Primary biliary cirrhosis

Genetic factors have an impact on PBC pathogenesis that is stronger than that in nearly any other autoimmune disease [85‐87]. Accordingly, a concordance rate of about 60% was seen in monozygotic twins (as opposed to nearly 0% for dizygotic twins) [88, 89]. A significantly increased incidence of PBC is seen in relatives of PBC patients [89, 90]. The relative risk of a first-degree relative of a PBC patient is 50- to 100-fold higher than for the general population [91], yielding a prevalence rate up to 5–6% [19, 90, 92]. Interestingly, among affected monozygotic twins, though the age of disease onset is similar, progression and disease severity vary, emphasizing the role of epigenetic and probably environmental factors [89].

It has been difficult to identify distinct susceptibility genes in PBC so far. Genetic associations in PBC were shown with major histocompatibility complex encoded genes. PBC is apparently associated with the DRB1*08 family of alleles, although marked variation is observed between different ethnic groups. Association with the DRB1*0801-containing haplotype is seen in populations of European origin, whereas in populations of Asian origin, an association is seen with the DRB1*0803 allele. A protective association has been described with DRB1*11 and DRB1*13, but once again, significant population differences are observed [93‐102]. Both associations of PBC with the DRB1*08 allele as well as the protective association of DRB1*11 and DRB1*13 were recently confirmed in the largest series ever reported including 664 unrelated patients from Italy [103]. The odds ratio for developing PBC was 3.3 for DRB1*08-positive subjects, whereas it was reduced to 0.3 for subjects positive for DRB1*11, to 0.7 for DRB1*13, and to 0.1 for carriers of both DRB1*11 and DRB1*13. This study again highlighted the relevance of geographic variation, with marked differences in allele association between northern and southern Italy.

Associations have also been reported with polymorphisms of genes involved in innate or adaptive immunity. Allelic variations of tumor necrosis factor α (TNFα) and of cytotoxic T lymphocyte antigen 4 (CTLA-4), a key regulator of the adaptive immune system, have been repeatedly associated with susceptibility to different autoimmune diseases (such as type I diabetes mellitus and systemic lupus erythematosus) and were also associated with PBC [104‐109]. Although an association of CTLA-4 variants and susceptibility to PBC could not be demonstrated in all studies [110, 111], Poupon et al. [112] most recently confirmed a potential role of TNFα and CTLA-4 variants in the pathogenesis of PBC. In 258 PBC patients and two independent control groups of 286 and 269 healthy volunteers, the authors investigated distribution of newly identified htSNPs of 15 selected candidate genes: two related to immunity encoding for CTLA-4 and TNFα, ten genes related to bile formation encoding hepatobiliary transporters, and three related to adaptive response to cholestasis encoding nuclear receptors. In a case–control analysis, only haplotype-tagging single nucleotide polymorphisms (htSNPs) in CTLA-4 and TNFα showed differences in distribution between PBC and controls, confirming their potential role in the pathogenesis of PBC. In contrast, htSNPs of the ten transporter genes as well as the three nuclear receptor genes under study were equally distributed, confirming previous studies of htSNPs in key transporters without major impact [1, 113, 114]. A strong association of the allelic variant TNFα rs 1799724 (C/T) with disease progression was shown. Most interestingly, a strong association with disease progression was also shown for AE2 rs 2303932 (T/A), a gene encoding for the apical anion exchanger 2 (AE2) in cholangiocytes and hepatocytes. In both cases, presence of the variant was associated with delayed disease progression. In a multi-variate Cox regression, the AE2 variant rs2303932 (T/A) was an independent prognostic factor for disease progression in PBC under UDCA treatment, in addition to serum bilirubin, alkaline phosphatase, and serum albumin levels which are established surrogate markers of prognosis in PBC [112]. Most recently, a landmark genetic association study was published [332]. Associations with the risk of disease were unravelled for 13 loci across the HLA class II region, two SNPs at the interleukin 12 alpha (IL12A) locus, one SNP at the interleukin 12 receptor beta 2 locus, and one previously described SNP at the CTLA4 locus. Associations with more than 10 further loci were described. Various associations with other loci have been described in individual populations, mostly of limited size. However, the vast majority have not been confirmed in independent cohorts, and to date, none of the genetic associations described in PBC have been proven sufficiently [84, 88, 108].

Future genetic linkage studies in affected families as well as association studies in large cohorts of unrelated patients may disclose genetic variants conferring susceptibility or influencing progression and severity of disease. Such linkage studies are awaited for PBC [115].

It remains speculative whether the female preponderance (gender ratio up to 10:1) reflects an X-chromosome-linked locus of susceptibility. Alternatively, a protective role of Y-linked genes could be assumed, or just a gender-specific exposure to environmental triggers like cosmetics [116] or nail polishers [117], as discussed below. However, speculation on a pathomechanistic role for X chromosomal genes was supported by the observation of an increased frequency of X chromosome monosomy in PBC as well as in other autoimmune diseases. This increase in X chromosome monosomy might lead to haploinsufficiency for specific X-linked genes and thereby increase disease predisposition [118]. Case reports of PBC in patients with Turner syndrome (45, X0) also supported this hypothesis [119].

Estrogen signaling has also been proposed to play a role in the homeostatic proliferative response of cholangiocytes in PBC. Accordingly, studies on polymorphisms in estrogen receptor genes revealed associations with the disease, at least in some populations [120]. At the tissue level, cholangiocytes from PBC patients in the earliest disease stages (but not cholangiocytes from normal controls) express estrogen receptors [121]. Agents able to modulate estrogen-receptor-mediated responses (such as tamoxifen) have therefore been proposed as novel, BEC homeostasis targeting therapies, and case reports support this hypothesis [122, 123], but as yet, this potentially interesting therapeutic approach has not undergone formal assessment in clinical trials.

Most recently, altered expression of hepatic microRNA (miRNA) has been described in liver tissue of PBC patients [124]. Certain miRNA negatively regulate protein coding gene expression and may play a critical role in various biological processes. However, a causal link between altered miRNA expression and the development of PBC still remains unproven.

Despite strong evidence for a genetic background in PBC, epidemiological studies have early suggested a role for environmental factors in triggering and/or exacerbating PBC [20, 125, 126]. A significant role for environmental factors was supported by the identification of geographic disease “hot spots,” as first reported in the northeast of England, using formal cluster analysis. The original UK analysis reported an increased frequency of PBC in former industrial and/or coal mining areas [127]. Another recent study from New York examined the prevalence of PBC and PSC near superfund areas and reported significant clusters of PBC surrounding toxic sites [128]. In synopsis, these observations gave rise to the hypothesis of a chemical environmental factor, potentially associated with contaminated land, which could either trigger disease or cause disease through a direct toxic effect [84]. This hypothesis would also provide one possible explanation for the tissue tropism of PBC if the toxin or toxins are excreted into bile (and thereby concentrated in the biliary tree) [84]. The observation that hormone replacement therapy and frequent use of nail polish are linked to the risk of developing PBC further supports the potential impact of environmental factors in the pathogenesis of PBC [117]. Smoking also seems to be a risk factor for PBC and has been demonstrated to accelerate progression [117, 129]. Associations of exposure to chemical environmental compounds and xenobiotics (including drugs, pesticides, or other organic molecules) with various human autoimmune diseases have been described as summarized in [86].

Xenobiotics may contribute to the pathogenesis of PBC by triggering autoimmune reactions. Different mechanisms for the induction of autoimmunity by xenobiotics have been proposed [86, 130]. A potential direct toxic effect of xenobiotics my cause cell death by apoptosis or oncosis, inducing the generation of immunogenic autoepitopes. In addition, chemical modification of native cellular proteins by removal and/or exchange of a hapten has been shown to change processing in antigen-presenting cells and may lead to the presentation of cryptic, potentially immunogenic peptides. Furthermore, xenobiotics may have the potential to modify host proteins to form neoantigens. Neoantigen-specific T cells and B cells, once primed, may cross-react with the formerly inert native autoantigens. In accordance with this hypothesis, Amano et al. [131] studied a number of xenobiotics with a structure similar to lipoic acid, a residue on the E2 epitope of the pyruvate dehydrogenase complex (PDC-E2), the main autoreactive antigen identified in PBC so far in PBC. Replacement of lipoic acid by certain xenobiotics enhanced the reactivity of PBC sera against the PDC-E2 epitope. Particularly, one of the xenobiotics, 2-nonynoic acid, induced reactivity of PBC sera stronger than that of the native lipoic acid residue. Interestingly, the methyl ester of 2-nonynoic acid has a viol-/peach-like scent and is used as an ingredient in perfumes. It is ranked 2,324th out of 12,945 chemical compounds in terms of occupational exposure with an 80% female preponderance due to its use in cosmetics.

Among the environmental factors that have been suggested as potential causative agents in PBC, particularly different bacteria have been discussed. In early histologic lesions in PBC, non-caseating granulomas are observed, as seen in other granulomatous liver diseases including sarcoidosis [1], drug reactions, and, most interestingly, infections. Furthermore, non-caseating granulomas are unique to PBC when compared to other autoimmune pathologies. This has led to suspicion of a microbial basis for PBC [132]. In support of this hypothesis, certain bacteria were found to contain PDC components fully cross-reactive with the mammalian form. It was proposed that exposure to these homologues could trigger cross-reactive immunity. In favor of a bacterial etiology, recent data suggest that Toll-like receptor ligands induce an augmented inflammatory response in PBC. In combination, presence of cross-reactive antigens in a pro-inflammatory environment would theoretically be able to break tolerance [133, 134].

With this theoretical background in mind, early studies associating various bacteria with PBC re-attract interest: E. coli has been reported to be present in excess in the feces of patients with PBC. In addition, the incidence of urinary tract infections often induced by E. coli is high in PBC patients [73, 135], and history of urinary tract infections increases the risk of PBC [117]. Another microorganism that has been proposed as a candidate for the induction of PBC is Novosphingobium aromaticivorans [136]. Titers of antibodies against lipoylated bacterial proteins of this ubiquitous organism, which metabolizes organic compounds including estrogens, were 1,000-fold higher compared to those against E. coli in patients with PBC, but no antibodies were observed in a large cohort of healthy subjects. Lactobacilli and Chlamydia, which show some structural homology with the autoantigen (although reactivity against them is considerably less than that against either E. coli or N. aromaticivorans), have also been implicated as putative pathogens, as have Helicobacter pylori and Mycobacterium gordonae [137‐140]. Recently, a case of PBC following lactobacillus vaccination for recurrent vaginitis was reported. The vaccine contained Lactobacillus salivarius, which exerts a high homology to the beta-galactosidase of Lactobacillus delbrueckii (LACDE BGAL_266–280), and cross-reactivity of patients' autoantibodies against the human PDC-E2_212–226 epitope and LACDE BGAL_266–280 was found. Affinity to the Lactobacillus epitope was higher than to the native mammalian, suggesting that antimicrobial reactivity may have preceded that to the self-mimic [141]. However, the AMA status of this patient before repetitive lactobacillus vaccination could not be assessed and causal relation of lactobacillus exposure and development of PBC remains speculative also in this study.

Despite these intriguing associations, no compelling data have been provided to show that one individual infectious agent can reproducibly be detected in patients with PBC. Although attractive, the model of bacterial infections as cause of PBC is thus supported by little direct evidence. Further objective data are warranted, obtained either from prospectively followed cohorts or through case–control epidemiological approaches, confirming a role for bacteria in triggering PBC [84].

An alternative infectious agent has recently been proposed as trigger of PBC when a human retrovirus was identified both in liver tissue and hilar lymph nodes from PBC patients. EM analysis of liver tissue obtained from PBC patients revealed retrovirus compatible particles in BECs. In periportal lymph nodes, mouse mammary tumor virus (MMTV) was detected and correlated with aberrant distribution of PDC-E2 in perisinusoidal cells. Homogenates of these periportal lymph nodes also had the capacity to infect BEC cultures inducing marked phenotypic change, and this effect could be abolished by irradiation of the culture media, suggestive of an infectious agent [142, 143]. Retroviral infection hypothetically could cause BEC damage either through a direct viral cytopathic effect, through cross-reactivity between viral protein and self-PDC, a “molecular mimicry” model, or virus-induced apoptosis [84]. Retroviral infection would also provide explanations for some key phenomena in PBC. PBC can recur rapidly after transplantation with all of the clinical manifestations including the detection of AMA in serum [144], the aberrant expression of the AMA-reactive protein on BEC [145], and histologic evidence of disease in up to 45% of patients [146]. In this respect, the observation of an association of more potent immunosuppressive therapy following transplant and earlier and more aggressive recurrence of PBC [147] is also of interest. MMTV replication is regulated in part by a progesterone-responsive glucocorticoid regulatory element in the promoter region, offering an alternative explanation for the female preponderance seen with PBC [148].

These findings attracted attention in the field and were acknowledged by other investigators who pointed out the need for clinical trials with antiretroviral therapies [149]. Subsequently, in a small non-randomized pilot study, therapy with Combivir (lamivudine + zidovudine) improved inflammatory scores, normalized AP, and reduced bile duct injury in patients with PBC [150]. These findings await confirmation in a randomized, controlled trial.

Unfortunately, major findings of the outlined in vitro studies could not be reproduced by independent groups, and others raised concerns that these findings might mainly reflect contamination or technical artifacts [151]. In an independent study, a large number of sera of PBC patients and healthy controls did not show reactivity against MMTV encoded protein, and no detectable immunohistochemical or molecular evidence for MMTV was found in liver specimens or peripheral blood lymphocytes [152]. It was also speculated that beneficial effects of antiretroviral therapy could be partly explained by anti-apoptotic properties of nucleoside analogs [151, 153]. Furthermore, mechanisms by which human betaretrovirus would enter human cholangiocytes are also not identified.

A more recent study, however, strengthened the case for involvement of retroviruses in (immune-mediated) liver disease: Sera of 179 patients with diverse chronic liver diseases and 31 controls were tested for reverse transcriptase activity and presence of human betaretrovirus by polymerase chain reaction. Reverse transcriptase activity was detected in 73% of autoimmune hepatitis patients, 42% of PBC subjects, 35% of patients with viral hepatitis, 22% of liver patients without viral or autoimmune pathogenesis (non-alcoholic fatty liver disease and alcoholic liver disease), and 7% of control subjects. In polymerase chain reactions, 24% of PBC samples were positive for human betaretrovirus compared to 13% in autoimmune hepatitis, 5% in other liver diseases, and 3% in non-liver disease control subjects [154]. If these data can be confirmed, a retroviral compound in the pathogenesis of immune-mediated and viral liver disease seems attractive, though not specific for PBC. Thus, despite some intriguing findings, the pathogenetic relevance of retroviruses in the development of PBC remains enigmatic.

Appendectomy, other abdominal surgeries, and tonsillectomy were significantly more frequently reported in patients with PBC in an epidemiological study in North America [155]. However, an earlier population-based case control study conducted in England did not show such associations [156]. More recently, another case–control study provided evidence that there was at least no association between PBC and the occurrence of appendectomy and pointed out the selection bias present in the previous study done in North America [157]. The linkage to appendectomy was theoretically attractive, since a PBC-specific immune response to the highly conserved caseinolytic protease P of Yersinia enterocolitica in 40% of patients with PBC was reported [158]. It is noteworthy, that infection with Y. enterocolitica is one of the major causes of acute terminal ileitis mimicking acute appendicitis [159].

Cholestasis leads to retention of bile salts and other potentially toxic constituents of bile not only systemically but particularly in hepatocytes. This promotes hepatocellular damage and, in a chronic state, can induce the development of fibrosis and cirrhosis of the liver.

As recently summarized by Paumgartner and Pusl [225], the general biochemical and physiologic principles in treating cholestatic liver disease can be broken down to “(1) reduction of hepatocellular uptake of bile salts and other organic anions; (2) stimulation of the metabolism of hydrophobic bile salts and other toxic compounds to more hydrophilic and less toxic metabolites; (3) stimulation of orthograde secretion into bile and (4) stimulation of retrograde secretion of bile salts and other potentially toxic cholephils into the systemic circulation for excretion by the kidney; (5) protection of injured cholangiocytes against toxic effects of bile salts; (6) inhibition of apoptosis caused by elevated levels of bile salts; and (7) inhibition of fibrosis” [225].

Treatment with UDCA addresses most of these postulated principles as recent clinical and experimental data suggest [224, 226, 227]. Still, no relevant effect of UDCA on (1) bile salt uptake and (2) bile salt metabolism [228, 229] has been demonstrated in man.

UDCA is currently considered the mainstay of therapy for PBC [28, 29]. Randomized, double-blinded, placebo-controlled trials have consistently shown that UDCA, administered today in standardized doses of 13 to 15 mg/kg/day, improves serum biochemical markers including bilirubin [215‐218, 220], an important prognostic marker in PBC [267]. A number of studies demonstrated an improvement of histological features by UDCA. In a combined analysis of four clinical trials including a total of 367 patients [268] and in an independent in 103 patients [221], UDCA therapy was associated with delayed progression and a marked decrease in progression rate from early-stage disease to late histologic stages. Despite these effects on disease progression and severity, the assessment of the effect of UDCA treatment on long-term survival has been more difficult. A combined analysis of three randomized controlled trials including 548 patients in total treated with UDCA for up to 4 years revealed improved survival free of liver transplantation in patients with moderate or severe disease [222]. Another combined analysis of five studies, all with a follow-up of at least 4 years, quantified the reduction in the risk of death or liver transplantation in patients treated with UDCA to 32% [269]. A long-term follow-up study in a cohort of 225 patients, in part overlapping with the aforementioned analyses, reported 10-year survival without liver transplantation to be significantly higher in UDCA-treated patients compared with survival predicted by the Mayo model [270]. Recent long-term trials from France, Spain, and The Netherlands have shown comparable results [11‐13, 271]. The survival rate of patients in early stages of the disease who biochemically responded to therapy was similar to that in the control populations. “Response” here is defined as a decrease in AP to <40% of pretreatment levels or normalization at 1 year (Barcelona criteria) [12] or serum bilirubin <1 mg/dl, AP ≤ 3 × N and AST ≤ 2 × N after 1 year of UDCA treatment (Paris criteria) [13, 14]. An additional non-randomized controlled study from Greece confirmed that in most patients with PBC treated with UDCA (particularly those who are in early stage of disease), the 10-year survival is comparable to that in the general population [272]. Despite these intriguing findings, the beneficial effects of UDCA on survival have repeatedly been questioned: A large randomized Swedish trial failed to confirm an effect of UDCA on disease progression and survival at a dose of approximately 8 mg/kg/day [273]. It was concluded from this and other studies that doses of UDCA lower than 10 mg/kg/day are of little benefit. In the follow-up of patients in a US [274] and a Canadian trial [217], UDCA had no significant influence on incidence of endpoints liver transplantation or death. Also, a meta-analysis of 11 randomized trials could not confirm a significant effect of UDCA on survival and incidence of liver transplantation [275]. In this meta-analysis, however, six studies with only 2 years of follow-up were included, as were two studies administering UDCA at low doses of 10 mg/kg/day or less. Other meta-analyses suffered from similar shortcomings [276, 277] and may therefore have missed a beneficial effect of UDCA. Accordingly, meta-analyses which only included long-term trials with follow-up of at least 2 years and those using an effective dose of UDCA of more than 10 mg/kg/day verified that treatment with UDCA significantly improves quality of life and transplant-free survival and delays histologic progression in early-stage patients [278, 279]. Current guidelines therefore recommend to treat PBC with UDCA using doses of 13 to 15 mg/kg/day and to start treatment early [28, 29, 280].

About two thirds of patients treated according to these recommendations respond adequately as defined by the Barcelona or the Paris criteria and may have a normal life expectancy. The most recent study from The Netherlands comprising 375 patients with a mean follow-up of 9.7 years again stressed the importance of early treatment showing a clear survival benefit for patients treated in early stages of disease with normal serum bilirubin and albumin levels at the start of therapy [14].

For the remaining one third of patients who fail to achieve biochemical response according to Paris and Barcelona criteria or who are at an advanced histological stage at start of medical treatment, therapeutic options are limited to date and novel approaches are needed.

UDCA is an accepted treatment of cholestatic pruritus in ICP (intrahepatic cholestasis of pregnancy). However, its effect on pruritus is variable in PBC [217]. At present, no convincing data for an antipruritic effect of UDCA in PBC are available [298].

Both peripherally acting pruritogens and central nervous dysfunction have been implicated in the pathogenesis of cholestatic pruritus [9]. According to these two concepts, most therapeutic interventions currently under study for cholestatic pruritus are either directed towards elimination of so far undefined pruritogens or modulation of central neurotransmission. The pharmaceutic options currently recommended include [29]: (1) anion exchange resins, such as cholestyramine and colestipol, which bind anions and amphipathic molecules and reduce their intestinal absorption and systemic accumulation. Despite extensive clinical experience, suggesting a beneficial effect in up to 90% of patients, no large clinical trials have evaluated the efficacy of exchange resins [298]. In patients lacking adequate improvement, treatment with (2) rifampicin may be helpful [298‐301]. Rifampicin is a semisynthetic antibiotic and as a potent pregnane X receptor (PXR) agonist leads to the induction of hepatic microsomal enzymes. Thereby, it may promote metabolism of potential pruritogens. As a third-line option, (3) opioid antagonists (naloxone, nalmefene, naltrexone) have been found to reduce itch severity in patients with PBC [298, 302‐304]. (4) The selective serotonin reuptake inhibitor sertraline [305, 306] has been reported to improve cholestatic pruritus in small clinical trials and has most recently been considered as a fourth-line treatment option [29, 280]. Other experimental approaches include 5-hydroxytryptamine receptor type 3 antagonists, cannabinoids, subhypnotic doses of propofol, plasmapheresis, albumin dialysis, and nasobiliary drainage in desperate cases, although adequate trials are lacking [29, 63]. Liver transplantation should be considered in serious cases in which all other strategies have failed, even if liver function is still conserved [29].

Specific medical therapies of fatigue associated with chronic cholestasis have not yet been defined. UDCA treatment seems to have no or limited beneficial effects on fatigue in PBC [217, 307]. Oral supplementation with antioxidants showed promising results in a pilot study, but had no beneficial effect in a randomized, placebo-controlled, crossover trial [308]. Independent of the underlying disease, altered serotoninergic neurotransmission has been implicated in the development of fatigue [309]. The 5-HT3 serotonin receptor antagonist ondansetron, however, did not significantly reduce fatigue compared with placebo in a randomized, placebo-controlled trial [310]. The selective serotonin reuptake inhibitors fluvoxamine and fluoxetine also failed to exert beneficial effects on fatigue in this patient group [311, 312].

In a series of 21 PBC patients with excessive daytime sleepiness [313], the centrally acting agent modafinil was investigated in an open-label study [314]. Significant improvement was seen in Epworth Sleepiness Scale scores and fatigue severity as assessed by PBC-40 fatigue domain score, but the drug had to be stopped due to side effects in a considerable part of the patients before the end of the study.

In contrast to pruritus, fatigue may not improve significantly following liver transplantation.

Independent of the etiology, bone mineral loss is a recognized complication of cholestatic liver diseases. In PBC, it moderately increases absolute and relative fracture risk compared to the general population [315, 316]. Treatment of the underlying liver disease with UDCA does not prevent bone loss in PBC [317, 318]. Calcium and vitamin D supplementation is usually recommended in osteoporosis, although their role in preventing osteoporosis and fracture in chronic liver disease is unclear [319, 320]. Alternative interventions to prevent and treat osteoporosis in liver disease have only been tested in a few studies. Parenteral calcitonin, tested in one controlled trial for 6 months, was ineffective in halting bone loss in patients with PBC [321], whereas hormone replacement therapy was demonstrated to improve vertebral bone density in PBC [319, 322, 323]. In large clinical trials, however, hormone replacement therapy was associated with serious side effects, including the risks of breast cancer, coronary artery disease, and thromboembolism, limiting its use for the prevention of fractures in osteoporosis [324‐326].

Use of bisphosphonates is controversial. In a placebo-controlled trial, cyclical etidronate did not significantly improve bone density [327]. Also, a 4-year treatment with clodronate plus calcium/vitamin D did not improve osteopenia in women with PBC in a prospective study [328]. In two controlled trials however, alendronate significantly improved spine and femoral bone mineral density compared with placebo and had greater antiresorptive activity than etidronate [329, 330].

According to current guidelines [29], supplementation with calcium (1,000–1,200 mg/day) and vitamin D (400–800 IU/day), though not evidence-based, should be considered in all patients with cholestatic liver disease. Alendronate or other bisphosphonates are recommended at a T score < −2.5 (DEXA) or following pathological fracture.