Primary biliary cholangitis (PBC) is a chronic cholestatic liver disease characterized by immune-driven destruction of small intrahepatic bile ducts leading a proportion of patients to hepatic failure over the years. Diagnosis at early stages in concert with ursodeoxycholic acid treatment has been linked with prevention of disease progression in the majority of cases. Diagnosis of PBC in a patient with cholestasis relies on the detection of disease-specific autoantibodies, including anti-mitochondrial antibodies, and disease-specific anti-nuclear antibodies targeting sp100 and gp210. These autoantibodies assist the diagnosis of the disease, and are amongst few autoantibodies the presence of which is included in the diagnostic criteria of the disease. They have also become important tools evaluating disease prognosis. Herein, we summarize existing data on detection of PBC-related autoantibodies and their clinical significance. Moreover, we provide insight on novel autoantibodies and their possible prognostic role in PBC patients.

Primary biliary cholangitis (PBC), known until 2014 as primary biliary cirrhosis, is a chronic autoimmune cholestatic liver disease, its main features being presence of anti-mitochondrial antibodies (AMA), female predominance and progressive destruction of small intrahepatic bile ducts[1]. A proportion of PBC patients progresses over the years to fibrosis and eventually to cirrhosis leading to hepatic failure. Diagnosis at early stages in conjunction with ursodeoxycholic acid (UDCA) treatment has been linked with prevention of disease progression in the majority of cases. Over the years, significant progress has been achieved in the armamentarium used for disease diagnosis and prognostication, including autoantibody testing and on-treatment prognostic indicators[1].

Originally PBC was reported in 1851 in a woman with jaundice and xanthomata and its clinical phenotype was described in 1949[2,3]. The first association of PBC with autoantibodies was reported by Mackay[4] in 1958 in a case with high titers of circulating complement-fixing antibodies to liver, kidney and other human tissue antigens[4]. Later on, in 1965, Walker et al[5] have reported the presence of AMA by indirect immunofluorescence (IIFL) in patients with PBC[5].

The discovery of AMA in conjunction with advances in their diagnostic testing and increased disease awareness has led to diagnosis at earlier stages. Moreover, treatment with UDCA, which is the standard of care for naïve PBC patients has been associated with improved long-term survival. All these factors have significantly contributed to the increase in prevalence rates throughout the years[6,7]. The highest incidence and prevalence rates have been reported in Europe and the United States (incidence and prevalence rates range from 3.3 to 32 per million person-years and 19 to 402 per million respectively)[8-11]. Discrepancies between regions may reflect differences in patients’ accessibility to healthcare services, increased disease awareness depending on physician’s expertise and widespread use of AMA and antinuclear antibody (ANA) testing during diagnostic work-up. Still, it is debatable whether increases in prevalence and incidence rates represent true changes over the years.

PBC is considered a prototype autoimmune disease, based on the abundance of AMA, female predominance and increased rate of other autoimmune diseases. Genetic and environmental factors are regarded as key players in the induction of immune tolerance loss to biliary epithelial autoantigens, a notion well appreciated also on work performed in animal models of the disease[12-15].

Higher disease rates between family members, especially siblings, formed the initial pathogenetic link between genetics and PBC. Genome wide association studies from Europe, North America, Japan and China have identified human leukocyte antigens (HLA) and non-HLA alleles that confer susceptibility to PBC, though discrepant results between ethnicities are apparent[16]. The majority of reported loci are encoding proteins implicated in the control of immune response mechanisms, including HLA, interleukin (IL)-12 production, B and T cell activation, interferon (IFN)-γ production and production of immunoglobulins. The contribution of genetics is also emphasized by the fact that monozygotic twins display significantly higher concordance rates compared to dizygotic twins (63% vs 0%)[17]. Still, lack of remarkable concordance rates between identical twins in PBC strongly supports the implication of environmental factors and epigenetics.

Numerous studies have revealed the important contribution of environmental exposures, including chemical xenobiotics, pollutants, cosmetics and also infectious agents, in immune tolerance loss and in the initiation of disease process[18-21].

Studies on animal models of the disease have underlined the likely input of specific environmental factors in the initiation and perpetuation of disease progression, adding further support for their significant input in genetically prone individuals[15].

The cornerstone for PBC diagnosis is the detection of AMA and/or PBC-specific ANA when investigating a patient with cholestasis (Table 1)[6]. It is important to stress that AMA are one of the few autoantibodies being included in diagnostic criteria of a disease[6]. Liver histology is essential when AMA are not detected in a patient with cholestasis with high suspicion of PBC, or when indications of variant forms exist[6,22].

AMA is the most characteristic feature of PBC, as up to 95% of patients are tested positive for these autoantibodies. A 2014 meta-analysis including 24 studies showed that the pooled sensitivity and specificity of AMA in the diagnosis of PBC is 84.5% and 97.8%, respectively[58]. The specificity of AMA for PBC has been initially revealed in two small longitudinal studies, where the majority of AMA positive patients with no evidence of cholestatic liver disease developed full blown PBC[59,60]. Still, recent studies have shown a small proportion of AMA positive individuals to develop PBC through the years[61,62]. These data should be interpreted with caution, as follow up duration might have been not long enough in these latter studies to allow clinical presentation of PBC that is known to progress slowly.

Data from Sun et al[63], though, have shown that 80% of AMA positive individuals with normal alkaline phosphatase have histological features of PBC, stressing that cholestasis is not a prerequisite for the establishment of PBC[63]. The pathogenic role of AMA is further supported by the case of two newborns where liver disease had developed after transfer of AMA from their mothers via the placenta, while liver pathology subsided when the autoantibodies disappeared.

However, AMA have been reported in up to 1% of healthy individuals[64,65]. Considering the substantially lower PBC prevalence, only a small percentage of those AMA positive individuals is going to advance to PBC. Moreover, longitudinal data on AMA kinetics in patients transplanted for PBC show these autoantibodies to persist, though biochemical or histological features of PBC recurrence have been reported in 36% 10 years post liver transplantation[66,67]. As autoantibodies can arise years before disease presentation only long-term observational studies spanning a period of decades will be able to elucidate this issue further.

Kisand et al[68] have suggested that development of PBC might arise in those who have or will develop high titer AMA overtime of various specificities and subclasses compared to AMA positive individuals with low titer antibodies of only one Ig class reactivity[68].

Whether AMA titers is a prognostic indicator for PBC needs to be assessed further, as several studies up to the present have produced conflicting data. Early studies from the 80s and 90s have shown AMA titers to correlate to disease activity and progression[69,70]. Another study demonstrated that PBC patients had significantly higher AMA titers, tested by IIFL and higher anti-PDC-E2 avidity compared to AMA positive individuals with normal biochemistry[71]. Moreover, Gabeta et al[39] have demonstrated IgG and IgA AMA titers to positively correlate with the Mayo risk score[39].

Whether testing by IIFL for individual IgG AMA subclasses could assist in identifying prognostic features of PBC patients remains obscure. Another study stressed that AMA of the IgG3 subclass positively correlate with more advanced liver disease, as manifested by higher frequency of cirrhosis and higher Mayo risk score[36].

Furthermore, treatment with UDCA was associated with a decrease in AMA titers in one of the first trials of UDCA in PBC patients, while in a recent study from China, response to UDCA treatment at 1 year was linked to decreased AMA titers[72,73]. However, several other studies failed to prove AMA titers or their longitudinal changes as prognostic markers of PBC progression[74-76]. In line with this, a small study on 9 asymptomatic PBC patients with inadequate response to UDCA, who continued with combination of UDCA and fenofibrate, showed a decrease in AMA titers in 4 of these patients. These data could suggest that AMA production may be regulated by peroxisome proliferator-activated receptor α in PBC patients. Still, this is a hypothesis that needs to be confirmed in the future, as it was not reproduced by other studies[77].

Anti-M2 AMA can also occur in the context of overlapping diseases, such as the autoimmune hepatitis/PBC variant, metabolic associated fatty liver disease/PBC variant, viral hepatitis (hepatitis B virus, hepatitis C virus)/PBC variant and other rheumatic diseases, such as systemic sclerosis[6,78,79].Their presence should be evaluated in the appropriated clinical context.

Various ANA specificities have been reported in up to 70% of PBC patients[80]. HEp-2 cells are preferred compared to the triple tissue substrate for the detection of ANA, as large nuclei and high rate of mitosis of these cells permit the discrimination of different staining patterns[34]. A variety of staining patterns, often coinciding, have been described in PBC patients, including rim-like (RLM), nuclear dot, speckled, homogeneous and centromere staining pattern[81,82]. For diagnostic purposes, ANAs are categorized into those not specific for PBC and those specific for the disease.

According to the official categorization of ANA patterns issued by the International Consensus on ANA Patterns initiative, 2 IIFL patterns are defined as PBC-specific, namely the multiple nuclear dot (MND) pattern (AC-6) and the punctuate nuclear envelope pattern (AC-12)[83].

Several studies have evaluated whether PBC-specific ANAs have a prognostic role in PBC patients. Discrepant results between studies can be justified by differences in study design (i.e. methods to evaluate the presence of antibodies, different antigenic preparations used), as well as heterogeneity of patients investigated[86,94].

Early studies, based mainly on IIFL detection of autoantibodies, had failed to establish a link between presence of PBC-specific ANAs and disease’ prognosis[99,100]. During the last 20 years, several reports relaying mostly on ELISA, western blot and line blot data, delineated the correlation of PBC-specific ANAs with more severe disease and worse outcome, though differences between studies exist. Most studies underline that positivity for anti-NPC antibodies and especially for anti-gp210 are associated with more severe disease and worst disease outcome[81,87,91,92,97,101].

One of these studies using IIFL for the identification of autoantibodies, revealed anti-RLM of the IgG3 isotype to correlate with more severe liver disease[91]. A meta-analysis encompassing 5 studies in Asian patients demonstrated anti-gp210 antibodies to be associated with worst disease outcome and suggested their use as optimal predictors of PBC prognosis at the time of diagnosis[102].

Nakamura et al[103] reported that persistently positive anti-gp210 antibodies during follow up in Japanese PBC patients under UDCA treatment confer a strong risk for progression to end-stage liver disease, whereas these patients had histologically more severe interface hepatitis, lobular inflammation, and ductular reaction[103]. These findings were not confirmed in a study assessing autoantibody patterns during follow up in Greek PBC patients[104].

Anti-np62 antibodies are detected in 22%-31% οf PBC patients, though their significance is not well documented and their diagnostic relevance remains unclear[105,106]. A Japanese study suggested that anti-np62 antibodies are associated with advanced disease. These results need to be confirmed by other studies[105].

Anti-sp100, the most frequent amongst antibodies displaying the MND pattern, is detected in 8%-44% of PBC patients, its specificity ranging from 63.8% to 100%[81,107-109]. The pooled sensitivity of anti-sp100 for PBC has been estimated at 23.1% and the pooled specificity at 97.7%[98]. Importantly, subgroup analysis could not identify significant differences in pooled specificity across the strata of geographical regions or across various methods used for their identification[98].

The prevalence of anti-lamin B receptor antibodies have been reported in 9%-15% of PBC patients and are considered highly specific[105,110]. Given their rarity and their unknown prognostic role, they are not routinely tested in every day clinical practice.

Anti-MND and anti-sp100 antibodies tested by IIF and by ELISA respectively have been associated with more severe disease, as attested by biochemical and histological features, faster disease progression and worse outcome[91,107]. Moreover, anti-MND of the IgG3 isotype was associated with longer disease duration and more severe histological picture[91].

Autoantibodies against PML are less frequently detected in PBC patients (12%-19%), and largely co-exist with anti-sp100, though cases with single reactivity to PML have been also reported[93,107]. Double reactivity to sp100 and PML was suggested to be associated with unfavorable outcome of PBC patients[93,107]. SUMOs have been exclusively linked to presence of anti-sp100 and anti-PML, suggesting antigen spreading as a possible mechanism for anti-SUMO generation[90].

Züchner et al[107] reported anti-sp100 levels to remain stable during the course of the disease. Of interest, alteration in the sp100 epitope recognition pattern in some patients during the natural course of the disease was noted in a proportion of patients under UDCA treatment, which indicates that UDCA can modify immunoglobulin expression[107]. In accordance with these findings are those from a study in 110 Greek patients, where serial determination of autoantibodies titers demonstrated that a decrease in anti-sp100 levels was associated with response to UDCA treatment and improvement of the Mayo risk score[104]. Akin to AMA, PBC-specific ANA are shown to persist after liver transplantation without signs of disease recurrence[106,111].

Few studies have suggested infectious agents to have acted as potential triggers for the development of anti-MND antibodies in PBC patients. Bogdanos et al[112] have demonstrated a strong correlation of anti-sp100 reactivity and presence of AMA only in those PBC patients with recurrent urinary tract infections[112]. Molecular mimicry between T-cell epitopes, first implicating mitochondrial and afterward nuclear proteins, has been proposed[113].

A study conducted in a Han Chinese population demonstrated significant genetic predisposition for sp100 but not for gp210. In detail, HLADRB1*03:01, DRB1*15:01, DRB1*01, and DPB1*03:01 alleles were associated with antibody production against sp100[114]. Whether this difference is indicative of diverse roles of anti-sp100 and anti-gp210 in PBCs pathogenesis needs to be elucidated in the future. Besides, as data of this study is confined to Chinese patients, these results need to be confirmed in other PBC cohorts.

The percentage of AMA negativity by IFL widely varies amongst studies, ranging from 5%-20%. Soon after the identification of the molecular targets of AMA and the development of molecular based assays, it has become apparent that such assays can considerably increase the sensitivity for AMA detection, leading to a significant drop of “true” AMA-negative cases[37,39,40,45,46,115]. By immunoblotting testing, IgG antibodies against PDC-E2 are detectable in 97% of AMA positive PBC and in 66% of AMA negative PBC[116].

Others have reported even higher prevalence of antigen-specific AMA in IFL AMA-negative PBC patients, reducing even more the percentage of AMA negative PBC[117-119]. In our hands, only 16% of IFL AMA-negative PBC patients had reactivity against mitochondrial autoantigens using a sensitive immunoblot technique[55].

Nevertheless, even when the most sensitive AMA tests are used, still a proportion of PBC patients may not have AMA. While some studies indicated that seronegative cases may convert to AMA seropositive over the years, some others have elegantly shown that there are still patients that will never develop these autoantibodies over the course of the disease. This is of fundamental importance as it implies that breaking of immunological tolerance to the mitochondrial autoantigens of AMA is not the sole cornerstone of PBC pathogenesis, and other non-mitochondrial antigen-driven mechanisms are potentially involved. It is also of diagnostic importance because it highlights the importance for the proper identification and routine use of diagnostic surrogate markers which can assist the diagnosis of PBC in AMA negative cases. It also raises the question as to whether AMA positive and AMA negative PBC are two entities with distinct characteristics or not.

It must be noted, that over the years and with the wealth of data provided, a consensus has been reached that “true” AMA-negative PBC is an entity indistinguishable from AMA-positive PBC, in terms of demographic, biochemical, clinical and histopathological features[117,118,120]. Though the total number of AMA negative cases with PBC included in various studies can be considered small, and despite the lack of multi-centre studies investigating in greater detail the features of AMA negative PBC, the current evidence supports the notion that the natural history as well as the prognosis of AMA-negative PBC is analogous to AMA-positive PBC[121,122]. Current guidelines indicate the therapeutical and clinical management of AMA negative PBC should be the same with AMA positive PBC. Early studies have reported that aside AMA status, some immunoserologic features may be different between AMA-negative and AMA-positive PBC, such as lower IgM and higher gamma globulin in the former rather than the latter group[117,118,120,123].

Several studies have published data reporting a significantly higher rate of positivity for ANA in AMA-negative compared to AMA-positive PBC. Original studies were based on IFL testing, while most recently such assays included ANA-specific antigen testing against sp100 and gp210. The wide range of ANA positivity is attributed to technical reasons, such as the sensitivity of the assay used for autoantibody detection and cohort selection bias or geographic/ethnic disparities. Early IFL studies reported the prevalence of ANA to range between 71% to 100%, in AMA-negative PBC patients, compared with 18% to 33% in AMA seropositive PBC patients[117,118,120,123].

The first point that at times is overlooked by the clinicians is that PBC-specific ANAs are not confined to AMA-negative PBC and therefore disease-specific ANA is by no means a characteristic feature of this form of the disease. It has now become apparent that this striking over-representation was due to the practical limitations imposed by IFL on tissue sections because ANA positivity could be obscured by the simultaneous presence of AMAs, which could perplex the reading by the immunodiagnostician. This is indeed true in week AMA positive cases were ANAs become more visible and can be present in frequencies comparable to AMA negative PBC cohorts[81]. With the advent of assays using as antigenic substrate recombinant sp100 and gp210, a more accurate estimation of the respective autoreactivities has been achieved, though great variation is still seen amongst studies. In a consecutive cohort, Muratori et al[81] found ANA of any pattern in 53% of patients, including 27% with anti-Sp100, and 16% with anti-gp210 antibody reactivity[81].

Most studies fail to report significant differences of anti-gp210 and anti-sp100 between AMA-positive and AMA-negative PBC cohorts[124].

The relatively recent identification of novel autoantigens in AMA negative patients with PBC has provided further hints regarding the diagnostic and clinical challenges of AMA-negative patients with PBC and their relevance to the presence of disease-specific autoantibodies.

During the last decade, antibodies against kelch-like protein (KLHL12) and hexokinase 1 (HK-1) were proposed as novel biomarkers in PBC patients, especially those lacking AMA[125,126]. While KLHL12 is a nuclear protein that is implicated in collagen export and ubiquitination of various proteins, HK-1 is an enzyme that is localized in the outer mitochondria membrane, phosphorylates glucose and modulates susceptibility to cellular apoptosis. Both display high specificity (≥ 95%), while their combination improved their overall sensitivity from 48.3% to 68.5% by ELISA and from 55% to 75% by immunoblot in AMA-negative PBC patients[125]. Recently, a study in Spanish PBC patients has confirmed the aforementioned data on the utility of these autoantibodies to diagnose AMA-negative patients by demonstrating anti-HK1 or anti-KLHL12 in a third of AMA negative and in 40% those negative for AMA, anti-gp210 and anti-sp100[127]. Furthermore, anti-HK1 positivity was associated with significantly higher possibility of liver decompensation and lower liver-related survival free of transplantation[127]. A more recent study assessed the prevalence of anti-KLHL12 and anti-HK-1 antibodies by ELISA at 5 sites within Europe and North America and documented the presence of these antibodies in patients with PBC in all geographies, irrespectively of the origin of the sera failing to identify geographic factors which could imply that the genetic make-up of the patients is responsible for the induction of these autoantibodies[128]. A Polish study reported the highest prevalence of anti-KLHL12 antibodies so far (36%) in both AMA-positive and AMA-negative patients and confirmed their high specificity for PBC diagnosis[110]. Their presence was linked to higher bilirubin levels and advanced fibrosis[110]. The prognostic significance of these autoantibodies needs to be evaluated in larger cohorts of PBC patients.

Bombaci et al[129] have proposed SPATA31A3 and GARP, as novel autoantigens in PBC, as PBC sera demonstrated high reactivity to these antigens irrespective of AMA and PBC-specific ANA presence[129]. Their combination with PDC-E2 assisted in discrimination of PBC from other diseases with high sensitivity and specificity. The authors offer in addition pathophysiological evidences on GARP expression in human cholangiocytes, which underlines the potential implication of this autoantigen in the induction of the disease[129].

Considering that the repertoire of autoantibodies related to several autoimmune diseases is increasing, multiplex assays have become available in order to facilitate their simultaneous detection in a reliable, automated fashion, without consuming substantial amount of time. Such an example is that reported by Liu et al[130] in 2010 on a new PBC-screening assay that allows simultaneous detection of MIT3, sp100 and gp210 of IgG and IgA isotype by ELISA and compared these results with those from distinct IgG ELISAs[130]. Testing of 1175 PBC patients and 1232 non-PBC controls revealed sensitivity and specificity of PBC screen to be 83.8% and 94.7%, respectively, which is similar to that found from the evaluation of the combined data from individual MIT3, sp100, and gp210 IgG ELISAs[130].

Furthermore, assessment of 253 AMA-negative PBC patients by IIFL with this assay, resulted in positivity for PBC-specific autoantibodies in almost half of them (44.7%). Based on these data, PBC screen has been proposed as a reliable first line test for the diagnosis of PBC[130]. Comparable were the results from an Italian study, where examination of 100 AMA-negative PBC patients by IIFL showed reactivity in 43% of these patients with the use of the PBC screen[131].

Data from a study using a multiplex line blot assay, designed for autoimmune liver diseases, that contained AMA-M2, M2-E3 (a recombinant fusion protein including the E2 subunits of PDC, BCOADC and OGDC), as well sp100, PML and gp210 recombinant proteins had an overall sensitivity and specificity for PBC of 98.3% and 93.7% respectively. IIFL displayed lower sensitivity (86.2%), though comparable specificity (97.9%)[132]. Recently, a new automated PMAT assay on the Aptiva instrument, which includes MIT3, sp100, gp210, HK1 and KLp, facilitated the recognition of higher frequency of AMA- negative PBC patients compared to conventional immunoassays[57]. Even though multiplex technology seems to be a promising tool in the diagnosis of PBC, its use in the diagnostic algorithm of PBC needs to be further evaluated.

In conclusion, thorough discussion of the available literature reveals important aspects of the diagnostic, clinical and prognostic significance of disease-specific autoantibodies in AMA positive and AMA negative PBC patients[133]. Overall, identification of AMA and PBC-specific ANA is the cornerstone for the diagnosis of PBC. PBC-specific ANA, including anti-gp210 and anti-sp100, have a pivotal role in PBC diagnosis in AMA-negative individuals with high suspicion of the disease. Moreover, they have an established role as predictive factors for more advanced disease and worse outcome. The role of novel autoantibodies as diagnostic and prognostic tools in the management of patients with PBC needs to be assessed further.