Primary biliary cirrhosis (PBC) is an autoimmune liver disease characterized by portal inflammation and immune-mediated destruction of intrahepatic bile ducts that often leads to cirrhosis and liver failure.1 Although twin and family studies suggest that genetic factors contribute to disease susceptibility and severity,2, 3 the cause of PBC remains poorly understood.4 Significant associations of genetic factors, including HLA alleles, tumor necrosis factor alpha,5–8 and cytotoxic T-lymphocyte antigen 4,8–14 with PBC have been reported. Among these, only HLA has consistently been associated with PBC susceptibility.15 The HLA-DRB1*08 family of alleles has been the most frequently described determinant for susceptibility to PBC16–21; HLA-DRB1*08:03 has been associated with PBC in the Japanese.22–26 However, HLA DQB1 alleles and haplotypes have not been fully investigated, and large cohort studies have indicated that HLA-DRB1*11 and DRB1*13 alleles are, in fact, protective against PBC.20, 21, 26 Because recent genome-wide studies of PBC have shown the strongest association signals in the HLA region,27–30 we sought to determine whether particular HLA alleles or haplotypes or DRB1 allele amino acid alignments were associated with susceptibility to PBC or disease progression in the Japanese population.
Along with twin and family studies, recent genome-wide association studies suggest that genetic factors contribute to the susceptibility and severity of primary biliary cirrhosis (PBC). Although several reports have demonstrated that the human leukocyte antigen (HLA) DRB1*08:03 allele is associated with disease susceptibility in Japan, the precise analysis of HLA haplotypes and the role of amino acid alignment have not been fully clarified. We investigated HLA class I A, B, and C and HLA class II DRB1 and DQB1 alleles and haplotypes in 229 Japanese patients with PBC and compared them with the published data of 523 healthy subjects. Significant associations were found with PBC susceptibility for the DRB1*08:03-DQB1*06:01 (13% versus 6%; P = 0.000025; odds ratio [OR] = 2.22) and DRB1*04:05-DQB1*04:01 haplotypes (17% versus 13%; P = 0.044; OR = 1.38). Conversely, there were significant protective associations with the DRB1*13:02-DQB1*06:04 (2% versus 5%; P = 0.00093; OR = 0.27) and DRB1*11:01-DQB1*03:01 haplotypes (1% versus 4%; P = 0.03; OR = 0.37). The frequency of the DRB1*09:01-DQB1*03:03 haplotype was significantly higher in patients who had received orthotopic liver transplantation (33% versus 11%; P = 0.0012; OR = 3.96). Furthermore, the frequency of serine at position 57 (P = 0.0000015; OR = 1.83) of the DRβchain differed the most in patients with PBC, compared with healthy subjects. Conclusion: This study established the role of HLA haplotypes in determining PBC susceptibility and progression in the Japanese population. Further resequencing of the HLA region is required to more precisely identify the genetic components of PBC. (HEPATOLOGY 2012)
Patients and Methods
Clinical and biochemical features of 229 PBC patients who were enrolled for this study between January 2005 and September 2010 are shown in Table 1. The racial background of all patients was Japanese. HLA class I and II allelic genotypes from 523 healthy subjects obtained in a previous study were available as controls.31 In addition, HLA class II allelic genotypes from 130 patients with chronic hepatitis C virus infection were adopted from another study as comparison cases having another liver disease.32 The diagnosis of PBC was based on criteria from the American Association for the Study of Liver Diseases.33 Serum antimitochondrial antibody was determined using indirect immunofluorescence, and a titer of ≥1:40 was considered to be positive.34 Our serological protocol did not include testing for particular antinuclear antibodies, such as ani-gp210 antibody reactivity, or antimitochondrial antibody titration. All patients were negative for hepatitis B surface antigen, antibody to hepatitis B core antigen, antibody to hepatitis C virus, and antibody to human immunodeficiency virus. Patients were classified into two stages of PBC, based on their most recent follow-up: Early-stage patients were histologically Scheuer stage I or II35 or of unknown histological stage without liver cirrhosis, and late-stage patients were histologically Scheuer stage III or IV or clinically diagnosed with liver cirrhosis or hepatic failure.14 Liver cirrhosis was diagnosed by histological examination and/or characteristic clinical signs of advanced liver disease.36 All subjects provided written informed consent for this study, which was approved by the institutional ethics committee.
|Characteristics||PBC (n = 229)|
|Median age, years (range)||57||(27-86)|
|Female, n (%)||204||(89)|
|Late stage of disease, n (%)||50||(22)|
|Cirrhosis, n (%)||42||(18)|
|OLT, n (%)||15||(7)|
|Serum AMA positive, n (%)||209||(91)|
HLA Class I and II Typing.
Genomic DNA from patients and controls was isolated by phenolic extraction of sodium dodecyl sulfate–lyzed and proteinase K–treated cells, as described previously.37 HLA typing was carried out using a Luminex multianalyzer profiling system (Luminex, Austin, TX) with a LAB type SSO One Lambda typing kit (One Lambda, Inc., Canoga Park, CA), which is based on polymerase chain reaction sequence-specific oligonucleotide probes. HLA genotypes were determined by sequence-based typing. Peptide sequences of all HLA-DRB1 alleles in the IMGT/HLA database release 3.4.0 (April 2011) were aligned.
Phenotype frequencies were estimated by direct counting for each HLA allele. The significance of an association was evaluated by determining the standard P values after chi-squared analysis or Fisher's exact test. A P value of less than 0.05 was considered statistically significant. Association strength was estimated by calculating the odds ratio (OR) and 95% confidence interval (CI).
Distribution of HLA A, B, and C Alleles.
Among HLA class I alleles, the frequencies of A*02:01 and C*03:03 were significantly increased in patients with PBC, compared with healthy subjects (16% versus 11%, P = 0.0029, and 18% versus 13%, P = 0.012, respectively) (Table 2). In contrast, patients had significantly lower frequencies of A*02:06 (6% versus 9%; P = 0.038), A*33:03 (4% versus 8%; P = 0.0025), B*44:03 (2% versus 7%; P = 0.0011), C*08:01 (5% versus 10%; P = 0.005), C*14:03 (3% versus 7%; P = 0.0018), and C*15:02 (2% versus 4%; P = 0.03) alleles, compared with controls (Table 2). No other HLA A, B, or C alleles differed significantly between the groups.
|HLA||Frequency (%)||P Value||OR (95% CI)|
|Patients With PBC (2n = 458)||Healthy Subjects (2n = 1,032-1,046)|
Distribution of HLA-DRB1 and DQB1 Alleles.
Among DRB1 alleles, DRB1*04:05 and DRB1*08:03 were significantly associated with PBC, compared with healthy subjects (17% versus 13%, P = 0.044, and 13% versus 6%, P = 0.000025, respectively) (Table 2). Patients with PBC had a significantly lower frequency of DRB1*11:01 (1% versus 4%; P = 0.02) and DRB1*13:02 (3% versus 6%; P = 0.029) allele carriage, compared with controls (Table 2). Among DQB1 alleles, the DQB1*04:01 and DQB1*06:01 alleles were significantly associated with an increased risk of PBC (18% versus 13%, P = 0.02, and 23% versus 15%, P = 0.000091, respectively) (Table 2). Conversely, DQB1*03:01 (6% versus 12%; P = 0.00027), DQB1*06:02 (7% versus 12%; P = 0.019), and DQB1*06:04 (2% versus 5%; P = 0.0041) all conferred a reduced risk of PBC occurrence (Table 2). No other HLA DRB1 or DQB1 alleles were significantly associated with PBC, compared with healthy subjects. We also examined the influence of DRB1 and DQB1 allele homozygosity with PBC susceptibility and protection, but found no significant associations. However, the DRB1*08:03 and DQB1*06:01 alleles were significantly associated with PBC, compared to comparison cases with chronic hepatitis C (13% versus 5%, P = 0.0017, and 23% versus 16%, P = 0.02, respectively) (Supporting Table 1). Conversely, DQB1*03:01 and DQB1*06:04 had significantly lower frequencies in patients with PBC than in chronic hepatitis C controls (6% versus 12%, P = 0.0076, and 2% versus 5%, P = 0.041) (Supporting Table 1).
Distribution of Haplotypes Among PBC Patients and Controls.
The frequency of the DRB1*08:03-DQB1*06:01 haplotype in patients with PBC was 13% and significantly higher than the 6% observed in healthy subjects (P = 0.000025; OR = 2.22) (Table 3). However, there was no significant difference between the groups regarding the DRB1*15:02-DQB1*06:01 haplotype (10% versus 9%; P = 0.47). There was also a modest relationship between carriage of the DRB1*04:05-DQB1*04:01 haplotype and disease susceptibility (17% versus 13%; P = 0.044; OR = 1.38). In contrast, protective effects were seen for the DRB1*13:02-DQB1*06:04 haplotype (2% versus 5%; P = 0.00093; OR = 0.27) and DRB1*11:01-DQB1*03:01 haplotype (1% versus 4%; P = 0.03; OR = 0.37) in our cohort.
|Allele at Each Locus||Patients With PBC (%)||Healthy Subjects (%)||P value||OR (95% CI)|
|DRB1||DQB1||2n = 458||2n = 1,032|
|*08:03||*06:01||60 (13)||66 (6)||0.000025||2.22 (1.53-3.20)|
|*04:05||*04:01||79 (17)||136 (13)||0.044||1.38 (1.02-1.87)|
|*13:02||*06:04||7 (2)||56 (5)||0.00093||0.27 (0.12-0.60)|
|*11:01||*03:01||6 (1)||36 (4)||0.03||0.37 (0.15-0.88)|
|*15:02||*06:01||47 (10)||92 (9)||0.47|
|*09:01||*03:03||58 (13)||138 (13)||0.77|
Association Between HLA and Clinical Findings.
PBC patients were stratified according to history of orthotopic liver transplantation (OLT) and disease progression. The HLA-DRB1*09:01 and DQB1*03:03 alleles (33% versus 11%, P = 0.0012, and 33% versus 12%, P = 0.0022, respectively) and the DRB1*09:01-DQB1*03:03 haplotype (33% versus 11%; P = 0.0012; OR = 3.96; 95% CI: 1.75-8.95) were all significantly associated with OLT (Table 4). Homozygosity for the DRB1*09:01 and DQB1*03:03 alleles (43% versus 4%, P = 0.0012, and 43% versus 4%, P = 0.00076, respectively) and the DRB1*09:01-DQB1*03:03 haplotype (43% versus 4%; P = 0.0012; OR = 16.50; 95% CI: 2.10-129.63) was significantly correlated with OLT. When PBC patients with cirrhosis (n = 42) were compared to those without (n = 187), similar significant genetic associations of the DRB1*09:01 and DQB1*03:03 alleles (23% versus 10%, P = 0.0043, and 23% versus 11%, P = 0.0094, respectively) and the DRB1*09:01-DQB1*03:03 haplotype (23% versus 10%; P = 0.0043; OR = 2.51; 95% CI: 1.37-4.62) with disease progression were found (Table 4). Homozygosity for the DRB1*09:01 and DQB1*03:03 alleles (27% versus 3%, P = 0.007, and 27% versus 2%, P = 0.0049, respectively) and the DRB1*09:01-DQB1*03:03 haplotype (27% versus 3%; P = 0.007; OR = 13.45; 95% CI: 1.36-133.18) was significantly correlated with cirrhosis, as well. No other HLA class I or II alleles or haplotypes were significantly associated with disease progression.
|Allele at Each Locus||OLT (%)||Non-OLT (%)||P Value||Cirrhosis (%)||Noncirrhosis (%)||P Value|
|DRB1||DQB1||2n = 30||2n = 428||2n = 84||2n = 372|
|*09:01||10 (33)||48 (11)||0.0012||19 (23)||39 (10)||0.0043|
|*03:03||10 (33)||51 (12)||0.0022||19 (23)||42 (11)||0.0094|
|*09:01||*03:03||10 (33)||48 (11)||0.0012||19 (23)||39 (10)||0.0043|
|*08:03||*06:01||6 (20)||54 (13)||0.38||8 (10)||52 (14)||0.37|
|*04:05||*04:01||5 (17)||76 (18)||0.92||9 (11)||72 (19)||0.09|
Distribution of DRB1 Amino Acid Residues.
The amino acid sequence encoded by the second exon of HLA-DRB1 was determined for each subject. The prevalence of glycine at position 13 (P = 0.0013; OR = 1.60), tyrosine at positions 16 (P = 0.0013; OR = 1.60) and 47 (P = 0.00017; OR = 1.62), serine at position 57 (P = 0.0000015; OR = 1.83), and leucine at position 74 (P = 0.0000069; OR = 2.01) was significantly higher in patients with PBC, compared with healthy subjects (Table 5). In contrast, serine at position 13 (P = 0.000037; OR = 0.51), histidine at position 16 (P = 0.0029; OR = 0.66), and phenylalanine at position 47 (P = 0.000096; OR = 0.61) conferred protection against the disease.
|Residue||Amino Acid||PBC (%) 2n = 458||Healthy Subjects (%) 2n = 1,032||P Value||OR (95% CI)|
|13||Glycine||98 (21)||150 (15)||0.0013||1.60 (1.21-2.12)|
|Serine||55 (12)||214 (21)||0.000072||0.52 (0.38-0.72)|
|16||Tyrosine||98 (21)||150 (15)||0.0013||1.60 (1.21-2.12)|
|Histidine||346 (76)||850 (82)||0.0029||0.66 (0.51-0.86)|
|47||Tyrosine||344 (75)||672 (65)||0.00017||1.62 (0.26-2.07)|
|Phenylalanine||114 (25)||364 (35)||0.00017||0.62 (0.48-0.79)|
|57||Serine||157 (34)||224 (22)||0.0000004||1.88 (1.48-2.40)|
|74||Leucine||90 (20)||112 (11)||0.0000069||2.01 (1.48-2.72)|
Analysis of the amino acid residues encoded by DRB1*09:01 revealed six unique differences from those encoded by other DRB1 alleles: lysine at position 9, aspartic acid at position 11, tyrosine at position 26, histidine at position 28, glycine at position 30, and valine at position 78 (Table 6).
|Glutamic acid||Valine||Phenylalanine||Aspartic acid||Tyrosine|
The present study examined HLA class I and II alleles and haplotypes and amino acid residues in patients with PBC in the Japanese population. Our key findings were as follows: (1) The HLA DRB1*08:03-DQB1*06:01 haplotype was significantly associated with disease pathogenesis, which was in agreement with several Japanese studies linking DRB1*08:03 with PBC; (2) Japanese PBC patients had significantly lower frequencies of HLA DRB1*13:02-DQB1*06:04 and DRB1*11:01-DQB1*03:01 haplotypes, suggesting protection by these haplotypes to the disease, as indicated by recent reports in Europe; (3) the existence of a relationship between HLA haplotype and OLT and disease progression; and (4) PBC-associated alleles have specific antigen presentation profiles.
The HLA- DRB1*08:03 (P = 0.000025) and DQB1*06:01 (P = 0.000091) alleles were strongly associated with PBC susceptibility. Although a relationship between DRB1*08:03 and PBC has already been reported in the Japanese population, an association with the DQB1*06:01 allele has not been investigated in a large cohort like ours. DQB1:06:01 is known to be in linkage disequilibrium with DRB1*08:03 or DRB1*15:02 in the Japanese population. Our data clearly show that the DRB1*08:03-DQB1*06:01 haplotype was significantly associated with PBC (P = 0.000025), but the DRB1*15:02-DQB1*06:01 haplotype was not. This suggests that the DRB1*08:03 allele and/or the DRB1*08:03-DQB1*06:01 haplotype might play a crucial role in PBC development in Japan. However, because DRB1*08:03 was found in only 13% of PBC patients in this study, other candidate genes and environmental factors require further study. The DRB1*04:05-DQB1*04:01 haplotype was also found to be weakly associated with susceptibility to PBC. Because our previous reports showed that this haplotype was strongly associated with autoimmune hepatitis and autoimmune pancreatitis in the Japanese population,38, 39 deeper evaluation of DRB1*04:05-DQB1*04:01 with regard to autoimmune diseases and PBC may uncover key relationships of clinical value. Recently, genome-wide association studies showed that HLA and other non-HLA genes were associated with susceptibility to PBC in Europe and North America.27–30 Accordingly, similar studies are now being performed to clarify the genes responsible for PBC in Japan.
This study shows, for the first time, that the DRB1*13:02-DQB1*06:04 and DRB1*11:01-DQB1*03:01 haplotypes played protective roles against PBC in the Japanese population. Our data support the recent consensus that DRB1*11 and *13 confer resistance in Europe and Japan,20, 21, 26 although we cannot exclude the possibility that these associations are only linkage markers for a yet undefined gene for PBC. Multiple lines of evidence show that DRB1*11 and DRB1*13 alleles are also protective against several infectious diseases. Because bacterial infections have been reported as possible causes of PBC,40, 41 HLA alleles or haplotypes that are resistant to such agents might influence protection against PBC development.15
Interestingly, this study revealed a novel association between the DRB1*09:01-DQB1*03:03 haplotype and PBC progression. Although Nakamura et al.26 reported that DRB1*09:01 was associated with disease progression of non-jaundice-type PBC, there have been no reports of a connection between HLA haplotypes and OLT or cirrhosis in Japan. Several studies from the United Kingdom and Sweden19, 42 have reported that DRB1*08:01 is associated with both susceptibility and progression to the disease, but a study from Italy could not confirm this.21 Homozygosity of the DRB1*09:01-DQB1*03:03 haplotype was also associated with disease progression in our cohort. The reasons for this observation are unknown; however, the association of this particular HLA haplotype and disease progression is striking. Because only 15 (7%) and 42 (18%) of our 229 patients had OLT and cirrhosis, respectively, further longitudinal follow-up studies in larger cohorts from different ethnicities are required. A recent study uncovered that anti-gp210 and anticentromere antibodies may be risk factors for the progression of PBC.43 It would be of interest to assess associations between these autoantibodies and HLA haplotypes in the future.
Last, the present study determined and analyzed the amino acid sequence encoded by the DRB1 allele in relation to disease susceptibility. The incidence of glycine-13, tyrosine-16, and leucine-74 encoded by DRB1*08:03 was higher and that of serine-13, histidine-16, and phenylalanine-47 encoded by DRB1*11 and DRB1*13 was lower in PBC patients. These data are consistent with a report by Donaldson et al.20 Serine-57 had the highest frequency among patients in our cohort (P = 0.0000004), likely because it is encoded by DRB1*04:05 and DRB1*08:03, which are both significantly associated with PBC susceptibility in the Japanese population. Serine-57 relevance was not found in a European study,20 probably because frequencies of the DRB1*04 and DRB1*08 alleles therein were found in 10% and 7%, respectively, of patients.21 The amino acid residue at position 57 influences the binding of antigen side chains associated with the 9th pocket of the expressed DR molecule, which might factor predominantly in susceptibility to PBC in Japanese cases. Interestingly, amino acid residues lysine-9, aspartic acid-11, tyrosine-26, histidine-28, glycine-30, and valine-78 were encoded by DRB1*09:01 only, suggesting that some or all of these may contribute to disease progression in Japanese patients.
In conclusion, the DRB1*08:03-DQB1*06:01, DRB1*13:02-DQB1*06:04, and DRB1*11:01-DQB1* 03:01 haplotypes are associated with either PBC susceptibility or protection in the Japanese population. DRB1*09:01-DQB1*03:03 is a novel haplotype associated with the progression of PBC that has several uniquely expressed amino acids. Other specific amino acid residues in the DRβchain appear to contribute to susceptibility or resistance to PBC. Genome-wide analysis and resequencing of the entire HLA region will be necessary to provide more precise genetic information on susceptibility to PBC in Japan.
The authors thank Yuki Akahane and Asami Yamazaki for their technical assistance, and Trevor Ralph for his editorial assistance.