Objective: To compare a series of commercial ELISA tests with an indirect immunofluorescent antibody (IFA) test for the detection of antinuclear antibodies (ANA) in children with juvenile rheumatoid arthritis (JRA).
Methods: Sera from 178 patients with JRA (88 pauciarticular, 68 polyarticular, 22 systemic) were compared with 26 healthy pediatric subjects. Twenty-one samples from patients with systemic lupus erythematosus (SLE) were also tested. All samples were analyzed by IFA and by 3 commercial ELISA methods. Concordance of ELISA results with IFA results (selected standard) were used as a measure of performance. Sensitivity and specificity were calculated for each test and likelihood ratios (LR) were established for IFA and ELISA in pauciarticular and polyarticular JRA sera. The increment in pretest probability was then obtained for each test as an additional measure of test performance.
Results: IFA rendered positive results on 18-77% of the JRA sera depending upon the subset, 100% of SLE sera, and 15% of normal patient sera. Using IFA as the standard, correspondence with positive results among patients with JRA ranged from 0 to 74% for the 3 ELISA tests, while it ranged from 5 to 73% in IFA negative sera. IFA tests showed intermediate range likelihood ratios (0.3, 0.5, 3.5, and 5) and increments in pretest probability ranging from 25 to 45%. While one of the ELISA tests attained 50% of increment in pretest probability for the positive test, it showed 0% increment as a negative test. The other 2 ELISA tests incremented the pretest probability from 0 to 25%.
Conclusion: Our findings indicate that in JRA, the lack of correspondence with the historic standard IFA precludes the use of ELISA tests for detection of ANA. In addition, IFA out-performs ELISA by a substantial degree when "clinical utility" analysis of test performance is utilized. Detection of ANA in children with JRA should either continue to rely on IFA or be based on a different set of antigens if an ELISA format is chosen.