.5, degrees of freedom = 511, P < .0001). The likelihood of being in METAVIR fibrosis stage F2-F4 increased with increasing values of ALT ratio and longer infection periods, and the odds of non-Caucasians being in stage F2-F4 relative to Caucasians was 1.48 (95% CI 0.99-2.20). We found 4 significant predictors of FPR: HCV genotype, HLA-A zygosity, age at infection, and duration of infection (F4,510 = 28.3; P < 10−5; adjusted R2 = 0.18). FPR (METAVIR units/yr) averaged 0.13 ± 0.23 for HCV genotype non-1 vs. 0.10 ± 0.18 for HCV genotype 1; 0.11 ± 0.18 for heterozygotes at HLA-A vs. 0.15 ± 0.28 for homozygotes; and 0.11 ± 0.17 for individuals infected before 40 years of age vs. 0.33 ± 0.48 for those infected at age 40 or older.
HLA Zygosity and Advanced Cirrhosis.
A separate sample of 155 patients with CHC and advanced cirrhosis listed for transplantation was also evaluated in this study. As expected, there was a male predominance in this group (n = 112; 72%) with an overall mean age of 49.2 ± 7.6 years at the time of biopsy. Although there were a greater number of heterozygotes than homozygotes at HLA-A (133 vs. 22) and HLA-B (140 vs. 14 [undetectable blanks = 1]), these differences were not apparent at HLA-C (21 vs. 21 [undetectable blanks = 113])—and, importantly, the homozygous versus heterozygous frequencies did not differ between controls and patients with advanced cirrhosis patients (Table 2).
This retrospective study of over 900 patients represents the largest clinical study to have examined the role of HLA class I allelic diversity in disease progression and severity of liver injury in CHC infection. Differences in zygosity at HLA-A and HLA-B loci resulted in minor differences in FPRs. HLA-A homozygosity predicted slightly higher FPRs compared with HLA-A heterozygous patients. There were no significant differences noted in allelic expression for the subgroup of patients with advanced cirrhosis.
Prior HLA disease association studies in patients with CHC have mostly investigated the role of HLA class II polymorphisms in persistent infection or response to interferon therapy. Although certain major histocompatibility complex class II alleles such as DRB1*11 have been implicated in reducing the severity of liver injury in patients with CHC, these studies have been relatively inconsistent in their findings, in part because they reflect differing geographic and racial populations. Furthermore, with regard to liver injury, certain differences in class II polymorphism expression have only been noted when comparing broad categories of patients with and without cirrhosis.26 By comparison, there is a paucity of studies in the published literature that have evaluated associations between HLA class I and disease severity or outcome in CHC.27 There appears to be no clear explanation for this discrepancy, given the number of specific HLA class 1–restricted CTL responses that have already been determined28 and the potential role of cytotoxic T cells in mediating liver injury in HCV infection.29 This discrepancy may reflect a preponderance of negative studies that have not been published, or that following early association studies in HLA and hepatitis B viral persistence, investigators have preferentially targeted specific and closely related groups of alleles based on molecular techniques for class II typing. Certainly the microlymphocytotoxicity type assay, as used in our study for HLA class I typing, is less specific and defines a broader range of alleles, thus making it difficult to determine weaker disease associations.
Heterozygote advantage at class I loci has been shown to confer a selective advantage for delaying progression to AIDS and reducing viral burden in human T-lymphotropic virus type 1 infection.30 Our study demonstrated only weak links between FPRs and heterozygosity at HLA-B and homozygosity at HLA-A. The reasons for these observed differences between zygosity patterns and disease progression are unclear. Certainly the use of FPRs in HCV disease models has inherent problems relating to liver biopsy staging and sampling errors, the assumption that liver injury follows a linear course, and inaccuracies in estimating the duration of infection.31 The use of patient self-reporting for duration of infection may have led to a potential bias in our Kaplan-Meier baseline estimates for projected time to progression to cirrhosis, but this result would have affected the homozygous and heterozygous subgroups equally. Approximately half of the patients in this study had minimal fibrosis with slow FPRs that would make it difficult to detect any weak associations that may influence disease progression. Because of extensive polymorphisms at class I loci, there may be differences in the presentation or binding affinity of HCV peptides to HLA molecules amongst individuals. In addition to allelic heterogeneity, differential expression of cytokines such as interferon γ that may upregulate HLA class I expression could affect peptide binding or presentation and play a role in mediating hepatic injury.32 There may be specific downregulation of HLA-A and HLA-B as has been observed in HIV-1 infection33 that may potentially limit the CTL response, and thus the degree of liver injury. However, other studies have failed to demonstrate an interaction between HCV proteins and major histocompatibility complex class I processing and antigen presentation.34 Another factor that could obscure the effects of the broader range of peptides presented by heterozygotes is the presence of altered-peptide ligands that bind to HLA molecules without stimulating CTL responses.35 Further studies are also needed to determine differences in CTL responses based on allelic heterogeneity in CHC infection. In addition, selective heterozygote advantage may not be evident in a population for an infection that results in a slowly progressive disease state unlikely to lead to death before reproduction. Even in our small group of patients with with CHC advanced cirrhosis, no differences were noted in the frequency of heterozygotes or individual alleles compared with the control population. Genetic influences may be difficult to predict in hepatitis C infection, where there are likely to be several confounding variables that influence outcome.36 Our findings suggest a relatively minor role for HLA class I alleles in liver injury and fibrosis in CHC, particularly in comparison to host factors that influence disease severity such as age, sex, and alcohol. It is possible that non-HLA class I loci may influence disease progression in CHC. Certain polymorphisms in the “transport associated with antigen processing” genes in the HLA class II region have been shown to influence disease progression.37 However, compared with host factors, the relative risk conferred by specific HLA alleles for disease severity is likely to be low, particularly when taking into account the significant degree of major histocompatibility complex polymorphisms that are also present in a heterogeneous population. Furthermore, the clinical relevance of any potential novel associations between HLA and disease severity would also need to be determined.
Certain HLA-C alleles may be important in HCV infection, because they may serve as ligands for natural killer cell receptors that influence viral persistence,38 thus leading to ongoing hepatocellular injury. We did not find any significant associations between HLA-C zygosity and disease progression or severity. Cell surface HLA-C expression using nonmolecular typing is approximately 10% that of HLA-A or HLA-B,39 whichwould have certainly limited assessment of allele-specific disease associations in this study.
HLA class I loci may occur in tight linkage disequilibrium in certain geographic populations, and any effect of zygosity on disease progression may be due to a single class I locus effect rather than individual loci. Certainly no heterozygote selection advantage was evident in terms of disease progression at either two or all three loci in our population. Furthermore, there were no associations between zygosity at any of the HLA class I loci and hepatic inflammation, HCV RNA level, genotype, and response to interferon-based therapy (data not shown).
In conclusion, the results of this study suggest that differences in zygosity at HLA class I alleles result in minor differences in FPRs and do not influence overall disease severity in CHC infection. Hepatic injury is more likely to be determined by a complex interplay between other host, genetic, and viral factors. Use of molecular typing techniques, or studies in other patient groups, such as the HIV/HCV-coinfected population, may yield more allele-specific disease associations.