Autoantibodies as Prognostic Markers in Autoimmune Liver Disease
- First Online:
- Cite this article as:
- Czaja, A.J. Dig Dis Sci (2010) 55: 2144. doi:10.1007/s10620-010-1268-4
- 336 Views
Certain autoantibodies in autoimmune liver disease have prognostic implications that are under-utilized and under-developed. The goals of this review are to indicate progress in characterizing the autoantibodies with prognostic connotations and to indicate the feasibility and importance of discovering other markers. Prime source and review articles in English were selected by a Medline search through 2010. Antibodies to soluble liver antigen, actin, liver cytosol type 1, asialoglycoprotein receptor, chromatin, cyclic citrullinated peptide, and uridine glucuronosyltransferases have been associated with the occurrence, severity, and progression of autoimmune hepatitis, and antibodies to Sp100, gp210, and centromere have had similar implications in primary biliary cirrhosis. Antibodies to soluble liver antigen have shown the most promise in autoimmune hepatitis as they have been associated with severe histological changes, long durations of treatment, relapse after drug withdrawal, and high frequency of liver failure. Antibodies to the nuclear rim pore protein, gp210, have shown the most promise in primary biliary cirrhosis as they have been associated with severe interface hepatitis, lobular inflammation, and progression to liver failure. The major limitations of the autoantibodies have been their lack of standardized assays, low negative predictabilities, and fluctuating levels. Performance parameters will improve as critical pathogenic pathways, comprehensive testing batteries, and standardized assays through international exchange workshops are developed. Progress has been made in identifying the serological markers of prognosis in autoimmune liver disease, and they promise to reflect critical disease mechanisms and enhance patient management.
KeywordsAutoantibodiesAutoimmune hepatitisPrimary biliary cirrhosis
Autoantibodies are the serological hallmarks of autoimmune liver disease . Antinuclear antibodies (ANA) [2, 3], smooth-muscle antibodies (SMA) , and antibodies to liver kidney microsome type 1 (anti-LKM1)  constitute the standard repertoire for the diagnosis and classification of autoimmune hepatitis [6–8], and antimitochondrial antibodies (AMA) are the diagnostic markers of primary biliary cirrhosis (PBC) [9–11]. Antinuclear antibodies and SMA rarely coexist with anti-LKM1 (4% concurrence) in white North American and northern European adults [12, 13], and this exclusivity defines two distinct subpopulations .
White North American and northern European patients with ANA and SMA (type 1 autoimmune hepatitis) are older and have a different human leukocyte antigen (HLA) phenotype than patients with anti-LKM1 (type 2 autoimmune hepatitis) [15–17]. In the first instance, HLA DRB1*0301 and DRB1*0401 are the principal genetic risk factors [18, 19], and in the second instance, DQB1*0201 (in strong linkage disequilibrium with HLA DRB1*07 and DRB1*03), is the principal genetic determinant . These conventional serological markers of autoimmune hepatitis have been invaluable in directing the diagnosis, but they have been deficient as prognostic indices. Serum titers have not correlated with disease activity, and the individual markers have not predicted treatment outcomes [21–23].
Similarly, early reports that the serum titer of AMA correlates with the stage and severity of PBC have not been confirmed , and AMA have mainly a diagnostic rather than prognostic connotation . The principal target of AMA is the E2 subunit of the pyruvate dehydrogenase complex [9, 10], and this reactivity may have a genetic basis . Neither the principal antigenic target nor its serological imprint, however, has reflected disease severity or treatment response in PBC [25, 27], and in this regard, AMA are similar to the conventional serological markers of autoimmune hepatitis .
Autoantibodies are immunoglobulins directed at normal host proteins, and when they are disease-related they constitute a pathological reaction [28, 29]. Pathological autoantibodies can be pathogenic if they deposit in critical organs and trigger severe inflammation . Antibodies to glomerular basement membrane (anti-GBM) in Goodpasture’s syndrome (anti-GBM disease) are prototypic pathogenic autoantibodies [30, 31]. The pathological autoantibodies in autoimmune liver disease lack pathogenic properties. Some autoantibodies, such as anti-LKM1 in autoimmune hepatitis  and AMA in PBC , can interfere with enzyme activity and potentially impair cell metabolism, but this interference has not affected the clinical course of the disease [21, 25]. Furthermore, the conventional pathological autoantibodies of autoimmune hepatitis [33–35] and PBC [36, 37] are not required for the illness.
The goals of this review are to indicate progress in characterizing autoantibodies in autoimmune liver disease that have prognostic implications and to underscore the feasibility and importance of discovering other markers with similar properties. These advances promise to improve current management strategies by identifying problematic patients early, prompting individualization of treatment schedules, and optimizing treatment end points.
Promising Prognostic Markers in Autoimmune Hepatitis
Serological markers of prognosis in autoimmune hepatitis
Anti-soluble liver antigen
Severe histological changes 
Long treatment duration 
Treatment dependence 
Hepatic death or transplantation 
Present when other markers absent 
Microfilaments (F-actin) of the cytoskeleton 
Treatment dependence in children 
Liver failure or liver transplantation 
Severe clinical and histological disease if reactivities to actin and α-actinin 
Anti-liver cytosol type 1
Type 2 autoimmune hepatitis 
Present when other markers absent 
Concurrent other immune diseases 
Severe liver inflammation 
Rapid progression to cirrhosis 
Adequacy of treatment response 
Active inflammation 
Anti-cyclic citrullinated peptide
Citrullinated peptides 
Cirrhosis and liver failure 
Erosive rheumatoid arthritis 
Predictive of rheumatoid arthritis 
Anti-liver kidney microsome type 3
Present when other markers absent 
Antibodies to Soluble Liver Antigen
Antibodies to soluble liver antigen were discovered in separate laboratories and initially given different names [38, 39]. The single designation, antibodies to soluble liver antigen (anti-SLA), is now applied, and it refers to highly specific markers of autoimmune hepatitis that have acquired a prognostic connotation [40–44]. These autoantibodies are present in 11–22% of white North American and northern European adults with type 1 autoimmune hepatitis [41, 42, 45, 46] and 18–44% of children with type 2 disease . They have also been described in 14–20% of patients with cryptogenic chronic hepatitis [41, 42, 45]. Antibodies to soluble liver antigen have a diagnostic specificity of 99% for autoimmune hepatitis , and their presence in chronic hepatitis of unknown nature compels consideration of an autoimmune basis for the inflammatory activity. Antibodies to soluble liver antigen were originally proposed as a marker of a “type 3 autoimmune hepatitis” [38, 39], but the absence of a distinctive clinical phenotype and the common concurrence of anti-SLA with the serological markers of type 1 and type 2 autoimmune hepatitis have not justified this separate designation [8, 45, 46, 48].
Clinical and Molecular Aspects
Differences in the frequency of anti-SLA in autoimmune hepatitis probably reflect the nature of the assay [47, 49, 50] and the ethnicity of the patients . Assays for anti-SLA can be based on partially purified or prokaryotically expressed antigen  or eukaryotically expressed antigen that preserves conformational epitopes . Reactivity to the antigen can then be assessed by radioligand assay, enzyme immunoassay, Western blot, or immunoprecipitation, and different estimates of prevalence and clinical significance can occur [47, 49, 50]. Furthermore, the production of anti-SLA has been associated with HLA DRB1*0301, [50, 51], and the frequency of these autoantibodies in different ethnic populations may be affected by this genetic factor. In Japanese patients with autoimmune hepatitis in whom HLA DRB1*0301 is unusual [52, 53], the frequency of anti-SLA is only 7% [42, 54]. Antibodies to soluble liver antigen have been described in 10% of patients with chronic hepatitis C, especially in those with anti-LKM1 (27%) . Other studies have emphasized their absence in chronic viral hepatitis [41, 42, 56].
The molecular target of anti-SLA has been characterized as a 422-amino-acid protein [57, 58]. It has been identified as a transfer ribonucleoprotein involved in selenocysteine insertion (tRNP(ser)sec), and this antigen has been recently named SEPSECS [Sep (O-phosphoserine) tRNA:Sec (selenocysteine) tRNA synthase] [40, 59] (Fig. 1). Antibodies to soluble liver antigen have exquisite epitope specificity to the peptide sequence p 395–414 of the target antigen. They are mainly of the immunoglobulin G1 subtype, and they exhibit no cross-reactivity with homologous virus-derived peptides . Furthermore, mice immunized with the antigen have developed HLA DRB1*0301-restricted T cells that target peptides within the antigen . These features have supported the hypothesis that anti-SLA have a role in the pathogenesis of autoimmune hepatitis, and they may account for the prognostic value of these serological markers .
Antibodies to soluble liver antigen identify individuals who have more severe histological changes, require longer durations of treatment, invariably relapse after drug withdrawal, and have a higher frequency of liver transplantation or death from liver failure than patients without these antibodies [43, 50, 51] (Table 1). Only 15% of white North American patients with autoimmune hepatitis express anti-SLA, but their early detection can forecast treatment difficulties [45, 51]. Their high specificity for autoimmune hepatitis, frequent occurrence in cryptogenic chronic hepatitis, and close association with HLA DRB1*0301, severe histological disease, and treatment dependence have made anti-SLA the most promising serological markers of prognosis in autoimmune hepatitis.
Antibodies to Actin
Antibodies to actin are a subset of smooth-muscle antibodies [62, 63], and they are characterized by reactivity against filamentous actin (F-actin) . Multiple assays have been used for their detection, many are institution-specific, and none has been universally endorsed. Antibodies to actin can be demonstrated by indirect immunofluorescence of microfilaments in HEp-2 cells or fibroblasts treated with vinblastine or colchicine [65, 66]. They can also be detected by indirect immunofluorescence of liver sections from rats chronically injected with phalloidin  or by assessing peritubular and glomerular structures in rat kidney . Counterimmunoelectrophoresis with purified muscle actin  and enzyme immunoassays based on purified F-actin [64, 68–70] have also been advocated, and an actin-myosin functional assay in which actin activates meromyosin ATPase in skeletal muscle has been proposed . A thermolabile F-actin depolymerizing factor may limit detection of the antibodies and may affect assay performance .
Enzyme immunoassays based on purified F-actin are less labor-intensive than assays based on indirect immunofluorescence. They are not subject to inter- or intra-observer variation; their performance parameters as diagnostic indices have been satisfactory, and a commercial kit is readily available [69, 70]. Accordingly, this assay for anti-actin has been preferred in clinical practice . Antibodies to actin that are determined by enzyme immunoassay have a sensitivity of 71%, negative predictability of 94%, specificity of 90%, and positive predictability of 58% for autoimmune hepatitis . These performance parameters are better than those associated with SMA [69, 70].
Comparison studies between the various methods have indicated that the indirect immunofluorescence assays have a higher specificity for the diagnosis than the enzyme immunoassays , but the presence of anti-actin by enzyme immunoassay in normal individuals  and in patients with chronic viral hepatitis or other liver diseases [73, 75] has not compromised its value in securing the diagnosis within the proper clinical context . Furthermore, the performance parameters may be enhanced by measuring reactivity against the amino acid sequence 351–362 of the synthetic actin peptide , quantifying the immunoglobulin G subclasses associated with anti-actin , or re-defining cut-off values to exclude low-level results [66, 73, 75].
The prognostic value of anti-actin, like its diagnostic prowess, is assay-dependent. Children with anti-actin by indirect immunofluorescence have been characterized by treatment dependence and progressive liver failure resulting in death or requirement for liver transplantation . Similarly, adults with anti-actin by at least two available assays have had an earlier age of disease onset and poorer response to corticosteroid therapy than patients without anti-actin . They have died of liver failure or required liver transplantation more frequently than patients with ANA (19% versus 0%, p = 0.03) .
These results have not been confirmed by studies based on enzyme immunoassays , but they have been supported by assays assessing reactivity against α-actinin . Double reactivities against actin and α-actinin identify patients with severe clinical and histological disease compared to patients with antibodies to filamentous actin . Antibodies to α-actinin are associated with antibodies to single-stranded DNA, and this double reactivity correlates with clinical and histological activity . The prognostic value of testing for anti-actin probably relates to a small, highly selective, molecular epitope, and it is this epitope that must be characterized and identified so that a finely tailored assay for actin can be developed.
Antibodies to Liver Cytosol Type 1
Antibodies to liver cytosol type 1 (anti-LC1) are directed against formiminotransferase cyclodeaminase, which is a cytosolic enzyme involved in the conversion of histidine to glutamic acid (Table 1) [81–83]. The enzyme is free in the cytosol of the hepatocyte and also reversibly bound to the Golgi membranes where it is active in the secretory pathway of the cell  (Fig. 1). Anti-LC1 constitute a polyclonal, antigen-driven B cell response against conformational and discontinuous linear epitopes of formiminotransferase cyclodeaminase , and titers fluctuate with disease activity . Consequently, anti-LC1 may be markers of residual liver inflammation or imprints of an autoantigen associated with disease severity. Since anti-LC1 may interfere with critical secretory activities within the cell, they may be close to the pathogenic bases for the disease . Anti-LC1 commonly (24–32%) occur in conjunction with anti-LKM1, and they are considered a second marker of type 2 autoimmune hepatitis [85–87]. Like anti-LKM1, they are expressed mainly in children and young adults (20 years old or less) [86, 87].
Antibodies to liver cytosol type 1 have been associated with concurrent immune diseases, marked liver inflammation, and rapid progression to cirrhosis. They have been the sole serological markers of autoimmune hepatitis in children with acute, acute severe, and chronic hepatitis, and they may be useful in establishing the diagnosis of autoimmune hepatitis in a subgroup negative for ANA, SMA, and anti-LKM1 [88, 89]. Such patients typically respond to corticosteroid therapy, and the presence of anti-LC1 is the key to their recognition . Anti-LC1 may co-exist with SMA and ANA in autoimmune hepatitis , exist in asymptomatic patients , and occur in 12% of patients with chronic hepatitis C [91–93]. Like anti-LKM1, anti-LC1 are rare in white North American adult patients with autoimmune hepatitis .
The prognostic value of anti-LC1 is limited by continuing controversies regarding its association with severe disease in children  and its independence from hepatitis C infection [56, 91, 93]. Assays for anti-LC1 include indirect immunofluorescence on snap-frozen sections of rat liver , counterimmunoelectrophoresis using human liver cytosol as the antigenic source , immunoprecipitation [86, 88], immunoblotting with chemiluminescence directed against human cytosolic proteins separated by gel electrophoresis [87, 91, 92], and enzyme immunoassay using recombinant formiminotransferase cyclodeaminase . Only one study has made head-to-head comparisons between the assays, and counterimmunoelectrophoresis has been favored over immunodiffusion and immunoblotting since it provided more economical and unambiguous results . The enzyme immunoassay based on recombinant antigen remains investigational .
Antibodies to Asialoglycoprotein Receptor
Antibodies to asialoglycoprotein receptor (anti-ASGPR) are directed against a transmembrane glycoprotein on the hepatocyte surface that captures, transports and displays foreign and self antigens (Table 1) [95–99]. The asialoglycoprotein receptor has two membrane locations within the hepatocyte, including the cell surface and the endocytic membrane network, and each location is regulated independently . The receptors present on the luminal surface of the sinusoidal cells have extracellular epitopes that can trigger anti-ASGPR [98, 100, 101] (Fig. 1). They consist of H1 and H2 subunits [99, 102], and the antigenic sites are located mainly on the H1 subunit . A monoclonal antibody that is specific for the H1 subunit has been developed to investigate further the biological importance of the asialoglycoprotein receptor in immune responses .
Clinical and Molecular Aspects
The asialoglycoprotein receptor has been implicated in both the humoral and cellular immune responses of autoimmune hepatitis . Anti-ASGPR occur in most patients with autoimmune hepatitis (67–88%) compared to those with chronic hepatitis B (7%), chronic hepatitis C (14%), acute hepatitis B (35%), alcoholic liver disease (8%), and PBC (14–22%) [105–110]. They occur in 75% of children with type 1 autoimmune hepatitis and 40% of those with type 2 disease , and they can be present in patients who are negative for the conventional autoantibodies . Peripheral blood lymphocytes produce anti-ASGPR when stimulated in vitro with asialoglycoprotein receptor [113, 114], and liver infiltrating T helper lymphocytes from patients with autoimmune hepatitis show a specific proliferative response to asialoglycoprotein receptor that is HLA class II restricted [113, 114].
The expression of the asialoglycoprotein receptor increases with apoptosis , and the asialoglycoprotein receptor is present at high density on hepatocytes in the periportal areas of the liver lobule where the characteristic histological lesion of interface hepatitis develops . Furthermore, asialoglycoprotein receptor induces dose-dependent agglutination and hemolysis in alcoholic cirrhosis , and pre-existent anti-ASGPR may predispose to the development of chronic hepatitis after experimental viral hepatitis . These observations indicate that the asialoglycoprotein receptor may be a common target in the humoral and cellular immune responses of autoimmune hepatitis or be involved directly or collaterally in the mechanisms of cell injury. Anti-ASGPR may thereby reflect the causes or the consequences of the inflammatory process, and for these reasons they have acquired a prognostic connotation in autoimmune hepatitis.
Patients with anti-ASGPR are distinguished from patients without these antibodies by having higher serum levels of γ-globulin and immunoglobulin G at presentation [119, 120]. They also have a greater frequency of relapse after drug withdrawal (88% versus 33%, p = 0.01) [120, 121]. Antibodies to ASGPR are associated with histological activity [121, 122], and their persistence or disappearance during corticosteroid therapy reflects the adequacy of the treatment response [121, 123]. These attributes suggest that anti-ASGPR may be useful in defining end points of treatment. Disappearance of anti-ASGPR during therapy has been associated with sustained remission after drug withdrawal, whereas its persistence or reappearance has heralded relapse [121, 123]. Anti-ASGPR may be the only prognostic markers of autoimmune hepatitis which have importance when they are absent.
Assays for Anti-ASGPR
The assays for anti-ASGPR are solid-phase enzyme immunoassays based on human-, rabbit-, or rat-derived ASGPR [95, 96, 106, 107] or a radioimmunofiltration assay (RIFA)  using purified ASGPR from rabbit liver. Assays based on human-derived ASGPR are more sensitive and specific for autoimmune hepatitis than assays based on non-human-derived antigens, but the differences are small and unlikely to have clinical importance . Concordance between the various assays is 78%, and the frequencies of positivity are similar between the rabbit-derived RIFA and the human-derived enzyme immunoassay in autoimmune hepatitis (82% versus 88%).
Standardization and wide distribution of the assay for anti-ASGPR has been difficult to achieve, largely because selection of the target antigen and methods for its utilization and commercialization in a test kit have been uncertain. Interest in anti-ASGPR has been re-kindled recently after the development of a readily available and specific enzyme immunoassay based on purified rabbit antigen . The assay has a sensitivity of 78% and specificity of 99% for autoimmune hepatitis, and it has re-confirmed earlier observations relating anti-ASGPR to inflammatory activity . It promises to extend the prognostic prowess of anti-ASGPR by quantifying results and correlating the strength of the serological reactivity with disease severity and outcome.
Antibodies to Chromatin
Antibodies to chromatin (anti-chromatin) are directed against a linear array of nucleosomes, each of which is composed of two molecules containing four histones interwoven with DNA [124–126] (Table 1). Chromatin is a macromolecular octameric complex that has multiple epitopes which account for its immunogenic property [124, 127]. It has been proposed as the stimulus for the loss of self tolerance in early stage systemic lupus erythematosus [128, 129], and the processing of chromatin by B lymphocytes can generate antibodies to histones (anti-histones) and double-stranded DNA (anti-dsDNA) [124, 127, 129]. Chromatin is the only target antigen associated with prognostic properties that has an intra-nuclear location in autoimmune hepatitis (Fig. 1). Antibodies to chromatin are found in 39% of patients with autoimmune hepatitis [130, 131], and they are associated with high serum levels of γ-globulin and immunoglobulin G .
Antibodies to chromatin occur more commonly in patients with active than inactive autoimmune hepatitis (32% versus 19%, p = 0.01) , and relapse after drug withdrawal develops more frequently in patients with anti-chromatin than in those without these antibodies (91% versus 66%, p = 0.002) [131, 132]. Antibodies to chromatin coexist with ANA in 94% of instances in autoimmune hepatitis, but ANA coexist with anti-chromatin in only 68% of instances. Antibodies to chromatin, therefore, can have a prognostic connotation that is not shared by ANA, and they may define a subgroup of ANA-positive patients with severe disease who are treatment-dependent [94, 131].
Antibodies to chromatin have a sensitivity for relapse after drug withdrawal of 42%, but their specificity (100%) and positive predictability (100%) for this outcome is absolute . Like the other prognostic markers, their negative predictability for relapse after drug withdrawal is low (22%) and comparable to that of anti-SLA (17%) and anti-actin (22%) . Antibodies to chromatin have the added attribute of reflecting inflammatory activity in autoimmune hepatitis and a disease severity that is difficult to suppress with conventional corticosteroid treatment . In this fashion, they may be useful in identifying problematic patients early. Their major limitation as a predictor of treatment failure is their restriction to patients with ANA.
Antibodies to Histones and Double-Stranded DNA
Antibodies to histones are present in 35% of ANA-positive patients with autoimmune hepatitis [133–136], and anti-dsDNA are detected in 23–34% of similar patients depending on the nature of the assay and substrate used for their detection . These reactivities may be collateral manifestations of the immune response against chromatin which is comprised of histones and DNA. Patients with anti-histones are not distinguished by the severity and aggressiveness of their liver disease . In contrast, patients with anti-dsDNA by an enzyme immunoassay against highly purified dsDNA derived from a bacterial system fail corticosteroid treatment more commonly than those without these antibodies (24% versus 3%, p = 0.04) . The enzyme immunoassay for chromatin is based on long soluble chromatin from calf thymus [94, 131], and the clinical correlations by this assay may reflect the antigenic complexity of the chromatin molecule and the primacy of the immune response against chromatin. These results may supersede those obtained by measuring collateral reactivities against the individual components of the nucleosome (i.e., histones and dsDNA).
Antibodies to Cyclic Citrullinated Peptide
Antibodies to cyclic citrullinated peptide (anti-CCP) are highly specific for rheumatoid arthritis (RA) [138, 139], and their presence predicts the development of RA  and erosive joint disease [141–143]. These associations with disease occurrence and severity suggest that anti-CCP may be pathogenic [140–142, 144]. Antibodies to cyclic citrullinated peptide occur in 9–11% of patients with autoimmune hepatitis [145, 146], and they commonly occur in the absence of RA (75%) . Such patients may develop RA later, but the more immediate association with anti-CCP is the propensity to develop cirrhosis and liver failure  (Table 1).
Patients with autoimmune hepatitis and anti-CCP have a significantly greater occurrence of histological cirrhosis at presentation than patients without these antibodies (47% versus 20%, p = 0.01), and patients with autoimmune hepatitis, anti-CCP and concurrent RA have cirrhosis more commonly than others (100% versus 21%, p = 0.005) . Death from hepatic failure also occurs more frequently in patients with anti-CCP than in patients without anti-CCP (25% versus 9%, p = 0.04) . These findings suggest that anti-CCP reflect a heightened immune reactive state and an aggressive disease propensity.
Citrullination of peptides is accomplished by peptidylarginine deiminase (PAD) , and the presence of the mRNA of PAD isotype 4 in fetal liver suggests that citrullination can occur within this organ  (Fig. 1). Extravascular citrullinated fibrin is one of the citrullinated proteins that triggers production of anti-CCP in patients with RA , and chronic liver injury such as cirrhosis is characterized by the accumulation of fibrin . The fibrin and fibrin fragments of cirrhosis may be citrullinated within the liver, and these citrullinated antigens may then activate CD4 T helper cells and stimulate production of anti-CCP (Fig. 1). In this fashion, anti-CCP may be surrogate markers of advanced hepatic fibrosis.
Alternatively, the production of anti-CCP may be indicative of a heightened immune reactive state in which epitope spread and molecular mimicry result in multiple serological reactivities, concurrent immune disorders, and an aggressive liver disease . Most patients with autoimmune hepatitis and anti-CCP have other autoantibodies, including anti-chromatin (60%), anti-actin (90%), anti-SLA (27%), and rheumatoid factor (56%) . These findings suggest that an aggressive immune reactive state resulting in cirrhosis may be characterized by the concurrence of multiple autoantibodies and extra-hepatic immune-mediated disorders like RA. By this hypothesis, anti-CCP are just one of several expressions of immune reactivity that together reflect the severity of the pathogenic process.
Patients with anti-CCP have anti-chromatin more commonly than patients without anti-CCP (60% versus 33%, p = 0.02), whereas the frequencies of anti-actin, anti-SLA, and rheumatoid factor are not greater in patients with anti-CCP than in other patients . Consequently, anti-CCP, anti-chromatin, or both autoantibodies may be the principal serological markers of disease severity and outcome. Future studies must determine if individual markers or certain constellations correlate most closely with prognosis. Antibodies to cyclic citrullinated peptide are detected by an enzyme immunoassay based on highly purified synthetic peptides containing modified (citrullinated) arginine residues, and it is generally available .
Antibodies to Uridine Diphosphate Glucuronosyltransferases
The uridine diphosphate glucuronosyltransferases (UGT) catalyze the glucuronidation of dietary constituents, drugs and environmental mutagens, and they are divided into two families, UGT1 and UGT2 [150, 151] (Table 1). These enzymes are located in the rough endoplasmic reticulum of the hepatocyte (Fig. 1) and in the small intestine where they can process ingested antigens that might subsequently induce autoreactivity . The UGT interact with cytochromes P450 (CYP), and they may form functionally intact complexes with CYP . The UGT and CYP enzyme systems each contribute to the phases of drug metabolism designated as phase I (based on CYP) and phase II (based on UGT), and each system has been the target of autoantibodies in different types of liver disease [153–157]. Only antibodies to liver kidney microsome type 2 (anti-LKM2) have had a prognostic connotation, but experiences with autoantibodies within the UGT-CYP axis have been limited. Antibodies to LKM2 may be prototypic of a new serological family whose prognostic importance will expand.
Antibodies to LKM2 were described by indirect immunofluorescence on male mouse liver and male rat kidney, and in each instance, the pattern of indirect immunofluorescence was associated with hepatitis due to tienilic acid . Subsequent studies suggested that the drug-induced hepatitis was associated with immune mechanisms related to drug-altered liver cell antigens . The principal antigenic target of anti-LKM2 was later identified as CYP2C9 , and the liver toxicity was associated with impairment of glutathione biosynthesis . The frequency and severity of hepatic injury mandated removal of tienilic acid from clinical practice. Antibodies to LKM2 no longer have a clinical application, but they exemplify a serological family that must be explored for prognostic relevance.
Antibodies to liver kidney microsome type 3 (anti-LKM3) are serological markers within the family of UGT-CYP reactivities that warrant focused attention (Table 1). These antibodies are found in 13% of patients with chronic hepatitis D, 8% of patients with type 2 autoimmune hepatitis, and rare patients with chronic hepatitis C [162–165]. They are directed against the UGT1 enzymes [166, 167], and they are distinguished from the anti-LKM1 occasionally recognized in chronic hepatitis C by their substrate specificity . Unlike anti-LKM2, they have not been associated with drug toxicity, and they may exemplify virus-induced autoimmunity against a microsomal enzyme system. Their prognostic relevance in type 2 autoimmune hepatitis is uncertain since the disease and the autoantibodies are each uncommon. Nevertheless, their association with viral hepatitis and a critical enzyme system involved in peptide transformations offers a unique opportunity to clarify pathogenic mechanisms of autoreactivity, molecular mimicry, and disease severity.
Promising Prognostic Markers in Primary Biliary Cirrhosis
Serological markers of prognosis in primary biliary cirrhosis
Associated with advanced age 
Increased serum γ-globulin levels 
Rapid histological progression 
May be sole marker of primary biliary cirrhosis 
Anti-promyelocytic leukemia protein
Concurrent with anti-Sp100 
Weak reactivity 
Severe cholestasis 
Rapid progress to hepatic failure 
High serum alkaline phosphatase levels 
Severe bile duct injury 
Titers correlate with stage and severity but unconfirmed 
Titers do not correlate with progression 
Diagnostic function only 
The antibodies to Sp140 that have been recently described are directed against another nuclear protein in 15% of patients with PBC, and they are more common in AMA-negative patients with PBC than AMA-positive patients (53% versus 9%, p < 0.0001) . These antibodies are found concurrently with anti-Sp100 in 90% of instances, and their independent diagnostic and prognostic roles have yet to be defined. They suggest the existence of a multi-antigenic nuclear complex in PBC that is targeted during the disease and whose disrupted nuclear functions may contribute to disease susceptibility or severity .
Early descriptions of anti-neutrophilic cytoplasmic antibodies (ANCA) in PBC included antibodies against nuclear rather than cytoplasmic antigens. These antigenic targets were lamin B, gp210 [174, 177–181], and Sp100 , and their nuclear location justified a name change from ANCA to ANNA (anti-neutrophilic nuclear antibodies) . As the clinical implications of each nuclear antibody have been clarified, designations by the name of the individual nuclear antibody rather than by the category of ANNA have been preferred.
Antibodies to Sp100
Antibodies to Sp100 are directed against a novel nuclear protein that has a molecular mass of 95–100 kDa and a dot-like distribution within cell nuclei by indirect immunofluorescence (Fig. 2) [183, 184]. The nuclear protein, Sp100, co-localizes with promyelocytic leukemia protein (PML) in the dot-like nuclear domains, and the antibodies to Sp100 and anti-PML typically co-exist in PBC  (Fig. 2). Promyelocytic leukemia protein is a suppressor of cell transformation and growth in promyelocytic leukemia cells, and antibodies to recombinant PML in PBC have weaker reactivity than anti-Sp100, recognize fewer epitopes, and prefer conformational epitopes . Antibodies to Sp100 and anti-PML do not cross-react, and their individual substrate specificities suggest that the two co-localized nuclear dot proteins in PBC may contribute synergistically to the occurrence of the disease (Table 2).
Antibodies to Sp100 have high specificity for PBC among patients with chronic liver disease (94%) [186, 187]. Only 5% of patients with classical autoimmune hepatitis have anti-Sp100, whereas 13% of patients with autoimmune hepatitis and histological features of bile duct injury have these antibodies . Nuclear dot patterns have been described in patients with rheumatic diseases that are unrelated to Sp100 , but some patients, especially those with systemic lupus erythematosus, have nuclear dot patterns and anti-Sp100 [189, 190]. Within the appropriate clinical context of chronic liver disease, the performance parameters for anti-Sp100 (sensitivity, 27%; specificity, 94%) support their diagnostic role in PBC, especially in AMA-negative patients [187, 191].
Antibodies to Sp100 have a prognostic role that is less established than their diagnostic property. They occur in older patients with PBC, and they are associated with higher serum levels of γ-globulin . They are also associated with PBC that progresses more rapidly from early to late histological stages during a 24-month interval of observation . T cell clones selected by their reactivity to peptides of the E2 subunit of the mitochondrial pyruvate dehydrogenase complex cross-react with peptides from Sp100 and gp210, and this cross-reactivity between mitochondrial and nuclear antigens in PBC suggests that molecular mimicry may sensitize T cells and generate an autoreactive response .
The molecular mimicry between mitochondrial antigens and Sp100 may also include bacterial antigens. Women with recurrent urinary tract infections have been described who have AMA without liver disease, and these AMA-positive women also express anti-Sp100 . Furthermore, the frequency of anti-Sp100 in patients with PBC and recurrent urinary tract infections has been higher than in patients without recurrent urinary tract infections (74% versus 5%). The induction of PBC-specific autoimmunity may extend from the community to the individual patient via a sequence of molecular mimicries that triggers the production of anti-Sp100. Additional studies are needed to determine if anti-Sp100 define an etiologically distinct subgroup of PBC with a particular clinical behavior. Antibodies to Sp100 are detected by enzyme immunoassay ; they can recur after liver transplantation without heralding recurrent PBC; and they can persist in low titer for up to 13 years .
Antibodies to gp210
Antibodies to gp210 occur in 16–25% of patients with PBC, and they can co-exist with anti-Sp100 [171, 172, 195] (Table 2). Their reactivity is recognized by indirect immunofluorescence as punctuate dots on the nuclear rim, and their target is an integral protein localized to the nuclear pore membranes  (Fig. 2). Most antibodies to gp210 recognize a 15-amino acid sequence in the carboxyl-terminal end of the molecule that faces the nuclear pore complex, and reliable enzyme immunoassays based on recombinant gp210 expressed in bacteria or polypeptides that have been chemically synthesized are available for their detection . The nuclear pore membrane is involved in the trafficking of nuclear proteins into the cytosol and the maintenance of critical cell functions. Theoretically, a disruption of function at this site could affect cell performance and contribute to disease severity. Antibodies to gp210 have been more closely associated with the severity and outcome of PBC than anti-Sp100.
Two types of disease progression have been described in PBC, and they include a hepatic failure type and a portal hypertension type [170, 196]. Antibodies to gp210 are associated with severe interface hepatitis and lobular inflammation, and they are risk factors for the hepatic failure type of progression (odds ratio, 33.77; 95% confidence interval, 5.93–636.74) . Antibodies to gp210 have been more frequent in patients with severe cholestasis, and impaired liver function than in patients without these markers , and they have occurred more often in patients with PBC and a poor outcome than in patients who have done well [175, 197]. Antibodies to gp210 afford an exciting opportunity to gain insight into the pathogenic mechanisms of PBC that influence disease severity and progression as well as expand the clinical resources for predicting disease behavior in individual patients.
Antibodies to Centromere
Antibodies to centromere (anti-centromere) are directed against the kinetochore proteins (commonly assayed as anti-Cenp-A or anti-Cenp-B) that develop during mitosis (Fig. 2). These proteins are the locations where spindle fibers attach so that sister chromatids can move to opposite poles of the dividing cell nucleus [198, 199] (Table 2). Antibodies to centromere occur in 9–60% of patients with PBC, depending on the presence or absence of concurrent systemic sclerosis, Sjögren’s syndrome, CREST (calcinosis, Raynaud’s phenomenon, esophageal dysfunction, sclerodactyly, and telangiectasia) syndrome, or other autoimmune disorders [197, 200–206]. The specificity of anti-centromere for PBC has been limited since these antibodies have been described in diverse rheumatic conditions [203, 207–209], chronic hepatitis C , and autoimmune hepatitis [2, 211]. Furthermore, anti-centromere are commonly found with multiple other concurrent antibodies to nuclear antigens [2, 211], and the presence of anti-centromere has not characterized a clinical phenotype in PBC other than to suggest the likelihood of another concurrent autoimmune disease [200, 201, 204]. Recent studies in PBC using enzyme immunoassays for the detection of anti-centromere have indicated that in the proper clinical context, anti-centromere have prognostic value [170, 196, 197, 212].
Patients with PBC and anti-centromere progress more rapidly to liver failure (48% versus 26% after 8.9 years) , have higher serum alkaline phosphatase levels , more severe bile duct injury on histological evaluation , and greater frequency of portal hypertension [170, 196, 197] than patients without these antibodies. The actual bases for these clinical associations remain unclear, especially since the contribution of other concurrent autoantibodies and autoimmune diseases to the severity and progression of PBC remain uncertain and all studies have not confirmed the associations . Nevertheless, the assessment of the antinuclear antibodies in PBC is an important frontier to explore as it may aid in identifying individual patients early who require close supervision.
Pitfalls in the Use of Serological Markers for Prognosis
Pitfalls and promises in the serological markers of prognosis in autoimmune liver disease
Over-interpretation of non-pathogenic serological profile and unnecessary treatment modifications 
Low negative predictability for poor outcome 
Negative and positive predictabilities of disease severity and outcome will improve as critical disease mechanisms are clarified 
Variable serum levels during course of disease and its treatment 
Tighter correlations will emerge between serological behavior of critical markers and disease course 
Serological findings will be associated with host-specific genetic factors that influence clinical phenotype and disease behavior 
Batteries of serological markers will compensate for individual variations in disease expression and reflect multiple pathogenic components 
No confident minimum level of positivity 
Distinctions between collateral and critical autoantibody production will be determined 
Lack of standardized assays 
International serum exchange with calibrated reference sera will ensure standardized testing 
Limited access to serological tests 
Appropriate testing algorithm will be codified and individual tests will be commercially available 
Autoantibody levels may vary during the course of the disease and its treatment , and the disappearance and reappearance of these markers constitute another pitfall in their clinical application (Table 3). The strength of autoantibody production may be host-specific rather than disease-dependent [42, 54], and autoantibody titers or levels have not been closely correlated with disease severity or outcome [22, 23, 25]. There is no confident minimum level of positivity that can be discounted; the disappearance of a particular autoantibody does not preclude the possibility of disease progression; and the reappearance of the serological marker is of unknown clinical significance [22, 214]. A positive result at any level and at any time in the course of the disease should be interpreted similarly until future investigations refine that interpretation.
Another pitfall in the serological diagnosis is the lack of standardized assays for each serological marker (Table 3). Commercial kits of enzyme immunoassays based on recombinant antigens are replacing indirect immunofluorescence for the detection of most autoantibodies, and many assays are institution-dependent. The enzyme immunoassays eliminate intra- and inter-observer variation, but many clinical correlations that have been made by indirect immunofluorescence may not be duplicated by other assays .
Promises in the Use of Serological Markers for Prognosis
The pathogenic pathways of autoimmune hepatitis are being unraveled, and the critical autoantigens that sensitize liver-infiltrating cytotoxic lymphocytes and antibody-producing plasma cells are being identified and characterized [149, 216–220]. The serological markers that reliably reflect these critical pathways will undoubtedly improve, and the negative and positive predictabilities of these new markers for disease severity and outcome will increase  (Table 3).
Genetic factors that influence the clinical phenotype of autoimmune hepatitis will continue to be clarified, and the host-dependent elements that influence disease expression will be better defined . Antibodies to soluble liver antigen are closely associated with HLA DRB1*0301 [50, 51], and anti-LKM1 are associated with HLA DRB1*07 . Autoantibodies of prognostic relevance may be surrogate markers of host-related genetic factors that in turn influence disease severity and treatment response. The autoantibodies of prognostic relevance may not only reflect immune responses to key antigenic triggers but also genetic factors that shape the clinical phenotype and consequences of the disease (Table 3).
Standardized methods of serological testing will be established (Table 3), and the International Autoimmune Hepatitis Group has already proposed an international serum exchange with calibrated reference sera . In this fashion, testing methods can be standardized and discrepancies between laboratories and clinical experiences minimized. A similar exchange workshop has been in place since 1986 for the detection of autoantibodies associated with type 1 diabetes mellitus. The serological assays in diabetes must be validated under international workshop conditions before publication of results.
Lastly, batteries of the implicated serological markers must be studied to determine if a cluster of autoantibodies has more prognostic value than any single marker  (Table 3). Performance parameters for one test may be influenced by factors, including genetic predisposition, patient age, concurrent autoimmune diseases, treatment strategy and disease stage, that do not affect other markers. Furthermore, each marker may reflect a pathogenic component not uniformly present or prominent in every patient. Constellations of the serological markers may compensate for individual variations in disease expression, and they may provide a more comprehensive impression of the disease process than any single test. Autoantibodies have been invaluable in the diagnosis of autoimmune liver disease, and their promise as prognostic indices constitutes the exciting new horizon.
This work received no financial support from a funding agency or institution.