Anti-gp210 and anti-centromere antibodies are different risk factors for the progression of primary biliary cirrhosis

Authors


  • Potential conflict of interest: Dr. Norman and Dr. Shums are employees of Inova Diagnostics, Inc.

Abstract

The predictive role of antinuclear antibodies (ANAs) remains elusive in the long-term outcome of primary biliary cirrhosis (PBC). The progression of PBC was evaluated in association with ANAs using stepwise Cox proportional hazard regression and an unconditional stepwise logistic regression model based on the data of 276 biopsy-proven, definite PBC patients who have been registered to the National Hospital Organization Study Group for Liver Disease in Japan (NHOSLJ). When death of hepatic failure/liver transplantation (LT) was defined as an end-point, positive anti-gp210 antibodies (Hazard ratio (HR) = 6.742, 95% confidence interval (CI): 2.408, 18.877), the late stage (Scheuer's stage 3, 4) (HR = 4.285, 95% CI:1.682,10.913) and male sex (HR = 3.266, 95% CI: 1.321,8.075) were significant risk factors at the time of initial liver biopsy. When clinical progression to death of hepatic failure/LT (i.e., hepatic failure type progression) or to the development of esophageal varices or hepatocellular carcinoma without developing jaundice (Total bilirubin < 1.5 mg/dL) (i.e., portal hypertension type progression) was defined as an end-point in the early stage (Scheuer's stage 1, 2) PBC patients, positive anti-gp210 antibodies was a significant risk factor for hepatic failure type progression [odds ratio (OR) = 33.777, 95% CI: 5.930, 636.745], whereas positive anti-centromere antibodies was a significant risk factor for portal hypertension type progression (OR = 4.202, 95% CI: 1.307, 14.763). Histologically, positive anti-gp210 antibodies was most significantly associated with more severe interface hepatitis and lobular inflammation, whereas positive anticentromere antibodies was most significantly associated with more severe ductular reaction. Conclusion: These results indicate 2 different progression types in PBC, hepatic failure type and portal hypertension type progression, which may be represented by positive-anti-gp210 and positive-anticentromere antibodies, respectively. (HEPATOLOGY 2007;45:118–127.)

Primary biliary cirrhosis (PBC) is a chronic, progressive, cholestatic autoimmune liver disease characterized by destruction of intrahepatic bile ducts, portal inflammation, and development of cirrhosis and hepatic failure.1, 2 Although the serological hallmark for diagnosis of PBC is the presence of antimitochondrial antibodies (AMAs) which is found in more than 90% of patients, antinuclear antibodies (ANAs) are also specifically detected in 30% to 50% of PBC patients.3–5 Among these ANAs, anti-gp210 and anti-sp100 antibodies are highly specific for PBC and are useful for the diagnosis of PBC patients who are negative for AMAs.4–6 In addition, whereas AMAs are not associated with the disease progression, ANAs are associated with disease severity and therefore are a marker of poor prognosis.7–14

Itoh et al.7 reported for the first time the association between the presence of anti-gp210 antibodies and the prognosis of PBC. Thereafter, Invernizzi et al.,8 Muratori et al.,9 and Miyachi et al.10 reported the association between autoantibodies against nuclear pore complexes and disease severity in PBC. In addition to these cross-sectional studies, we recently found, by a long-term follow-up study of 80 PBC patients in National Hospital Organization Study Group for Liver Disease in Japan (NHOSLJ), that PBC patients in whom anti-gp210-C terminal peptide antibodies are sustained at a high level are at high risk for the progression to end-stage hepatic failure as compared with those in whom anti-gp210-C terminal peptide antibodies are negative or initially positive but decreased to a low level after treatment with ursodeoxycholic acid.12 Wesierska-Gadek et al.14 recently reported the significance of antibodies to nuclear pore complexes in the long-term outcome of PBC. These reports may indicate that antibodies to nuclear pore complex, especially antibodies to gp210-C terminal peptides, can be surrogate markers for the progression of PBC to end-stage hepatic failure. However, most of these studies were retrospective, and confirmation is still needed by properly designed prospective studies.

As for the predictive value of other ANAs, Yang et al.13 recently reported that ANAs in general, and anti-centromere antibodies in particular, are associated with hepatic failure in PBC, whereas other reports did not show any significant effect of anti-centromere antibodies on the prognosis of PBC.15, 16 However, the relative risk of each ANAs for the progression of PBC was not simultaneously investigated by multivariate analysis in these studies. Thus, the significance and predictive value of ANAs seen in other autoimmune diseases, such as type 1 diabetes, rheumatoid arthritis, systemic lupus erythematosus, and inflammatory bowel diseases, remains to be elucidated in PBC.17–21

The objective of this study was to determine the clinical significance of ANAs (i.e., anti-gp210, anti-centromere, anti-sp100, anti-chromatin) in relation to the progression of the disease. By the analysis of 276 PBC patients who have been registered in retrospective/prospective cohort study of NHOSLJ, we confirmed the significance of anti-gp210 antibodies as a risk factor for the progression to end-stage hepatic failure. In addition, we identified the significance of anti-centromere antibodies as a risk factor for the progression to the development of esophageal varices or HCC without developing jaundice (serum total bilirubin level < 1.5 mg/dl). Based on these results, we propose 2 different progression types in PBC. One is a hepatic failure type progression which is represented by positive-anti-gp210 antibodies, and the other is a portal hypertension type progression, which is represented by positive-anti-centromere antibodies. This new clinical classification of PBC patients, based on ANAs in relation to long-term outcome, should be useful for monitoring and selecting PBC patients in whom treatment is necessary.

Abbreviations

AIH, autoimmune hepatitis; AMAs, anti-mitochondrial antibodies; ANAs, anti-nuclear antibodies; CHC, chronic viral hepatitis C; CREST, calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, and telangiectasia; ELISA, enzyme-linked immunosorbent assay; HR, hazard ratio; LAD, liver-associated diseases; LC, liver cirrhosis; LT, liver transplantation; NHOSLJ, National Hospital Organization Study Group for Liver Disease in Japan.

Materials and Methods

Patients and Serum Samples.

A total of 463 PBC patients were registered to the PBC cohort study in NHOSLJ from January 2003 to January 2006. Serum samples were serially bled between August 1982 and January 2006 and have been stored at −20°C at each institution until use. Among these PBC patients, the following patients were excluded from the present analysis: those in whom liver biopsy was not performed at the time of diagnosis, those in whom serum samples were not available at the time of initial liver biopsy, those who had episodes of acute or autoimmune hepatitis (AIH) (alanine aminotransferase > 500 IU/l, aspartate aminotransferase > 500 IU/L), those who had been given a maintenance dose of prednisolone more than 5 mg/body for concomitant AIH, and those who had serological markers for persistent hepatitis virus B or C infection. Application of these criteria resulted in a total cohort of 276 PBC patients (35 male, 241 female, age 30-83, median 58 years of age, mean ± SD 57.8 ± 9.8) being analyzed (Table 1).

Table 1. Demographic and Immunological Features of PBC Patients with Early or Late Stage at the Time of Initial Liver Biopsy
ParameterTotal (n = 276)Early Stage (n = 217)Late Stage (n = 59)P Value*
  • *

    P value for t test of means or Fisher's exact test of proportions. Abbreviations: LAD, liver associated disease; LT: liver transplantation

Age (Range, median)57.8 ± 9.8 (30–83, 58)57.5 ± 9.3 (32–79, 57.5)59.0 ± 11.8 (30–83, 61)0.333
Male (%)35 (12.7%)20 (9.2%)15 (25.4%)0.00187
anti-gp210 +ve72 (26.1%)43 (19.8%)29 (49.2)1.79 × 10−5
anti-centromere +ve72 (26.1%)57 (26.3%)15 (25.4%)1.0 a
anti-sp100 +ve24 (8.7%)18 (8.3%)6 (10.2%)0.650
anti-chromatin +ve15 (5.4%)12 (5.5%)3 (5.1%)1.0
MIT3 +ve227 (82.2%)176 (81.1%)51 (86.4%)0.442
Observation Period (months) (Range, median)75.9 ± 59.9 (1–292, 60.5)76.2 ± 60.8 (2–292, 63)74.8 ± 59.6 (1–235, 63)0.882
Progression to death by LAD, or LT28 (10.1%)8 (3.7%)20 (33.9%)1.82 × 10−9

All of these patients showed abnormal serum chemistry, suggesting bile duct injury at the time of initial liver biopsy. Histological diagnosis was performed based on the internationally accepted criteria for PBC, and the histological staging was performed according to Scheuer's classification.22, 23 The 217 patients were at early stage (Scheuer's stage 1 and 2), and 59 patients were at late stage (Scheuer's stage 3 and 4) at the time of initial liver biopsy (Table 1).

Among 276 patients, 21 patients (7.6%) had concomitant autoimmune diseases: Sjögren's syndrome (n = 8), Hashimoto's thyroiditis (n = 3), Sjögren's syndrome + Hashimoto's thyroiditis (n = 2), Sjögren's syndrome + systemic sclerosis (n = 1), CREST syndrome (n = 1), Raynaud's disease (n = 2), rheumatoid arthritis (n = 2), hyperthyroidism (n = 2). The treatment for PBC during observation periods was as follows: ursodeoxycholic acid (300-900 mg/d) alone (n = 210), bezafibrate (200-400 mg/d) alone (n = 6), ursodeoxycholic acid + bezafibrate (n = 34), ursodeoxycholic acid or bezafibrate + maintenance prednisolone less than 5 mg/day (n = 14), medication free (n = 12).

At least 2 serum samples obtained at different occasions were used for the measurement of autoantibodies during the course of observation period (range, 1-292 months; median, 60.5 months; mean ± SD, 75.9 ± 59.9 months). As disease controls, autoantibodies were measured using serum samples from 20 patients with AIH, 20 patients with chronic viral hepatitis C, and 40 patients with liver cirrhosis caused by hepatitis virus B or C infection. Sera from 30 healthy subjects were used as healthy controls.

The presence of esophageal varices, ascites, and HCC were evaluated by gastrointestinal endoscopy, ultrasound sonography, and computed tomography.

Enzyme-Linked Immunosorbent Assay.

The serum antibody titers to gp210-C terminal peptide a.a.1863-1887 (SPNALPPARKASPPSGLWSPAYASH) were measured by enzyme-linked immunosorbent assay (ELISA) as described elsewhere.12 In brief, an ELISA plate coated with 0.5 μg gp210 C-terminal peptide was incubated with 100 μL serum samples diluted 1:100 for 60 minutes at room temperature. The bound antibodies were detected by peroxidase-conjugated anti-human IgG. The antibody titers were calculated with reference to the standard serum and the antibody titers more than 6 units/ml (mean value + 5 SD of the titers of 30 healthy controls) were arbitrarily determined as positive for antibodies.12 The antibody titers to mitochondrial antigens MIT3 (recombinant proteins containing PDC-E2, BCOADC-E2, OGDC-E2), centromere A&B (recombinant centromere proteins), sp100 (synthetic peptides), and chromatin (H1-stripped chromatin) were determined using ELISA kits (INOVA Diagnostics, San Diego, CA), and antibody titers more than 25 U/ml were interpreted as positive according to the manufacturer's protocol and instructions.

Liver Biopsy Specimen and Histological Evaluation.

A total of 91 initial needle liver biopsy specimens (Scheuer's stage 1 n = 44, stage 2 n = 27, stage 3 n = 20) from 91 PBC patients (13 male, 78 female) who fulfilled the above diagnostic criteria for PBC were histologically analyzed.

The 4-μm-thick formalin-fixed and paraffin-embedded sections were routinely stained with hematoxylin and eosin, Azan Mallory, reticulin silver impregnation, and rhodamine. In addition, cytokeratin 7 was immunohistochemically stained in all specimens as described.24 In brief, formalin-fixed, paraffin-embedded sections were deparaffinized in xylene and subsequently rehydrated in sequential ethanols (100%-70%). After washing with PBS, antigen retrieval was performed in 10 mM citrate buffer, PH 6.0, by heating at 100°C for 10 minutes by microwave. Then, the sections were treated with Peroxidase Blocking (DAKO Japan, Kyoto, Japan) for 5 minutes, and incubated with anti-cytokeratin 7 monoclonal antibody (Dako Cytomation, Botany, Australia, dilution 1:100) for 60 minutes at room temperature. A standardized 2-step method with ENVISION+ (DAKO) was used for the visualization of bound antibody using 3, 3′-diaminobenzidine as a chromogen (DAKO), and counterstained with Mayer's hematoxylin (DAKO).

The histological variables examined included fibrosis (0-4) (0, absent; 1, expansion of fibrosis to parenchyma; 2, portal-central or portal-portal bridging fibrosis; 3, presence of numerous fibrous septa; 4, cirrhosis), portal inflammation (0-3) (0, absent; 1, mild: mild cell infiltration; 2, moderate: dense infiltration of one portal area; 3, severe: dense infiltration of more than 2 portal areas), interface hepatitis (0-3) (0, absent; 1, mild: interface hepatitis less than one third the circumference of one or two portal tracts; 2, moderate: interface hepatitis less than two thirds the circumference of one or two portal tracts; 3, severe: interface hepatitis more than two thirds the circumference of one or two portal tracts, lobular inflammation (0-3) (0, absent; 1, mild; 2, moderate; 3, severe), ductular reaction (1-3) (1, absent or mild; 2, moderate; 3, severe). On completion of the evaluation of each of these variables, histological staging using Scheuer's classification (stage 1-4) was determined.23

Each biopsy specimen was analyzed by 2 independent observers (M.I. and Y.T.). In case of initial disagreement in the assessment, consensus was reached on further review.

Ethics Board.

The study has been approved by the Ethics Board at each medical center affiliated with NHOSLJ and conducted after obtaining informed consent from each subject for use of their serum samples and liver biopsy samples to advance knowledge on the cause of PBC.

Statistical Analysis.

We used step-wise Cox regression analysis to determine the independent prognostic significance of ANAs values, sex, PBC stage, and age at the time of initial liver biopsy in all PBC patients. The end point of these analyses were (1) death of hepatic failure/liver transplantation (LT) or (2) death of liver-associated diseases (LAD)/LT. LAD includes end-stage hepatic failure, HCC, and bleeding from esophageal varices. Each ANA value was dichotomized into positive and negative. The observation period was calculated as the time from the date of initial liver biopsy until the date of death, date of LT, last contact, death of non–liver-associated diseases, or end of follow-up, whichever came first. We estimated hazard ratios (HR) and corresponding 95% confidence intervals (95%CI) of those variables for each endpoint. We checked the proportional hazards assumption through log plots.

We used unconditional stepwise logistic regression to determine the independent prognostic significance of ANAs values, sex, and age in early stage (Scheuer's stages 1, 2) PBC patients at the time of initial liver biopsy for two different types of PBC progression. These include hepatic failure type progression and portal hypertension type progression. Hepatic failure type progression was defined as the progression to end-stage hepatic failure/LT irrespective of the presence or absence of complications of portal hypertension, and portal hypertension type progression was defined as the clinical progression to develop HCC or complications of portal hypertension including esophageal varices without developing jaundice (serum total bilirubin level < 1.5 mg/dL). The risk of PBC progression in positive cases was compared with that in the negative cases.

Relative risks of histopathological findings in relation to ANAs values were similarly estimated as odds ratio (OR) and corresponding 95 % CI by unconditional stepwise logistic regression. The risk of histopathological findings in ANA-positive cases was compared with that in the ANA-negative cases.

All statistical tests were based on 2-sided probability, and significance levels were 0.05. Statistical analyses were conducted by SAS, version 9.3 (SAS Institute Inc., Cary, NC).

Results

Clinical, Immunological, and Histological Data in Early-Stage or Late-Stage PBC Patients at the Time of Initial Liver Biopsy.

As shown in Table 1, 217 and 59 PBC patients were in early (Scheuer's stage 1, 2) and late stage (Scheuer's stage 3, 4), respectively, at the time of initial liver biopsy. The frequency of male sex was significantly higher in late-stage patients as compared with early-stage PBC patients, whereas the age and observation period were similar between the 2 groups.

Seropositivity for anti-gp210, anti-centromere, anti-sp100, and anti-chromatin antibodies at the time of initial liver biopsy in PBC patients is also shown in Table 1. As for disease controls, seropositivity for anti-gp210, anti-centromere, anti-sp100, and anti-chromatin antibodies were 0.0%, 25.0%, 15.0%, and 45.0%, respectively, in AIH patients; 0.0%, 0.0%, 10.0% and 0.0%, respectively, in patients with chronic viral hepatitis C (CHC) patients; and 0.0%, 10.0%, 0.0% and 0.0%, respectively, in patients with cirrhosis caused by chronic hepatitis virus B or C infection. Seropositivity for these ANAs was all 0.0% in healthy controls. Thus, anti-gp210 antibodies were highly specific for PBC, and seropositivity for anti-gp210 antibodies was significantly higher in patients with PBC as compared to patients with AIH, CHC, or cirrhosis (P = 0.0056, P = 0.0056, P = 0.00038, respectively). Furthermore, the seropositivity for anti-gp210 antibodies was significantly higher in late-stage patients as compared with early-stage PBC patients (P = 1.79 × 10-5).

Seropositivity for anti-centromere antibodies in patients with PBC was similar to that in AIH patients (P = 1.0) but was significantly higher than that in patients with cirrhosis and CHC (P = .00526, P = .0285, respectively). Seropositivity for anti-sp100 antibodies was similar among PBC, AIH, and CHC patients, whereas seropositivity for anti-chromatin antibodies was significantly higher in AIH patients as compared with patients with PBC, CHC, or cirrhosis (P = 3.0 × 10-6, P = .00122, and P = 6.7 × 10-5, respectively). The seropositivity for anti-centromere, anti-sp100, and anti-chromatin antibodies was similar between early-stage and late-stage PBC patients.

Among 72 anti-centromere–positive PBC patients, 11 patients (15.3%) had the following concomitant autoimmune diseases, Sjögren's syndrome (n = 3), Sjögren's syndrome + Hashimoto's thyroiditis (n = 2), Sjögren's syndrome + systemic sclerosis (n = 1), rheumatoid arthritis (n = 1), Hashimoto's thyroiditis (n = 1), CREST syndrome (n = 1), Raynaud's phenomenon without definite rheumatic disorders (n = 2). Among 204 anti-centromere–negative PBC patients, 10 patients (4.9%) had the following concomitant autoimmune diseases: Sjögren's syndrome (n = 5), Hashimoto's thyroiditis (n = 2), rheumatoid arthritis (n = 1), hyperthyroidism (n = 2). The frequency of these concomitant autoimmune diseases is significantly higher in anti-centromere–positive patients as compared with anti-centromere–negative patients (P = 0.00808). There was no significant difference of the frequency of these concomitant autoimmune diseases between other ANA-positive and -negative patients (data not shown).

Clinical Course of Patients with PBC During Observation Periods.

Among 43 early-stage PBC patients who were initially positive for anti-gp210 antibodies, anti-gp210 antibodies became consistently negative in seven patients within 3 months after the initiation of ursodeoxycholic acid treatment. Among 174 early-stage PBC patients who were initially negative for anti-gp210 antibodies, anti-gp210 antibodies became consistently positive in eight patients. The titers of anti-gp210 antibodies fluctuated from positive to negative and vise versa in seven PBC patients (early stage n = 4, late stage n = 3 at the time of initial liver biopsy). By contrast, the titers of anti-centromere antibodies did not significantly change from positive to negative or vise versa throughout the observation period (1-292 months; median, 60.5 months).

Among 217 patients in early stage at the initial liver biopsy, one patient died of HCC and seven patients progressed to end-stage hepatic failure, in whom five patients underwent LT and two died of hepatic failure without undergoing LT during observation period (2-292 months; median, 63 months). Among 59 patients in late stage at the initial liver biopsy, four patients died of HCC and 16 patients progressed to end-stage hepatic failure; of these, five patients underwent LT and 11 died of hepatic failure without undergoing LT during the observation period (1-235 months; median, 63 months).

The frequency of patients who progressed to death of LAD including end-stage hepatic failure, HCC, and bleeding from esophageal varices or who underwent LT for end-stage hepatic failure was significantly higher in late-stage patients at the initial liver biopsy (33.9%) as compared with early-stage patients at the initial liver biopsy (3.7%) (P = 1.82 × 10−9).

Univariate Analysis of Transplantation-Free Survival.

As depicted in Fig. 1A, the frequency of the progression to death of LAD/LT was significantly higher in late-stage PBC patients as compared with early-stage PBC patients by Kaplan-Meier method (log-rank test P = 1.35 × 10−10). The frequency of the progression to death of LAD/LT was also significantly higher in male as compared with female patients (log-rank test, P = 7.38 × 10-6) (Fig. 1b).

Figure 1.

Probability of survival free of LT calculated by Kaplan-Meier method in 276 patients with biopsy-proven PBC. (A) Patients with PBC were classified into early stage (Scheuer's stage 1, 2) and late stage (Scheuer's stage 3, 4) at the time of initial liver biopsy. The frequency of death of LAD/LT was significantly higher in late-stage patients as compared with early-stage PBC patients (P = 1.35 × 10−10) by log-rank test. (B) PBC patients were classified into male and female at the time of initial liver biopsy. The frequency of death of LAD/LT was significantly higher in male as compared with female patients (P = 7.38 × 10-6) by log-rank test.

As for ANAs, the frequency of the progression to death of LAD/LT was significantly higher in the anti-gp210–positive patients as compared with anti-gp210–negative patients (log-rank test, P = 1.3 × 10-7)(Fig. 2A), whereas there was no statistically significant difference in the frequency of the progression to death of LAD/LT between anti-centromere–positive and –negative (Fig. 2B), anti-sp100–positive and –negative (Fig. 2C), and anti-chromatin–positive and –negative patients (Fig. 2D).

Figure 2.

Probability of survival free of LT calculated by Kaplan-Meier method in 276 patients with biopsy-proven PBC. (A) Patients with PBC were classified into those who were positive and negative for anti-gp210 antibodies at the time of initial liver biopsy. The frequency of death of LAD/LT was significantly higher in the anti-gp210–positive patients as compared with anti-gp210–negative patients (P = 1.3 × 10−7) by log-rank test. (B) Patients with PBC were classified into anti-centromere antibodies–positive and –negative patients at the time of initial liver biopsy. The frequency of death of LAD/LT was lower in the anti-centromere–positive patients as compared with anti- centromere–negative PBC patients, although the difference was not statistically significant (P = 0.0510) by log-rank test. (C) Patients with PBC were classified into anti-sp100 antibodies–positive and –negative patients at the time of initial liver biopsy. The frequency of death of LAD/LT was similar between the two groups (P = 0.69685) by log-rank test. (D) Patients with PBC were classified into anti-chromatin antibodies–positive and –negative patients at the time of initial liver biopsy. The frequency of death of LAD/LT was similar between the 2 groups (P = 0.40132) by log-rank test.

Multivariate Analysis of Anti-nuclear Antibodies in Relation to Death of Hepatic Failure/LT or Death of LAD/LT.

Step-wise Cox regression analyses showed that male sex, positive anti-gp210 antibodies, and late-stage (Scheuer's stage 3, 4) were significant and independent risk factors of the death of hepatic failure/LT or death of LAD/LT (Table 2). When death of hepatic failure/LT was defined as an end point, positive anti-gp210 antibodies was the strongest independent risk factor followed by late stage and male sex. When death of LAD/LT was defined as an end point, the late stage was the strongest risk factor followed by positive anti-gp210 antibodies and male sex. Positive anti-centromere antibodies and age were not significant independent risk factors when death of hepatic failure/LT or death of LAD/LT was defined as an end point.

Table 2. Hazard Ratios of Stepwise Cox Regression Analysis and 95% Confidence Intervals for Death of Hepatic Failure/LT** or Death of LAD*/LT
VariablesHazard ratio (95% confidence interval) for death of
hepatic failure/LTLAD/LT
  • *

    LAD: liver-associated diseases include end-stage hepatic failure, hepatocellular carcinoma and bleeding from esophageal varices

  • **

    LT: liver transplantation

  • ***

    only the results revealing significant independent predictor variables are shown.

Sex, male3.266 (1.321, 8.075)***3.097 (1.322, 7.254)
Age(one year−1)
anti-gp210 +ve6.742 (2.408, 18.877)3.390 (1.499, 7.665)
anti-centromere +ve
anti-sp100 +ve
anti-chromatin +ve
Stage 3 or 44.285 (1.682, 10.913)5.697 (2.406, 13.491)

Analysis of Antinuclear Antibodies in Relation to Hepatic Failure Type or Portal Hypertension Type Progression From Early Stage (Stage 1, 2).

During the observation period, 7 of 217 early-stage PBC patients showed hepatic failure type progression (2 patients died of end-stage hepatic failure and 5 patients underwent LT) and 16 of 217 early-stage PBC patients showed portal hypertension type progression (1 patient died of HCC in whom esophageal varices were positive and 15 patients developed esophageal varices without developing jaundice; total bilirubin <1.5 mg/dL)(Table 3). The frequency of PBC patients who showed hepatic failure type progression was significantly higher in anti-gp210–positive patients (14.0%) as compared with anti-gp210–negative patients (0.6%) (P = 0.000268). The frequency of PBC patients who showed portal hypertension type progression was significantly higher in anti-centromere–positive patients (17.5%) as compared with anti-centromere–negative patients (3.8%) (P = 0.00165) (Table 3). No significant difference was seen in the frequency of both hepatic failure type and portal hypertension type progression between anti-sp100–positive and –negative patients or between anti-chromatin–positive and –negative patients.

Table 3. The number of PBC Patients in Early Stage at the Time of Initial Liver Biopsy Who Showed Portal Hypertension Type–Progression or Hepatic Failure Type Progression
AntibodyNumber of PatientsNumber of Patients Who Showed Progression
TotalPortal Hypertension TypeHhepatic Failure Type
  • *

    p < 0.0001 for Fisher's exact test of proportions

  • **

    p < 0.001 for Fisher's exact test of proportions

anti-gp210posi. 4312 (27.9%)*6 (14.0%)6 (14.0%)*
 nega. 17411 (6.3%)10 (5.7%)1 (0.6%)
anti-centromereposi. 5712 (21.0%)**10 (17.5%)**2 (3.5%)
 nega. 16011 (6.9%)6 (3.8%)5 (3.1%)
anti-sp100posi. 183 (16.7%)3 (16.7%)0 (0.0%)
 nega. 19920 (10.0%)13 (6.5%)7 (3.5%)
anti-chromatinposi. 122 (16.7%)1 (8.3%)1 (8.3%)
 nega. 20521 (10.2%)15 (7.3%)6 (2.9%)

Unconditional logistic regression analyses showed that positive anti-gp210 and anti-centromere antibodies were significant and independent risk factors of PBC progression (Table 4). However, the same analysis, which was stratified by the progression type, showed that positive anti-gp210 antibodies was a risk factor of only hepatic failure type progression but not portal hypertension type progression, and that positive anti-centromere antibodies was a risk factor of only portal hypertension type progression but not hepatic failure type progression (Table 4). As for portal hypertension type progression, age was an independent risk factor in addition to positive anti-centromere antibodies. These results indicate that anti-gp210 and anti-centromere antibodies might be a risk factor of different progression types of PBC.

Table 4. Odds Ratios of Unconditional Step-wise Logistic Regression Analysis and 95% Confidence Intervals for Hepatic Failure Type Progression or Portal Hypertension Type Progression from Early Stage
VariablesOdds Ratio (95% Confidence interval) for Progression
TotalPortal Hypertension TypeHepatic Failure Type
  • *

    only the results revealing significant independent predictor variables are shown

Sex, male
Age (one year−1)1.078* (1.009, 1.159)
anti-gp210 +ve7.086 (2.649, 20.206)33.777 (5.930, 636.745)
anti-centromere +ve4.486 (1.659, 12.778)4.202 (1.307, 14.763)— a
anti-sp100 +ve
anti-chromatin +ve

Analysis of Antinuclear Antibodies in Relation to Histological Variables.

To dissect the predictive role of anti-gp210 and anti-centromere antibodies in the progression of PBC, the associations of ANAs with histological variables were assessed using 91 initial biopsy specimens (Scheuer's stage 1 n = 44, stage 2 n = 27, stage 3 n = 20) derived from 91 patients with PBC (male 13, female 78). Odds ratios for portal inflammation score more than 2, interface hepatitis score more than 1, ductular reaction score more than 2, fibrosis score more than 2, and lobular inflammation score more than 1 were estimated in association with ANAs by use of unconditional stepwise logistic regression models. These analyses showed that positive anti-gp210 antibodies was a risk factor of more severe interface hepatitis, lobular inflammation, and ductular reaction and that positive anti-centromere antibodies was a risk factor of more severe ductular reaction, interface hepatitis, and portal inflammation (Table 5). Positive anti-chromatin antibodies was also a risk factor of more severe interface hepatitis (Table 5).

Table 5. Odds Ratios of Unconditional Step-wise Logistic Regression Analysis and 95% Confidence Intervals for Histological Variables in Stage 1, 2 and 3 PBC Patients
VariablesOdds Ratio (95% Confidence Interval) for
Portal Inflammationinterface hepatitisDuctular ReactionLobular Inflammation
  • *

    only the results revealing significant independent predictor variables are shown

anti-gp2104.38* (1.6, 12.76)2.85 (1.02, 8.32)4.05 (1.30, 13.50)
anti-centromere2.66 (0.84, 8.17)3.11 (1.07, 9.84)5.70 (1.97, 17.55)
anti-sp100
anti-chromatin14.06 (2.29, 273.93)

Discussion

We simultaneously measured serum antibodies to nuclear antigens (gp210, sp100, centromere, and chromatin) by ELISA, the most quantitative and specific assay to measure each ANA, and analyzed the clinical significance of these ANAs in relation to the progression pattern of PBC in a retrospective and prospective NHOSLJ cohort of 276 patients with biopsy-proven PBC. This study confirmed and extended our previous observation by demonstrating that anti-gp210 antibodies are the strongest predictive factor among ANAs for the progression to end-stage hepatic failure (i.e., hepatic failure type progression).12 In addition, we demonstrated for the first time that anti-centromere antibodies are a significant predictive factor for the progression to develop a complication of portal hypertension such as esophageal varices without developing jaundice (portal hypertension type progression). Based on these observations, we hypothesize that there are at least two different types of progression in PBC. One is a hepatic failure type progression, which is represented by positive anti-gp210 antibodies, and the other is a portal hypertension type progression, which is represented by positive anti-centromere antibodies.

Anti-centromere antibodies are found in approximately 50% of CREST syndrome (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, and telangiectasia), a benign variant of systemic sclerosis.25 Anti-centromere antibodies are also found in other autoimmune diseases including the diffuse form of systemic sclerosis, Sjögren's syndrome, systemic lupus erythematosus, rheumatoid arthritis, and Raynaud's phenomenon without definite rheumatic disorders in lower frequency.26, 27 Interestingly, anti-centromere antibodies are also found in approximately 20% to 30% of patients with PBC.15, 16, 28, 29 As for the clinical significance of anti-centromere antibodies, Hossny et al.27 reported that it is linked to Raynaud's phenomenon whatever its cause. Parveen et al.15 reported that there is no correlation between anti-centromere antibodies and clinical, histological, and laboratory findings in PBC. Rigamonti et al.16 reported that PBC-alone patients with positive anti-centromere antibodies had the same rate of liver-related deaths as those with PBC-alone negative for anti-centromere antibodies, whereas PBC patients with concomitant systemic sclerosis had a significantly lower rate of progression to transplantation or death of liver-associated diseases.16 By contrast, Yang et al.13 reported that anti-centromere antibodies are associated with the development of liver failure in PBC. Thus, the predictive or diagnostic value of anti-centromere antibodies has been controversial in PBC. These results in the current study clearly show that patients with PBC who are positive for anti-centromere antibodies are at high risk for portal hypertension type progression, but not for hepatic failure type progression.

Anti-sp100 antibodies have been reported to be positive in approximately 20% to 30% of patients with PBC.5, 11 Anti-sp100 positivity is more common in patients with advanced stages of disease, and faster disease progression rates were observed among anti-sp100–positive PBC patients.5, 11 However, positive anti-sp100 antibodies were not a statistically significant risk factor for hepatic failure type progression nor portal hypertension type progression in the current study. The cause of this difference in results is unclear but could be a result of differences in the assay system or in the patient population. Conversely, anti-chromatin antibodies are positive in approximately 75% of systemic lupus erythematosus patients, up to 100% of drug-induced lupus patients, and 20% to 50% of patients with autoimmune hepatitis type 1.30 Correspondingly, positive anti-chromatin antibodies were shown to be an independent risk factor for more severe interface hepatitis in the current study, indicating the concomitant presence of AIH or AIH-like component of interface hepatitis in PBC patients enrolled in the current study. Further study is needed to elucidate the clinical significance of these ANAs.

Based on the histological observation of PBC liver, it has been assumed that two main mechanisms are involved in the progression of PBC.1, 22, 23, 31–35 The first is bile duct destruction leading to chronic cholestasis and development of fibrosis or cirrhosis of biliary pattern. The second is interface hepatitis, which tends to lead to cirrhosis, the pattern of which resembles cirrhosis following various forms of chronic viral hepatitis. The current study demonstrated that positive-anti-gp210 antibodies are a risk factor for more severe interface hepatitis, lobular inflammation, and ductular reaction. Furthermore, there was a trend of more severe ductopenia in late stage of anti-gp210–positive patients (data not shown). These results might indicate that the two main mechanisms of PBC progression, bile duct destruction, and interface hepatitis are more severe in PBC patients positive for anti-gp210 antibodies as compared with those negative for anti-gp210 antibodies, thus leading to the more frequent progression to end-stage hepatic failure. The current study also showed that positive anti-centromere antibodies is a risk factor for more severe ductular reaction, interface hepatitis, and portal inflammation. There was a trend of less severe ductopenia in late stage of anti-centromere–positive patients as compared with anti-gp210–positive patients (data not shown). Cytokines or growth factors from inflammatory cells rather than retention of bile constituents are critical for the induction of the ductular reaction in PBC.22 Ductular reaction is also known to promote fibrosis via the production of TGF-β, monocyte chemoattractant protein-1, and platelet-derived growth factor by proliferating ductal epithelium in chronic viral hepatitis.36–38 Therefore, more severe ductular reaction, rather severe bile duct damage, or ductopenia may play a critical role for portal hypertension type progression of PBC in anti-centromere–positive patients, although the mechanism of bile duct damage that tends to lead to chronic cholestasis and development of cirrhosis of biliary pattern is also operative in various degrees in anti-centromere–positive PBC patients.

The mechanisms underlying the production of ANAs in PBC remain poorly understood. The aberrant expression of self-antigen is considered to be one of the mechanisms for the breakdown of immunological tolerance. The enhanced expression of PDC-E2 was found on the luminal region of biliary epithelial cells of small bile ducts in PBC liver.39 The expression of sp100 antigens was reported to be induced by inflammatory cytokines such as type 1 IFN.40 We recently found that the expression of gp210 antigens is increased on the nuclear envelope of biliary epithelial cells of small bile ducts in PBC liver and that the intensity of gp210 staining is positively correlated with the intensity of inflammation around small bile ducts.13 Furthermore, the consistent presence of anti-gp210 antibodies is a strong risk factor for the progression to end-stage hepatic failure, whereas the disappearance of anti-gp210 antibodies after initiation of ursodeoxycholic acid treatment indicate the most favorable clinical course in PBC, as shown in this study and by Nakamura et al.12 These results may indicate that gp210 is possibly a target antigen in both bile duct destruction and interface hepatitis, two major pathological features that play a critical role for the progression of PBC.32–35

Apoptosis is also considered to be a mechanism for breakdown of immunological tolerance.41 In fact, enhanced apoptosis was reported to be operative in bile duct loss in PBC.42 The altered apoptosis-related antigen presentation, the disruption of proteosome-dependent or -independent antigen processing, is considered to play a role for the breakdown of immunological tolerance of nuclear autoantigens, including centromere proteins.43 Furthermore, molecular mimicry is considered another important mechanism for breakdown of immunological tolerance of self-antigens.44–46 In this hypothesis, the presence of the EIEXD or EXDK motif in bacterial and human PDC-E2, OGDC-E2, human gp210, sp100, and centromere proteins may play a critical role for cross-recognition by pathogenic autoreactive T cells, leading to epitope spreading among self-antigens followed by sustained and more severe T-cell–mediated inflammation.46 In this context, the risk of PBC progression should be determined by the presence or absence of anti-gp210 or anti-centromere antibodies in early stage of PBC. This may indicate that autoreactive immune repertoire is prepared in early stages of the disease process and can determine the long-term outcome of the disease. Similar mechanisms have been reported to be operative in other autoimmune diseases.17–21

In conclusion, we show that PBC can be classified into 2 clinically different types based on progression pattern and ANAs. Because autoantibodies to a certain nuclear autoantigen (i.e., gp210, centromere) predict the progression pattern of the disease in early stages of PBC, it is probably a matter of time before autoantibodies to gp210 and centromere will be added to the arsenal of tests for PBC classification and its treatment.

Acknowledgements

The authors thank Drs. Makoto Nakamuta, Hiroaki Hazama, Nobuyuki Senda, Hideo Morimoto, Masakazu Yamada, Tomohiko Mannami, Yasuhiro Yamashita, Shin Tanaka, Yukio Ohara, Michio Kato, Taizo Hijioka in the National Hospital Organization Study Group for Liver Disease in Japan (NHOSLJ) for providing serum samples and clinical information from patients with PBC; thanks to Drs. Shinya Nagaoka, Kazuyuki Ohata, Takashi Kamihira, Eisuke Ozawa, Koji Yano, and Takehiko Matsumoto in NHO Nagasaki Medical Center for patients' care and technical assistance.

Ancillary