William Check, PhD
Many years ago there was a television commercial for a popular light beer in which two groups of football players shouted out what they liked about the product. “Tastes great!” one team yelled. “Less filling!” the other said. What was so wondrous, the advertisement implied, was that one beer could satisfy both groups.
In laboratory medicine, satisfying everyone’s requirements with one product is not so easy. Take antinuclear antibody (ANA) testing to aid in the diagnosis of autoimmune diseases such as systemic lupus erythematosus, or SLE. In a session at the 2009 meeting of the American Association for Clinical Chemistry, speakers addressed the question, “Are ELISA ANA and traditional immunofluorescent ANA results comparable?” While there were no shouts of “More sensitive!” or “More specific!,” it was clear that laboratorians find fault with the indirect fluorescence antibody (IFA) method because it is subjective and lacks specificity and that some rheumatologists are unhappy about what they perceive to be a lack of sensitivity of ANA testing by enzyme linked immunosorbent assay, or ELISA (also called enzyme immunoassay, or EIA).
“At a titer of 1:40, traditional ANA is positive in a large minority of healthy people, about 25 percent,” said John L. Carey III, MD, of the Department of Pathology, Henry Ford Health System. Dr. Carey summarized 10 publications in which, depending on the population tested, positives among those without an autoimmune disease ranged from 13 to 40 percent, clustering around 32 percent, which can lead to a poor positive predictive value. “This is not well known to physicians,” Dr. Carey said. For a condition with a prevalence of one in 1,000, fluorescence ANA ordered by a primary care physician in a patient without a history or symptoms consistent with an autoimmune disease will have a positive predictive value well below one percent. A consequence of the poor positive predictive value of fluorescence ANA in this population is increased testing and referral. In patients who have a history or symptoms suggesting an autoimmune disease, the type of patient rheumatologists see more often, the PPV of fluorescence ANA is much higher, perhaps 30 percent.
Rheumatologist Donald Bloch, MD, presented a case in which a primary care physician wrongly diagnosed a woman with fatigue and musculoskeletal pain as having fibromyalgia, after a negative ANA by ELISA. After further workup and failed attempts at therapy for fibromyalgia by several rheumatologists, the patient was referred to Dr. Bloch, who ordered repeat ANA, this time by IFA. It was positive at a titer of 1:5,120 in an “atypical-speckled” pattern, leading to a diagnosis of primary biliary cirrhosis. Because this autoantigen is not routinely included as a substrate in solid-phase assays, the patient’s autoantibodies were not detected. “I’m concerned that not all of the clinically important autoantigens are present in ELISA ANA tests,” said Dr. Bloch, who is assistant professor of medicine, Harvard Medical School, and associate physician in the Center for Immunology and Inflammatory Diseases, Massachusetts General Hospital.
“There are data on both sides of the question,” Dr. Bloch says, about whether an ANA screen by ELISA is more or less sensitive than by IFA. “Before the whole country decides to switch to solid-phase assays for screening, better studies are needed,” he adds. Many of Dr. Bloch’s colleagues agree. In February of this year, the American College of Rheumatology issued a position paper asserting that fluorescence ANA is the gold standard (http://www.rheumatology.org/publications/position/ana_position_stmt.asp?aud=mem).
False-positive results can also have unfortunate consequences, says David F. Keren, MD, moderator of the AACC session. In an interview with CAP TODAY, Dr. Keren said: “People [with false-positive results] are put on inappropriate therapies and physicians go down many blind alleys for expensive workups. Which is why the 30 to 40 percent false-positive rate that you can get with HEp-2 testing is a big problem. That’s a pretty heavy background for trying to interpret a clinical picture.”
Posing the issue as sensitivity versus specificity “is too broad a generalization,” Dr. Keren adds, and explains: “There is too much lumping going on. There are many different kits approved for ANA testing. Some EIA kits are just as sensitive as the HEp-2 assay [IFA is most often performed on HEp-2 cells fixed on slides]. The sensitivity of IFA depends on the screening dilution used.” At a cutoff of 1:80 or 1:160, which gives a false-positive rate between five and 10 percent, an IFA ANA assay “will perform much like the most sensitive EIAs,” says Dr. Keren, medical director of Warde Medical Laboratory. As for supposed false-negative results with EIA tests, Dr. Keren says, “Don’t forget—ANA-negative lupus is real.”
In addition to IFA and EIA assays for ANA, Bio-Rad Laboratories offers the newest form of solid-phase assay, a multiplex platform called the BioPlex 2200 based on Luminex xMAP bead technology. “Many labs currently offer our BioPlex multiplex system for ANA detection,” Steven R. Binder, who presented industry’s viewpoint in the AACC session, told CAP TODAY. “And many rheumatologists, after learning how these new tests work, are generally satisfied with the results.” It’s a “minority of rheumatologists who are creating a lot of concern about this issue,” says Binder, who is director of technology development in Bio-Rad’s clinical diagnostics group.
At St. Joseph’s Medical Center in Stockton, Calif., a BioPlex 2200 has been the sole ANA testing modality for three years, says laboratory manager Denise Facaros, BS, CLS. “We have had a rather favorable response from our clinicians, with not a lot of questions,” Facaros says. Before switching from IFA, she and medical director Stephen G. Connolly, MD, explained to the rheumatologists their reasons for adopting the new method, chiefly greater specificity, and offered them the opportunity to submit specimens from patients they were treating, to verify the accuracy of the new method. “Based on the reports we gave them, they were satisfied,” Facaros says. Now the lab reports out the multiplex ANA screen result and—when the screen is positive—the extractable nuclear antigens, or ENA, result from the multiplex assay.
A second multiplex ANA assay based on Luminex xMAP bead technology is the Athena Multi-Lyte, marketed in the U.S. by Inverness Medical. Athena Multi-Lyte ANA Test Systems contain 15 beads—nine coated with separate nuclear antigens and one bead coated with a nuclear extract of HEp-2 cell nuclei that can mimic IFA ANA, says Chris Howard, MSc, MBA, marketing manager at Inverness Medical. The remaining five beads make up the Intra-Well Calibration, in which a standard curve is calculated in each well on each run at the same time as the patient sample. Of the reluctance of rheumatologists to move away from IFA ANA, Howard says, “It’s a function of awareness. When the doctors were in medical school, they were trained from the fluorescence ANA perspective. That is what they are comfortable with. I believe that the data are out there to show that the product is more specific than fluorescence ANA.”
“This is a highly confused field right now,” says Henry A. Homburger, MD, who was not at the AACC session. Commenting on the argument over the equivalence of EIA and IFA, he says: “That battle has been going on for at least a decade. I don’t regard all ELISA assays to be suitable replacements for the screening IFA ANA test.” Dr. Homburger, who is professor emeritus of pathology and laboratory medicine at Mayo College of Medicine and former director of the Immunology Antibody Laboratories at Mayo Clinic, does consider at least one to be equivalent—about 10 years ago he replaced IFA with an ELISA as his firstline screen for ANA.
While the debate goes on, data collection continues. At ARUP’s Research Institute for Clinical and Experimental Pathology, Harry R. Hill, MD, executive director of the Institute, and Susan S. Copple, MS, MT(ASCP)SI, autoimmune laboratory technology supervisor, compared four FDA-approved ELISAs with IFA ANA on HEp-2 cells on almost 1,000 patient samples, including documented SLE patients, patients with rheumatoid arthritis, and controls from the CDC as well as samples sent to ARUP Laboratories for ANA testing. Three selected technologists read all IFA ANAs. Copple, who did the work as part of her master’s degree thesis in medical laboratory sciences, summarizes the main result: “We found that the ELISA ANAs were more sensitive in detecting ANAs in documented SLE patients while the ELISA-negative samples were also uniformly negative by IFA.”
Dr. Hill, group medical director of ARUP’s Laboratory of Immunology, concludes, “In contrast to the letter sent out by the American College of Rheumatology, ANA ELISA combined with multiplex screening for ENA to determine specificity of the ANA is probably more objective and reliable than subjective fluorescent ANA testing.”
Copple has personal knowledge of the negative impact of a false-positive IFA ANA result. Some years ago her mother went to the doctor with nonspecific symptoms. Because her IFA ANA was positive, she was diagnosed with SLE and put on high-dose prednisone. After a year, Copple says, “she ended up in the ER and subsequently required major abdominal surgery. After the operation it took more than a year for her to get off prednisone—she did not have an automimmune disease.” Since that experience, Copple says, she has heard similar stories from others. “At ARUP we are currently screening over 12,000 samples for ANA per month. We appear to be performing ANA tests on an increased segment of the population, which in turn increases the number of patients with ANA titers of 1:40 to 1:320 who may not have an autoimmune disease.” In some of these cases, she adds, the patients may be put on harmful drugs unnecessarily.
Dr. Keren says ANA testing has evolved considerably in the last half century. In the 1950s the LE test was used; Dr. Keren describes it as “fairly crude.” Sensitivity was improved by IFA staining on HEp-2 cells, which also allowed visualization of staining patterns, such as homogeneous, centromere, nucleolar, or speckled. In the past 10 to 15 years, EIA and multiplex bead assays have been introduced, which usually employ recombinant proteins or extracts. Although these tests do not give patterns, they can reveal the specific antigen(s) with which the patient’s serum reacts. (Multiplex ANA is basically an immunoassay that takes advantage of specific antigen-coated beads and flow cytometry to detect several ANA relevant autoantibodies simultaneously.)
Unfortunately, there are no standards for which nuclear antigens should be included in an ELISA or multiplex assay, nor for which concentration of each antigen to include. As a result, EIA ANA assays from different manufacturers react differently. With some specimens, one EIA kit will give a negative result while another yields a positive. “Yet all are called ‘ANA tests,’” Dr. Keren says.
It is estimated that between 100 and 150 antigens are present in a HEp-2 ANA test, providing greater variety than EIA kits. While no formal standards regulate IFA ANA substrates, Dr. Keren says, “in general, if you look at CAP Surveys, IFA substrates tend to perform more consistently than EIAs.” HEp-2 assays also provide patterns, though Dr. Keren believes that in many cases staining patterns don’t have as much clinical utility as previously thought. They are highly dependent on the experience of the observer, can vary depending on the dilution of serum used because a strong pattern can cloak a weaker pattern, and some patterns such as the centromere pattern are not readily detected on frozen section substrates, he says.
Guidelines have been published for the clinical use of ANA tests, including analysis of specific autoimmune diseases, and for assays for individual nuclear antigens, as well as on the utility of staining patterns (Kavanaugh A, et al. Arch Pathol Lab Med. 2000; 124: 71–81).
Because of the variation among ANA assays, Dr. Keren suggests, “it would almost seem to be preferable if the ANA test in a laboratory’s order sheet, or catalog, would list its reactivity against the particular antigens.” Even better would be the listing of conditions likely and unlikely to be detected by the specific test being used. “The term ‘ANA screen’ conveys expectations that vary from a complete general screen for autoimmune diseases to the ability to detect certain specific antibodies, such as anti-centromere, to help confirm diagnosis of the CREST syndrome,” he says.
In his laboratory Dr. Keren offers both EIA and HEp-2 assays. “Physicians can order whichever they want,” he says, “but if they order ‘ANA’ we screen with EIA.” If the EIA is positive, a HEp-2 assay confirms the positivity and provides both a titer and pattern. If the EIA is negative, they don’t reflex to HEp-2. In Dr. Keren’s experience, it is “very uncommon” to have a clinically significant positive on HEp-2 after a negative on EIA. Because of the variability of interpretation of HEp-2 assays, in Dr. Keren’s laboratory all slides are read in a blinded manner by two technologists who compare interpretations afterward. “It’s somewhat subjective, but in some cases so is surgical pathology,” Dr. Keren says.
Due to the variety of “ANA tests” available and the differences in titers, patterns, and antigenic specificities of the offered tests, interpreting the results is challenging for many primary care physicians. “I not infrequently get phone calls from internists where, for example, a sample was positive on EIA, then reflexed to fluorescence ANA, which gave a titer of 1:1,280 with a speckled pattern,” he says. In those cases Dr. Keren listens to the clinical history and asks for details about the patient. Many times the patient has only general complaints. “That’s a tough situation,” he says. “Many such ANA-positive patients have minimal nonspecific symptoms and do not have SLE. Yet the presence of a positive ANA compels some clinicians to order further costly testing or occasionally to try medication.”
In his AACC talk, Dr. Carey singled out two papers that define the false-positive rate of IFA ANA tests. In one publication, at a dilution of 1:40, fluorescence ANA had almost 100 percent sensitivity but a false-positivity of 32 percent; at 1:160 the false-positive rate dropped to five percent, but sensitivity also dropped, to about 90 percent (Tan EM, et al. Arthritis Rheum. 1997;40:1601–1611). The authors of the other paper, using a cutoff of 1:160, concluded, “ANA-EIA offers equivalent sensitivity and higher specificity compared to ANA-IFA” (Gniewek RA, et al. Clin Diagn Lab Immunol. 1997;4:185–188). As Dr. Keren said, at a realistic dilution, IFA ANA behaves much like a good EIA.
Because of the high false-positive rate of IFA ANA, there are a number of caveats in the literature about its use by nonspecialists. Dr. Carey cited two abstracts from the 1995 ACR meeting to make this point. One, from Lawlor, et al., said, “...The extremely low specificity and PPV of ANA ... suggests ANA use in primary care should be discouraged.” A similar warning about using IFA ANA as a general screen was sounded in an article about distinguishing fibromyalgia from lupus:
“Many physicians have been taught to screen for ‘connective tissue diseases’ with serum testing for ANAs, and then clarify the diagnosis with additional autoantibody testing. In practice, however, this approach does not work. The reason: a positive ANA test by [IFA], usually defined as detectable staining at a dilution of at least 1:40, is too sensitive and not sufficiently specific to be used as a screening test” (Blumenthal DE. Cleve Cl J Med. 2002;69:151–152).
In 1997, Dr. Carey reported results from a comparison of five EIAs versus fluorescence ANA at 1:40 in a group of patients whose clinical diagnoses had been established by three to five years of followup. Sensitivities for detecting systemic autoimmune diseases varied widely, he found: Fluorescence ANA was 96 percent sensitive, while the two best EIAs had sensitivities of 86 and 93 percent (Carey JL. Clin Lab Med. 1997;17:355–365). Dr. Carey implemented an EIA as his primary screen for ANA, preceded by consultation with rheumatology. “The educational component is important,” he said. He now performs about 15,000 EIA ANA tests annually. Specialists can order fluorescence ANA as a sendout; fewer than 1,000 are ordered per year. Dr. Carey cited a 1998 publication by Dr. Homburger as “one of the best papers on implementing ANA assays.”
In the early 1990s Dr. Homburger began looking for an EIA to replace HEp-2 for ANA testing, mostly because ELISA was faster and had less variability than IFA. “We had a very busy lab, having technologists rotate through stations, which made it incredibly difficult for them to maintain their proficiency reading ANA by IFA,” he says. “Most EIAs at that time were not very good.” When a kit from Helix Diagnostics became available that looked satisfactory, he and his colleagues tested it against IFA ANA. EIA ANA was positive in all 197 patients with clinically diagnosed systemic rheumatic diseases, compared with 95 percent positivity with IFA ANA (Homburger HA, et al. Arch Pathol Lab Med. 1998;122:993–999). Among 102 healthy control subjects, each of the two methods was positive in 15 percent. Discordance between the two assays appeared when results were weakly positive. “This tells you that when you are near the cutoff point, it is not unusual to get a positive in one assay and a negative in the other,” Dr. Homburger says. “We don’t consider that a problem. We couldn’t identify the specificity of the antibody in most of those cases anyway.” Dr. Homburger adds, “Not all ELISA kits we looked at then or since met that criterion of performance.”
After careful consultation with Mayo Clinic’s rheumatologists, Dr. Homburger converted entirely from IFA ANA to ELISA ANA. “I wanted to know if they would miss patterns and titers,” he says. “They told me they used ANA as a semiquantitative screening test. As long as there was a rough correspondence between response on ELISA and IFA ANA, that would be OK.” The rheumatologists said they paid relatively little attention to patterns. “In lupus they relied more on followup tests for specific antigens. Overall, they were quite content to support our switching to ELISA,” he says. Dr. Homburger offered a complete battery of individual followup tests to identify antigen specificity. Followup is now done with the BioPlex 2200 multiplex assay.
To Dr. Homburger, “the crux of the current question” is not just whether EIA is equivalent to IFA for ANA testing, but whether certain methods are de facto unsuitable for use in ANA screening. “ACR would maintain that an indirect fluorescence assay is the only suitable method for routinely testing for broad-spectrum ANA,” he says. “They doubt the validity of ELISA and multiplex methods.” He agrees that some ELISAs are not capable of picking up all specificities of ANA detected by IFA. However, he contends, “Some ELISA assays utilize a digest of HEp-2 cells, sometimes supplemented with purified antigens, and they do behave in an equivalent way to IFA and could be used as such.”
At this time, Dr. Homburger believes, multiplex assays are not equivalent to screening ANA by IFA or by a good ELISA. “By definition multiplex assays incorporate a cocktail of microspheres or beads each of which has a single antigen on its surface,” he says. “The sum of those antigens is less than what is in IFA ANA or a broad-spectrum ELISA.” For instance, nucleolar antigens are not present in multiplex assays. Makers of the assays would argue that absent antigens, such as nucleolar antigens, are not clinically significant, Dr. Homburger says. But he believes that detecting specificities of ANA such as nucleolar antibodies is in fact clinically important. “Antinuclear antibodies with specificity for nucleolar antigens are found in patients with scleroderma and SLE, and failure to detect these antibodies by multiplex methods can be a serious limitation when multiplex assays are used as a first order test for ANA in patients suspected of having a connective tissue disease,” he explains. “In some cases, this is the only specificity of ANA that is present, and multiplex methods will give false-negative results in such cases.”
Dr. Bloch attempts to educate clinicians about the possibility of false-positive fluorescence ANA results. “For many years I have taught that there is no reason to fear a positive ANA,” he said at the AACC session. “For a physician who is armed with the knowledge of potential false-positive results, a positive fluorescence ANA tells you to consider a differential diagnosis of possible ANA-associated diseases.” The list, he adds, includes a variety of systemic and organ-specific autoimmune diseases, as well as infections, and malignancies and medication-induced ANAs.
In an interview, Dr. Bloch expanded on this idea. “I was trying to make the point that ANA testing is performed as part of an overall evaluation, often of a patient who has symptoms that do not suggest a specific systemic rheumatic disease. The patient I described could have had one of many rheumatic diseases, as well as some nonrheumatic diseases. So even though she didn’t have any of the diagnostic criteria for lupus, an ANA was ordered because the doctor was trying to get a hint as to what was going on.” Over time, Dr. Bloch says, internists and rheumatologists have become accustomed to how IFA ANA assays work, including that these assays are positive in a certain proportion of healthy people. “Most know that a positive ANA by itself doesn’t mean a patient has lupus,” he continues. “It could be positive in a wide variety of diseases, as well as in a proportion of the normal population.”
Dr. Hill and Susan Copple of ARUP offer additional insights into ANA testing from their comparative study. Copple notes one important source of false-positive results in IFA ANA assays: “Some companies providing IFA ANA assays are not following the guidelines on using IgG-specific heavy chain conjugates. This increases the rate of false-positives due to IgM and IgA ANAs, which do not have the same diagnostic significance as IgG ANA.”
In Dr. Hill’s opinion, IFA ANA should be used only in a setting of clinical suspicion of a specific autoimmune disease, “not for overall screening.” He and Copple are working closely with a University of Utah rheumatologist, Allen Sawitzke, MD, who is a co-author on their paper, to work out an algorithm acceptable to the clinicians and to the laboratorians. “We feel that in patients with suspected autoimmune disease, initial screening should be carried out with a sensitive ELISA, then followed by multiplex testing to determine the extractable nuclear antigen specificities of the detected ANA,” Dr. Hill says. In their view, this offers optimal sensitivity combined with good specificity in diagnosing autoimmune disorders.
Evaluations of multiplex ANA testing have been published (Shovman O, et al. Ann NY Acad Sci. 2005; 1050:380–388; Binder SR. Lupus. 2006;15: 412– 421). Bead assays have from six to 13 individual autoantigens; some assays are supplemented by a bead coated with a HEp-2 extract. Bio-Rad’s Binder points out that the four multiplex tests are all different. Prospective purchasers have to be aware of how each test is constituted and how it performs.
In principle, one advantage of multiplex ANA assays is increased specificity. Unfortunately, Binder says, “Some perfectly healthy people will consistently test positive for autoantibodies. Maybe they have only one, and at a low level. Only a small percentage are on their way to developing SLE or another autoimmune disease.” In reality, it is impossible to develop a test with which all false-positive results can be eliminated. “Thirty percent is too high, while zero percent is unattainable,” Binder says. “Multiplex assays in this decade try to balance a low false-positive rate with the high sensitivity—above 90 percent—of EIA assays.”
There is debate now about whether multiplex assays really do match the sensitivity of IFA ANA. In Binder’s talk, he showed data from a study done several years ago at the Oklahoma Medical Research Foundation using archived samples from military personnel. That initial research showed that autoantibodies appear before clinical onset of SLE (Arbuckle MR, et al. N Engl J Med. 2003;349:1526–1533). In the current research, serum samples from 129 of the 130 patients, all taken five to 10 years before SLE classification, were analyzed by multiplex assay. Sensitivity was equal between multiplex and IFA ANA. Positives appeared earlier with multiplex in 30 samples, compared with 15 for IFA. Data have been submitted for publication.
Binder emphasizes that a valid comparison must use samples from patients who are being worked up to establish a diagnosis, as opposed to people who have been treated for some years. “As soon as you start treating, the body’s ability to produce antibodies changes,” he points out. In one survey, antibody positivity by fluorescence ANA dropped from 100 percent to 84 percent in 10 years (Sjowall C, et al. J Rheumatol. 2008;35:1994–2000). “We have tried to frame the sensitivity debate by focusing on patients presenting for initial workup,” Binder says. “In my opinion, a multiplex method can offer sensitivity similar to IFA in these patients.”
Arguments and data have gone back and forth regarding specific autoantigens, such as nucleolar antigens for scleroderma. Broader evaluations have also been published (Reveille JD, et al. Arthritis Rheum. 2003;49:399–412). Binder points to a 2002 ACR meta-analysis that has “a long list of indications for which the IFA is of no proven value” (Table 10 in: Solomon DH, et al. Arthritis Rheum. 2002;47:434–444). “Most of the ‘specificities’ that are not seen in multiplex methods are related to this list,” he says. “The problem is that rheumatologists don’t want to accept this meta-analysis.”
Meanwhile, Denise Facaros and Dr. Connolly have expanded multiplex testing for ANA from St. Joseph’s and other local hospitals in the Catholic Healthcare West system to several CHW hospitals in other states. “We had to continue to educate clinicians,” Facaros says. “We didn’t have complete success. Some doctors were reasonable and others were really stubborn. A few said, ‘I want my pattern and titer.’” Since the lab had stopped doing IFA ANA, they send these requests to a reference lab. Facaros says these requests are “few and far between.”
Current multiplex ANA volume is 2,000 samples per month. Facaros initially became excited about multiplex ANA when she attended a symposium put on by a manufacturer of another multiplex ANA platform. When she tried that vendor’s product, however, “it backfired on us,” she says. “It gave weak reactions on patients with no signs or symptoms.” When the BioPlex 2200 came out, Facaros says, “We started skeptical.” They did several comparisons between the BioPlex and the previous multiplex platform, as well as versus ELISA, before they were convinced that BioPlex works.
In addition to good performance, Facaros says, the BioPlex 2200 has software that is helpful. “It holds many patterns of patient data and gives the best fit,” she says. “It provides an interpretation for clinicians.” The report may say “Consistent with patient who has mixed connective tissue disease” or “Consider SLE” or “Consider scleroderma.”
“I think it has helped greatly,” she says. It can even be helpful with weak positives, saying “Doesn’t match any disease.”
In this period, when a laboratory may be using any of three very different types of ANA assay, reports should provide basic information, Dr. Carey says. Along with the method used, the report should define specified antigens, if that is relevant. For a negative IFA screen, the report should caution that systemic rheumatic disease is not excluded; for a positive, it should say that it is not diagnostic. “Systemic rheumatic disease is predominantly a clinical diagnosis,” Dr. Carey says.
Though Dr. Bloch remains a partisan of the HEp-2 ANA method, he will not be surprised if someday solid-phase assays can replace IFA. “Many additional autoantigens can be put on beads,” he says. “At some point there may be enough, different, clinically significant autoantigens on beads to permit this modality to replace indirect immunofluorescence as a screening test for antinuclear antibodies.” But, he says, we are not there yet.
William Check is a medical writer in Wilmette, Ill.