HLA IN PRIMARY BILIARY CIRRHOSIS: AN OLD STORY NOW REVIVING

Abstract

Primary biliary cirrhosis (PBC) is an autoimmune biliary disease characterized by injury of small and medium size bile ducts eventually leading to liver cirrhosis and death. While the causes remain enigmatic, recent evidence has strengthened the importance of genetic factors in determining the susceptibility to the disease. Besides the strong heritability suggested by familial occurrence and monozygotic twins concordance, for decades there has not been a clear association with specific genes, with the only exception of a low risk conferred by a class II human leukocyte antigen (HLA) variant, the DRB1*08 allele, at least in some populations. Only recently the story began to change when a strong protective associations between PBC and the HLA DRB1*11 and DRB1*13 alleles were found in Italian and UK series. But HLA genes fully returned to attract interest thanks to recent genome-wide association studies (GWAS) which clearly demonstrated that the major component of the genetic architecture of PBC are within the HLA region. As expected in a genetically complex disease, GWAS also identified several novel non-HLA variants, but it is to note that all of them are in immuno-related genes. In this review, the paradigmatic tale of what, and how, we learned about HLA genes in PBC will be retraced with particular focus on how GWAS are enabling us to rewrite the story of PBC pathogenesis. These recent discoveries will not only driving functional studies but will also held the promise of developing novel disease-specific treatments.

Primary biliary cirrhosis (PBC) is a chronic cholestatic liver disease characterized by progressive destruction of small and medium size intrahepatic bile ducts leading to cirrhosis and ultimately liver transplantation or death (1). PBC has an estimated prevalence of 1 in 1,000 women over the age of 40, and ursodeoxycholic acid is the only approved therapy (2). The pathogenesis of PBC is clearly autoimmune (3), as indicated by specific serum and cell-mediated responses against defined epitopes of self antigens, and by a striking female predominance (female to male ratio of about 10 to 1). In addition, epidemiological data indicate that family members of patients have an increased risk of developing PBC or another autoimmune disorder. Based on these considerations, the current hypothesis on the etiopathogenesis of PBC implies that this disease is the result of a genetic predisposition that is permissive for a still unknown environmental agent, possibly xenobiotic or infection (1).

For decades, PBC has been considered to have a unique genetic background when compared to other autoimmune diseases because of the strong familial clustering but weak associations with genetic polymorphisms (4). Indeed, despite numerous candidate-gene association studies have been performed no conclusive data on specific genes have been obtained. In addition, it is to note that linkage analysis was poorly feasible in PBC based on the advanced age at diagnosis and the rarity of the disease. In contrast, recent evidences have strengthened the importance of the genetic susceptibility in determining disease onset and severity, including a role for sex chromosome abnormalities in affected women (5, 6) and high concordance for disease in monozygotic twins (7).

The human leukocyte antigen (HLA) loci, located in the major histocompatibility complex, are the most genetically diverse loci in the human genome (8) (Figure 1). HLA genes encode cell-surface molecules that by means of peptides presentation, mediate key immunological events, such as definition of self-tolerance or cellular immune responses to tumours and pathogens (9, 10). Similar to other genetically complex diseases (11), HLA has been extensively studied in PBC, but for decades data have cumulatively suggested only a weak association with the class II HLA DRB1*08 allele (4). This was likely because early studies manifested several potential limitations. First, insufficient statistical power due to inadequate sample sizes. Second, lack of careful matching between cases and controls. Third, earlier studies did not rely on molecular analysis. Finally, multiple replications have rarely been carried out.

Figure 1. Human leukocyte antigen (HLA) complex on human chromosome 6.

The small region (6p21.31) in short arm of chromosome 6 is conventionally divided into three regions (Class I, II and III) and contains many loci that are involved in inflammatory responses. Some of them are shown. Abbreviations: MICA major-histocompatibility complex class I chain genes A; MICB major-histocompatibility complex class I chain genes B; LTA lymphotoxins A; TNF-a tumor necrosis factor a; LTB lymphotoxins B; HSP heat-shock protein; C2 complement component 2; BF complement factor B; C4A and C4B complement components 4A and 4B, respectively; TAP transporter associated with antigen processing; LMP large multifunctional protease; Tapasin TAP-binding protein.

The story began to change when, to overcome all of these flows, our group evaluated the HLA polymorphisms in the largest PBC series ever reported (12, 13), and provided evidence not only that PBC susceptibility is associated with the HLA DRB1*08 allele but also that DRB1*11 and DRB1*13 confer strong protection from the disease. Importantly, few months later, the same protective HLA alleles were confirmed to be associated with PBC in a large-scale case-control study from UK (14). Since both these protective alleles are known to influence the penetrance of infectious agents, they have implications in light of the proposed infectious theory in PBC origin. However, the reviving interest for HLA genes in PBC arising from these studies was shortly overcome by three recent genome-wide association studies (GWAS) in PBC which showed that the strongest associations are located at the HLA region (15–17).

This review does not aim at summarizing our knowledge of genetics of PBC, but it will mainly be focused on the old and recent associations with HLA variants obtained with candidate-gene large-scale studies and GWAS, and on how these data may change the genetic landscape of PBC.

HIGHLIGHTS OF GENETIC PREDISPOSITION IN PBC

In the past, a number of reports have reported an increased risk of developing PBC within family members of affected individuals, a scenario called ‘familial PBC’ (18). The main part of these studies as well as population-based epidemiological reports were performed in Great Britain (19–22). In this geographical area, the former reported rates of PBC prevalence within family members were about 1–2.4% (19, 20). Prevalence rates of familial PBC was later reported to be 6.4% in UK (22), and between 3.8% and 9.0% in a number of studies from North America, Europe, and Japan. A further estimate of the familial prevalence of PBC, the sibling relative risk, was found to be 10.5 in a UK study (22). In addition, a recent large US study indicated that having a first-degree relative with PBC was significantly associated with increased risk of disease, with an odds ratio at 10.7 (23). Of course, shared environmental factors by family members may well explain these findings, as suggested by data on prevalence and incidence of PBC.

A role for genetics is also suggested by the frequent coexistence with other autoimmune diseases in more than one third of patients with PBC (23). Diseases that may coexist in PBC patients or family members include rheumatoid arthritis, Sjogren syndrome, and autoimmune thyroid disease (23). The current explosion of new information about these autoimmune diseases is allowing the identification of a number of common genes underlying multiple diseases (11).

Disease concordance rate in monozygotic twin pairs (the proportion of affected pairs concordant for the disease) is another powerful tool to estimate the impact of genetic factors in susceptibility to complex disorders, including autoimmune disorders (11). In the past, PBC concordance rate has been limited to two reports (24, 25), one concordant and one discordant pair of twins, but monozygosity was not genetically proven. Thanks to a worldwide effort, we were able to identify 8 monozygotic and 8 dizygotic twin pairs in which at least one subject was affected by PBC and to find a concordance rate of 63%, the highest among autoimmune diseases (7). In the general attempt to dissect the effects of different exposure to environmental factors, we also explored classical epigenetic factors in sets of PBC twins, but excluded their major role in PBC development (26).

A role for genetics in PBC is also suggested by animal models of human PBC (27). Most of them are indeed spontaneous murine models due to a number of different genetic changes. The genetically-determined models of PBC include the IL-2 receptor alpha deleted (IL-2Ralpha^−/−), transforming growth factor beta receptor II dominant-negative (dnTGF-betaRII), scurfy, nonobese diabetic (NOD) c3c4, and AE2 gene-disrupted (AE2a,b^−/−) mice. For one of these, the IL-2 receptor alpha deleted, there has been a PBC-like disease reported in a child with IL-2 receptor alpha (CD25) deficiency (28).

The literature on PBC contains many publications that have attempted to identify genes with a role in disease susceptibility and progression by evaluating small numbers of variants in one or few specific candidate genes by means of case control study designs. Of course, most of these genes code for immune-related molecules and were already implicated in other autoimmune disorders, including tumor necrosis factor (TNF), cytotoxic T lymphocyte antigen-4 (CTLA-4), toll-like receptors, caspase 8, vitamin D receptor, interleukins (ILs) 1, 2, and 10 and numerous cytokine and chemokine receptors. However, such approaches have led to very few insights into the genetic basis of PBC, mainly for lack of robust replication. A paradigmatic example is that related to CTLA-4 gene association studies. While two earlier studies from the UK (29) and China (30) found a SNP associated with PBC, more recent data from Brazil (31), Italy (32), Germany (33), UK (34), and US (35) failed to confirm it. In addition, the follow-up study by the UK group (34) failed to replicate their original positive finding (29), while the follow-up study by the US group (36) found a novel SNP association in contrast with their original negative finding (35). Accordingly, caution is suggested when interpreting these findings. In the future, candidate gene studies, of course with appropriate size and replication, should focus on dissecting interaction between risk loci, on investigating variants frequencies in different geographical areas, and on risk loci influencing outcomes, treatment response, and symptoms. To this regard, laudable examples are the recent study by Lazaridis and coworkers which showed that a TNF gene variant amplifies a CTLA4 genotype risk for PBC by examining the possible interaction between these two risk variants (37). The same authors also suggested an interaction between CTLA-4 and programmed cell-death 1 gene variants (35). Additionally, because the genetic background may greatly varies among different ethnic groups, our group evaluated the association of other candidate genes in PBC by comparing multiple cohorts from different ethnic populations (38, 39). Finally, a recent French study has reported that the progression rate of liver disease under ursodeoxycholic acid therapy was significantly linked to variants of TNF and AE2 genes (40).

Although over the past 50 years clinical awareness of PBC has greatly increased, laboratory diagnosis far more precise, and therapy more effective, the main reason for the female preponderance of PBC have remained unclear (41). We have recently proposed that the presence of defects in sex chromosomes might explain both the female’s disproportionate affliction with PBC and the genetic predisposition to the disease Indeed we first reported an age-dependent enhanced monosomy X in the peripheral white blood cells of women with PBC (5), later that one X chromosome is preferentially lost (6), and finally that epigenetic factors influencing PBC onset are more complex than methylation differences at X-linked promoters (26).

HLA ASSOCIATIONS AND PBC

The HLA is one of the most widely studied regions in human genome and certainly contains valuable genetic information of many complex genetic diseases that have yet to be fully dissected (42–47). HLA genes are located on the short arm of chromosome 6 (6p21.31) with about 3.6 mega base pairs extension, and consists of three subregions: the two telomeric class I, class III, and the centromeric class II regions (Figure 1). The true role of the various HLA alleles in inducing autoimmune reactions remains largely unclear, and many might be the underlying mechanisms (48). Among them, it has been suggested that certain HLA alleles are less efficient at presenting self peptides to developing T cells in the thymus, with failing of the negative selection. In particular, it is possible that certain HLA molecules present peptide at an “intermediate level” thus being recognized by T cells without inducing tolerance. Indeed, most self peptides are presented at levels below that which is needed to engage effector T cells, while others induce clonal deletion and anergy as presented at high levels. Alternatively, it is possible that specific HLA alleles enhance the autoimmune activation by enhancing immunogenicity and influencing the expressed repertoire of T cells.

As a matter of fact, HLA region contains many loci that are largely involved in inflammatory responses, such as major histocompatibility complex class I chain genes A and B, tumor necrosis factor a, heat-shock proteins, complement component 2 and transporter associated with antigen processing (Figure 1). Variants of HLA genes have been found to be associated with almost every known complex genetic diseases. However, it has been difficult to identify genetic variants within HLA directly linked to the cause of diseases; The main reasons for these difficulties are listed and discussed in the next paragraphs.

The old evidences – Little interest for HLA

In the past, a number of studies has evaluated the association of HLA class I variants with PBC susceptibility (49–55), but no significant results were found (Table 1). Several reasons could explain the lack of associations. First, the small number of patients evaluated in each study (ranging between 21 and 75), which may account for an inadequate statistical power for comparisons. Second, it has to be remembered that in the past only limited members of HLA class I alleles could have been assessed due to the technical methods available at that time, with the risk to underestimate the existing associations. Finally, linkage disequilibrium may well explain why HLA class I gene associations with PBC, as well as with many other autoimmune diseases, are in general not striking (4, 56). Because of these major flaws, few years ago our group examined the association with HLA class I variants in a large Italian cohort of PBC and controls and reported that PBC is associated with various HLA-B alleles (57) (Table 1). However, being present only in a small proportion of our patients, these associations should be regarded as weak. In the future, HLA class I variants still need to be replicated in different ethnic groups, of course with adequate sample size and study design. Indeed, it could be assumed that similar to the epidemiological data, the genetic background in PBC could be associated with a geographical pattern. It is interesting to note that we are witnessing a resurgence of interest on these gene variants because of their critical function as ligands for killer immunoglobulin-like receptors on natural killer cells and various T lymphocytes (58).

Table 1.

Synopsis of HLA association studies in PBC

Country	Year	HLA investigated	Significant HLA associations	Prevalence in PBC (n)	Prevalence in Controls (n)	P (Corrected)	Ref.
Spain	1979	A, B, C, DRw	DRw3	57.1% (12/21)	14.8% (11/74)	<0.004	(49)
Japan	1983	A, B, DR	DR2	68% (15/22)	30% (15/50)	<0.042	(50)
UK	1985	A, B, DR	No associations	-	-	-	(52)
UK	1987	A, B, DR, C4A, C4B, Bf, C3	C4B2	45% (15/33)	17% (53/307)	0.014	(53)
US	1987	DR, DQ	DR8	30.1% (35/114)	4.7% (8/171)	<0.0001	(59)
			DR5 decreased in PBC	9.6%	25.2%	0.0118
US	1987	A, B, DR(1–7)	No associations (DR8 not tested)	-	-	-	(80)
US	1990	A, B, C, DR7, DRw8, DRw17, DQw2, DQw3	DRw8	18.4% (6/35)	4.7% (73/1546)	0.02	(81)
			DQw3 decreased in PBC	26.3%	53.5%	<0.001
Germany	1991	A, B, C, DR, C4A, C4B, Bf	DRw8	36% (9/25)	3.6% (6/169)	0.00013	(54)
			C4A-Q0	72%	34.5% (51/148)	0.0056
UK	1992	DRB, DQB	DR8	11% (18/159)	4% (6/162)	<0.01	(82)
			DR8/DQB1*0402	11% (10/89)	2.2% (4/181)	<0.001
Denmark	1992	A, B, C, DRB, DQA, DQB, DPA, DPB	DR3	52.2% (12/23)	24.6% (296/1204)	<0.05	(55)
Japan	1993	DR, DQ, DPB1	DQ3	80.9% (38/47)	51.3% (77/150)	<0.05	(66)
			DPB1*0501	85.1%	55.3%	<0.01
			DPB1*0402 decreased in PBC	2.2%	23.3%	<0.05
			DR52 (DRB3) decreased in PBC	34%	54%	<0.05
UK	1993	DRB1, DRB3, DQA, DQB	DR8	18.5% (24/130)	9.1% (33/363)	<0.005**	(83)
Japan	1994	A, B, C	DRB1*0803	35.5% (22/62)	7.4% (32/430)	<0.0001	(51)
		DRB1, B3, B5	DQA1*0103	43.5% (27/62)	18.4% (79/430)	<0.0001
		DQA1, B1	DQB1*0601	43.5% (27/62)	17.9% (77/430)	<0.0001
			DQA1*0102 decreased in PBC	3.2% (2/62)	16.3% (70/430)	0.0288
UK	1994	DRB1, DQB1, DPB1	No associations	.	.	.	(63)
US	1994	DRB1, DQA1, DQB1	DRB1*0801	7.8% (8/102)	1.9% (9/480)	<0.05	(60)
			DRB1*0901	3.9%	0.8%	<0.05
			DQA1*0401/0601	9.8%	2.7%	<0.001
			DQA1*0102 decreased in PBC	5.9%	18.8%	<0.001
			DQB1*0602 decreased in PBC	2.9%	12.1%	<0.025
Germany	1995	DPB1	DPB1*0301	50% (16/32)	12.8% (6/47)	<0.015	(64)
UK	1995	DPB1	No associations	.	.	.	(84)
UK	2001	DRB, DQA, DQB	DRB1*0801	15% (25/164)	2.9% (3/102)	0.0014	(65)
			DRB1*0801/DQA1*0401/DQB 1*0402 (late stage)	23% (21/88)	2.9%	4.4 E –6
Sweden	2002	DRB1, DQB1, DPB1	DRB1*08	29.3% (29/99)	11.4% (18/158)	0.001	(61)
			DQB1*0402	28.6% (28/98)	10.8%	0.001
Italy	2003	A, B, DRB1	B*15	8% (9/112)	3.4% (19/558)	0.0039	(57)
			B*41	3.1%	0.3%	8.8 E –15
			B*55	3.6%	1.3%	0.034
			B*58	1.8%	0.3%	0.00042
			DRB1*11 decreased in PBC	10.7% (16/149)	27.6%	1.6 E -6
Brazil	2003	DR, DQ	No associations	.	.	.	(31)
US	2005	DRB1, DQA1, DQB1	DRB1*08	19.4% (14/72)	8.7% (33/381)	0.011	(62)
			DRB1*1501	13.9%	26.6%	0.022
			DQA1*0102	16.7%	37.9%	0.0014
			DQB1*0602	12.7%	27.6%	0.012
UK	2006	DRB1, DQA1, DQB1	DRB1*08	12.0% (50/412)	4.0% (10/236)	0.0009	(14)
			DRB1*13 decreased in PBC	14.0%	20.0%	0.042
			DQA1*0401	14.0%	4.2%	0.0006
			DQB1*0301	26.0%	34.0%	0.045
			DQB1*0402	11.0%	4.0%	0.002
Italy	2008	DRB1	DRB1*02	12.0% (80/664)	12.0% (239/1992)	0.041	(13)
			DRB1*08	7.2%	2.3%	4.8 E -31
			DRB1*11 decreased in PBC	13.6%	30.0%	2.3 E -23
			DRB1*13 decreased in PBC	8.6%	11.2%	0.00004
Japan	2010	DRB1	DRB1*0405	18.9% (63/334)	13.2% (34/258)	0.005	(67)
			DRB1*0803	13.3%	6.4%	0.0001
			DQA1*1101	1.0%	3.7%	0.002
			DQB1*1302	2.2%	5.6%	0.003
			DQA1*1406	0.7%	2.1%	0.045
			DQB1*1501	6.9%	11.6%	0.005

Many studies have reported associations of HLA class II alleles and PBC in populations of Caucasian and Asian ethnicity (Table 1). The association with HLA DRB1*08 allele has been found most frequently among reported studies from Germany, US, Spain, and Sweden, thus indicating that this allele might constitute a risk factor for PBC among Caucasians (54, 59–62). However, it is to note that several European studies have failed to confirm an association with DRB1*08 (31, 52, 55, 57, 63). Other than DRB1*08 variant, it has been reported associations with DR3 (49, 55) or DPB1*0301 (64). In 2003, we suggested that the DRB1*11 allele has a protective effect against PBC in the Italian population (57). An UK study reported that the linkage of the DQA1*0401 allele and the DR8-DQB1*0402 haplotype is associated with disease progression but not initial susceptibility (65), while a more recent US study demonstrated an association between DRB1*08-DQA1*0401-DQB1*04 haplotype and PBC, albeit in a minority of patients (62). Among populations of Asian ethnicity, studies from Japan failed to find a consistent picture of HLA class II associations with PBC (50, 51, 66), with an association between PBC and DR2 in one (50), DPB1*0501 in another (66), and DRB1*0803 in a third (51). However, although the lack of consistent associations between specific DRB1 alleles and PBC in Japan, a recent study suggested that different HLA variants may relate to clinical features of disease. Indeed, Nakamura and colleagues reported a strong association of an HLA-DRB1*0405 and *0803 with disease only in the subset of patients positive for anti-sp100 (odds ratio = 1.61), a well-known PBC-specific serum anti-nuclear autoantibody, and anti-centromere antibodies (odds ratio = 2.30) (67). Interestingly, Hirschfield et al found similar data in Caucasian populations (68). Also because of the potential clinical implication, future association studies should address the link between different HLA variants and immunological phenotypes in PBC. Overall, we can conclude that the picture of HLA class II involvement in PBC was quite complex and uninteresting until recently.

More recent findings – Renewed interest for HLA

Based on the above data, it is clear why until recently HLA variants did not arouse great interest of basic and clinical researchers working to characterize the molecular mechanisms that contribute to disease development, and more specifically, for understanding the role of genetics in PBC. The story began to change when our group showed that beyond the consistent (but weak) positive association with HLA DRB1*08 allele, PBC was also strongly associated with two protective HLA variants, DRB1*11 and DRB1*13 (first reported in abstract form in 2005 (12)) (13). In particular, by typing for HLA class II polymorphisms a large cohort of 664 Italian patients with PBC and 1992 controls, we confirmed the known positive association with DRB1*08 (odds ratio = 3.3), whereas we reported for the first time the protective alleles DRB1*11 (odds ratio = 0.4), and DRB1*13 (odds ratio = 0.7); A weak association with HLA DRB1*02 was also found, and only the associations with DRB1*08 and DRB1*11 were common to all geographical areas, (Northern, Central, and Southern Italy) (13). These results were later confirmed in a large UK set of patients and controls in which protection against PBC was associated with DRB1*13 (odds ratio =0.65) along with a positive association with the class II MHC allele DRB1*0801 (odds ratio = 3.05) (14). The finding is of great interest since the two HLA variants found to be protective for PBC suggest possible disease mechanisms as having a protective role for multiple infectious diseases. Indeed, these studies suggest that HLA-DRB1*11 allele exerts a strong protective role against hepatitis C virus (69), human papilloma viruses (70), and human immunodeficiency virus (71). Similarly, HLA-DRB1*13 is protective for hepatitis B virus (72), hepatitis C virus (73), human papilloma viruses (74), human immunodeficiency virus (75), and malaria (74). Overall, these data indicate that these HLA class II alleles may influence the maintenance of immune tolerance as well as the penetrance of infectious agents, thus having implications in light of the proposed infectious theory in PBC etiology (1). In accordance, because the protective HLA alleles are associated with resistance to several infections, it can be hypothesized that the lack of such alleles might contribute to the molecular mimicry of infectious agents leading to immune tolerance breakdown in PBC (1).

Hints from GWAS – HLA turns out as the first association with PBC

The field of human genetics has rapidly changed since the recent completion of the human genome sequence and novel challenging theories has been proposed. Overall, thanks to dramatic advances in molecular technology linked to the field of genetics (76), we are now witnessing an explosion of new information about the allelic architecture of human complex diseases, such as PBC (77). In particular, the ability to evaluate the entire human genome for common polymorphisms (i.e. those present in more than 5% of the general population) has allowed us to disclose more than 80 disease-susceptibility loci. The National Cancer Institute (NCI)-National Human Genome Research Institute (NHGRI)’s catalog reports an updated list of published GWAS (http://www.genome.gov/26525384). It is of great interest that the recent GWAS approaches have allowed to identify an extended major histocompatibility complex, spanning about 7.6 Mb of the human genome (78). Indeed, many additional loci (most with putative immuno-regulatory role) were identified outside the well-known HLA class I, II, and III regions (78). A growing number of studies are providing evidence of genetic complexity within the MHC region in a number of disorders.

In PBC, the first GWAS was recently performed in cases from Canada and the US (15) and reported significant associations with HLA, as well as with other non-HLA loci including IL12A, and IL12RB2 polymorphisms. This first study manifested a sufficient statistical power by including 536 patients with PBC and 1536 controls typed for about 300,000 common variants, but more solid data were soon after provided by combining datasets from the Canadian-US GWAS with a separate Italian GWAS (16) (Figure 2 and Table 2). Over 610,000 common variants were examined in 457 Italian PBC cases and more than 1 million in 947 controls. When considered alone, the Italian cohort association dataset achieves genome-wide significance at the HLA locus with several other loci showing suggestive association signals (Figure 2a and 2b). Analysis of the combined dataset (998 cases and 8777 controls) showed many more loci to have reached the conservative genome-wide threshold p value (P < 5 × 10⁻⁸), most of these also showing p values less than 5 ×10⁻⁵ in the Italian-alone cohort. In particular, this meta-analysis allowed to confirm the finding that HLA regions had the strongest statistical association with PBC. At the HLA region, the variants showing the strongest associations with PBC were similar between the two datasets, with almost complete overlap of the strongest association observed between the DQB1 and DQA2 loci (Figure 2b). Importantly, the association with HLA region were also confirmed and strengthened by a third GWAS, recently conducted in a very large cohort of 1,840 UK PBC cases and 5,163 population controls (17) (Table 2).

Figure 2. Results of genome-wide association tests for PBC as reported in the Nature Genetics (16).

The ordinate shows the level of significance for each SNP along each chromosome (a) or for the HLA region on chromosome 6 (b). The Italian PBC subset (diamond symbols) and combined European dataset (circle symbols) are shown. The dashed line corresponds to P = 5 × 10⁻⁸. For HLA, the strongest association was for rs7774434 for both the Italian alone (P = 2.05 × 10⁻¹¹, odds aratio = 1.74) and for the combined data set (P = 1.31 × 10⁻²⁷, odds ratio = 1.71). Abbreviations: HLA-DQB1 denotes the gene encoding HLA class II DQ β chain 1, IL12A the gene encoding interleukin-12α, IL12RB2 the gene encoding interleukin-12 receptor β2, IRF5, the gene encoding interferon regulatory factor 5, SPIB, the gene encoding the SPi-B transcription factor, IKZF3, the gene encoding the IKAROS family zinc finger 3 and ORMLD3 encoding ORM1 like 2. Reproduced with permission of the Nature Genetics; © 2010. All rights reserved.

Table 2.

List of genes associated with PBC

Gene Loci	Previous studies	Canada/US GWAS (Ref. 15)	Italy-Canada/US GWAS (meta-analysis study) (Ref. 16)	UK GWAS(Ref. 17)
HLA	Yes	Yes	Yes	Yes
IL12A	-	Yes	Yes	Yes
IL12RB2	-	Yes	Yes	Yes
IRF5/TNPO3	-	Yes	Yes	Yes
ORMDL3/IKZF3	-	Yes	-	Yes
MMEL1	-	Yes	-	Yes
SPIB	-	Yes	Yes	Yes
DENND1B	-	-	Yes	Yes
CTLA-4		Yes	-	-
STAT4	-	Yes	-	Yes
CD80	-	-	-	Yes
NFKB1	-	-	-	Yes
IL7R	-	-	-	Yes
CXCR5	-	-	-	Yes
TNFRSF1A	-	-	-	Yes
RAD51L1	-	-	-	Yes
CLEC16A	-	-	-	Yes
MAP3K7IPI	-	-	-	Yes
PLCL2	-	-	-	Yes
RPS6KA4	-	-	-	Yes
TNFAIP2	-	-	-	Yes

Taken together, these three GWAS identified a number of non-HLA loci, with plausible candidate genes that indicate the involvement of the innate and adaptive immune systems in the etiopathogenesis of PBC (Table 2). In particular, these findings support the role for the TLR, TNF, and NF-kB pathways. Among the associations consistently reported are to note those with the IL12A and IL12RB2 loci, the gene encoding the SPi-B transcription factor (SPIB) as well as two other loci, the gene encoding interferon regulatory factor 5 (IRF5) and the gene encoding the IKAROS family zinc finger 3 (IKZF3) and that encoding ORM1 like 2 (ORMDL3) also implicated in risk for other autoimmune diseases such as lupus and asthma, respectively. Suggestive associations were also observed between PBC and two other loci associated with other autoimmune conditions, the signal transducer and activator of transcription 4 (STAT4) and DENND1B. Finally, the most recent UK GWAS identified novel associations between PBC and loci, such as CD80, NFKB1, IL7R, CXCR5, and TNFAIP2 (17) (Table 2). These studies clearly identified PBC association with several novel non-HLA loci and added evidence of overlaps in the risk loci predisposing to PBC and other autoimmune diseases. However, all these novel genetic data allow us to make some observations. First, the greater the number of studies and of subjects (cases and controls) studied, the greatest is the number of common genetic variants associated with PBC. This suggests caution in some way and the reasonable need to re-direct our future research studies for example to rare variants or to copy number variants or to gene expression. The second observation is the strong consistency among the findings of these three GWAS, thus suggesting the presence of a common genetic pattern for PBC. This finding is of course very positive but, again, caution is needed because the first three GAWS have been performed in populations of European ancestry, and it will be important to replicate the reported associations also in non European populations. Indeed, a recent study from Japan failed to confirm some GWAS associated variants (79).

CONCLUSIONS AND FUTURE DEVELOPMENTS

It is currently believed that the development of PBC requires that an environmental factor, particularly an infection, initiates an autoimmune reaction in a genetically predisposed individual. However, although strongly implicated by family and twin studies, no specific genetic factors involved in susceptibility to PBC were identified. The story began to change first with the demonstration of associated protective HLA variants by means of large-scale candidate-gene association studies. But the major role of the HLA region in the genetic architecture of PBC susceptibility became definitively clear thanks to the first GWAS in PBC. Based on these data, it will be challenging to perform a deep high-throughput analysis of this genetic region, although these effort has to consider the extensive linkage disequilibrium and variability across the HLA region makes difficult a further resolution of these associations. Moreover, being the HLA molecules tightly linked with the maintaining (or breaking) of the immune system homeostasis, functional studies on new candidate HLA variants have to be carried out with the final goal of understanding the PBC etiopathogenesis and developing novel disease-specific therapies.

Abbreviations

PBC

primary biliary cirrhosis

HLA

human leukocyte antigen

GWAS

genome-wide association studies

IL

interleukin

TNF

tumor necrosis factor

CTLA-4

cytotoxic T-lymphocyte antigen-4

SPIB

SPi-B transcription factor

IRF5

interferon regulatory factor 5

IKZF3

IKAROS family zinc finger 3

ORMDL3

ORM1 like 2

STAT4

Signal transducer and activator of transcription 4