Immunopathogenesis of cholestatic autoimmune liver diseases
Preclinical safety department, Novartis Pharma AG, Basel, Switzerland (J. Medina); Department of Gastrointestinal and Liver Diseases, Academic Medical Center, Amsterdam, The Netherlands (E. A. Jones); Liver Unit, Hospital de la Princesa, Universidad Autónoma de Madrid, Spain (C. García-Monzón, R. Moreno-Otero).Correspondence to: Ricardo Moreno-Otero, Liver Unit, Hospital de la Princesa, Diego de León 62, 28006-Madrid, Spain. Tel./fax: + 34 91 3093911; e-mail: email@example.com
Primary biliary cirrhosis and primary sclerosing cholangitis are well recognized chronic cholestatic liver diseases that are considered to have an autoimmune basis. Recent progress in the study of autoimmune liver diseases has improved the recognition and characterization of these conditions. An important component of this progress has been the identification of liver disease-associated autoantibodies and their respective target antigens, and the development of specific assays for these autoantibodies. In addition, some nonhumoral immunological findings imply an involvement of specific immunopathogenic mechanisms in the development of these conditions. Furthermore, immunogenetic factors associated with increased susceptibility to some of these diseases have been identified. This article reviews the most relevant information relating to the postulated autoimmune pathogenesis of these diseases, with special emphasis on their associated humoral and cellular immunological abnormalities and immunopathogenetic factors. Some of the remaining important unresolved issues relating to the pathogenesis of these diseases, that need to be addressed in further research, are highlighted.
Several cholestatic autoimmune liver diseases (AILDs) are recognised. Of these, primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) are the best characterized. Certain overlap syndromes that are more difficult to characterize, such as the overlap syndrome primary biliary cirrhosis/autoimmune hepatitis and the overlap syndrome primary sclerosing cholangitis/autoimmune hepatitis, can also be classified as cholestatic AILDs.
A number of factors have been considered as possible triggers for the chronicity of the disease process in AILDs, e.g. viruses, bacteria, chemicals, drugs, genes . It is most likely that the manifestations of AILDs arise as a consequence of multiple factors. In most of these diseases, autoantibodies have been identified, their molecular targets characterized and their epitopes mapped. Although these findings are important, the pathogenetic mechanisms that trigger the onset of these diseases, and the precise role of the characteristic autoantibodies associated with them remain unknown. Several risk factors for certain AILDs have been determined, such as immunogenetic background, sex and geographical region. This information needs to be extended and integrated with other relevant complementary information to facilitate the diagnosis and management of cholestatic AILDs. The aim of this article is to review currently available information relating to the etiology, serology, immunogenetics and pathogenesis of PBC and PSC, and to comment briefly on the possibly related cholestatic overlap syndromes.
Primary biliary cirrhosis
PBC, more accurately termed chronic nonsuppurative destructive cholangitis, is a disease characterized by chronic patchy destruction and obliteration of septal and intrahepatic bile ductules, associated with an infiltration of plasma cells and lymphocytes in the portal tracts, interface hepatitis, fibrosis, and slowly progressive intrahepatic cholestasis. Progression occurs over years or decades. The end stage is an established biliary cirrhosis [2,3].
Unexplained pruritus in a middle-aged subject (particularly a woman) with an elevated serum alkaline phosphatase (two- to five-fold) and/or an elevated serum 5-nucleotidase, suggest PBC. Patients may also present with no symptoms as a consequence of the incidental finding of biochemical index of cholestasis, such as an elevated serum levels of alkaline phosphatase, bile acids or cholesterol. Very rarely, patients present with clinical cholestasis. However, the most specific and sensitive marker of the disease is the presence of a circulating IgG antimitochondrial antibody (AMA), which is detectable in more than 90% of patients (titer > 1 : 40) and is only rarely present in other forms of liver disease . AMA autoantibodies can be found not only in serum but also in the bile  and in the saliva of PBC patients . A liver biopsy showing compatible histology is also necessary to make a confident diagnosis of PBC. Interestingly, neither the titer nor the tissue-binding pattern of AMA correlate with clinical disease . In addition to AMA, elevated serum levels of IgM and cryoproteins consisting of immune complex-like structures (not necessarily containing antigen) capable of activating the alternative complement pathway are found in 80–90% of patients .
Hyperbilirubinemia is usually only present when the disease has progressed to advanced stages and indicates a poor prognosis . Histologically, biliary epithelial cell damage, accompanied by a T cell rich mononuclear cell portal infiltrate and upregulation of various adhesion molecules (ICAM-1, VCAM-1, LFA, VLA, ELAM-1) on endothelial cells, epithelial cells and inflammatory cells in the portal tracts and hepatic parenchyma are found [10–12]. In the peripheral blood of PBC patients, reduced numbers of T cells may occur due to a selective intrahepatic sequestration of CD8 + T lymphocytes . In addition, impaired activation of CD4 + T cells has been reported in PBC .
The AMA autoantibodies recognize inner mitochondrial membrane proteins identified as: the E2 subunits of the pyruvate dehydrogenase complex (PDC-E2); the branched chain 2-oxo acid dehydrogenase complex (BCOADC-E2); the 2-oxoglutarate dehydrogenase complex (OGDC-E2); the E1a and the E1b subunits of PDC (PDC E1a and PDC E1b); and the dihydrolipoamide dehydrogenase-binding protein of the PDC (PDC-E3 BP) or protein X [15–23]. The immunodominant epitopes of PDC-E2 (the major autoantigen found in 90% of PBC patients), BCOADC-E2, OGDC-E2 and PDC-E3BP have been mapped within the lipoic acid binding domains. In contrast to spontaneous autoantibodies found in normal non-PBC sera , binding of autoantibodies in PBC sera to these epitopes results in inhibition of the enzymatic activity of PDC, BCOADC-E2 and OGDC-E2, respectively [25–27]. Studies using monoclonal antibodies to these autoantigens have shown a unique staining pattern in the apical region of biliary epithelial cells of PBC patients .
Other autoantibodies occur less frequently: a subset of patients with PBC was found to have autoantibodies directed against nuclear envelope proteins. The major autoantigens are the glycoprotein gp210, a transmembrane protein of the nuclear pore complex, whose predominant epitope is a 15 aminoacid sequence in the cytoplasmic, carboxyl-terminal domain of the protein, and p62, a nuclear pore glycoprotein [29,30]. Some patients have autoantibodies against the proteins Sp100 and PML present in nuclear dots . Other less specific autoantibodies in PBC are those against lamin B receptors, a transmembrane protein of the inner nuclear membrane. The mechanism of the organ-specific breakdown of B cell and T cell self-tolerance to these ubiquitous antigens remains uncertain.
AMA autoantibodies may play a pathogenic role in PBC . This requires extracellular (cell surface) expression of the relevant autoantigen (PDC-E2) on biliary epithelial cell surface membranes [33,34] of PBC patients. Regional lymph nodes have also been shown to express increased amounts of both E2 and X protein . The possibility that the antigen is released after tissue damage seems to be unlikely, as AMA autoantibodies are the first abnormality detected early in the course of the disease (before elevation of serum alkaline phosphatase occurs) and they continue to be detectable after liver transplantation for PBC in the presence of normal allograft histology [32,33]. Another possibility would be that autoantibodies might interfere with cellular function as a consequence of penetrating cells, either by endocytosis following binding to brush border myosin or by internalization of IgA AMA through the polymeric immunoglobulin receptor on the basal surface of the biliary epithelial cell . Finally, it has been suggested that a biliary epithelial cell membrane protein, sharing a cross-reactive epitope with PDC-E2, could act as a neoantigen [37 ]. However, the validity of these hypotheses remains to be confirmed. In addition, although biliary epithelial cells express HLA class I and II molecules , it seems unlikely that these molecules can present antigen to naive ymphocytes .
The possibility that T cells also play a role in the pathogenesis of PBC is suggested by the evidence that autoreactive T cells infiltrate the liver, as shown by immunohistochemical staining of biopsies. Characterization of this tissue infiltrate has been performed, showing that it contains a marked enrichment in CD8 + cytotoxic T cells as compared with peripheral blood ( 39). Recently, it has been shown through immunohistochemical techniques that intrahepatic CD30 + lymphocytes have a role in the IgA type B cell hyperactivity that contributes to the portal tract and bile duct lesions in PBC . The T cell epitope of PDC-E2 in PBC has been studied using T cell lines derived from the peripheral blood mononuclear cells of PBC patients. In a study by Shimoda et al. , the minimal epitopes of such T cell clones mapped to the aminoacid sequence 163–176, within the inner lipoyl domain of PDC-E2. This finding suggests an overlap in the PDC-E2 specificity of the T and B cell epitopes in PBC [42–45]. The pattern of lymphokine production of T cell clones derived from PBC patients undergoing liver transplantation indicates that both Th1 and Th2 lymphocytes are involved in the tissue damage that occurs in this disease.
Although the information available on the mechanisms of T cell autoreactivity in PBC is limited, results from investigations using animal models have contributed appreciably to our understanding of this process. In an elegant series of experiments using a murine experimental model of autoimmune liver disease, it has been shown that the breakdown of self-tolerance to PDC-E2 (particularly at the T cell level) represents at least one of the factors that could trigger the development of PBC. In this model, sensitisation of animals to PDC-E2 of mammalian origin resulted in a breakdown of both B cell and T cell tolerance to murine PDC-E2, as well as the development of a nonsuppurative destructive cholangitis [46,47].
The precise mechanisms by which T cells selectively infiltrate the portal tract and damage bile ducts are not clear. It has been shown that in the livers of PBC patients there is increased fibronectin expression on the biliary basement membrane. Integrin alpha4–fibronectin interaction may facilitate adhesion and the penetration of infiltrating alpha4-expressing lymphocytes into the biliary epithelial layer . This phenomenon might be an important step in the pathogenesis of the disease.
Destruction of bile ducts has been shown to involve local apoptotic cell death, a process which is mediated by apoptosis-regulating bcl-2 familial proteins . In this context, not only T cells seem to be involved in the pathogenesis of PBC, but Fas ligand expressing CD68 + monocytes have also been shown to play a role in the apoptotic biliary epithelial cell loss that takes place in this disease .
The factors that trigger the development of PBC are unknown. Viral and bacterial infectious agents, as well as certain drugs, have been identified in association with the disease. A high incidence of retroviral antibodies has been reported in patients with PBC, as well as a strong association between the presence of antibodies to double stranded DNA and antibody reactivity to a specific retroviral agent, human intracisternal A-type particle (HIAP). These findings indicate a possible autoimmune response to antigenically-related cellular proteins [51,52]. In addition, in patients with PBC, virus-like particles have been observed by electron microscopy in biliary epithelium, endogenous retroviral sequences have been cloned from liver samples, and antibody reactivity to HIAP has been detected. However, there is no definitive direct evidence to link the endogenous retroviral sequences in PBC patients to the retroviral antibody reactivity or virus like-particles.
Molecular mimicry between bacterial antigens and PDC-E2 has been proposed as a potential mechanism of autoimmunity in PBC [27,41]. However, there is no clear evidence that chronic bacterial infectious processes play a causative role in PBC .
Family studies suggest that genetic factors may play a role in determining susceptibility to PBC, and several polymorphic genes may encode factors predisposing to the disease. There is evidence from several different populations to support a role for the major histocompatibility complex (MHC) class II antigen, HLA DR8, in increasing the risk of PBC , but other authors have reported associations of PBC with HLA DR3 and possibly HLA DR4  in Caucasians, while in Japanese subjects an association with HLA-DR2 has been reported . The diverse results of these studies may be attributable to differences in the sensitivity of assays and methodology used in analyses, the diversity of ethnic groups studied, as well as possible errors in diagnosis.
Tumour necrosis factor (TNF) polymorphism may also be important. Production of this pro-inflammatory cytokine by lymphocytes of PBC patients is down-regulated. An association has been found between reduced carriage of the rare allele TNF*2 and PBC . In addition, the high frequency of -308 TNF1/TNF1 genotype seen in patients with advanced PBC indicates that this allele may be linked to disease progression. Furthermore, the CTLA-4 exon 1 polymorphism has been identified as the first non-major histocompatibility complex gene conferring susceptibility to develop PBC . Finally, some drugs (e.g. chlorpromazine, nonsteroidal anti-inflammatory drugs) may precipitate a PBC-like cholestatic syndrome .
Therapy of PBC patients with ursodeoxycholic acid (UDCA) seems to slow progression of the disease and to reduce AMA titer [59,60], but does not obviate the need for liver transplantation. There seems to be a rationale for immunosuppression, but the commonly used immunosuppressive drugs, including prednisolone, have unacceptable safety profiles, particularly for asymptomatic patients .
Primary sclerosing cholangitis
PSC is a cholestatic syndrome of unknown etiology with a chronic variable course. It is a slowly progressive hepatobiliary disease, characterized by patchy fibro-obliterative inflammation of intrahepatic and extrahepatic bile ducts . The syndrome occurs most frequently in young males and is characterized by: chronic cholestasis; frequent association (50%) with chronic autoimmune inflammatory bowel disease, usually ulcerative colitis; presence of perinuclear antineutrophil cytoplasmic antibodies (p-ANCA); and characteristic cholangiographic abnormalities. The disease usually progresses slowly over a decade, or longer, from an asymptomatic stage to clinical cholestasis and eventually an established biliary cirrhosis. Cholangiocarcinoma is a common late complication . The diagnosis is usually based on the characteristic cholangiographic changes (beading and strictures of bile ducts) and/or compatible hepatic histology (onion skin-like concentric fibrosis around small bile ducts) in a patient with serum biochemical features of mild cholestasis and, possibly, an associated chronic autoimmune inflammatory bowel disease .
Several possible etiological factors have been proposed, including infectious (bacteria, viruses), toxic, genetic and immunologic factors . Given that many PSC patients have concomitant ulcerative colitis, the possibility that portal bacteremia or an enterohepatic circulation of toxins exist has been considered by several investigators ( 64,65). Among the infectious factors, the presence of Helicobacter species in the liver in PSC, as well as antibodies to this microorganism in serum of PSC patients have been found with a high frequency [66–68]. However, portal vein bacteremia is rare in PSC patients, as well as among patients undergoing surgery for ulcerative colitis. Therefore, there is little evidence supporting the hypothesis that portal bacteremia may be involved in the etiology of PSC . There are isolated reports of associations between reovirus 3 and cytomegalovirus and PSC [70–72], but no histological features of cytomegalovirus infection have ever been demonstrated in bile duct epithelial cells or hepatocytes of PSC patients . Therefore, there is no convincing evidence that supports the hypothesis that PSC is linked to a viral infection.
Only a genetic predisposition has been shown to be strongly associated with the occurrence of this disease: the HLA haplotypes A1-B8-DRB3*0101-DRB1*0301-DQA1*0501-DQB1*0201 and DRB3*0101-DRB1*1301-DQA1*0103-DQB1* 0603 tend to be associated with PSC. This genetic susceptibility may be determined by polymorphism within the HLA class III region; in particular, a study conducted by Bernal and colleagues , showed an association with the TNF2 allele (-238 gene polymorphism) but not with other TNF alleles. TNF2 was significantly increased in the presence of B8 and DRB3*0101 only, and was independent of DRB1*0301. The associations with B8 and TNF2 were stronger than the associations with any of the other HLA class II alleles examined .
It has been proposed that both cellular and humoral immune mechanisms may be important in the pathogenesis of PSC. This hypothesis is supported by the occurrence of autoantibodies with several different specificities in PSC, as well as by altered cell mediated immunity in PSC patients. The most prominent autoantibodies in PSC are p-ANCA , which comprise a range of autoantibodies with different antigenic specificities including proteinase 3, myeloperoxidase, elastase, bactericidal/permeability-increasing protein, cathepsin G and lactoferrin . An interesting study by Terjung and colleagues  has also demonstrated that p-ANCA are antinuclear antibodies reactive with granulocyte-specific antigens present in the nuclear lamina (lamins A, C and B1, and lamin B receptor). Very recently, catalase and alpha-enolase have also been characterized as autoantigens in patients with PSC . Because catalase is an important antioxidant enzyme that prevents cell damage induced by highly reactive oxygen-derived free radicals, an impairment of the redox status of cells by catalase autoantibodies has been suggested as a possible pathogenic mechanism in PSC. A relationship has been established between the abundance of p-ANCA and particular clinical features of PSC . For instance, anticardiolipin antibodies correlate with the Mayo risk score and histologic stage of the disease, i.e. indices of the severity of the disease process ; thus, anticardiolipin antibodies might be a prognostic marker in PSC.
Cell mediated immune abnormalities in PSC include a decline in total number of circulating T cells, an increase in the ratio CD4 to CD8 cells and an increase in the number of B cells . There is some evidence to suggest that gamma delta T lymphocytes may be involved in immune-mediated tissue damage in PSC . Finally, it has been postulated that biliary epithelium, by expressing HLA class II antigens, may be able to present autoantigens to T cells [81,82], which may subsequently mediate tissue damage of PCS. For example, B7-2 expression on biliary epithelial cells has recently been shown to be increased in PSC . B7-2 is a costimulatory factor that binds to CD28 on T cells involved in antigen presentation. However, these findings have not been confirmed in an independent study , which found no evidence of an antigen-presenting role for the biliary epithelium. Further studies are needed to clarify the role of T cells in tissue damage in PSC.
At present no medical treatment has been shown to slow disease progression in PSC. However, immunosuppressive therapy, as well as UDCA, have been reported to have some efficacy in small uncontrolled studies . In addition, endoscopy with stenting is usually useful in the management of symptomatic dominant strictures involving large extrahepatic bile ducts. Liver transplantation is indicated for end stage disease.
Autoimmune cholangitis (AIC) is diagnosed in AMA-negative, ANA-positive individuals who have clinical and laboratory features of cholestasis [85–87]. Serum alkaline phosphatase is invariably elevated, and the serum transaminase and gamma-glutamyl transpeptidase levels are also usually increased. AIC is not considered to be an overlap syndrome of PBC and autoimmune hepatitis (AIH). Making a definitive diagnosis may be difficult .
Some authors consider that PBC and AIC are components of the same disease spectrum, and they claim that AMA negativity is attributable in part to technical problems with the assay method [32,89]. However, other authors have proposed that antibodies to carbonic anhydrase in AIC are a marker that enables AIC to be differentiated from PBC , whereas the antinuclear antibodies anti-Sp100 and anti-Gp210 have been shown to be markers of PBC in patients with features of AIC . Because antilactoferrin antibodies have been found in a large proportion of AIC patients , lactoferrin located in bile ducts and liver cells has been proposed as a candidate target antigen in AIC.
Usually, treatment with ursodeoxycholic acid is recommended, and prednisolone therapy is considered; however, results of treatment are uncertain, as no controlled data on therapy have been published.
Overlap syndrome primary biliary cirrhosis/autoimmune hepatitis
The overlap syndrome of PBC and AIH is an autoimmune liver disease characterized by the presence of biochemical, serological and histological features typical of both PBC and AIH. Other names for this syndrome are autoimmune cholangiopathy , mixed-type chronic aggressive hepatitis and PBC , and PBC, hepatitic form . Patients usually have florid piecemeal necrosis and high serum transaminase levels. The syndrome can be differentiated from AIC by the absence in PBC/AIH of histological changes of chronic nonsuppurative cholangitis and the presence of AMA autoantibodies . No specific AMA seems to be associated with this syndrome . Initially, this overlap syndrome was considered to be rare , but its prevalence has probably been underestimated .
Recent studies have suggested that a PBC-like process precedes the development of this syndrome, because patients at initial presentation have been found to have bile duct lesions histologically or anti-M2 positive AMA autoantibodies . Genetic susceptibility (mainly AIH-associated HLA type 8, DR3 or DR4) may be the reason for progression to a syndrome with many or all of the features of AIH. Combined treatment with immunosuppressive therapy and UDCA is usually recommended, although appropriate controlled therapeutic trials have not been reported.
Overlap syndrome primary sclerosing cholangitis/autoimmune hepatitis
The overlap syndrome of PSC and AIH was initially described in pediatric patients who fulfilled criteria for the diagnosis of both PSC and AIH . These patients showed typical bile duct strictures and dilatations on endoscopic retrograde cholangiography and typical histological features of chronic hepatitis (interface hepatitis). High titer ANA autoantibodies are present, together with elevated levels of aminotransferases, alkaline phosphatase, gamma-glutamyltranspeptidase and IgG, as well as AMA, and pANCA autoantibodies. Only a few cases have been reported [100–102]. The patients appear to benefit from immunosuppressive treatment.
Although our knowledge of the pathogenesis of cholestatic autoimmune liver diseases has increased substantially during recent years, there are still many questions that remain to be answered. In particular, the precise mechanisms that result in the expression of neoantigens (or loss of self tolerance), and that trigger a local autoimmune reaction against bile duct epithelial cells need to be determined. It is not known with certainty to what extent such phenomena are related to immunogenetic factors. Application of new technologies to the study of these diseases should help to identify risk groups, to improve their classification and, possibly, to determine factors that trigger their development and perpetuate the disease processes. Progress in the field of AILDs should be greatly facilitated by an increased understanding of the fundamental nature of autoimmunity.