SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES

Outcomes of infection with hepatitis C virus (HCV) vary widely, from asymptomatic clearance to chronic infection, leading to complications that include fibrosis, cirrhosis, hepatocellular carcinoma, and liver failure. Previous studies have reported statistical associations between human leukocyte antigen (HLA) heterozygosity and favorable outcomes of infection with either hepatitis B virus (HBV) or human immunodeficiency virus (HIV) (the “heterozygote advantage”). To investigate whether HLA zygosity is associated with outcome of HCV infection, we used data from the United States Organ Procurement and Transplantation Network database of 52,435 liver transplant recipients from 1995 through 2005. Of these, 30,397 were excluded for lack of HLA data, retransplantation, known HIV infection, or insufficient information regarding HBV infection. The remaining cases were analyzed for associations between HCV infection and HLA zygosity with 1-sided Fisher's exact tests. Results show significantly lower proportions of HLA-DRB1 heterozygosity among HCV-infected than uninfected cases. The differences were more pronounced with alleles represented as functional supertypes (P = 1.05 × 10−6) than as low-resolution genotypes (P = 1.99 × 10−3). No significant associations between zygosity and HCV infection were found for other HLA loci. Conclusion: These findings constitute evidence for an advantage among carriers of different supertype HLA-DRB1 alleles against HCV infection progression to end-stage liver disease in a large-scale, long-term study population. Considering HLA polymorphism in terms of supertype diversity is recommended in strategies to design association studies for robust results across populations and in trials to improve treatment options for patients with chronic viral infection. Access to deidentified clinical information relating genetic variation to viral infection improves understanding of variation in infection outcomes and might help to personalize medicine with treatment options informed in part by human genetic variation. (HEPATOLOGY 2007.)

Broad-spectrum responses by CD4+ and CD8+ T cells are essential mechanisms of cell-mediated immunity to viral infection,1–4 and increasing evidence supports a central role of T cell responses in the clearance and control of hepatitis C virus (HCV) infection.5 T cells interact with infected cells when T cell receptors bind to human leukocyte antigen (HLA) molecules, glycoproteins that present antigen peptide fragments, or epitopes on the surfaces of infected cells.4, 6, 7 HLA genes exhibit extreme allelic polymorphism.4, 6, 7 This polymorphism is important both within an individual and between individuals. At the population level, HLA polymorphism provides a mechanism to limit the spread of pandemic pathogens, which might otherwise propagate unchecked throughout a host population with homogeneous defenses.8 Within an individual, inheritance of duplicate—or homozygous—HLA alleles may result in a disadvantage against chronic viral infection due to a diminished repertoire of epitope presentation to T cells.

Despite the polymorphism of HLA alleles, variation in only a limited number of amino acid residues known as anchor or binding motifs strongly influences the epitopes to which an HLA molecule will bind.4, 6 In addition to this primary anchor recognition, evidence supports a secondary anchor recognition component,9 and mutations outside of the primary anchor motif leading to escape from T cell response have been reported.10 HLA alleles with largely overlapping binding motifs are grouped together into supertypes.11, 12 An individual carrying 2 different alleles of the same supertype is able to present a narrower range of antigens than an individual carrying 2 alleles of different supertypes. A likely consequence for the restricted T cell repertoire that results from having 2 loci with the same supertype is reduced ability to impede escape mutants from further replication at the expense of the host.

Heterozygote advantage has been reported for human immunodeficiency virus (HIV) and hepatitis B virus (HBV) infections.13, 14 To increase understanding of the genetic components of differences in response to HCV infection, we compared HLA zygosity between HCV-infected cases and uninfected controls using a database of over 50,000 liver transplant recipients from the past decade. This database is a uniquely large-scale collection of information regarding HLA alleles and viral infection status. HLA typing is uncommon in medical practice until organ transplantation is indicated, when compatibility between donor and recipient tissues becomes a concern.15 Consequently, abundant information about HLA alleles is rare. Using a data set of this magnitude allowed us to test the hypothesis of heterozygote advantage against HCV, which posits that HCV-infected carriers of heterozygous HLA alleles benefit from reduced progression to end-stage liver disease and liver transplantation than HCV-infected carriers of homozygous HLA alleles. We also tested for associations between either individual HLA alleles or HLA-B-DRB1 haplotypes and presence of HCV infection.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES

Organ Procurement and Transplantation Network Data.

To facilitate the process of matching organ donors with recipients, the United States Organ Procurement and Transplantation Network (OPTN) maintains a repository of HLA alleles and clinical data. We used this information to test for associations between HLA alleles and HCV infection status in a retrospective study based on OPTN data as of August 1, 2006, with 52,436 liver transplant recipient cases reviewed from 1995-2005. All data were fully anonymous. A summary of the demographic characteristics of these cases is available elsewhere.16

Several criteria were required to include a case for analysis. The availability of at least 1 HLA allele was mandatory. Furthermore, retransplantation could bias allele counts, because duplicate genotypes are overrepresented in multiple database records, so retransplants were excluded. HIV infection status is not always recorded in the OPTN database because of differences between states' legislation regarding permission to disclose this information. Patients known to be infected with HIV were excluded to minimize confusion with effects due to coinfection with HIV. Thus, the HIV infection status of the included patients was unknown. Additionally, information regarding both HBV and HCV infection status was required to control for hepatitis coinfection.

HCV infection status was determined from at least 1 of 4 database fields: clinical HCV presentation, HCV RNA assay, HCV antibody, and HCV recombinant immunoblot assay. Of these data attributes, knowing either of the last 2 could potentially resolve the distinction between cleared and chronic infections. Unfortunately, due to missing values for antibody test and recombinant immunoblot assay results in the database, we could distinguish reliably between cleared and persistent infections for only a small number of cases (68 cleared HCV infections were confirmed); consequently, all HCV-infected cases were pooled together rather than stratified by outcome.

HBV infection was interpreted similarly from the following database fields: HBV DNA, HBV clinical presentation, HBV core, and HBV surface antigen. Any positive test result was interpreted as evidence for HBV infection. At least 1 negative result was required for evidence of no HBV infection. Missing data were interpreted as no result.

Methods for HLA typing are based on either serological or genotypic assays.6, 17 This provides HLA types for 5 loci (A, B, C, DR, and DQ) in the OPTN database. Homozygosity was defined as the presence of 2 identical alleles for a locus, or the presence of a single allele and an indication in the database that no second allele was detected. (This second criterion is distinct from cases with no data for a second allele.) Because viruses can attenuate the host immune response, homozygosity in this sense could indicate either inherited homozygosity or a phenotypic trait acquired in the disease state.18, 19 False negatives among laboratory assays could also cause failure to detect a second allele. Consequently, we also evaluated implications for a stricter definition of homozygosity, which excluded cases without information in the database for 2 alleles per locus.

Statistical Inference.

We tested for significant associations between HLA alleles and HCV infection status as categorical data, whereby each case was assigned to 1 cell in a 2-by-2 contingency table (e.g., for presence/absence of HCV infection versus presence/absence of homozygosity for a particular HLA locus). We applied statistical inference to the resulting matrix with a series of 1-sided Fisher's exact tests. In this approach, the odds ratio (OR) is the test statistic, which quantifies the prevalence of correct inferences when testing for an association between binary categorical variables.20 We tested the null hypothesis that the resulting OR is not greater than unity, with an alternative hypothesis that the OR is greater than unity. Rejecting the null hypothesis at an experiment-wide false-positive rate of 0.05 indicates that the proportion of HLA heterozygosity differs significantly between HCV-infected and uninfected groups, with 95% confidence.20

The Fisher's exact test requires that observations be independent, random samples from a clearly defined study population.20 We designed experiments to exclude pseudoreplication and excessive multiple testing. We preserved the overall rate by minimizing the number of statistical tests and applying the Bonferroni correction, which reduces the acceptable per-comparison error rate according to the total number of tests.20 Test results were calculated with R, version 2.4.0. Separate tests were performed for an association between zygosity and HCV infection status for each of 5 HLA loci, with alleles represented as 2-digit genotypes and as supertypes where possible.

Associations between individual HLA alleles and chronic HCV infection were evaluated similarly, though 2-sided tests were performed because insufficient information exists to motivate a priori the hypothesis that any given allele is either beneficial or detrimental in response to HCV infection. Multiple testing was accommodated with a revised approach to interpretation of P values, because of the number of hypothesis tests performed. This approach deployed a transformation of P values to Q values following methods previously described.21 The Q value also represents an error rate, described as the false-discovery rate. Test results with Q values below 0.05 were considered statistically significant.

We inferred HLA-B-DRB1 haplotype phases in R with the haplo.stats package, version 1.2.2. Cases with missing or unresolved alleles were excluded, leaving 13,825 HBV-negative cases. Only phased haplotypes having 90% or greater posterior probability were considered further. Haplotypes were tested for associations with viral infection status using 2-sided Fisher's exact tests as described above.

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES

In light of the inclusion criteria stated above, 30,397 cases were excluded and 22,038 cases were subsequently analyzed (Fig. 1). Among included liver transplant recipients, 11,728 (53.2%) were infected with neither HBV nor HCV, 1708 (7.8%) were infected with HBV only, 5901 (26.8%) were infected with HCV only, and 2701 (12.2%) were coinfected with both HBV and HCV. Infection with hepatitis virus was strongly associated with HCV and HBV coinfection (OR = 3.143, 95% confidence interval >2.966, P < 2.2 × 10−16). Because of the known heterozygote advantage for HBV, HBV-infected cases were withheld from association tests.

thumbnail image

Figure 1. Case review process for OPTN liver transplant recipient data, summarizing cases excluded with the reason for exclusion and viral hepatitis infection status among cases included. Numbers in parentheses are case counts.

Download figure to PowerPoint

Heterozygote Advantage.

With alleles represented as 2-digit genotypes, we find a significantly greater prevalence of HLA-DRB1 heterozygosity among uninfected than HCV-infected cases (P < 0.002), and no significant differences for any other genes (Table 1). This result is sustained by representing alleles as supertypes, where the proportion of carriers of different HLA-DRB1 supertypes differs significantly between HCV-infected and uninfected groups (P = 1.05 × 10−6) (Table 1).

Table 1. Contingency Table Counts and Results from Fisher's Exact Test for Associations Between HCV Infection and HLA Zygosity for Alleles Represented as 2-Digit Genotypes and as Supertypes
HLA LocusNo. of Caseshom−/HCV−hom+/HCV−hom−/HCV+hom+/HCV+Odds Ratio95% Confidence IntervalP Value
  • HBV-infected cases were excluded. The definition of homozygosity (hom+) includes cases with failure to detect a second allele.

  • *

    P < 0.05/20.

Genotype alleles        
A15,690982460849852730.8849>0.77970.9533
B16,42210,54138053151860.9708>0.83160.6433
C6749374474618713881.0401>0.92710.2915
DQB112,034724283534944631.1493>1.03650.01313
DRB115,657996549748913041.2462>1.09830.001986*
Supertype alleles        
A15,6708933148644787731.0377>0.95770.2272
B16,4229436148546638381.1419>1.05620.00253
DRB115,657971874447114841.3420>1.21121.05 × 10−6*

Applying a stricter definition of homozygosity does not drastically alter the pattern of significance across loci when alleles are analyzed in terms of supertype diversity (Table 2). We observe strong support for advantage among carriers of 2 different HLA-DRB1 supertypes (P = 4.05 × 10−5). The association between HCV infection and HLA-DRB1 zygosity becomes insignificant, with alleles represented as 2-digit genotypes and sampling of fewer homozygotes in the strict sense (Table 2). Weak support for an association between allelic diversity and HCV infection appears in carriers of different HLA-B supertypes, though the test outcomes differ depending on the inclusion of cases in which no second alleles were detected (compare P = 0.00253 in Table 1 with P = 0.0007255 in Table 2).

Table 2. Contingency Table Counts and Results from Fisher's Exact Test for Associations Between HCV Infection and HLA Zygosity with Alleles Represented as 2-Digit Genotypes and as Supertypes
HLA LocusNo. of Caseshom−/HCV−hom+/HCV−hom−/HCV+hom+/HCV+Odds Ratio95% Confidence IntervalP Value
  • HBV-infected cases were excluded. The strict definition of homozygosity (hom+) was used, requiring that 2 alleles be detected per locus, and excluding cases where no second allele was detected.

  • *

    P < 0.05/20

Genotype alleles        
A14,8699824404985200.9854>0.59820.5698
B15,89510,540295315110.7522>0.38450.8347
C56613744311871150.9683>0.54010.5958
DQB110,98372421633494841.0681>0.84470.3363
DRB114,9229965424891241.1642>0.73360.3179
Supertype alleles        
A14,849893391844785201.1300>1.02550.01896
B15,89594351,13446636631.1829>1.08420.0007255*
DRB114,922971828947112041.4561>1.24334.05 × 10−5*

Overall, we find no evidence for an association between HLA zygosity and HCV infection for HLA-A, HLA-C, or HLA-DQ loci. We find strong support for an advantage of HLA-DRB1 supertype diversity against HCV infection, because cases of different HLA-DRB1 supertypes are relatively more rare among HCV-infected patients than in the uninfected sample of liver transplant recipients.

Alleles, Supertypes, and Haplotypes.

Alleles known to be associated with favorable HCV infection outcome include B*27, DRB1*11, and DQB1*03.1, 7, 22–24 These alleles were significantly underrepresented among HCV-infected cases, relative to uninfected cases (Q < 0.01) (Tables 3 and 4). Other alleles associated with favorable HCV infection outcome are A*02, B*08, B*39, B*62 (Q < 0.01) (Table 3), DRB1*04, DRB1*08, DQB1*03, and DQB1*07 (Q < 0.01) (Table 4). Supertypes associated with favorable HCV outcome are A2S and DR3S (P < 0.0007) (Table 5).

Table 3. Prevalence of Class I HLA Alleles Sampled and 2-Tailed Fisher's Exact Test Results for Allele Associations with HCV Infection Status
HLA AlleleHBV− Carriers (%)HCV−/HBV− Carriers (%)HCV+/HBV− Carriers (%)Odds Ratio (95% Confidence Interval)P Value
  • For brevity, only alleles with significant associations or over 10% population prevalence are shown.

  • *

    Q (false discovery rate) < 0.01.

A*014825 (15.0)3255 (15.2)1570 (14.6)1.06 (1.0–1.1)0.097
A*028060 (24.3)5470 (24.7)2590 (23.3)1.12 (1.0–1.2)0.00079*
A*034119 (12.9)2734 (12.9)1385 (13.0)1.00 (0.9–1.1)0.91
A*291369 (4.36)816 (3.91)553 (5.25)0.72 (0.6–0.8)3.4 × 10−8
B*073653 (11.0)2510 (11.3)1143 (10.3)1.13 (1.0–1.2)0.0024
B*083767 (11.3)2595 (11.7)1172 (10.5)1.14 (1.1–1.2)0.00068*
B*271011 (3.07)726 (3.32)285 (2.59)1.30 (1.1–1.5)0.0002*
B*38639 (1.94)369 (1.69)270 (2.45)0.68 (0.6–0.8)3.0 × 10−6*
B*39836 (2.54)609 (2.79)227 (2.06)1.37 (1.2–1.6)5.6 × 10−5*
B*444176 (12.6)2626 (11.9)1550 (13.9)0.82 (0.8–0.9)5.0 × 10−8*
B*621663 (5.05)1183 (5.40)480 (4.36)1.27 (1.1–1.4)2.4 × 10−5*
Cw*032273 (15.6)1573 (16.1)700 (14.5)1.16 (1.0–1.3)0.0057
Cw*042596 (17.4)1676 (16.9)920 (18.4)0.88 (0.8–1.0)0.015
Cw*061746 (12.4)1120 (12.0)626 (13.2)0.89 (0.8–1.0)0.038
Cw*074860 (30.2)3319 (30.7)1541 (29.0)1.15 (1.0–1.3)0.0036
Cw*08748 (5.46)445 (4.90)303 (6.55)0.72 (0.6–0.8)4.1 × 10−5*
Table 4. Prevalence of Class II HLA Alleles Sampled and 2-Tailed Fisher's Exact Test Results for Allele Associations with HCV Infection Status
HLA AlleleHBV− Carriers (%)HCV−/HBV− Carriers (%)HCV+/HBV− Carriers (%)Odds Ratio (95% Confidence Interval)P Value
  • For brevity, only alleles with significant associations or >10% prevalence are listed.

  • *

    Q (false discovery rate) < 0.01.

DRB1*044856 (15.2)3360 (15.8)1496 (14.1)1.17 (1.1–1.3)2.4 × 10−5*
DRB1*074454 (14.0)2570 (12.1)1884 (17.6)0.59 (0.5–0.6)<2 × 10−16*
DRB1*081445 (4.60)1050 (5.00)395 (3.79)1.35 (1.2–1.5)6.0 × 10−7*
DRB1*10383 (1.22)222 (1.06)161 (1.55)0.68 (0.5–0.8)0.00028*
DRB1*112284 (7.24)1711 (8.11)573 (5.49)1.57 (1.4–1.7)<2 × 10−16*
DRB1*134130 (13.0)2727 (12.9)1403 (13.3)0.95 (0.9–1.0)0.22
DRB1*153734 (11.8)2510 (11.9)1224 (11.6)1.03 (0.9–1.1)0.51
DQB1*013625 (14.3)2369 (14.0)1256 (15.0)0.91 (0.8–1.0)0.019
DQB1*026465 (25.1)4044 (23.6)2421 (28.0)0.70 (0.7–0.8)<2 × 10−16*
DQB1*033429 (13.8)2390 (14.3)1039 (12.7)1.17 (1.1–1.3)0.0002
DQB1*064176 (16.8)2749 (16.4)1427 (17.4)0.91 (0.8–1.0)0.028
DQB1*072440 (9.97)1892 (11.5)548 (6.86)1.88 (1.7–2.1)<2 × 10−16*
Table 5. Contingency Table Counts and Results from 2-Tailed Fisher's Exact Test for Associations Between HLA Supertype Alleles and HCV Infection Status
HLA SupertypeNo. of CasesAllele−/HCV−Allele+/HCV−Allele−/HCV+Allele+/HCV+Odds Ratio95% Confidence IntervalP Value
  • *

    P < 0.05/12.

A1S16,09558833058482423301.0761.01–1.150.03015
A2S16,59355622945548925971.1191.05–1.190.0006534*
A3S15,97059202928469224300.9550.89–1.020.1721
A24S15,85973083576322817470.9040.84–0.970.005263
B7S16,78161023122507324841.0450.98–1.120.1828
B27S16,51582434130273614060.9750.90–1.050.5058
B44S16,70959842939510126850.9330.87–1.000.03569
B58S16,4639786498011675301.1211.00–1.250.03904
B62S16,4738977457319819421.0710.98–1.170.1148
DR1S15,83183904216219010351.0630.98–1.160.1482
DR3S16,11570673012364123950.6480.61–0.69<2.0 × 10−16*
DR4S15,97669953623367416841.1301.05–1.210.0006368*

Alleles overrepresented among HCV-infected cases are: A*29, B*38, B*44, Cw*08 (Q < 0.01) (Table 3), DRB1*07, DRB1*10, and DQB1*02 (Q < 0.01) (Table 4). Supertype DR3S is disproportionately common among HCV-infected cases (P < 2 × 10−16) (Table 5).

For brevity, alleles carried by less than 10% of the HBV-negative cases sampled are not listed in Table 3 unless they were significantly associated with HCV status. These “rare” class I HLA alleles are: A*09, A*10, A*11, A*19, A*23, A*24, A*25, A*26, A*28, A*30, A*31, A*32, A*33, A*34, A*36, A*43, A*66, A*68, A*69, A*74, A*80, B*05, B*12, B*13, B*14, B*15, B*16, B*17, B*18, B*21, B*22, B*35, B*37, B*40, B*41, B*42, B*45, B*46, B*47, B*48, B*49, B*50, B*51, B*52, B*53, B*54, B*55, B*56, B*57, B*58, B*59, B*60, B*61, B*63, B*64, B*65, B*67, B*70, B*71, B*72, B*73, B*75, B*77, B*78, B*81, Cw*01, Cw*02, Cw*05, Cw*09, Cw*10, Cw*12, Cw*13, Cw*14, Cw*15, Cw*16, Cw*17, and Cw*18. Similarly, class II HLA alleles not listed in Table 4 are: DRB1*01, DRB1*0103, DRB1*02, DRB1*03, DRB1*05, DRB1*06, DRB1*09, DRB1*12, DRB1*14, DRB1*16, DRB1*17, DRB1*18, DQB*04, DQB*05, DQB1*08, and DQB1*09.

Haplotype phases were determined with high posterior probabilities (>90%) for 6653 HBV-negative cases, of which 2251 were HCV-positive. These yielded 721 2-locus haplotypes, 374 of which we tested for associations. (We opted not to test for associations among the other 347 haplotypes to eliminate multiple testing where the contingency tables contained zero-valued entries.) Only 1 result yielded P < 0.0001, which was considered significant in light of 374 statistical tests (Table 6). The B*44-DRB1*07 haplotype is significantly more common among HCV-positive than HCV-negative cases (P = 1.0 × 10−13) (Table 6). Conversely, the B*44-DRB1*11 haplotype is less common among HCV-positive than HCV-negative cases (P = 1.4 × 10−4) (Table 6). Haplotypes that include B*44 alleles can have either protective or detrimental associations with HCV infection status, contingent on the DRB1 allele. This corroborates the significant effect of HLA-DRB1 against HCV found when loci were analyzed separately, though an alternative interpretation is discussed below.

Table 6. HLA-B-DRB1 Haplotype Associations with HCV Infection Status
HLA-B-DRB1 HaplotypeHBV− CarriersHCV−/HBV− CarriersHCV+/HBV− CarriersOdds Ratio (95% Confidence Interval)P Value
B*44-DRB1*07709 (5.33%)379 (4.30%)330 (7.33%)0.55 (0.5–0.6)1.0 × 10−13
B*44-DRB1*1155 (0.41%)49 (0.56%)6 (0.13%)4.2 (1.8–12.1)0.00014
B*61-DRB1*1336 (0.27%)13 (0.15%)23 (0.51%)0.29 (0.1–0.6)0.00027

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES

Access to data from HLA allele types and viral infection status from thousands of clinical cases is rare. To our knowledge, this is the largest number of samples investigated for statistical associations between HLA variation and response to chronic viral infection. The OPTN cases present a rare opportunity for analytic insight, despite data-related concerns. Before discussing the implications of our findings, we address issues related to the data and how they might affect our findings.

The study population is limited to liver transplant recipients in the United States, where HCV genotype 1 is most common. Outcomes and end points other than liver transplantation were not considered, whether earlier, such as resolved infection,25–27 fibrosis, cirrhosis, and hepatocellular carcinoma,28, 29 or later, such as postoperative success.30 No sufficient information regarding HIV infection status was available, though we were able to control for HBV coinfection. Patient age, sex, ethnic background, viral type, and viral titer were unknown but can all influence allele frequencies or infection outcomes.31–35 Treatment histories of infected cases were not known but are presumed to differ widely. Because linkage disequilibrium is known to complicate genetic disease-association studies, 2-locus haplotypes were inferred from the OPTN data with alleles represented independently of chromosomes. Major histocompatibility complex loci are clustered together on the short arm of chromosome 6, producing an association between heterozygosity at observed and unobserved loci.6 Potential bias from matching donors to recipients should be negligible, because HLA matching requirements are less stringent for liver compatibility than for other organs.15

As with all association studies, particularly in genetic analysis of complex disease, interpretation may be confounded by hidden covariates; for example, some epitopes allow promiscuous (cross-locus) HLA presentation across supertypes and loci.36 A conspicuous shortcoming of the experimental design used here is that homozygous protective alleles were not accommodated in testing the hypothesis of heterozygote advantage.

Further evidence for involvement of class II HLA alleles in response to infection with HCV is not surprising. The heterozygote advantage against HCV has previously been suggested for a class II HLA locus.37 By contrast, relatively few studies have reported associations between class I genes and response to HCV infection.29, 38–41 Previously published studies also report failure to detect an association between HLA class I heterozygosity and favorable outcome of HCV infection, though they included a much smaller number of samples and did not deploy the supertype representation.38

A recent review of HLA associations with HCV infection outcome presents a meta-analysis from results appearing in previously published studies of associations between the HLA DRB1*11 and DQB1*0301 alleles with resolved HCV infections.22 Though the end points of those studies were much earlier in the course of HCV infection, the meta-analysis results are in accord with our findings. For DRB1*11, the summary estimate OR from 8 published investigations was 2.5 (95% confidence interval, 1.7-3.7) and a review of 7 investigations for DQB1*0301 associations yielded an estimated OR of 3.0 (95% confidence interval, 1.8-4.8). Both alleles have protective associations against HCV infection. Our OR values are lower but yield the same patterns of significance. Consistent associations between DRB1*11 alleles and less severe liver disease are widely reported.22

HLA-DRB1 is a class II locus and is the second most polymorphic.4, 6 It presents exogenously derived epitopes to CD4+ T helper cells on the surfaces of B cells, macrophages, and immune-activated cells. Involvement of CD8+ cytotoxic T lymphocytes via interaction with class I HLA molecules presenting endogenously derived antigen fragments is expected in chronic viremia.1–4 HLA-B is a class I locus and the most polymorphic HLA locus.4, 6 It is ubiquitously expressed, presents endogenously derived antigen peptide fragments on infected cell surfaces, and is known to be associated with differences in response to infection with HIV.42, 43

Response to viral infection is mediated by many different genes. The cell-mediated immune response is a concerted effort requiring interactions among multiple gene products. Rather than any single individual locus or allele in a solo effort to block viral replication, multiple loci play specialized roles as a team.1, 5 This is illustrated by considering B*44 alleles with low-resolution designations. Taken alone, B*44 alleles are seen as detrimental in association with HCV infection. However, their role in broader HLA-B-DRB1 haplotype context merits reconsideration, where the direction of the association is influenced predominantly by the DRB1 allele. Alternatively, unresolved linkage associations may exist among high-resolution variants of B*44 and DRB1 alleles, and both the specific B*44 and the specific DRB1 allele present could modify the course of HCV infections. While it is known that B*4402 is frequently inherited with DRB1*0401, and B*4403 with DRB1*0701,44 many other high-resolution B*44-DRB1 haplotypes are known,45 yet no such high-resolution allelic information was available from the OPTN data. Knowledge of high-resolution alleles is required for detailed understanding of how carriers of B*44 alleles respond to HCV infection.

Furthermore, HLA diversity is not the sole explanation for genetic variation in response to HCV infection. For example, interaction with natural killer cells via killer cell immunoglobulin-like receptors is another important source of variation in response to HCV infection.3, 38, 39 Unfortunately, killer cell immunoglobulin-like receptor variants are not represented in OPTN data.

Very few published studies have investigated associations between HLA supertypes and disease response.43, 46 We find the supertype classification scheme to be a useful representation of HLA polymorphism. Alleles of the same supertype present related epitopes.11, 12 Frequencies of supertypes are more stable than genotype frequencies when compared between populations composed of different ethnic backgrounds.11 Thus, reproducibility of results from HLA–disease association studies may be enhanced by employing the supertype representation.

Outcome of chronic viral infection results from an interaction between host and viral genotypes in an environment maintained by the host.47 Detailed data about variation in both host and viral genotypes, and clinical attributes of the infection outcome are essential to understanding pathosystem complexity.47, 48 Prospective trials to evaluate whether more aggressive treatment options are indicated for cases known to carry homozygous HLA-DRB1 alleles might be considered. Data repositories of anonymous host immunogenetic and viral genetic variation, coupled with information regarding infection outcome and appropriate control samples, are important tools to understand and predict variation between individuals in disease outcome.48–52

The fact that HLA genotype is not routinely known or considered presents an opportunity to apply knowledge of human immunogenetic variation. Evaluating HLA genotype earlier in treatment could ultimately help guide treatment decisions. Availability of more detailed yet anonymous HLA genotype information, case histories, and HCV subtypes are all helpful sources of information that will be required in abundance before repeatable, reliable predictions can be applied in clinical settings.

The diverse spectrum of outcomes resulting from HCV infection results from variation in interacting genetic processes between the virus and the host. Impaired immune response by CD4+ T cells is a requirement for chronic HCV infection, leading to transient, rather than sustained, control of viremia and, in time, increased viral loads.1, 5 We found a significant association between genetic diversity of the CD4+ T cell repertoire and progression of HCV infection to end-stage liver disease. These findings suggest considering HLA-DRB1 genotypes among other factors might be considered helpful in selecting therapeutic treatment options. HCV-infected patients carrying homozygous HLA-DRB1 loci or alleles associated with poor HCV immune response suffer greater risk for progression to end-stage liver disease than cases heterozygous at HLA-DRB1 or with protective alleles against HCV infection. Related findings of associations between variation in infection outcomes with HIV or HBV and HLA allelic diversity further corroborate clinical implications of this study. Applying enhanced understanding of human genetic variation in clinical settings could lead to treatment options with greater overall efficacy than current standards of care, and might help yield progress toward personalized solutions for medicine.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES

We thank T-10 and the HIV and HCV teams therein. Sarah Taranto at United Network for Organ Sharing provided access to the OPTN data and helpful explanations. Professor William Klitz kindly provided guidance for interpretation of haplotype phasing and high-resolution haplotype frequencies. Bill Bruno, Will Fischer, Bette Korber, Thomas Leitner, Alan Perelson, and Jim Szinger also provided scientific advisement. In memoriam: Roger A. Morris. LA-UR 06-5447.

REFERENCES

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. REFERENCES