-
PDF
- Split View
-
Views
-
Cite
Cite
Keri N Althoff, Kelly A Gebo, Sheri D Schully, Reply to Steele et al, Clinical Infectious Diseases, Volume 76, Issue 9, 1 May 2023, Pages 1698–1699, https://doi.org/10.1093/cid/ciad005
Close - Share Icon Share
To theEditor—Steele and colleagues [1] suggest 2 mechanisms for naturally arising high titer immunoglobulin (Ig) M (IgM) and IgG antibodies, resulting in random reactivity patterns that could give rise to false seropositive antibody detection results. We agree that attention must be paid to the likelihood of false seropositive antibody results, particularly during times of low infection prevalence such as the start of a pandemic when an antigen is circulating but not yet causing substantial clinically recognized illness.
Steele and colleagues [1] point to three severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) seroprevalence studies with a healthy skepticism of results, including our study findings, which they describe as “more represented of the ‘norm’ than” other studies [2–4]. We ask three questions of the noted studies by Steele and colleagues [1] to explore their differing estimates of seroprevalence prior to the “start” of the pandemic:
Which antibodies were detected? In our study of 24 079 individuals with an available serum specimen collected via the All of Us study protocol from 1 January to 18 March 2020, we only tested for IgG antibodies (via the Abbott Architect SARS-CoV-2 enzyme-linked immunosorbent assay [ELISA] and the EUROIMMUN SARS-CoV-2 ELISA) [2]. By not testing for IgM antibodies (unlike the other 2 studies [3, 4]), we could not classify individuals as seropositive prior to elevated IgG.
What testing approach was used? We used a sequential testing approach where specimens testing seropositive on the higher sensitivity assay targeting the nucleocapsid (Abbott) were subsequently tested on the more specific assay targeting the spike protein (EUROIMMUN). After estimating the sensitivity and specificity of the individual assays, we estimated that the sequential testing algorithm had a sensitivity of 90.7% and a specificity of 100.0%; given the high specificity, there was a low probability of a false seropositive in our sequential testing algorithm [2]. Sequential testing during times of low disease prevalence is an epidemiologic principal [5] and was the US Centers for Disease Control and Prevention's recommended approach for SARS-CoV-2 antibody testing prior to widespread illness. Basavaraju and colleagues [3] also used additional tests on specimens with a first positive result, but did not report estimated sensitivity and specificity of a sequential testing approach and acknowledged “no clear delineation between potentially cross-reactive specimens and those that were obviously from SARS-CoV-2–infected individuals.” Apolone and colleagues [4] used a proprietary receptor-binding domain (RBD)–specific ELISA to classify individuals as “positive cases” in their main analysis; however, only 6 of the seropositive cases had evidence of functional neutralizing antibodies upon further testing. The strength in the sequential testing is not only in reducing the potential for a false seropositive but also the ability to estimate the sensitivity and specificity of the sequential testing algorithm to quantify the probability of false seropositives.
How robust are the findings? Given that the prevalence of coronavirus disease 2019 (COVID-19) in the United States had not yet reached a threshold to confirm the first symptomatic case, we estimated the mean probability of a false seropositive from our sequential testing algorithm via a simulation study. We reported a 93% probability that at least 1 of our 9 individuals was falsely classified as seropositive, but that the probability that all 9 were false seropositives was 0.019% [2].
Studies using specimens collected during the weeks and months prior to the first confirmed symptomatic cases in a geographic area preceded the nasal specimen collection for SARS-CoV-2 polymerase chain reaction (PCR) testing and genomic analysis. Because no test or testing algorithm is perfect, well-designed serology study findings must be triangulated with phylogenetic genomic sequence analyses of the evolution of the virus to determine the extent of community transmission occurring prior to the first confirmed symptomatic COVID-19 cases.
Notes
Acknowledgments. The All of Us Research Program would not be possible without the partnership of its participants.
Financial support. This work was supported by the National Cancer Institute and by the Office of the Director, National Institutes of Health.
References
Author notes
Potential conflicts of interest. K. N. A. reports grants or contracts from the National Institutes of Health (NIH) and consulting fees paid to the author from TrioHealth for serving on the Scientific Advisory Board. K. A. G. initiated this work during her role as the Chief Medical and Scientific Officer of the All of Us Research Program. K. A. G. also reports grants or contracts paid to their institution from Octapharma, US Department of Defense's Joint Program Executive Office for Chemical, Biological, Radiological and Nuclear Defense; NIH National Institute of Allergy and Infectious Diseases (3R01AI152078-01S1); NIH National Center for Advancing Translational Sciences (U24TR001609-S3 and UL1TR00309); Mental Wellness Foundation; HealthNetwork Foundation; Bloomberg Philanthropies; Moriah Fund; Shear Family Foundation; and the State of Maryland Defense Health Agency (W911QY2090012). Additional royalties or licenses from UptoDate were paid directly to K. A. G., and additional consulting fees were paid directly to K. A. G. from Spark HealthCare, Teach for America, and Aspen Institute. K. A. G. serves as an uncompensated participant of Pfizer's Scientific Advisory Board. S. D. S. reports no potential conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
