OUP user menu

Enzyme-Linked Immunosorbent Assay Screening Then Indirect Immunofluorescence Confirmation of Antinuclear Antibodies
A Statistical Analysis

Susan S. Copple MS, MT(ASCP) SI, Allen D. Sawitzke MD, Andrew M. Wilson MStat, Anne E. Tebo PhD, Harry R. Hill MD
DOI: http://dx.doi.org/10.1309/AJCP6R8EELGODAYW 678-684 First published online: 1 May 2011


The purpose of this study was to analyze antinuclear antibody (ANA) screening by enzyme-linked immunosorbent assay (ELISA) followed by indirect fluorescent antibody (IFA) testing to confirm and characterize the pattern and titer of the antibody. We evaluated 4 ANA ELISAs and 1 HEp-2 IFA substrate in 224 clinically defined serum samples consisting of 30 from systemic lupus erythematosus (SLE) cases, 94 from rheumatoid arthritis cases, and 100 from healthy donors plus 495 serum samples submitted for routine ANA testing and 12 reference serum samples distributed by the Centers for Disease Control and Prevention. IFA tests were read independently by 2 certified medical technologists.

ELISA sensitivities ranged from 90% to 97% compared with 80% by IFA in the SLE serum samples. The ELISAs had specificities of 36% to 94%, whereas the IFA had 99% specificity.

Overall, ELISAs for ANA assays demonstrated better sensitivity and good specificity, suggesting ELISA is a more cost-effective alternative to IFA testing for initial ANA screening. Samples positive by ANA ELISA should be tested on HEp-2 to determine the titer and pattern.

Key Words:
  • Antinuclear antibodies
  • HEp-2 immunofluorescence
  • ANA ELISA testing
  • ANA screening
  • Multiplexing

An increasing volume of antinuclear antibody (ANA) testing has resulted in the development of alternative screening methods to indirect fluorescent antibody (IFA) testing for ANA detection. IFA testing requires experienced certified technologists to read and interpret the assay, and there is a lack of standardization of IFA materials. This study was designed to evaluate the diagnostic performance and usefulness of ANA testing by enzyme-linked immunosorbent assay (ELISA) to screen for ANAs compared with traditional HEp-2 ANA using IFA testing. To do this, serum samples from healthy subjects, serum samples from patients with clinically defined disease, standard control samples, and a large number of samples sent to a reference laboratory for routine ANA testing were examined.

Patients with several connective tissue diseases (CTDs), including systemic lupus erythematosus (SLE), Sjögren’s syndrome, scleroderma, polymyositis, and others are often positive for ANA. Although a negative ANA by IFA testing has been said to exclude a diagnosis of SLE, a positive ANA result is not specific for any CTD.16 Numerous studies have indicated that healthy people, as well as patients without a CTD, can have ANA IFA titers as high as 1:320 without signs of clinical disease.714 Difficulties inherent in clinically diagnosing CTDs have led to increased volumes of ANA testing.

The IFA test is currently considered the “gold standard” for testing for ANAs in clinical practice. By using HEp-2 cells as the substrate, the IFA test permits the detection of antibodies to more than 30 different nuclear and cytoplasmic antigens comprising more than 50 autoantibodies.15 ANA IFA test results are usually reported based on 4 basic patterns: homogeneous, speckled, nucleolar, and centromere. In addition, titers are usually reported for each pattern seen in IFA testing. However, it is a labor-intensive assay and is highly dependent on the skills of the reader.1,5,6,1518

ANA testing by ELISA has emerged as a screening tool to address the marked increase in ANA test volumes. ELISAs for ANA screening typically include the following antigenic specificities: double-stranded DNA (dsDNA), histone, Smith, U1 ribonucleoprotein (RNP), SSA (Ro), SSB (La), centromere B, Jo-1, ribosomal-p (Ribo-p), scleroderma-70 (Scl-70), and mitochondria; but they can vary in their use of native or recombinant antigens Table 1. In most ANA screening ELISAs, disrupted HEp-2 cells are also included in the assay, to supply the same antigenic sites found in the HEp-2 IFA test. Some investigators have thought that including the HEp-2 substrate increases sensitivity, while others argue it only lowers specificity. The majority of serum samples being tested for ANA by clinical laboratories are negative; therefore, by screening with ELISA, fewer ANA IFA tests can be run at a lower cost to laboratories and patients. We compared the performance of 4 ANA ELISAs with an ANA IFA test on HEp-2 cells.

Materials and Methods

Selection of Patient and Control Samples

A total of 731 samples were analyzed for this study. Of the samples, 30 were from patients with clinically defined SLE and 94 from patients with clinically defined rheumatoid arthritis (RA) based on the American College of Rheumatology (ACR) criteria. A set of 12 well-defined human reference serum samples distributed by the Centers for Disease Control and Prevention (CDC) and 100 samples from healthy donors from the Decades of Life normal study at the ARUP Laboratories, Salt Lake City, UT, were also tested. The healthy donor serum samples were from a population of 70% women and 30% men ranging in age from 19 to 59 years. The self-reported racial/ethnic groups were as follows: Caucasian, 83%; Hispanic, 7%; Asian, 5%; Mediterranean, 4%; and African American, 1%. A group of 495 clinical samples sent to ARUP Laboratories for routine ANA testing was also evaluated. Of the 495 samples, 50 initially tested negative by the ARUP screening method and 445 were positive. All samples in this study were deidentified, and the study was approved by the University of Utah Institutional Review Board (Salt Lake City).

View this table:
Table 1

Detection of ANAs

IFA testing and ELISA were the methods used for ANA testing. The IFA assay used was the NOVA Lite HEp-2 ANA (INOVA Diagnostics, San Diego, CA), which uses an IgG heavy chain–specific conjugate. The serum samples were tested on different days, and slides were read by 2 medical technologists using 1 of 2 fluorescent microscopes, each fitted with a light-emitting diode light source to minimize the effects of waning bulb life. The medical technologists were blinded to each other’s readings. Both have American Society for Clinical Pathology certification and 4 to 8 years of IFA reading experience. An ANA IFA consensus value was determined by assigning a positive result if both readers reported the sample as having a titer of 1:40 or greater. The consensus value was used to compare ANA IFA results with the results of the other assays. By using ANA ELISAs from 4 manufacturers (Aeskulisa ANA HEp-2, Aesku Diagnostics, Wendelsheim, Germany; Bio-Rad ANA Screen, Bio-Rad, Hercules, CA; Phadia Varelisa ANA 8 Screen, Phadia, Uppsala, Sweden; and QUANTA Lite ANA ELISA, INOVA Diagnostics), 731 samples were tested. None of the 4 ANA ELISAs was formulated in the same way. However, all contained dsDNA, Smith, RNP, SSA (60 kDa), SSB, Scl-70, centromere B, Jo-1, and Ribo-p antigens and HEp-2–disrupted substrate except for the Phadia assay, which did not contain the HEp-2 substrate. The specific manufacturer’s published mix of antigens is shown in Table 1. All assays were performed following the manufacturer’s recommendations by technicians competent in performing ELISA.


Analytical and Diagnostic Performance of IFA in the Detection of ANA

The study was initiated by first evaluating IFA testing for detecting ANA in serum samples from our patient groups. Because the interpretation of IFA testing is highly dependent on the observer, 2 certified, experienced medical technologists were asked to read and interpret these results. For statistical analysis, a sample was assigned 1 if it was reported by both readers to have a titer of 40 or more or 0 if one or both readers reported less than 40 (<1:40).19 The overall analysis showed significant agreement between results obtained by both readers (Cronbach α of 0.86). This was true for positives in the SLE and CDC samples (Cronbach α of 0.98 and 0.93, respectively). Increased variability in interpretation was observed among the 495 daily serum samples sent for ANA testing. In general, excellent agreement was observed for titers less than 1:20 and titers 1:320 or greater in the daily serum samples (data not shown).

Evaluation of Healthy Donor Serum Samples to Determine Assay Specificity

Figure 1 shows a comparison of the 5 ANA assays (4 ELISAs and 1 IFA test, all using IgG-specific conjugate) performed on a group of 100 serum samples from healthy donors. With the IFA test, all healthy donor serum samples were reported as less than 1:40 for ANAs, demonstrating a negative predictive value of 1.0 or 100% specificity. The Phadia, Aesku, and INOVA ELISAs demonstrated specificities of 94%, 96%, and 80%, respectively, resulting in false-positive rates of 6%, 4%, and 20%, respectively, in the healthy donor serum samples when compared with IFA results as the standard. The Bio-Rad ELISA demonstrated 36% specificity, resulting in a 64% false-positive rate.

Determination of Assay Specificity in the Diseased Control Serum Samples

Figure 1 shows test characteristics when using 3 of the 4 ELISAs and the IFA assay to test the RA serum samples. These serum samples were selected from a rheumatologist’s clinic for meeting the diagnostic criteria for RA.20 Owing to limited sample volume, the Bio-Rad assay was not used to evaluate the RA serum samples. The ANA ELISAs, Phadia, Aesku, and INOVA, detected 22%, 9%, and 45% positives, respectively, when testing for ANA in the 94 RA specimens. Only 6 of the 94 samples had a positive ANA result by the IFA test. Overall, 89 of the 94 serum samples were reported as less than 1:40. Five samples had a consensus titer of 1:160 with homogeneous patterns. The 5 samples positive by IFA testing were also positive by the 3 ELISAs.

Sensitivity Using Clinically Defined Serum Samples

Sensitivity was determined by analyzing the 30 clinically defined SLE serum samples that met ACR criteria.21 As shown in Figure 1, the ANA IFA test had only an 80% sensitivity for the 30 confirmed SLE serum samples, while the Bio-Rad, Phadia, Aesku, and INOVA ANA ELISAs demonstrated excellent screening sensitivities of 96.6%, 96.6%, 90%, and 96.6%, respectively. Of the 30 SLE serum samples testing negative by the Aesku ELISA, 2 were near the assay’s positive threshold value and tested positive by the other 3 ELISAs. One of these samples had a titer of less than 1:40, and the other a 1:40 titer and a mixed pattern by HEp-2 IFA testing.

Figure 1

Comparison of the percentage of positive samples using the 4 enzyme-linked immunosorbent assays (ELISAs) and human epithelial cell (HEp)-2 indirect fluorescent antibody (IFA) testing among the serum sample populations. Owing to limited sample volume, the RA samples were not run on the Bio-Rad assay. HC, healthy control serum samples; RA, rheumatoid arthritis serum samples; and SLE, systemic lupus erythematosus serum samples. Clinical samples were selected serum samples previously sent to ARUP Laboratories (Salt Lake City, UT) for antinuclear antibody (ANA) testing (50 negative and 445 positive by the Bio-Rad ANA ELISA).

CDC-Defined Samples

The CDC provides a set of 12 well-defined serum samples containing antibodies to specific antigens to manufacturers of ANA test kits, laboratories testing for ANAs, and others who want to determine the performance characteristics of their assays. Table 2 shows the antibodies included in the individual CDC sample and assay results. The Bio-Rad, Phadia, and Aesku ELISAs were positive in 12, 9, 10, and 10 of these 12 control samples, respectively, and the IFA assay was positive in 11 of the 12 serum samples. Three of the ELISAs did not detect fibrillarin or polymyositis-scleroderma. The antibody to fibrillarin, which is found in fewer than 3% of patients with SLE, was detected at a titer of 1:80, producing a clumpy nucleolar pattern on HEp-2 cells by IFA testing. The polymyositis-scleroderma antibody was detected by 2 of the 5 assays. It is a nucleolar antibody found in approximately 15% of patients with polymyositis.

View this table:
Table 2

Samples Submitted for Routine ANA Testing

This group consisted of 495 samples sent to ARUP Laboratories for routine ANA testing. The first 445 samples that were positive and the first 50 samples that were negative by the Bio-Rad ANA ELISA, used for screening ANAs at ARUP Laboratories, were chosen for further study. Performance of the other ANA ELISAs (Phadia, Aesku, and INOVA) was compared with the HEp-2 ANA IFA results.

Most samples were positive by the ELISA and the IFA methods or negative by both methods Figure 2. We further examined the samples that produced ANA IFA titers of 1:40 to 1:160 but were negative by the ELISAs. These samples consisted of 29, 26, and 17 for the Phadia, Aesku, and INOVA ELISAs, respectively. In this group, 20 samples had an ANA IFA titer of 1:40 with a homogeneous or mixed pattern, and another 4 samples produced titers of 1:80 with a homogeneous pattern. These 24 samples are consistent with low-titer autoantibodies found in the healthy population. Only 1 sample that was negative with the INOVA ELISA and 2 samples that were negative with the Phadia and Aesku assays produced titers of 1:160 with homogeneous patterns. The ANA IFA result is generally considered significant at a titer of 1:160 or greater when screening for autoantibodies. Thus, the INOVA, Aesku, and Phadia ELISAs demonstrated sensitivities of 98.3%, 96.4%, and 92.3%.

Figure 2

Of 495 serum samples, 50 negative and 445 positive samples were selected by screening with the Bio-Rad antinuclear antibody enzyme-linked immunosorbent assay used at ARUP Laboratories (Salt Lake City, UT). The percentage of positive results using the Phadia, Aesku, and INOVA assays as compared with human epithelial cell (HEp)-2 indirect fluorescent antibody (IFA) test is shown.

This group of samples (IFA+ and ELISA–) was further tested with 4 different multiplex immunofluorescent extractable nuclear antigen (ENA) assays. Each of the 4 assays consisted of combinations of SSA (60 kDa), SSA (52 kDa), SSB, Smith, Smith/RNP, recombinant RNP, Scl-70, centromere B, dsDNA, chromatin, histone, Ribo-p, and Jo-1. In this negative ELISA sample group, all except 2 of the 29 samples were negative for all analytes included in the 4 multiplex ENA assays. The 2 negative samples, which were negative by the Aesku ELISA and positive at 1:80 by ANA IFA testing, were positive for SSA (60 kDa) by all 4 ENA assays and also positive by the Bio-Rad, Phadia, and INOVA ELISAs, suggesting these were true misses by the Aesku ELISA. These results also suggest that low titers, 1:40 to 1:160, demonstrate reader bias.


Testing for ANAs is an initial logical step in evaluating for CTDs in patients with manifestations suggestive of such a diagnosis. The sensitivity, specificity, and predictive value of the test vary owing to the specifics of the assay selected by the laboratory. ANA IFA testing is also affected by many variables, such as the specificity of the substrate, the conjugate, the microscope bulb, and, especially, the reader. High-volume laboratories, such as ARUP, which runs 14,000 or more ANAs per month, need a platform to screen these serum samples.

In this study, we sought to screen for ANAs by ELISA followed by selective use of IFA testing as a routine approach for screening and confirming the presence of ANAs. The data showed higher sensitivities, 90% to 97%, for the ELISAs evaluated compared with 80% sensitivity for the HEp-2 IFA test in the clinically defined SLE samples. The high sensitivity of the ELISA allows it to be used for screening for ANAs so that negative serum samples can be reported directly, whereas serum samples positive by ELISA can be confirmed by IFA testing. Owing to the lower specificity of the ELISAs (36% to 94%), all positive samples would undergo further testing by IFA testing to confirm the presence of ANAs and at what titer and pattern. Using an ELISA with 95% or greater sensitivity would allow a laboratory to report the majority of patient samples, which are negative, at lower cost and with a shorter turnaround time. In contrast, the ELISA with 36% specificity would result in an unacceptable number of unnecessary ANA IFA tests.

In 2009, Qin et al22 reported their comparison of HEp-2 IFA testing with ELISA for ANA and DNA testing. They concluded: “ELISA prescreening combined with IIFA [indirect immunofluorescence assay] can obtain the information of the nuclear pattern and allow the observation of the titer alterations. The combination of two or more testing methods can greatly enhance the accuracy of the results.”22 A study by Tonuttia et al23 demonstrated that commercially available ANA ELISAs show different degrees of sensitivity and specificity and that some have a diagnostic accuracy that is comparable to or, in some cases, higher than IFA testing.

In December 1996, the National Committee for Clinical Laboratory Standards published a guideline for quality assurance of indirect immunofluorescence testing for ANAs, which offers a voluntary standard developed by consensus of the clinical laboratory testing community.17 This guideline examines the use of an antihuman IgG-specific conjugate vs an antihuman polyconjugate when testing for ANA by IFA on HEp-2 cells. It states that the use of a polyconjugate will detect the IgM class ANAs associated with RA, medications, and age, which are usually not of diagnostic significance. The guideline favored the use of an antihuman IgG-specific conjugate to enhance the positive predictive value of these ANA assays.17,24 Recently, the ACR released a position statement stating that the IFA test is the preferred method for screening for ANAs.25 Although this recommendation can be said to be prudent given the poor performance of certain commercially available ANA multiplex immunofluorescent microsphere assays, several studies have indicated limited usefulness of IFA testing as a screening tool.5,16,26,27 Indeed, the detection of specific antinuclear reactivities such as SSA and Ribo-p antibodies has been reported in patients with negative ANA IFA screening results due to low expression of these antigens and/or poor fixation of cells.15,17,25,27

Another common problem associated with the IFA test is the use of nonspecific IgG (heavy and light chains) or polyvalent conjugate that detects IgA, IgG, and IgM ANAs. In January 2000, the College of American Pathologists published a special article regarding the use of polyvalent conjugates in ANA testing.24 While the use of IgG-nonspecific or polyvalent conjugates enhances the positive predictive values of these assays, it is often associated with the detection of clinically insignificant antibodies.17,24 Results from a recent survey, CAP S-A 2008,28 indicated that titers varied from 1:40 to 1:1,280 in testing of aliquots of the same sample using different manufacturer’s HEp-2 assays in part because of these polyvalent or IgG-nonspecific conjugates.

The increasingly common practice of using the ANA test to rule out autoimmune disease, combined with the subjectivity of reading IFA tests and the failure to standardize the manufacturing of ANA IFA assays, has brought into question its usefulness as the preferred screening method.1,16,26,29 For our study, we used readings from 2 experienced, certified medical technologists as the standard for comparison in each of the study groups. ANA IFA testing demonstrated good sensitivity and specificity. However, the random sampling of daily specimens sent to the ARUP Laboratory for ANA testing resulted in a 13.5% chance of having the ANA IFA test produce a positive result. Patients whose serum samples had false-negative results, in this study 4.5%, most likely would have ongoing symptoms and eventually obtain follow-up testing and receive a proper diagnosis. However, 9% of IFA positives may be false-positives. Laboratories performing ANA IFA testing using a polyconjugate against IgG, IgM, and IgA could potentially produce even higher numbers of potential false-positive results. Patients with results falling into the false-positive category often are referred to a specialist, require additional testing with the associated costs, or are prescribed medications that are potentially unnecessary and harmful.30,31 Even if the false-positive ANA IFA rate is 1%, it could potentially clog rheumatologists’ clinics, causing delay in diagnosis and treatment of patients with actual autoimmune disease.29

Because an ANA assay with 100% sensitivity and specificity does not exist, clinicians must look to balance sensitivity and specificity. Based on this study, clinicians should test for ANAs only when a CTD is suggested by the patient’s history and physical examination findings. It is cost-effective to use a sensitive ANA ELISA when screening thousands of patient samples each month. Owing to its low specificity, the ANA ELISA should be used only for screening, and positive results should be sent for confirmation by HEp-2 IFA testing or further ENA testing to determine the presence and specificity of the antibodies. Screening for ANAs by a sensitive ELISA reduces subjectivity, cost, and turnaround time. However, a thorough investigation and validation of a screening assay for ANA is essential in choosing the right assay for the laboratory. With a vast number of assays and methods to choose from, it is extremely important for clinicians and laboratory staff to know and understand the limitations of all assays being used for ANA testing.1,3,5,18 Our data support the routine use of ANA ELISA screening of patients suspected of having SLE or other autoimmune diseases followed by IFA on positive samples for confirmation of antibodies, pattern, and titer.


We thank Diana Knapp, MT(ASCP), Jeanette Enbody, MT(ASCP), and Carri Craig, NCA I for their many hours of reading HEp-2 IFA slides. All kits used in this study were kindly provided free of charge by the manufacturers who had no role in the collection, analysis, or interpretation of data or the writing of the manuscript.


  • This study received funding from the ARUP Institute of Clinical and Experimental Pathology, Salt Lake City.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
View Abstract