Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The article by Pollock and Toh1 about the detection of antinuclear antibodies (ANAs) using HEp-2 cells transfected with Ro/SS-A raises the important question of the optimal method for screening of sera from patients with suspected autoimmune connective tissue disease. We have directly compared the performance of the same Hep-2 transfected cells (Hep2000; Immuno Concepts, California, USA) with Hep-2 untransfected cells (Quantafluor; Sanofi Diagnostics Pasteur Inc, Minnesota, USA). The results from our study combined with a reassessment of Pollock and Toh's data cast doubt on their conclusion that Hep-2 transfected cells are more reliable than other substrates for detecting clinically meaningful ANAs.
Sera from 258 patients referred to our laboratory for ANA testing were analysed for the presence, titre, and pattern of ANAs using Ro/SS-A transfected and untransfected HEp-2 cells. Indirect immunofluorescence was performed at a screening dilution of 1/40 and titred to 1/2560, and slides were reviewed independently by at least two scientists. In general, the correlation between the two substrates for detection of ANAs was good (table 1); discrepancies that are unlikely to be of clinical relevance were found for low titre positives. The only significant difference between Hep-2 transfected and Hep-2 untransfected cells was that the latter were more sensitive for ANAs at high titres.
Pollock and Toh report that seven of 110 Ro/SS-A positive sera did not have a positive ANA pattern in the background non-hyperexpressing cells; in contrast, we found speckled ANA staining using the untransfected cells in 10 of 11 anti-Ro/SS-A positive samples. One patient with alcoholic liver disease was repeatedly negative on Hep-2 untransfected cells but Ro/SS-A positive on Hep-2 transfected cells. This patient's serum was analysed for antibodies to extractable nuclear antigen (ENA). Enzyme linked immunosorbent assay (ELISA) (ENA RELISA; Immuno Concepts) detected anti-SS-A antibodies at a concentration of 33 ENA units. However, the patient does not have any features of an autoimmune connective tissue disease and testing of his serum by counterimmunoelectrophoresis and line immunoassay (INNO-LIA ANA; Innogenetics NV, Ghent, Belgium) confirmed the absence of anti-Ro/SS-A antibodies.
Pollock and Toh also conclude that the positive predictive value and specificity of detection of anti-Ro/SS-A antibodies on Hep-2 transfected cells is 100% and 91%, respectively, by comparing detection by immunofluorescence with detection by ELISA. This conclusion is questionable, because the specificity and predictive value of the test should depend on a correlation with the clinical diagnosis rather than another diagnostic test. Although it is useful to know the concordance between Ro/SS-A detected by Hep-2 transfected cells and ELISA, the true clinical usefulness of the test lies in its specificity and positive predictive value for the associated clinical conditions. In fact, the true specificity of the test can be calculated from the data provided in table 3 of their paper, and is only 77% (64 of 83 patients positive for Ro/SS-A) for the diagnosis of systemic lupus erythematosus and Sjogren's syndrome, with a sensitivity of 89% (64 of 72 patients).
Together with our data, these figures indicate that the Hep2000 substrate in fact performs suboptimally, and like ELISAs that detect anti-Ro/SS-A antibodies in up to 17% of normal sera,2 appears too sensitive for reliable diagnosis in autoimmune connective tissue disease. Furthermore, the advantage of using the transfected cell line for initial screening is likewise questionable, because not only do anti-Ro/SS-A positive sera detected by immunofluorescence need to be checked for additional ENAs by other methods, but 15 sera testing positive for anti-Ro/SS-A antibodies by immunodiffusion were negative by immunofluorescence in Pollock and Toh's study.
The authors reply
First, we would like to point out that the cohort of sera screened for antinuclear antibodies (ANAs) by Williams et al (258) is considerably smaller than that screened by us (10 500).1 Second, the incidence of ANA negative Ro detected on conventional Hep-2 cells (one in 11) is comparable with our report of seven in 110 ANA negative Ro. Third, although the one ANA negative Ro reported by Williams et al did not appear to have autoimmune connective tissue disease, it cannot be implied from this very small sample study that all ANA negative Ro do not have this disease. Fourth, we have made no claims to the specificity of the assay as stated in the commentary by Williams et al. We were at pains to avoid the use of this term. Clearly, Williams et al have misread our text. We have commented on the sensitivity of the assay and not its specificity. They also appear to have misread our findings with respect to the 15 sera that tested positive by immunodiffusion, which they refer to in their letter. Ro was missed in 14 of these 15 sera not because the test was negative by immunofluorescence but because they gave strong immunofluorescent staining and high titre ANAs, which we suggested might have masked the presence of Ro. Fifth, table 1 as presented in their letter is not informative because it does not give the number of tests that are Ro positive using the Hep2000 substrate. Finally, the conclusions of Williams et al that the Hep2000 assay performs suboptimally and is too sensitive for reliable diagnosis of autoimmune disease is unfounded. The specificity of the assay of 77% as calculated for us by Williams et al based on our data is respectable. Furthermore, there are currently no published data (and Williams et al certainly have not provided any data, other than the single ANA negative Ro) that the test is too sensitive for routine diagnosis. In closing, we stand by our conclusion that Hep2000 provides a suitable substrate for the routine and ready detection of Ro positive sera by immunofluorescence.