Pathogen infections and primary biliary cholangitis

Summary

Primary biliary cholangitis (PBC) is a multi‐factorial disease caused by the interaction of both genetic predisposition and environmental triggers. Bacterial infection has been investigated most intensively, both epidemiologically and experimentally, as a prime environmental aetiology in PBC. The association of recurrent history of urinary tract infection (UTI) with PBC has been frequently confirmed by several large‐scale, case–control studies, despite variation in geographic area or case‐finding methods. Escherichia coli is a predominant pathogen in most cases with UTI. Animal studies and molecular mimicry analysis between the human and E. coli E2 subunit of the 2‐oxo‐acid dehydrogenase complexes demonstrated that E. coli infection is a key factor in breaking immunological tolerance against the mitochondria, resulting in the production of anti‐mitochondrial autoantibodies (AMA), the disease‐specific autoantibodies of PBC. Novosphingobium aromaticivorans, a ubiquitous xenobiotic‐metabolizing bacterium, is another candidate which may be involved in the aetiology of PBC. Meanwhile, improved environmental hygiene and increased prevalence of PBC, especially in males, may argue against the aetiological role of bacterial infection in PBC. Multiple mechanisms can result in the loss of tolerance to mitochondrial autoantigens in PBC; nonetheless, bacterial infection is probably one of the dominant pathways, especially in female patients. Notably, there is a rising prevalence of male patients with PBC. With increasing exposure to environmental xenobiotics in both genders, studies directed towards identifying the environmental culprit with systematically designed case–control studies are much needed to further determine the environmental factors and role of bacterial infections in PBC.

Introduction

Primary biliary cholangitis (PBC) is a chronic cholestatic liver disease that predominantly affects adult women, most commonly beginning in middle age. It is characterized as a chronic non‐suppurative destructive cholangitis with granuloma formation in the liver, degeneration of biliary epithelial cells (BECs) and disappearance of small‐ or medium‐sized intrahepatic bile ducts, leading to chronic and progressive cholestasis 1. Extensive evidence such as dense infiltration of mononuclear cells into the intrahepatic small bile ducts, the presence of high‐titre disease‐specific autoantibodies, i.e. anti‐mitochondrial autoantibodies (AMA), and a high prevalence of autoimmune diseases as co‐morbidities, indicate that autoimmunity plays a critical role in the disease pathogenesis 2, 3, 4.

PBC is a multi‐factorial disease: individuals having predisposed genetic factors could develop PBC as a consequence of environmental triggers 5, 6. Bacterial infection is regarded as one of the most important environmental factors that contribute to the breaking of tolerance to mitochondrial autoantigens in PBC. This is evident by various case series and case–control studies that demonstrated the high incidences of bacterial infection in patients with PBC 7, 8, 9. Here, we will delineate the aetiological role of pathogen infection in PBC, a model autoimmune disease.

Contribution of environmental triggers to PBC

As in many other autoimmune diseases 10, 11, 12, genetic predisposing factors have been shown to be associated with disease progression in PBC. For example, epidemiological data indicated an increased prevalence of patients with PBC among first‐degree relatives and siblings of an index patient, in what is known as familial clustering of PBC 13, 14, 15, 16. In addition, the concordance rate of PBC is 63% in monozygotic twins, which is substantially higher than that of many other autoimmune diseases 17. However, other studies have indicated that the susceptibility of PBC does not depend solely upon genetic predisposition. Although studies of monozygotic twins revealed a high concordance rate, discordant pairs were also identified 17. Recent epidemiological studies revealed a relatively low risk of PBC in first‐degree relatives of indexed patients during 8 years of follow‐up, suggesting that genetic predisposition alone does not define the risk of PBC 18. In discordant monozygotic twins, DNA methylation profiles, copy number variation and gene expression differed significantly between affected and unaffected twins 19. Aberrant epigenetic modification of regulatory T cells is also involved in disease activity and tissue inflammation in autoimmune disorders 20, 21. These observations show clearly that environmental factors as well as epigenetic modifications also play significant roles in the development of PBC 22. Environmental factors‐mediated activation of autoimmune response could be elicited by means of molecular mimicry, bystander activation, or both 23, 24 and have been examined in epidemiologically and in experimental animal models of autoimmune disease, such as the non‐obese diabetic mouse for type 1 diabetes and experimental autoimmune encephalomyelitis for multiple sclerosis 25, 26, 27. Among the possible environmental factors, the contribution of bacterial infection in PBC has been investigated intensively, both epidemiologically and experimentally.

Epidemiological evidence suggesting the link of bacterial infection with PBC

Urinary tract infection

Determination of the aetiological association of bacterial infection with PBC has been largely driven by epidemiological, case–control studies of PBC, with robust evidence pointing to a crucial role of urinary tract infection (UTI) in increasing the risk of PBC. This observation further directed the focus on Escherichia coli, a predominant pathogen isolated frequently from patients with UTI, in the aetiopathogenesis of PBC.

The possible association of UTI with PBC was first reported in 1984, when Burroughs described that significant bacteriuria was found in 17 of 87 women with PBC (19%) and six in 89 women with other types of chronic liver disease (7%) (P < 0·05) 28. Further, E. coli was the most frequently isolated organism, detected in 70% of infections. Since then, the role of UTI in the development of PBC and AMA production has caught significant attention. Conversely, a study published by Floreani et al. in 1989 could not demonstrate a significant association of UTI with PBC, in that the overall prevalence of bacteriuria was 11·2% in PBC, 12·1% in other chronic liver diseases (CLD), 18·4% in the 65 female CLD patients and 10·7% in Sjögren’s syndrome patients 29. Although these studies were small in size, lacked age‐matched controls and had conflicting results, these results were enough to encourage the researchers to include UTI as one of the important environmental risk factors of PBC in subsequent case control studies.

In the 21st century, several large‐scale, case–control studies have been performed in the United States and European countries to elucidate environmental triggers implicated in the aetiopathogenesis of a disease 14, 30, 31, 32, 33 (Table 1). The first study, by Howel et al., was a population‐based case–control study in North East England. This study enrolled 100 patients with PBC, with all cases occurring between 1993 and 1995 and including age‐ and sex‐matched controls. Their data did not indicate a significant association of UTI with development of PBC, in that 58 of 100 (61%) of cases and 109 of 223 (51%) of controls never had UTI [odds ratio (OR) = 1·7, 95% confidence interval (CI) = 0·96–3·0) 31. However, other case–control studies thereafter demonstrated consistently that a history or recurrence of UTI were associated significantly with increased risk of PBC, with adjusted ORs ranging from 1·511 to 2·7 14, 30, 32, 33. Among these, Prince et al. performed a case–control study (n = 318) in the same geographical area conducted by Howel et al. In contrast, their data demonstrated a significant association of UTI with PBC (adjusted OR = 2·4, 95% CI = 1·7–3·4) 33. They attributed the difference between the two studies to the small sample size (100 versus 318), as well as the large number of risk factors investigated in the former study. They recruited another cohort from all over the country through the UK–PBC foundation, the national advocatory group for patients with PBC. In this cohort, the association of UTI with PBC was also significant (OR = 1·7, 95% CI = 1·5–2·1) 33. There are two case–control studies on PBC in the United States. In 2001, Parikh‐Patel et al. enrolled 241 patients with PBC with 261 siblings, and 141 friends without PBC. They assessed the association of both urinary tract and vaginal infection in females only and the results were statistically significant (adjusted OR = 2·12, 95% CI = 1·10–4·07) 32. Another study in the United States was the largest ever, including 1032 PBC patients and 1041 controls matched for age, sex, race and geographical location. While this study identified frequent use of nail polish as a novel environmental trigger, it again identified history of UTI as a significant risk factor of PBC (adjusted OR = 1·511, 95% CI = 1·192–1·195) 30. The most recent study was carried out in France, enrolling 222 patients with PBC from 2006 to 2007 14. A number of risk factors, including UTI and other non‐autoimmune diseases, were investigated. The results showed that a history of recurrent UTI in 104 of 218 cases (48%) and 157 of 509 controls (31%) was associated significantly with PBC after multiple logistic regression analyses (P = 0.0010), with the highest OR ever (2·7, 95% CI = 2·0–3·7). Taken together, the importance of a recurrent history of UTI in the aetiopathogenesis of PBC has been well confirmed by several large‐scale, case–control studies, regardless of the geographic area (North East England, the United States and France) or case‐finding methods.

Table 1.

Large‐scale case–control studies and urinary tract infection as a risk factor for pyruvate dehydrogenase (subunit PBC)

Authors	Region	Number (cases/controls)	Adjusted odds ratio (95% confidence interval)
Howel et al.*	North East England	100/223	1.7 (0.96–3.0)
Parikh‐Patel et al.**	United States	241/402	2.12 (1.10–4.07)
Gershwin et al.	United States	1032/1041	1.511 (1.192–1.915)
Prince et al.***	North East England	318/2438	2.4 (1.7–3.4)
Prince et al.****	United Kingdom	2258/2438	1.7 (1.5–2.1)
Corpechot et al.	France	222/509	2.7 (2.0–3.7)

^*Univariate analysis, not adjusted; ^**female subjects only, and vaginal infection was included; ^***cases recruited from North East England; ^****cases were selected from the membership of the UK PBC Foundation.

Other infections

Besides UTI, vaginal infection is also known to be caused frequently by E. coli infection. The association of vaginal infection in PBC was investigated by Parikh‐Patel et al. 32 and Gershwin et al. 30. In Parikh‐Patel’s study, the incidence of vaginal infection history was significantly higher in patients with PBC compared to controls (P < 0·01). Using multivariate analysis, they combined UTI and vaginal infection, and demonstrated again that they were associated significantly with increased risk of PBC. Gershwin et al. 30 also reported an elevated incidence of PBC associated with a history of vaginal infection (unadjusted P‐value = 0·0018). However, only UTI, not vaginal infection, was identified as a possible risk factor in PBC with multiple logistic regression models, probably because vaginal infection was regarded as a confounding factor of UTI in this study.

To date, no other bacterial infections or events possibly related to bacterial infection have been identified as a risk of PBC. The association with a history of tonsillectomy, appendectomy and cholecystectomy has revealed inconsistent results.

Mitochondrial autoantigens – the link between E. coli infection and PBC

UTI appears to be the only bacterial infection that has been identified as a risk factor for PBC by case–control studies. What is the link between UTI and development of PBC? Why and how does bacterial infection of the urinary tract lead to PBC? E. coli is a predominant pathogen in most cases with UTI. Evidence from a number of experimental studies suggests that E. coli infection is a key factor in breaking immunological tolerance against the mitochondrial autoantigen, resulting in the production of AMA, the disease‐specific autoantibodies of PBC 34. Our laboratory has also unravelled that the tolerance breakdown against mitochondrial autoantigens through E. coli infection possibly led to the recognition of autoantigens by autoreactive T lymphocytes as well as B lymphocytes 35, 36, 37, 38.

AMA epitopes

AMA are the most disease‐specific autoantibodies in human immunopathology and are detected in 90–95% of patients with PBC 39, 40. Although a high titre of autoantibody in the sera of patients with PBC was observed by Mackay more than 60 years ago 41, the immunodominant epitopes of AMA were not determined until our group identified the pyruvate dehydrogenase complex E2 subunit (PDC‐E2) as the mitochondrial autoantigen of PBC using cDNA cloning 42. AMA recognize a family of enzymes located in the inner membrane of mitochondria, coined the 2‐oxo‐acid dehydrogenase complexes (2‐OADC), which mainly include PDC‐E2, branched chain 2‐oxo‐acid dehydrogenase complex (BCOADC‐E2), 2‐oxo‐glutaric acid dehydrogenase complex (OGDC‐E2) and dihydrolipoamide dehydrogenase binding protein (E3BP) 43. These enzymes have a common structure consisting of an N‐terminal domain with a single or multiple attachment sites to lysine (¹⁷³K in mammalian PDC‐E2) of the lipoic acid (LA). The dominant epitope sites recognized by AMA are in contiguity with the LA attachment site(s) as the lipoyl domains of these target antigens 44, 45, 46, 47. The amino acid residues critical to maintaining the structural integrity of the PDC‐E2 lipoyl domain have been revealed by site‐directed mutagenesis 48. The high specificity of AMA for PBC suggests that AMAs are not simply serological markers for diagnosis but are important in the immunopathology of PBC.

CD4⁺ and CD8⁺ T cell epitopes

The histological signature of PBC includes dense infiltration of mononuclear cells in the portal tracts near small‐ or medium‐sized bile ducts. Immunohistochemical examination of these lymphocytes reveals a predominance of CD4⁺ and CD8⁺ T cells with B cells and natural killer (NK) cells 49, 50. BECs and hepatocytes in the liver of PBC patients also express large amounts of human leucocyte antigen (HLA) classes I and II molecules 35, 51. Therefore, both CD4⁺ and CD8⁺ autoreactive T cells play a crucial role in the pathogenesis of PBC.

The mapping of the T cell autoepitopes on PDC‐E2 was an important discovery that allowed for further studies of autoreactive cellular immunity in PBC. First, the availability of overlapping recombinant PDC‐E2 peptides enabled the demonstration that a subpopulation of liver‐infiltrating T cells from PBC were specifically directed against PDC‐E2 52. Furthermore, amino acid residues 163–176 (GDLLAEIETDKATI) within the inner lipoyl domain of PDC‐E2 were identified as the minimal T cell epitope using a panel of 33 overlapping synthetic peptides spanning the entire PDC‐E2 sequence 36 (Table 2). Further studies have shown that the HLA restriction molecule for such PDC‐E2 epitopes is HLA‐DR53 (B4*0101) 36 and that amino acid residues E, D and K (at positions 170, 172 and 173, respectively) are essential for recognition by the PDC‐E2 specific T cell clones. Interestingly, this ExDK motif can also be found within the outer lipoyl domain of PDC‐E2 and is recognized by T cell clones specific for PDC‐E2 163–176. Similarly, these CD4⁺ T cell clones also recognize the other mitochondrial autoantigens, including OGDC‐E2, BCOADC‐E2 and E3BP. More specifically, T cell clones cross‐reacted with amino acid residues 100–113 within OGDC‐E2, 90–103 within BCOADC‐E2 and 34–47 within E3BP, all of which are located in the lipoyl domain of the respective subunit 37, 53 (Fig. 1). Importantly, the frequency of PDC‐E2‐specific CD4⁺ T cells was 100–150‐fold higher in the liver and hilar lymph nodes than in the peripheral blood 53. Our laboratory also characterized a major histocompatibility complex (MHC) class I (HLA‐A2)‐restricted epitope for CD8⁺ T cells as PDC‐E2 peptide 159–167 (KLSEGDLLA), which again mapped to the same region of the autoantigen PDC‐E2 49 (Table 2, Fig. 1). Taken together, AMA and autoreactive CD4⁺ helper and CD8⁺ cytotoxic T cells contain a shared peptide sequence of the inner lipoyl domain of human PDC‐E2, and the T and B cell epitopes on PDC‐E2 and other enzymes of the 2‐OADC can be considered a cluster of overlapping epitopes (Fig. 1).

Table 2.

Molecular mimicry and immunodominant epitopes of human pyruvate dehydrogenase E2 subunit (PDC‐E2) 155–185*

Human PDC‐E2	KVGEKLSEGDLLAEIETDK*ATIGFEVQEEGY
B cell	KVGEKLSEGDLLAEIETDK*ATIGFEVQEEGY
CD4+ T cell	KVGEKLSEGDLLAEIETDK*ATIGFEVQEEGY
CD8+ T cell	KVGEKLSEGDLLAEIETDK*ATIGFEVQEEGY
E. coli PDC‐E2**	K-G--------L-EIETDK----------G-
Novosphingobium aromaticivorans PDC‐E2	--------GD-LAEIETDKAT--FE---EG-

^*K denotes ¹⁷³lysine which is the attachment site of lipoic acid (LA). The epitope recognized by each cell type is underlined.

^**ExDK sequence was preserved in both human and Escherichia coli PDC‐E2.

Figure 1.

Overlapping 2‐oxo‐acid dehydrogenase complexes (2‐OADC) E2 subunit (E2) T and B cell epitopes in primary biliary cholangitis (PBC), schematic representation of autoreactive B cells, CD4⁺ and CD8⁺ T cells epitopes. CD4⁺ T cell clones specific for amino acid residues 163–176 of the inner lipoyl domain of PDC‐E2 also recognize amino acid residues 36–49 of the outer lipoyl domain of pyruvate dehydrogenase (PDC)‐E2, and cross‐react with amino acid residues 90–103 within BCOADC‐E2, 100–113 within OGDC‐E2 and 34–47 within E3BP, all of which are located in the lipoyl domain of the respective subunit. CD8⁺ T cell clones recognize PDC‐E2 peptide 159–167, which again mapped to the inner lipoyl domain of PDC‐E2.

Tolerance breakdown against mitochondrial autoantigens

PDC is a complex of three enzymes (E1, E2 and E3 component) that converts pyruvate into acetyl‐co‐enzyme A, which is then used in the citric acid cycle to carry out aerobic cellular respiration. PDC links the glycolysis metabolic pathway to the citric acid cycle, and thus is extremely vital for cell survival. In this regard, it is reasonable that PDC‐E2 is a well‐preserved molecule among various species, including bacteria. As shown in Table 2, human PDC‐E2 shares a significant homology with E. coli PDC‐E2, especially in the region of immunodominant epitope of AMA. This molecular mimicry between human PDC‐E2 and E. coli PDC‐E2 may explain the breaking of tolerance to mitochondrial autoantigens and the development of AMA. In fact, the cross‐reactivity of AMA against several prokaryotic antigens has been reported for a number of microbes, including E. coli, Klebsiella pneumoniae, Proteus mirabilis, Staphylococcus aureus and Salmonella Minnesota. However, the target components of AMA in these bacteria have not been fully determined 54. Sera from patients with PBC have been shown to react with both human PDC‐E2 and E. coli PDC‐E2, and the epitope of E. coli PDC‐E2 also maps to similar lipoyl domains 55, 56, 57. Furthermore, the entire ExDK sequence, which was reported to be the essential sequence of human PDC‐E2 for recognition of CD4⁺ PDC‐E2‐specific T cells, is shared by both human and E. coli PDC‐E2 36. Taken together, these studies suggest clearly that E. coli infections may trigger the breaking of immunological tolerance against human PDC‐E2, not only at the B cell but also at the T cell‐level.

Other possible microbial candidates

It is of note, however, that PDC‐E2 is phyolgenetically highly conserved from prokaryotes to higher organisms, and E. coli is not the only candidate bacterium in the breaking of tolerance to mitochondrial autoantigens. In fact, serum titres to E. coli PDC‐E2 were lower and appeared later in the disease than those to human PDC‐E2 56. Thus, it is possible that other bacteria may be involved in the aetiology of PBC through molecular mimicry.

Novosphingobium aromaticivorans

Among a number of candidate bacteria, N. aromaticivorans has drawn significant attention. N. aromaticivorans is a Gram‐negative aerobe, classified with the Sphingomonas genus 58. This microorganism is environmentally ubiquitous and found in soil, water and coastal plain sediments. It also has the capacity to metabolize ubiquitous xenobiotics, and therefore is used in bioremediation wastage 59. Our laboratory has paid attention to this microorganism 60 because the amino acid sequences of two proteins from N. aromaticivorans display a high degree of homology with the dominant immunogenic lipoylated domain of human PDC‐E2 (Table 2). This is the highest level of homology between any known microorganism and this mitochondrial autoantigen. Accordingly, we were able to demonstrate high cross‐reactivity of AMA with two known lipoylated bacterial proteins from N. aromaticivorans: 100% of anti‐PDC‐E2 positive sera reacted, with titres approximately 100–1000‐fold higher than against E. coli. The reactivity was also detected in sera from AMA‐negative patients with PBC, while no control sera reacted against the bacterial proteins. Moreover, N. aromaticivorans was detected by polymerase chain reaction (PCR) in faecal specimens of approximately 25% of patients with PBC and controls 60.

Further, we focused on this bacterium because of its the capacity to ubiquitously metabolize xenobiotic, including its cleavage activity on 17β‐oestradiol, transforming the inactive conjugated form into the free active form 61. However, animal studies using infection with N. aromaticivorans could not clearly conclude its specificity in the induction of PBC 38, 62.

Lactobacillus delbrueckii

L. delbrueckii, subspecies bulgaricus, is a probiotic microorganism for starter cultures in yogurt production. Bogdanos et al. reported that IgG3 antibodies directed to the SxGDL[ILV]AE motif of L. delbrueckii‐galactosidase were present in AMA‐positive PBC sera and cross‐react with humanPDC‐E2. L. delbrueckii subspecies bulgaricus is thus implicated as a possible environmental trigger for breaking tolerance against mitochondrial autoantigens 63. Interestingly, a clinical case of symptomatic PBC with antibodies to the L. delbrucekki subspecies bulgaricus galactosidase following lactobacillus vaccination for recurrent vaginitis has also been reported 64.

Betaretroviruses

A role for retroviruses in the aetiology of PBC has also been suggested 65. PBC sera reacted with proteins from human immunodeficiency virus 1 and human intracisternal A‐type particle 66. Retroviral particles related closely to the mouse mammary tumour virus were detected in BECs and lymph nodes from patients with PBC 67, 68, and co‐culture of non‐PBC BECs with PBC lymph nodes generated the expression of PDC‐E2‐like antigens on the apical surface of BECs 67. However, further intensive investigations failed to recapitulate these results; i.e. no detectable reactivity against mouse mammary tumour virus proteins were detected in PBC sera, and no detectable immunohistochemical or molecular evidence for MMTV was found in the liver specimens or PBLs of patients with PBC 69. Thus, the association of betaretroviruses with PBC is unlikely.

Is bacterial infection the only plausible trigger?

Although there is substantial evidence clearly indicating the aetiological implication of bacterial infection in tolerance breakdown against mitochondrial autoantigen and the development of PBC, we should keep in mind two important epidemiological facts regarding PBC – an improved environmental hygiene and an increased prevalence of PBC, especially in males. Environmental hygiene has improved greatly in the last few decades worldwide, and the magnitude of UTI has been definitely reduced, although due partly to the prevailed use of antibiotics. By contrast, both incidence and prevalence of PBC seem to be increasing. A systematic review in 2012 based on 29 epidemiological studies of PBC indicated that incidence and point prevalence of PBC ranged from 0·39 to 5·8 per 100 000 populations and from 1·91 to 40·2 per 100 000 populations, respectively. In particular, eight studies depicting yearly prevalence rates for several consecutive years reported increased prevalence rates over time 70. The increase in the prevalence of PBC argues against a ‘bacterial infection hypothesis’, based on the improved environmental hygiene.

Also, it is notable that the male : female ratio seems to be changing in recent decades, and the number of male patients with PBC is increasing. Previous large‐scale population‐based studies demonstrated that the male : female ratio was almost 1 : 10 71, 72, 73, 74. However, recent data suggested that the proportions of male patients increased to around 15%, lowering the male to female ratio to 1 : 6 70, 75, 76, 77. Surprisingly, Lleo et al. reported that male patients consist of 21% PBC cases in Denmark, and even up to 33% in Lombardia, Italy 78. Another example of the increase in male patients with PBC comes from a very recent unpublished epidemiological study in Japan, which suggested a male : female ratio as low as 1 : 4.2. This alteration in the male : female ratio may be explained by increased recognition of the disease in the male population. Hence, the increased prevalence of PBC in male population, if true, may suggest that in addition to E. coli infection, other environmental factors also are involved in causing PBC.

Concluding remarks

As mentioned previously, PBC is a multi‐factorial autoimmune‐mediated disease and both genetic predisposing factors and environmental triggers influence susceptibility of the disease. Bacterial infection, especially an E. coli or N. aromaticivorans infection, may be associated with PBC through molecular mimicry between human and bacterial PDC‐E2. However, multiple pathways could lead to the loss of tolerance in PBC 79, 80, 81, 82, 83. The possible role of retroviruses has been discussed earlier. UTI mediated by E. coli infection is probably one of the dominant pathways in female patients, but other alternative pathways certainly exist and may be dominantly operative in male patients.

Previous case–control studies also suggest a xenobiotic‐mediated aetiology in PBC. Geographically uneven distribution of PBC patients in a particular region are reported, especially near toxic waste sites 84, 85, 86. These epidemiological data prompted researchers to identify environmental mimotopes in the form of xenobiotics. The conjugate derived from 2‐octynoic acid (2‐OA) present in cosmetics and some chewing gums was unique in both its quantitative structure–activity relationship analysis and reactivity with PBC sera 87. Furthermore, another xenobiotic, 2‐nonyamide, provided an optimal chemical structure of the xenobiotics modified epitope, which demonstrated enhanced recognition by AMA‐positive PBC sera 88. These findings illustrate that xenobiotic modification of PDC‐E2 by chemical exposure in everyday life can generate immunogenic neoantigens that result in the loss of tolerance to human PDC‐E2 in subjects who are genetically susceptible to PBC 89. This hypothesis is supported by animal studies in which animals immunized with 2OA‐BSA developed AMA and autoimmune cholangitis 90, 91. Furthermore, cross‐reactive monoclonal human antibodies that are reactive to both native PDC‐E2 and 2‐OA also recognize lipoic acid 92, further suggesting that xenobiotically modified lipoic acid is an initial target of autoimmunity in PBC.

Case–control studies are powerful tools for determining environmental triggers and indeed have identified a number of risk factors among a variety of demographics, lifestyle, comorbidities and past history for development of PBC. However, environmental factors contributing to PBC are complex. Furthermore, it is also of note that risk factors must have been assumed a priori in a case–control study, and as a result unexpected risk factors cannot be identified. In this regard, we need to seek environmental triggering factors, such as bacterial infections or other factors, with an open mind using newly designed case–control studies together with the available state‐of‐the‐art methodologies. Furthermore, as all previous case–control studies have been performed in European countries and the United States, similar approaches to identify environmental triggers should be attempted in other geographical areas, including Asia.

Disclosure

None.