INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The detection of autoantibodies specific for eight common nuclear antigens has proved clinically useful for patients with systemic lupus erythematosus (SLE), Sjögren's syndrome, mixed connective tissue disease (MCTD), and scleroderma. Most of these eight antibodies were initially described using the Ouchterlony technique, but a variety of immunological methods have been used. Unfortunately, results from different methods have not shown good agreement. Efforts to standardize results from all methods have been hampered by low volumes of the sera used in the original characterizations and a lack of large volumes of other well-characterized positive control sera. Until recently most available reference sera have been characterized only by antinuclear antibody (ANA) and Ouchterlony methods.

The World Health Organization in conjunction with the International Union of Immunological Societies developed several standards for evaluation of ANAs, including WHO 66/233 for immunoglobulin G ANA, WHO/IUIS 480010 fluorescein isothiocyanate-conjugated anti-human immunoglobulin, and Wo/80 for antibodies to double-stranded DNA (dsDNA) (2). Standardization of other autoantibodies was initiated during a 4-year study by the European Consensus Study Group (11, 12). In 1980, the Arthritis Foundation (AF), in collaboration with the Centers for Disease Control (CDC), established a Committee on Antinuclear Antibody Serology in the United States. This organization prepared a panel of five AF-CDC reference sera, made available in 1982, that included specificities for ANA, dsDNA, SS-B, RNP, and Sm antibodies (7). Several international organizations joined to support the activities of the Committee on Antinuclear Antibody Serology: the International League Against Rheumatism, the International Union of Immunological Societies and the World Health Organization (8). This cooperative effort expanded the band of AF-CDC sera to a total of 10 different sera covering the following spectrum: five fluorescent ANA patterns (diffuse, nucleolar, centromere, and two speckled patterns) and seven ANA specificities (SS-A, SS-B, U1-RNP, Sm, Scl-70, and Jo-1). Although the European Consensus Study Group standards were found to be suitable for enzyme immunoassay (EIA) methods, some of the AF-CDC standards were found to yield inconsistent results with newer methods (4). The AF-CDC standards were recently reevaluated to define their usefulness for immunoblotting techniques, and most of the sera were found to produce the expected band patterns with the exception of the anti-SS-A(Ro) sample, which did not show a consistent band pattern (5). The AF-CDC serum panel was recently used to evaluate performance of EIA kits from nine manufacturers. The study demonstrated good performance with SS-A, SS-B, Scl-70, centromere, and Jo-1 kits while the dsDNA and Sm kits performed less well (10).

Beginning in 1994 a large number of manufacturers began producing EIA kits utilizing a variety of nuclear extracts, purified antigens, and recombinant protein preparations containing the most common nuclear antigens. Because of the lack of good reference materials to evaluate the large number of new EIA kits, it was determined by the Standards Committee of the Association of Medical Laboratory Immunologists (AMLI) in the summer of 1995 that a consensus panel of well-characterized serum samples should be produced. The immediate goals were (i) to produce a battery of consensus sera that would be produced in sufficient quantity to be used by both reagent manufacturers and clinical diagnostic laboratories and (ii) to gather information concerning the performance of EIA kits from different manufacturers. As EIAs have been used for nuclear autoantibody testing for a relatively short time, the committee felt that the development of a consensus serum panel was the most realistic approach to the initial efforts. As more laboratories adopt EIA techniques for routine testing and additional clinical information is gained about the performance of these assays, the long-term goal is to develop a reference serum panel for the most commonly encountered nuclear antigens. A specific battery of autoantibodies was selected for study because of their clinical relevance and because the antigens they are directed against are well characterized. The target antibodies included those against dsDNA, SS-A, SS-B, Sm, RNP, Scl-70, centromere, and Jo-1. Although the goal was to produce sera that would perform well in EIAs, evaluation of the sera was also performed with methods being currently used in the clinical laboratories carrying out the evaluation. The methods used included immunofluorescence assay (IFA) on Crithidia lucilliae for anti-dsDNA, IFA on HEp-2 cells for anti-centromere, and double immunodiffusion (DD) and hemagglutination (HA) for the extractable nuclear antigens (ENAs). During the course of the study, several manufacturers who were in the process of developing blotting methods (Western blotting [WB], dot blotting [DB], or immunoblotting [IB]) also evaluated the sera by these methods.

All clinical laboratories are experiencing severe fiscal restraints that dictate that testing be done as efficiently as possible. EIA methods can be automated to provide more cost-effective autoantibody testing than the classic manual methods of IFA and double immunodiffusion (4, 7, 8), but the lack of a reference panel suitable for EIA testing has been a hindrance to many laboratories adopting EIA methods. The project undertaken by the AMLI Standards Committee was intended to develop an autoantibody consensus serum panel suitable for the assays currently on the market. The evaluation studies described in this report show the successful production of eight sera and some comparison results for the different reagent manufacturers.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Reference preparations. Approximately 1 liter of plasma from each of three normal individuals and 11 patients was obtained by plasmapheresis. The patients were selected based on preliminary evaluations and samples obtained after review of the diagnosis in conjunction with the clinical rheumatologist treating the patients. Diagnostic criteria as outlined by the American College of Rheumatology (formerly American Rheumatism Association) were used to establish the diagnosis for each patient (1, 8, 9).

Participant laboratories. Nine AMLI members and 21 vendors participated in this study. AMLI members volunteered to participate in response to announcements in the AMLInteractions member newsletter. AMLI member participants' laboratories were geographically diverse, representing various regions of the United States and Canada (listed as A-1 through A-9 in tables). Vendors invited to participate included those who exhibit at AMLI annual meetings or advertised as supplying products used in autoantibody testing (1a). During the testing phase of the study, additional vendors who offer WB or IB assays for autoantibodies were requested to participate.

Methods. Methods used in this study included EIA, IFA, HA, DD, counterimmunoelectrophoresis, immunoelectrophoresis, radiolabeled immunoprecipitation (IP), and WB, IB, and DB as shown in Table 1. AMLI members used methods representative of all of the above methods except CIE, WB, and DB. Vendor methods were primarily EIA, but also included DD, HA, CIE, WB, IB, and DB.

Antigen sources. IFA methods for dsDNA used C. lucilliae as the substrate. EIA methods used calf thymus DNA, plasmid dsDNA, phage , or human recombinant DNA. One AMLI member performed a unique (unpublished) assay that captures the immunoglobulin G from patient sera and detects dsDNA by using intrinsically labeled, synthesized M13-dsDNA. One vendor for IB listed human K-562 as the antigen source for dsDNA; one vendor used human recombinant dsDNA. The EIA methods for SS-A used extracts of calf thymus, bovine spleen, or sheep spleen cells. The DD methods for SS-A used extracts of human spleen cells, Wil-2 whole cells, and calf thymus. The EIA methods for SS-B, Sm, RNP, Scl-70, and Jo-1 used extracts of calf or bovine thymus, rabbit thymus, bovine spleen, or sheep spleen. The EIA method that detects the U1-RNP antibody also detects the Sm antibody, while the Sm EIA method measures only Sm. Therefore, the RNP results will be termed RNP/Sm results in this paper. The DD methods for SS-B, Sm, RNP, Scl-70, and Jo-1 used extracts of calf thymus or rabbit thymus.

Data analysis. All of the data sheets were received and recorded by one of the study coordinators, and then the recorded data were compared to the raw data sheets by a second study member. Coded data were then sent to each participant and reviewed for correctness prior to the final analysis. Any methods that only contained results from one laboratory were not further analyzed because of the difficulty of comparing method specificities and sensitivities. Results for IP and HA methods are grouped together because of the low number of results from HA. For the purpose of the EIA data analysis, the few AMLI members' laboratory results were excluded from the summary tables because they all used kits from the same vendor. In all cases, the results from the AMLI laboratories were identical to those of the reagent vendor results.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Results for all 14 samples evaluated by the participating laboratories and manufacturers are contained in Tables 2 to 5 for the IP, HA, IFA, and EIA methods. Table 6 contains the limited results for the samples when tested by WB or IB. A summary of the data will be discussed both for each sample studied and for each antibody evaluated.

Summary by serum sample. Sample A was from an individual with the clinical diagnosis of CREST syndrome. All test participants found centromere antibody in this sample using IFA (14 of 14) and EIA (2 of 2). The sample was completely negative for reactivity for SS-A, SS-B, Sm, RNP/Sm, dsDNA, and Jo-1 by all methods. Two laboratories out of eight, one using IP and one using HA, gave false-positive results for Scl-70. Only one of three vendors who tested this specimen for centromere antibodies by IB had a positive result.

Summary by antibody using IP or HA and EIA. SS-A antibody was found in samples C, D, E, F, G, and J for essentially all EIAs (119 of 126 tests). A majority of IP or HA methods were also positive but less reactive; only 38 of 49 assays were positive. More problematic were the results for sample H, containing rheumatoid factor, for which 6 of 21 EIAs were positive while all 9 IP or HA results were negative. The explanation for the results of sample H is not known but may be a result of rheumatoid factor interference in some of the assays, an increased sensitivity to SS-A in a few EIA kits, or false-positive results due to some other interference.

Method summary. Results from evaluations to determine the number of false-negative, false-positive, and inconclusive results for the IFA, IP or HA, and EIA methods are found in Table 7. The EIAs had much lower numbers of false-positive and false-negative results than did the IP or HA assays. However, the EIAs had a significant number of inconclusive results, especially for the Sm and RNP assays. Interestingly, the assay that performed best was the IFA for centromere. The centromere EIA was only available from two vendors, so a good comparison evaluation was not possible. The assays with the largest number of false negatives were the IP or HA for SS-A and the IFA for dsDNA. The largest number of false-positive results was found for the Scl-70 IP or HA assay, the Jo-1 EIA, and the IP or HA SS-B assays.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Results from WBs performed at SLR Research Corporation on the serum samples in the AMLI consensus panel. The numbers correspond to the locations of the bands consistent with the characteristic antigens that are detected by the characterized antisera (Table 8).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This study can be used to compare results from the Ouchterlony method used by the majority of AMLI laboratories and the EIA results performed at the vendor laboratories. A total of 17 negative Ouchterlony results were seen with samples that were positive with essentially all EIA kits. For the SS-A antibody samples negatives were seen with samples C, D, E, F, and G; for the SS-B samples a negative was seen with sample G; for the U1-RNP antibody samples negative were seen with samples D and I; for the Scl-70 antibody samples a negative was seen with sample B; and for the Jo-1 antibody samples a negative was seen with sample F. The most likely explanation for these results is that the EIAs are slightly more sensitive than the Ouchterlony methods as performed by some of the testing laboratories. These false-negative results could also be a result of technical errors made during the testing. An alternate explanation could be that precipitation lines seen when multiantigen preparations were used were identified incorrectly. This type of error could result in either false-negative or false-positive results.

The Ouchterlony method also gave a total of 11 of false-positive results, when the majority of other Ouchterlony results were negative and essentially all EIA results were negative. This type of error could be a result of technical errors or incorrect identification of precipitation lines. In summary, a larger number of false positives and false negatives were seen with the Ouchterlony method that with the EIA method even though the EIA method had more than twice as many results. The Ouchterlony results that were the most problematic were the high rate of false-negative results with SS-A and the number of inconclusive results with U1-RNP.

The IFA method had mixed performance in this study. It was excellent for the detection of the centromere antibody in this panel. No false-positive or false-negative results were seen. In contrast, the IFA method for dsDNA gave some false negatives, for two of the three positive samples in the panel. Sample C was particularly problematic; it was negative by IFA in 8 of 13 laboratories while positive in 12 of 16 EIAs. Results for this sample by IFA could be a result of lack of sensitivity of some of the IFA slides or a problem with interpretation of the pattern present. Sample C was a low-titer antibody near the cutoff for many of the EIAs. The conflicting IFA and EIA results for samples C and I may also be a result of sensitivity difference between the two methods. These results were consistent with many previous studies of dsDNA assays, in which discrepant results have often been seen with various methods (3). In general, agreement is usually seen when high-titered, high-affinity, and high-avidity antibodies are present, but discrepant results are more often seen when testing lower-titered, lower-affinity, or lower-avidity antibodies. In addition, care must be taken when evaluating dsDNA assays to ensure that the assay is not reactive with ssDNA antibodies that are often in patient sera. We did not evaluate the study samples for ssDNA because assay kits for ssDNA are only available from a few vendors.

The EIA method in general had better performance than the Ouchterlony method. For samples that could be clearly determined, false negatives were only seen with the SS-A testing, and false positives were seen in only a few assays, with the highest number being for Jo-1 antibody results. The problem in the EIA testing was the very high number of inconclusive results in samples with significant numbers of positive EIA results and mostly negative Ouchterlony results. Of the 14 samples tested, the inconclusive EIA results were seen with five samples for U1-RNP, two samples for Sm, two samples for Scl-70, and one sample for SS-A. Most of the samples with high numbers of inconclusive results had lower reactivity in the assays than the samples that had consensus positive results. These indeterminant results illustrate the significant difference in the EIA kits from the different vendors and could reflect differences between the kitse.g., different buffers affecting the binding of the antibodies, different antigen epitopes present, antigen coating concentration differences, different positive-negative cut-off levels, or the presence of other contaminating antigens in some of the antigen preparations. Similar results were seen in the recently published study by Tan et al. that demonstrated differences in kit performance between available kits (10). The EIAs with positive results used a variety of antigen sources so a single antigen preparation method could not be implicated. More study is necessary to determine the source of these discrepant results. For the consensus panel materials we did not use any of the samples with discrepant results as prototype sera. From a clinical perspective, the most problematic of these discrepant samples are the Sm and Scl-70 testing which should have disease specificity. At present there is no consistent method of standardizing result units from the EIAs. Because of this lack of standardization no direct comparison of results can be made between kits. Additional studies need to be done to make it possible to compare the quantity of antibody being measured. Until this is done the sensitivities and disease specificities of the EIA kits from the different manufacturers cannot be compared accurately.

The most surprising data was the lack of consensus results with the WB and DB data. Because these methods have been used extensively in the research laboratories to characterize and study the antigenic epitopes of many autoantigens, many researchers have begun using blotting methods as their gold standard method. The results of these studies demonstrated a significant difference in the results from the commercially available blotting materials. Our data are similar to those obtained during the European Consensus Studies of ENA antibody detection by immunoblot performed in 1991 and 1992 (11, 12). Further work must be done to standardize the details of the blotting methods prior to their use in routine clinical testing.

This study has demonstrated that consensus sera can be produced for EIA, Ouchterlony, and IFA testing methods. Sera are now available for use by vendors and clinical laboratories for detection of SS-A, SS-B, Sm, U1-RNP, Scl-70, Jo-1, dsDNA, and centromere antibodies. It is hoped that these sera will be used to improve the clinical laboratory testing for these antibody specificities.