The demonstration of antimitochondrial antibodies (AMAs) has been a serological hallmark for the diagnosis of primary biliary cirrhosis (PBC) since the mid-1960s.1 They can be demonstrated by an immunofluorescence test (IFT) using cryostat sections from rat or mouse organs showing typical bright uniform cytoplasmic staining of the entire cytoplasm in distal renal tubules, gastric parietal cells, thyroid epithelial cells and a more granular staining on cardiac muscle.2 3 These antibodies have been shown to react with an inner mitochondrial membrane antigen which has been named “M2”4 5 and consists of five subunits: M2a (70 kDa), M2b (56 kDa), M2c (45 kDa), M2d (36 kDa), and M2e (31 kDa).6 Subsequently, the target antigens have been identified as subunits of the 2-oxoacid dehydrogenase complex (OADC), namely the E2 component of the pyruvate dehydrogenase complex (PDC-E2) corresponding to M2a, the protein X or E3 binding protein of PDC (M2b), the 2-oxoglutarate dehydrogenase complex (OGDC) and branched-chain 2-oxoacid dehydrogenase complex (BCOADC) corresponding to M2c, and the E1α and E1β subunits of PDC (M2d, e).7^–10 Applying OADC-related antigens in sensitive methods such as fluorimetric immunoassay, enzyme-linked immunoassay (ELISA; including bead assays) or western blotting, the relevant antibodies can be detected in about 95% of patients with PBC.6 11^–13 In most patients (about 90%) antibodies to M2a (PDC-E2) can be detected either alone or in association with antibodies to at least one of the other subunits. About 4% of PBC sera react exclusively with the OGDC. Since the M2 antigen prepared from beef heart mitochondria contains all the relevant 2-OADC enzyme components its application in ELISA and western blot allows the detection of PBC-specific anti-M2 antibodies with high sensitivity and specificity. In contrast, using only recombinant PDC-E2 or commercially available 2-OADC components, antibodies against the unusual components may be missed.

However, in the last few years we have observed an increasing incidence (15–20% of patients with newly diagnosed PBC) of sera from PBC patients which show the typical AMA pattern in the IFT but are negative for all subunits of the OADC when tested by ELISA and western blotting against the M2 antigen and commercially available OADC component. Therefore, we postulated that further, hitherto undefined, AMA specificities might exist. Using mitochondrial sub-fractions from different organs in ELISA and western blot we found that these sera predominantly reacted with two proteins of molecular weights 60 and 80 kDa present in a 100 000 g supernatant from bovine heart mitochondria, which could be enriched by discontinuous sucrose density gradient centrifugation accumulating at densities of 1.14–1.16.14 The aim of the present study was, therefore, the further analysis of the specificity of this reaction and the identification of the target antigen(s).

PATIENTS

Anti-M2 negative and anti-M2 positive patients with primary biliary cirrhosis

Between 2002 and 2007 we received sera from 20 140 patients with chronic liver diseases from different hospitals and physicians in Germany for serological diagnosis or exclusion of an underlying autoimmune process. In 1850 of them AMAs were present in the immunofluorescence test (IFT), 1609 were additionally anti-M2 positive by ELISA and western blotting, ie, 241 (13%) had AMAs only by IFT but were anti-M2 negative.

Sera were defined as AMA positive but “anti-M2 negative” when they showed the typical AMA pattern in the IFT but reacted neither by ELISA nor by western blotting with any of the M2-related antigens (the M2 antigen prepared from beef heart mitochondria containing all five antigens of the 2-oxoacid dehydrogenase complex6 and the commercially available pyruvate dehydrogenase (PDC) and 2-oxoacid dehydrogenase complex (2-OAGDC) (see below)). Sera were labelled “anti-M2 positive” when they reacted with M2 in the ELISA and western blot (sera reacting with commercially available PDC or 2-OAGDC but not with the M2 fraction did not exist).

Only 51 of the 241 AMA positive/anti-M2 negative patients could be selected for further analysis based on the following criteria: availability of (1) sufficient amounts of serum for further serological investigations; and (2) complete clinical and laboratory parameters as well as histological data as far as possible (group 1 patients). Ten of the sera were kindly analysed by Dr E Gershwin, Davis, California, and confirmed to be anti-M2 negative.

In 27 of the 51 patients PBC had been confirmed histologically revealing lesions either typical or compatible with PBC (group 1a, “true” PBC). Twenty-six of them had features of early PBC (granuloma formation, bile duct proliferation or destruction); in one patient PBC stage IV was diagnosed. In a further 10 of the 51 patients with AMA positive/anti-M2 negative chronic cholestatic liver disease liver biopsy had either not been performed (n = 7) or revealed features of steatosis (n = 1), chronic active hepatitis (n = 1) or no pathological alterations (n = 1) (group b, “probable PBC”). Besides the AMAs the remaining 14 patients had PBC-specific anti-nuclear antibodies (ANAs) such as antibodies to nuclear dots (sp100; n = 10), antibodies to nuclear membrane (gp210; n = 1), antibodies to both, sp100 and gp210 (n = 1) or antibodies to centromeres (n = 1)15 (group 1c, “ANA positive/AMA positive/anti-M2 negative PBC”).

Clinical and biochemical data of these three groups of patients are given in table 1. Comparing the three groups, no significant differences were observed, ie, in this respect patients with “probable” PBC resembled patients of group 1a or 1c.

Seventeen of the patients of group 1a, eight of the patients of group 1b, and six patients of group 1c were untreated: the remaining patients had already received the standard ursodeoxycholic acid (UDCA) therapy.

Furthermore, in the study we included sera from 77 patients with clinically well-defined anti-M2 positive PBC (group 2) who had been seen between 2002 and 2007 at the outpatient department of the Department of Internal Medicine I, University of Tübingen (72 females, five males; mean age 56.5 years, range 33–81 years). Fourteen patients were untreated, 63 patients had received UDCA, and nine of them were treated additionally with low-dose steroids and immunosuppressive drugs due to the association of PBC with autoimmune hepatitis (n = 3), systemic lupus erythematosus (n = 4) or rheumatoid arthritis (n = 2). In 28 patients liver biopsy had been performed (22 patients had PBC stage I/II, six patients had PBC stage III/IV).

Further clinical and laboratory parameters of these group 2 patients are given in table 1. Analysing the data with the Kruskal–Wallis test, a non-parametric test for unpaired groups, no statistical significances were observed. However, comparing group 1a and group 2 with the Mann–Whitney test, there were significant differences with respect to transaminases (table 1).

Control groups

As controls, sera from 87 patients with AMA negative disorders were included:

• 20 patients with autoimmune hepatitis (15 females, five males; mean age 43.9 years, range 6–76 years)

• 20 patients with primary sclerosing cholangitis (five females, 15 males; mean age 47.8 years, range 6–76 years)

• 20 patients with alcoholic liver disease (12 females, eight males, mean age 57.4 years, range 38–80)

• 20 patients with ANA positive collagen disorders (18 females, two males; mean age 43.5 years, range 11–75 years)

• seven patients with rheumatoid arthritis (six females, one male; mean age 56.3 years, range 46–71 years)

Sera from 30 healthy subjects (18 females, 12 males; mean age 44.5 years, range 20–56 years) were also used. All sera had been stored at −20°C. All serological analyses were performed with the understanding that the patients had been informed by their physicians prior to drawing the blood about the storage of their sera and probable further serological investigations.

METHODS

Preparation of antigens

For the immunofluorescence test (IFT) cryostat sections from rat liver, kidney, stomach and heart were used.2 16

Whole mitochondria, sub-mitochondrial particles (SMPs) obtained by sonication of the mitochondria, as well as the supernatant obtained after centrifugation of these SMPs at 100 000 g were prepared from beef heart, rat liver and pig kidney according to standard methods17 and used as antigens.

Discontinuous sucrose density centrifugation was performed as described recently18 using ten density gradients (1.04, 1.08, 1.10, 1.12, 1.14, 1.16, 1.18, 1.20, 1.24 and 1.28). The 100 000 g supernatants were layered onto the top of these gradients and centrifuged for 24 h at 100 000 g.19 Afterwards, gradients were combined resulting in five antigen fractions (G1, <1.04 and 1.04; G2, 1.08 and 1.10; G3, 1.12 and 1.14; G4, 1.16 and 1.18; and G5, 1.20–1.28).

The M2-antigen fraction containing all five determinants corresponding to the 2-oxoacid dehydrogenase complex6 was prepared from bovine heart mitochondria by chloroform release as described by Beechey et al.19

The pyruvate dehydrogenase complex (PDC) and 2-oxo-glutarate dehydrogenase complex (OGDC) (both from porcine heart) were obtained from Sigma-Aldrich (St. Louis, Missouri, USA).

Methods for detection of antimitochondrial antibodies

IFT on cryostat sections was performed according to standardised methods.2 16 Sera were diluted 1:10, and bound antibodies were visualised with fluorescein-conjugated polyvalent goat anti-human IgG, IgA and IgM antibodies at a dilution of 1:1000 (Dako, Hamburg, Germany).

In the ELISA we followed the method described previously.20 For screening AMA sub-specificities, antigen concentrations of 10 μg/ml were used for coating the microtitre plates; sera were diluted 1:1000. Peroxidase-conjugated goat anti-human IgG and IgM antibodies (DIANOVA, Hamburg, Germany) were used in parallel at a dilution of 1:3000.

The western blot was performed under denaturising conditions (sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE)) according to Laemmli21 with a 4.5% stacking and a 10% or 15% running gel. Twenty micrograms of the M2 antigen as well as 100 μg of PDC and OGDC were applied to each lane. Of the five gradient fractions, G1–G5, 50 μl were loaded on to the gel. After electrophoresis, the gels were either stained with Coomassie blue or the proteins were transferred to nitrocellulose sheets (Amersham, Freiburg, Germany) as described by Towbin et al22 for 2 h. The sheets were incubated with phosphate-buffered saline containing 1% bovine serum albumin to block free binding sites. Afterwards they were incubated with patients’ sera at a dilution of 1:50 for 90 min. Bound antibodies were visualised with peroxidase-conjugated rabbit antihuman IgG and IgM antibodies (Dako): 3-amino-9-ethyl-carbazole was used as the substrate. The reaction was stopped with 7% acetic acid.

Purification of new mitochondrial antigens and identification by mass spectrometry

In order to purify antigenic determinants from the gels, corresponding bands were excised from the Coomassie blue stained gels and the antigens were eluted with an electro-eluter (Bio-Rad, Hercules, California, USA) according to the manufacturer’s protocols. The antigen obtained was again applied to electrophoresis, and the determinant thus obtained was analysed by mass spectrometry. In gels, tryptic digestions were performed as described23 with some modifications.24 Briefly, the excised protein bands were fully destained and digested for 3 h with porcine trypsin (sequencing grade, modified; Promega, Madison, Wisconsin, USA) at a concentration of 67 ng/μl in 25 mmol/l ammonium bicarbonate, pH 8.1, at 37°C. Prior to peptide mass mapping and sequencing of tryptic fragments by tandem mass spectrometry, peptide mixtures were extracted from gels by 1% formic acid followed by two changes of 50% acetonitrile. The combined extracts were vacuum dried until only 1–2 μl was left and the peptides were purified by ZipTip according to the manufacturer’s instructions (Millipore, Bedford, Massachusetts, USA). Matrix-assisted laser desorption/ionisation time-of-flight mass spectroscopy (MALDI–TOF) analysis from the matrix α-cyano-4-hydroxycinnamic acid/nitrocellulose prepared on the target using the fast evaporation method25 was performed on a Bruker Reflex III (Bruker Daltonik, Bremen, Germany) equipped with a nitrogen 337 nm laser and gridless pulsed ion extraction.

Sequence verifications of some fragments were performed by nanoelectrospray tandem mass spectrometry on a Q-Tof I mass spectrometer (Micromass, Manchester, UK) equipped with a nanoflow electrospray ionisation source. Gold-coated glass capillary nanoflow needles (Type Medium NanoES spray capillaries) were obtained from Proxeon (Odense, Denmark). Database searches (NCBInr, non-redundant protein database) were done using the MASCOT software from Matrix Science.25

Absorption studies

For absorption studies, antigens were coupled to cyanogen bromide (CNBr)-activated Sepharose 4B (Pharmacia Biotech, Freiburg, Germany) according to standardised protocols and then incubated with patients’ sera for 16 h at 4°C on a rotating disc. The supernatant was then re-tested by western blotting for the presence of specific reactions.

Statistical analysis

For comparison of clinical data in different groups of patients, SPSS version 15.0 was used applying the non-parametric Kruskal–Wallis test for overall analysis and the Mann–Whitney test for the comparison of two unpaired groups. Differences with p<0.05 were considered statistically significant.

Normal ranges for antibody reactivities with all antigens were determined by analysis of 30 healthy donors. Mean value of their optical density (OD) plus double the standard deviation was defined as the cut-off value.

RESULTS

Evidence for a novel antimitochondrial antibody in primary biliary cirrhosis by ELISA and western blotting

Sera from 27 histologically proven PBC patients being AMA positive in the IFT but anti-M2 negative in ELISA and western blotting (group 1a patients) were tested by ELISA and western blotting against the 100 000 g supernatant from mitochondria of rat liver, pig kidney and bovine heart. Strongest reactions were observed with the antigen fraction prepared from bovine heart. Seventeen of the 27 sera (63%) were positive in the ELISA, and by western blotting 14 (52%) showed a positive reaction revealing a common determinant at 60 kDa (fig 1), eight recognised a further determinant at about 80 kDa. Both bands did not correspond to the M2 determinants (fig 1). Eleven sera were negative in the western blot.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Reaction of an anti-M2 positive serum (serum 1) and three antimitochondrial antibody (AMA) positive/anti-M2 negative sera2^–4 with a 100 000 g supernatant from bovine heart mitochondria in the western blot. Serum 1 recognises the typical M2-related antigens a–d at 70, 56, 45 and 36 kDa; serum 2 reacts with a determinant at 60 kDa; serum 3 with two determinants at 60 and 80 kDa; and serum 4 a determinant only at 80 kDa. Serum 5, from a healthy blood donor, is negative.

Purification of the 60 and 80 kDa antigens

The 100 000 g supernatant from bovine heart mitochondria still contained a bulk of mitochondrial proteins including the M2 antigen (fig 1). It was, therefore, applied to discontinuous sucrose density gradient centrifugation. Ten gradients were obtained and used in the western blot. The determinant at 60 kDa was detected in gradient fraction 1.16 and the 80 kDa band in gradient fraction 1.18 (G4; fig 2) while the M2 antigen preferentially accumulated in gradients fractions 1.20–1.28 (G5; not shown) as already described in earlier studies.18

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Analysis of gradient fractions of the 100 000 g supernatant from sub-mitochondrial particles from bovine heart after sucrose density centrifugation with respect to their content of the 60 kDa and the 80 kDa determinants: using an AMA positive/anti-M2 negative PBC serum reacting with both determinants (serum 2, see fig 1), the 60 kDa protein was detected in the sucrose gradients 1.16 and the 80 kDa determinant in the 1.18 gradient. AMA, antimitochondrial antibody; PBC, primary biliary cirrhosis.

The 60 and 80 kDa bands were excised from the gel, and after electro-elution another SDS electrophoresis was performed which confirmed a reaction at 60 and 80 kDa, respectively. These bands were again excised from the gel and electro-eluted, concentrated by lyophilisation and then applied to ELISA and western blotting (“purified” 60 and 80 kDa antigens, respectively).

Reactivity of sera from PBC patients with the purified 60 and 80 kDa antigens as well as the M2 antigen

Sera from the 51 patients of group 1a–c and the 77 patients of group 2 were tested by western blotting against the excised 60 and 80 kDa bands as well as the M2 antigen. The incidence of the different reactivities is given in table 2, and examples for the reaction patterns are given in fig 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Reaction of an anti-60 kDa positive/anti-M2 negative (A), an anti-60 kDa and anti-M2 positive (B) and an anti-60 kDa negative/anti-M2 positive serum (C) with the M2 antigen and the 60 kDa band eluted from a previous sodium dodecyl sulfate gel. Serum A shows a positive reaction with the 60 kDa protein but is negative with M2, serum B recognises the 60 kDa antigen and in the M2 fraction additionally the M2a (PDC-E2) and M2b determinant (E3-binding protein of PDC) at 70 kDa and 56 kDa, and the anti-M2 positive serum C reacts with subunits of the M2 antigen (70 kDa and 56 kDa) but not with the 60 kDa antigen. PDC-E2, E2 component of the pyruvate dehydrogenase complex.

Of the 27 patients in group 1a 59% reacted with at least one of the two determinants at 60 and 80 kDa in the western blot. In patients with classical anti-M2 positive PBC the incidence of these antibodies was lower, ie, 32% reacted with the 60 or 80 kDa bands.

Identification of the 60 kDa antigen by MALDI–TOF

The 60 and 80 kDa antigens were excised from the gels and analysed by mass spectrometry. With the 80 kDa antigen no conclusive data were obtained. In the 60 kDa antigen, sequences corresponding to subunits of three different mitochondrial enzyme complexes were found: first, the core protein 1 of ubiquinone cytochrome c reductase; second, the β, δ and ϵ fragments of the F₁F₀-ATPase; and third, the acyl CoA dehydrogenase.

Fragments of the 2-oxoacid dehydrogenase complex were not observed in either the 60 or the 80 kDa band.

Demonstration of antibodies to the 60 kDa antigen by ELISA

The whole 100 000 g supernatant from bovine heart, and the purified 60 kDa antigen were tested by ELISA against the 27 sera from patients with PBC who were AMA positive in the IFT but anti-M2 negative. As already mentioned, 17 of them were positive with the whole 100 000 g supernatant. However, applying the purified 60 kDa antigen excised from the SDS gel, only two (7%) were positive and antibody titres were low (data not shown). Of the 77 anti-M2 positive patients 18 (23%) reacted with the 60 kDa antigen. The 80 kDa antigen was not tested by ELISA.

These data indicate that at this stage of investigation ELISA is not a suitable method for screening for antibodies to the 60 kDa antigen.

Absorption studies

One AMA positive/anti-M2 negative (group 1a) and one AMA positive/anti-M2 positive serum (group 2), both reacting with the 60 kDa band were incubated with the G4 fraction and the M2 antigen coupled to CNBr-activated Sepharose 4B. Absorption with G4 containing the 60 kDa protein abolished the 60 kDa reactivity in the western blot of both sera but not the anti-M2 reactivity of the AMA positive/anti-M2 positive serum. In contrast, absorption with the M2 antigen had no effect of the anti-60 kDa reactivity of both sera while the anti-M2 reactivity of the group 2 serum became completely negative (data not shown).

Specificity of anti-60 kDa antibodies for primary biliary cirrhosis

In order to analyse the specificity of the novel anti-60 kDa and anti-80 kDa antibodies for PBC, sera from 87 patients with different AMA negative disorders and from 30 healthy individuals were tested by western blotting against the G4 fraction and the eluted 60 and 80 kDa antigens. However, none of these sera revealed a positive reaction with these antigens.

Correlation of anti-60 kDa antibodies with laboratory parameters

Comparing within the 27 patients with histologically proven PBC being AMA positive/anti-M2 negative those being anti-60 kDa positive (n = 14) and those being anti-60 kDa negative with respect to laboratory parameters (alkaline phosphatase, alanine aminotransferase, aspartate transaminase, gamma glutamyl transferase, bilirubin, and IgG , IgA and IgM globulins) no significant differences were observed (data not shown). The patients also did not differ in sex ratio (F:M, 11:3 and 12:1, respectively) and age (52.3 (SD 14.2) years vs 55.1 (SD 15.2) years).

DISCUSSION

In this study we show that in some sera from patients with clinically and histologically well-defined PBC AMAs can be detected by IFT, which do not react with several M2-related antigens (M2 fraction, PDC, OGDC) in ELISA or western blot. In these patients we observed novel antibodies to mitochondrial proteins at 60 and 80 kDa. These antigens were present in a mitochondrial antigen fraction obtained from bovine heart mitochondria. Applying a 100 000 g supernatant from bovine heart mitochondria to sucrose density centrifugation, the two antigens accumulated in gradients 1.16–1.18 (G4) in contrast to the M2 antigen which was predominantly found in gradients 1.10–1.28 (G5), as previously described.18 Sixteen (59%) of 27 sera from AMA positive/anti-M2 negative PBC sera recognised at least one of these two determinants. Absorption studies confirmed that the 60 kDa band did not correspond to M2-related antigens. Interestingly, this novel AMA occurred also in association with anti-M2, ie, 20 (26%) of the 77 anti-M2 positive patients also had antibodies to the 60 kDa and seven (9%) to the 80 kDa band. This observation, as well as the fact that the anti-60 or 80 kDa antibodies occurred also in association with PBC-specific antinuclear antibodies, are strong arguments towards their relation to PBC. We also found these novel antibodies in 50% of ten AMA positive/anti-M2 negative patients in whom PBC was clinically suspected but had not been confirmed histologically. This group did not differ significantly from patients with histologically proven PBC with respect to biochemical and clinical data, ie, these antibodies may, indeed, be taken as further marker antibodies for the serological diagnosis of PBC. This is also underlined by our findings that sera from patients with other hepatic and non-hepatic disorders did not recognise any of the two determinants.

Interestingly, mass spectrometry revealed that the 60 kDa band contained fragments of three different mitochondrial enzymes, namely the F1 part of the F₁F₀-ATPase, the core 1 protein of the ubiquinone cytochrome c reductase, and the acyl CoA dehydrogenase, which are all located at the inner mitochondrial membrane and involved in electron transport and energy metabolism. Unfortunately, for the 80 kDa determinant we did not obtain conclusive results. M2/OADC-related sequences were not observed in both determinants, underlining the distinct entities of the novel AMA and anti-M2.

Until now, we have not been able to clone the acyl CoA dehydrogenase, but meanwhile we have succeeded in expressing the core 1 protein of the ubiquinone cytochrome c reductase in Escherichia coli. However, applying this recombinant antigen in ELISA or western blotting we did not obtain positive reactions with anti-60 kDa positive sera, and also absorption with this antigen did not influence the anti-60 kDa reactivity (own unpublished observations). An identity of the novel mitochondrial antigen with this enzyme can, therefore, be excluded with high probability. The third enzyme being considered, namely F₁F₀-ATPase, is a large enzyme complex that can be visualised by electron microscopy as stalked particles attached to the inner face of the inner membrane. Each head group (F₁) has a stalk and a protein base piece (F₀) extending right across the inner membrane, which allows protons to return in a controlled fashion to the matrix space. Despite its considerable intricacy, the F₁F₀-ATPase was one of the first enzymes to be utilised by living cells; it has scarcely changed in the last 2000 million years. The F₀ part consists of three subunits a, b and c, and the F₁ part is a water-soluble complex consisting of five subunits (α, β, γ, δ and ϵ).26 27 We have tried to express these subunits in E coli in order to obtain recombinant antigens. Preliminary data indicate that the F1α subunit is not a major target antigen recognised by sera from AMA positive/anti-M2 negative PBC patients; but at least some of them react with the β and probably also the γ or δ subunit,14 28 and under certain conditions absorption with the β subunit reduced the anti-60 kDa reactivity in the western blot. However, because of methodological problems we have not yet been able to prepare these proteins in large amounts in a stable and pure form, and data are, therefore, still inconsistent. This may also be due to the fact that antibodies to conformational and linear epitopes on these subunits exist – as also known for the M2 antigen29 30 – which further complicates the issue. Further studies on the identification and verification of F1-ATPase as another target antigen of AMAs in PBC have, therefore, still to be performed.

It is our experience that other laboratories using commercially available slides for the IFT quite frequently fail to detect especially those AMA positive/anti-M2 negative sera as described in the present study. Those slides are in most instances treated with organic chemicals for stabilisation. In contrast, we use native cryostat sections, and, indeed, treating them with those chemicals, the AMA fluorescence of the anti-60 kDa positive/anti-M2 negative sera was abolished while the AMA fluorescence of the anti-M2 positive sera was not affected (unpublished observation).

The existence of a particular subgroup of patients with PBC who are anti-M2 negative is well known, and in these patients, predominantly, antibodies to defined nuclear proteins such as sp100 or gp120 have been found.5 15 Also, our patients, who showed typical AMA staining in the IFT but were anti-M2 negative, seem to represent a further subgroup since clinical, biochemical and also histological features were typical for PBC, although these data have certainly to be confirmed in further studies. The novel antibodies to the 60 or 80 kDa protein may, therefore, be considered as a diagnostic marker for this subgroup because we have not observed them in any other disease. However, the fine specificity of the target antigen(s) and the clinical relevance of these antibodies are still under investigation. Nevertheless, the data presented indicate that there still is a potential to improve the serological diagnosis of PBC, either by increasing the sensitivity of assays for AMA as recently outlined13 or by applying newly identified mitochondrial autoantigens. Furthermore, they are a strong argument for the postulated heterogeneity of PBC with respect to aetiology and pathogenesis.31 32