The presence of circulating anti-mitochondrial antibodies (AMAs) represents a key diagnostic feature of primary biliary cirrhosis (PBC). Using highly sensitive assays with recombinant antigens, up to 95% of PBC patients have sera autoantibodies directed against the ubiquitous 2-oxo-acid-dehydrogenase complexes involved in mitochondrial energy metabolism. AMA positivity with its unique prevalence rate among autoimmune diseases1 represents one of the three diagnostic criteria for PBC, the others being liver histology compatible with the disease and a persistent elevation of alkaline phosphatase.2 Approximately 5% of well-documented PBC patients do not react with any of the mitochondrial antigens using currently available assays. Over the past decade, the “AMA-negative” population has been progressively reduced in our laboratory by the development of more sensitive detection techniques.3 Nonetheless, the persistence of some AMA-negative subjects raises the question of whether these patients represent a sub-population with distinct clinical features or simply have antibody titers and specificities undetectable with available technology.4 Recent technological advances have led to the development of increasingly automated detection methods associated with more rapid, accurate, and reliable assaying,5 which could address the question. The development of such methods not only decreases the likelihood of human error in biological test performance, importantly assisting standardization; it contemporarily decreases time requirements for single-assay performances. Moreover, the possibility of investigating multiple antigen reactivities additionally reduces detection time. In the attempt to validate a novel detection method and to better define the reactivity of AMA-negative patients, we developed a bead assay using the Luminex principle. Glutathione-S-transferase–conjugated fusion proteins of the mitochondrial autoantigens in PBC were used to investigate the AMA serum positivity of a well-defined cohort of patients with PBC and healthy and diseased ethnically matched controls.
The antimitochondrial response in primary biliary cirrhosis (PBC) is the most highly directed and specific self-reacting antibody in human immunopathology. Originally, antimitochondrial antibodies (AMAs) were detected by indirect immunofluorescence (IIF) and found in approximately 90% of well-documented patients with PBC. The introduction of recombinant autoantigens and the use of immunoblotting have increased the sensitivity and specificity of AMAs, and they are now considered positive in approximately 95% of patients with PBC. Clearly, accurate autoantibody detection represents one of the fundamental requirements for reliable diagnostics in autoimmunity. To address the 5% of AMA-negative patients with PBC, we have generated and validated a bead assay for the detection of AMA. We enrolled 120 patients with PBC, including a non-random group of 30 rigorously proven AMA-negative patients, 50 healthy subjects, and 74 controls with autoimmune diseases (18 with primary sclerosing cholangitis, 16 with autoimmune hepatitis, and 40 with systemic lupus erythematosus). Individual bead assays were done with the three mitochondrial autoantigens, PDC-E2, BCOADC-E2, and OGDC-E2. As expected, 90 of 90 previously known AMA-positive patients remained positive with this assay but, interestingly, 20% of the rigorously defined AMA-negative patient group had antibodies to one or more of the mitochondrial autoantigens. Furthermore, 100% of these newly detected AMA-positive patients were anti-nuclear antibody (ANA) positive. Conclusion: The development of this assay reflects the potential for automated detection with rapid and reliable assaying and further highlights the diminished number of truly AMA-negative PBC patients. (HEPATOLOGY 2007;45:659–665.)
Patients and Methods
After informed written consent, blood was drawn from 244 subjects at the San Paolo Hospital of Milan (Italy), including 120 patients with PBC, 50 healthy controls, 18 patients with primary sclerosing cholangitis, 16 patients with autoimmune hepatitis, and 40 patients with systemic lupus erythematosus. The PBC population included 90 AMA-positive patients and 30 previously well-defined AMA-negative patients. AMA-negative patients were determined by absence of antimitochondrial reactivity in three consecutive indirect immunofluorescence (IIF), immunoblot, and enzyme-linked immunosorbent assay (ELISA) assays using recombinant autoantigens. All AMA-negative patients were negative at presentation and at all times thereafter. The diagnosis of PBC was defined by internationally accepted criteria,2, 6 identified by two of the following three: histological evidence of PBC by liver biopsy, presence of antimitochondrial antibodies, and an otherwise unexplained 1.5-fold or greater increase of alkaline phosphatase levels for at least 6 months. Patients with serum positivity for hepatitis B surface antigen and hepatitis C virus antibodies were excluded from the study. Serum biochemical and immunological tests, including alanine aminotransferase, alkaline phosphatase, bilirubin, albumin, prothrombin time, serum immunoglobulins, hepatitis B surface antigen, and anti-hepatitis C antibodies, were assessed by routine laboratory techniques.
Preparation of Recombinant PDC-E2, BCOADC-E2, and OGDC-E2.
We used our previously well-defined recombinant autoantigens, all originally cloned at UC Davis,7–10 expressed in plasmid PGEX 4T-1. Overnight cultures of expression clones were diluted 1:10 with fresh Lauria-Bertani medium (50 μg/ml ampicillin) until the optical density (OD) was 0.7 to 0.8 and induced with 1 mM isopropyl-β-thiogalactopyranoside for an additional 3 to 4 hours at 37°C. Cells were pelleted, resuspended in PBS containing 1% triton X-100 and 1% Tween 20 (Sigma Chemical Co., St. Louis, MO), and sonicated. The sonicated extract was centrifuged at 10,000g for 15 minutes at 4 °C and the supernatant collected. Glutathione-agarose-beads were washed 3 times with PBS and the fusion protein eluted by competition with 50 mM Tris HCl pH 8.0 containing 20 mM reduced glutathione (Sigma, St. Louis, MO). Protein concentrations of the eluates were determined by a Bradford assay, and specificity of the purified recombinant proteins was verified by immunoblotting with monoclonal antibodies, including positive and negative controls in all cases.11
Immunoblotting for AMA Testing.
Established optimal amounts of antigen, namely, 16 μg, 30 μg, and 30 μg, respectively, of the purified recombinant antigens PDC-E2, BCOADC-E2, and OGDC-E2 were loaded on a 10% mini protein gel (Bio-Rad Laboratories, Hercules, CA), fractionated at 170 V for 1 hour, and transferred onto nitrocellulose membranes overnight, which were then cut into strips. After blocking for 1 hour with 5% skim milk, samples were added at a dilution of 1:1,000, 1:500, and 1:200 in milk and incubated for 1 hour. Strips were washed 5 times for 5 minutes with PBS with 0.5% Tween 20 (PBS-T). Strips were then incubated for 1 hour with goat anti-human IgA/G/M antibody (Zymed, South San Francisco, CA), washed, and assayed with Pierce Pico Luminol Fluorescent Substrate (Pierce, Rockford, IL). Blots were exposed on photographic membranes and images digitalized with a FluorTech 8900 gel doc system (Alpha Innotech, San Leandro, CA) equipped with a chemiluminescent filter.
Indirect Immunofluorescence for Serum Autoantibody Analysis.
Serum anti-nuclear (ANA) and anti-mitochondrial (AMA) antibodies were determined blindly by IIF as described.12 Briefly, sera were tested for the presence of ANA using human Hep-2 cells and HeLa cells at different dilutions using FITC-conjugated anti-IgA/G/M as a secondary antibody. Positivity and immunofluorescent pattern were evaluated by confocal microscopy (Orthoplan, Leitz, Wetzlar, Germany). Titers of 1:40 or greater were considered positive.
Ninety-six-well ELISA plates were coated overnight with MIT3 antigen at a concentration of 2 μg/μl. Plates were blocked with 1% bovine serum albumin in PBS for 1 hour. After two washes with PBS-T, 100 μl serum at a dilution of 1:200 in PBS was added and incubated for 1 hour at room temperature. Plates were washed three times with PBS-T, and goat anti-human IgA/G/M antibody (Zymed, South San Francisco, CA) was applied. After four washes with PBS-T, 100 μl ABTS Peroxidase Substrate solution (KPL, Gaithersburg, MD) was added to each well. After color development, plates were read on a microplate reader (Molecular Devices, Sunnyvale, CA).
LiquiChip beads were purchased from Qiagen (Valencia, CA) and proteins conjugated to the activated beads at a concentration of 0.2 μg/μl. First, to assess the validity of the assay and baseline fluorescence values, we took advantage of our specific monoclonal antibodies to the mitochondrial antigens11, 13; positive and negative controls were tested in duplicate or quadruplicate (data not shown). Once the assay was validated, PBC sera were diluted at a final concentration of 1:10−3, 1:10−4, 1:10−5, and 1:10−6 in Blotto blocker (Pierce, Rockford, IL) and added to the pre-mixed bead solution (1,250 beads/10 μl). All samples were blindly assayed in duplicates; several samples were routinely assayed in all experiments as controls. Samples were incubated for 2 hours on a shaker at room temperature. Plates were then washed 3 times with PBS-Tween and incubated with biotin-Sp–conjugated goat anti-human IgA/G/M antibody (Jackson ImmunoResearch, West Grove, PA) for 45 minutes. Plates were washed again and incubated with R-phycoerythrin-conjugated streptavidin (Caltag, Burlingame, CA) for 30 minutes in the dark. Plates were washed once and beads resuspended in PBS-Tween and read on a Luminex 100 reader (Luminex Corp., Austin, TX). Known positive and negative controls were added to each plate.
Differences in median values among two groups were assessed using the Mann-Whitney U test; differences among means, using Student t test. Differences in proportions were tested using Fisher's exact test. All analysis was two-sided, and P values below 0.1 were considered statistically significant.
For the bead assay, glutathione-S-transferase reactivity was defined as common background and thus subtracted as median fluorescence value from the median fluorescence values of the other analyzed antigens. To establish a cutoff value for the pathological samples, 50 healthy control samples were analyzed as a reference population. Mean fluorescence value (mfv) ± 3 SD was used as a cutoff value. Cutoff values were 1560 mfv for OGDC-E2, 7230 mfv for BCOADC-E2, and 5400 mfv for PDC-E2. Means of the duplicate sample values were used for analysis. All statistical analysis was performed using GraphPadPrism software, version 4.0 (GraphPad Software Inc., San Diego, CA).
All patients were tested for AMA status by IIF, ELISA, and immunoblotting with recombinant antigens before enrollment in this study. Ninety AMA-positive patients, who reacted with at least 1 of the 3 epitopes, and 30 known AMA-negative patients with PBC were selected. For the purpose of this study, the largest possible number of AMA-negative subjects was chosen, generating a study population that obviously differs in its AMA reactivity distribution from the general PBC population. In fact, 30 AMA-negative patients could be representative for a random population of approximately 600 patients, considering the average prevalence of 5% AMA negativity. The enrolled cohorts of AMA-negative and AMA-positive patients displayed no significant differences, as illustrated in Table 1. In fact, median age, duration of disease, and biochemical status of the patient populations were similar, as well as histological stage defined by liver biopsy.
|AMA-Positive (n = 90)||AMA-Negative (n = 30)||P|
|Male: female ratio||3:87||1:29||NS|
|Median age (years)*||67 [35–82]||64 [38–74]||NS|
|Median age at diagnosis (years)*||51 [19–70]||51 [33–71]||NS|
|Duration of disease (years)*||15 [1–36]||14 [3–28]||NS|
|ALT (UI/l)†||53 ± 5||71 ± 9||NS|
|Alkaline phosphatase (UI/l)†||461 ± 29||475 ± 40||NS|
|Bilirubin (mg/dl)†||0.8 ± 0.1||0.7 ± 0.1||NS|
|Albumin (g/dl)†||4.2 ± 0.1||4.2 ± 0.1||NS|
|PT (INR)†||1.00 ± 0.01||1.00 ± 0.01||NS|
|Stage I-II||37 (41%)||12 (40%)||NS|
|Stage III-IV||53 (59%)||18 (60%)||NS|
|Mean IgM (mg/dl)†||316 ± 19||266 ± 37||NS|
|Mean IgG (mg/dl)†||1359 ± 54||1515 ± 121||NS|
|Mean IgA (mg/dl)†||294 ± 18||311 ± 33||NS|
|ANA status||43 (47.8%)||17 (56.7%)||NS|
|UDCA therapy||63 (70%)||22 (73%)||NS|
After validation of the assay with PBC sera, control sera, and known positive monoclonal reagents (not shown), a subpopulation of 56 AMA-positive patients was selected to determine the persistence of antibody reactivity and robustness of the assay. Progressive serum dilutions (1:10−3, 1:10−4, 1:10−5, and 1:10−6) were analyzed. Several patient sera tested consistently positive at a dilution of 1:10−5, significantly reducing the amount of serum needed for a single assay and confirming the increased sensitivity in comparison with earlier techniques (Fig. 1).
Overall AMA Reactivity.
After validation of the assay, data were acquired from the patient populations using healthy controls as a reference. Ninety patients were AMA positive with 98% reacting with PDC-E2, 57% with BCOADC-E2, and 26% with OGDC-E2 (Table 2). AMA-positive subjects display an impressive increase over threshold values especially for PDC-E2, the major autoantigen in PBC. Distribution of the patient populations according to the measured fluorescence intensities is illustrated in Fig. 2. As expected, all of the previously “AMA-positive” subjects reacted with at least one of the antigens (Table 3). The systemic lupus erythematosus, primary sclerosing cholangitis, and autoimmune hepatitis control populations, which were included to confirm specificity using sera from patients with different autoimmune diseases characterized by elevated numbers of circulating autoantibodies, tested negative for all the analyzed antigens.
|PBC, AMA-Positive (n = 90)||PBC, AMA-Negative (n = 30)||Healthy Controls (n = 50)||PSC (n = 18)||AIH (n = 16)||SLE (n = 40)|
|PDC-E2||88 (98%)||5 (17%)||0||0||0||0|
|BCOADC-E2||51 (57%)||3 (10%)||0||0||0||0|
|OGDC-E2||23 (26%)||2 (7%)||0||0||0||0|
|Any positive||90 (100%)||6 (20%)||0||0||0||0|
|PDC-E2 only||37/90 (41%)||3/30 (10%)|
|BCOADC-E2 only||1/90 (1%)||0|
|PDC-E2 + BCOADC-E2||29/90 (32%)||1/30 (3%)|
|PDC-E2 + OGDC-E2||2/90 (2%)||0|
|BCOADC-E2 + OGDC-E2||1/90 (1%)||1/30 (3%)|
|PDC-E2 + BCOADC-E2 + OGDC-E2||20/90 (22%)||1/30 (3%)|
We subsequently focused our attention on the “AMA-negative” population, because this subpopulation of patients has been progressively reduced by the development of novel and improved detection methods. Six of 30 (20%) sera of the analyzed AMA-negative patients reacted with at least one of the antigens using the bead assay (Table 2). More specifically, 5 samples reacted with PDC-E2 (17%), 3 of 30 with BCOADC-E2 (10%), and 2 of 30 (7%) with OGDC-E2. Reactivity distributions among the samples varied, including 3 samples that reacted with PDC-E2 alone, 1 with all 3 antigens and 2 with a combination of 2 autoantigens (BCOADC-E2 + PDC-E2 and BCOADC-E2 + OGDC-E2, respectively) (Table 3). Antibody titers detected within the subpopulation also varied, being relatively lower for PDC and higher for BCOADC and OGDC when compared with AMA-positive patients. All patient sera were not reactive by Western blot (Fig. 3); however, reactivity toward the antigens did not correlate between the two assays. As exemplified in Fig. 3, strong Western blot responsiveness does not necessarily translate into the proportional bead assay response.
Clinical Specificity of the AMA-Negative Population.
We further analyzed the AMA-negative subpopulation, attempting to identify unique characteristics associated with the diverse reactivity. Clinical and biochemical features of bead assay positives and negatives are presented in Table 4. No significant differences can be detected between the 2 populations comparing by age, disease duration, and biochemical features. Particular attention was paid to serum immunoglobulin values and the presence of antinuclear antigens, which have been described as particularly abundant in the AMA-negative subpopulation and potentially associated with a different disease phenotype. In fact, 6 of 6 patient sera that reacted with the bead assay were also ANA-positive, compared with 11 of 23 (48%) within the bead assay–negative population (P = 0.023, chi-squared test).
|Bead Assay AMA-Negative (n = 24)||Bead Assay AMA-Positive (n = 6)||P|
|Male: female ratio||1:23||0:6||NS|
|Median age (years)*||64 (50–74)||64 (38–73)||NS|
|Median age at diagnosis (years)*||47 (35–71)||58 (33–65)||NS|
|Duration of disease (years)*||13 (3–28)||16 (4–20)||NS|
|ALT (UI/l)†||66.0 ± 12||89 ± 26||NS|
|Alkaline phosphatase (UI/l)†||510 ± 55||416 ± 67||NS|
|Bilirubin (mg/dl)†||0.8 ± 0.1||0.7 ± 0.1||NS|
|Albumin (g/dl)†||4.2 ± 0.1||4.0 ± 0.2||NS|
|PT (INR)†||1.00 ± 0.1||0.95 ± 0.1||NS|
|Stage I-II||9 (37.5%)||3 (50%)||NS|
|Stage III-IV||15 (62.5%)||3 (50%)||NS|
|Mean IgM (mg/dl)†||199 ± 34||350 ± 66||NS|
|Mean IgG (mg/dl)†||1528 ± 123||1684 ± 470||NS|
|Mean IgA (mg/dl)†||311 ± 36||395 ± 105||NS|
|ANA status||11 (45.8%)||6 (100%)||0.023|
|Homogeneous||4/11 (36.4%)||3/6 (50%)||NS|
|Speckled||6/11 (54.5%)||2/6 (33.3%)||NS|
|Rim-like||1/11 (9.1%)||1/6 (16.7%)||NS|
The identification of the specific mitochondrial autoantigen of PBC in 19877 led to an exponential increase of diagnostic AMA assays. Serum AMAs have high specificity for PBC being virtually undetectable in healthy individuals14 and are defined as a product of clonal selection,15 highlighting their fundamental role as disease markers and potentially active participants.16 AMA testing, which had been long limited to IIF, was more recently improved by the generation of a series of animal and human recombinant antigens17, 18 available for ELISA and Western blotting.19 The new techniques reduce the necessary amount of serum for assay performance and are more sensitive than previous methods,20 leading to an increase in the detection of AMA reactivity.
The coupling of protein antigens to beads for diagnosis has been developed to improve sensitivity associated with spatial presentation, essential for conformational epitopes, allowing the detection of multiple antigens at once.21–23 The assay presents important advantages because of the inclusion of a large and variable number of individual bead sets and the possibility of customization. Moreover, it incorporates the advantages of a fully automated procedure that minimizes operator variability and allows high standardization.
We developed the presented diagnostic assay to offer a valid, fast, and reliable diagnostic alternative in PBC. The number of analyzed antigens was limited to three to standardize the technique and define applicability, preserving the method's advantages but also reducing potential sources of bias. The 3 antigens (PDC-E2, BCOADC-E2, and OGDC-E2) were purified, applied on the activated beads, and assayed in a standardized manner with PBC sera. Excellent correlation with previous analysis was seen in 100% of patients defined as AMA positive by Western blot. Strikingly, several sera displayed consistent reactivity up to 1:100,000 dilutions, 100 times the average dilution for Western blot, undoubtedly proving the advantages potentially connected to the different spatial presentation allowed by the bead structure.24
The technological improvement was also reflected in the detection of AMAs within nonreactive sera. Twenty percent of the previously defined AMA-negative patient sera reacted with at least 1 antigen and 10% with more than 1, excluding a structural bias within the assay as a cause of broader reactivity. Reactivity against the mitochondrial antigens varied in its intensity, reaching levels attributable to highly AMA positives in some cases. Interestingly, degree of immunoblot intensity and bead reactivity, although overlapping in detection, did not correlate. The different spatial presentation of the antigen on the bead, which allows the preservation of a tridimensional structure, might be responsible for this phenomenon, as well as the absence of denaturing conditions required for the immunoblotting process.
Interestingly, all of the newly detected AMA-positive patients were also ANA positive; a feature shared only by 48% of the remaining AMA negatives whose prevalence appears in line with that of previous reports.25 No additional biochemical or immunological differences could be detected among the two populations. ANA-positive patients have been reported as a sub-population characterized by more severe disease26 and worse prognosis.27 ANA have also been detected more frequently in AMA-negative patients, a condition that was attributed to the possible masking of antinuclear reactivity by high titer mitochondrial positivity detected by IIF.4 No differences in ANA patterns or titers could be detected between the 2 populations. Given our results, we cannot predict whether any association might exist between ANA positivity and the development of AMA positivity. Nonetheless, if an improved assay could aid in the detection of low-titer AMA reactivity, which might represent earlier phases of disease, it would be valuable for patient management and research as well as to identify otherwise symptomless cases. The dynamics of anti-mitochondrial generation have been elusive, and the possibility of defining initial AMA production would represent a great advantage in the definition of causative factors involved in PBC.
The feasibility and reliability of the bead assay allows speculations on potential future applications of such an assay. Two minor mitochondrial antigens—PDC-E1α and E1β—were not included in the assay, but reactivity to these antigens is rarely found in patients who do not have reactivity to PDC-E2. Nevertheless, addition of these or other minor mitochondrial antigens into the currently formulated assay might help identify other AMA-negative patients. Initially the confirmation of the assay on a wide random patient population would be advisable, considering that 30 AMA-negative patients are representative for 600 patients with PBC. Subsequently, several minor autoantigens,28–30 including a vast array of nuclear antigens,12, 31 could be added to the bead panel in the attempt to generate a comprehensive PBC assay or an autoimmune liver assay. The development of such a panel would permit the definition of autoantibody status with a single assay, importantly simplifying and accelerating diagnostics in autoimmune liver conditions.
We thank Jenelle Fraser for her help with the assay reading.