Serological assays for delayed SARS-CoV-2 case identification – Author's reply

We read with interest the insightful comments put forward by Kay Weng Choy, raising important considerations for clinicians planning to use point-of-care serological assays for delayed case identification of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in response to those presented in our Article.¹

We agree that along with evaluating assays for cross-reactivity of IgM and IgG with common infectious diseases, further benefit could be derived by assessing the potential of assay performance in those with autoimmune disease and immunodeficiency. Indeed, previously reported work in SARS-CoV-1 would suggest the potential for cross-reactivity of autoantibodies for SARS-CoV-2 IgG.² Similarly, consideration has also been given to how age could affect viral load and the subsequent development of SARS-CoV-2 IgG antibodies,³ which is a focus of our ongoing work.

Our study was designed specifically to evaluate the use of point-of-care assays for frontline health-care workers directly involved in the clinical care of patients with SARS-CoV-2 infection; therefore, it was not possible to evaluate any difference in detection of SARS-CoV-2 IgG in young or older people. At a strategic level, health-care workers with autoimmune disease or known immunodeficiency were required to be actively shielding during the study period and so were unable to take part.⁴ An evaluation of the potential effect of immunodeficiency on assay performance was beyond the scope of our study; however, we strongly agree that this is an important issue for future studies where consideration can be given to testing in different populations.

Additionally, Kay Weng Choy correctly highlights that whole blood is likely to be the primary sample type at point-of-care and, therefore, evaluation of diagnostic performance is warranted for whole blood and serum samples. Further research involving our group has been reported in August, 2020, comparing not only serum with whole blood samples in the laboratory, but also with finger-prick testing across a number of different point-of-care assays.⁵ Observed test sensitivity was broadly similar; however, the reported variation in assay performance across these three methods highlights the need for robust evaluation of individual kits (as they become available) in specific populations. This variation is particularly relevant if consideration is being given to use with finger-prick blood. In our study, individuals were recruited on a single occasion and no repeat testing was considered in the performance evaluation.

Finally, Kay Weng Choy highlights the value of orthogonal testing algorithms, advocating for a second test, each with unique assay design characteristics with the aim of improving the positive predictive value. Indeed, within our own institutions we have developed a testing algorithm that uses an anti-nucleocapsid and anti-spike protein immunoassay. This algorithm has potential to increase diagnostic yield but it is worth noting that insufficient antibody target data provided by a considerable number of manufacturers could provide additional challenges to the design of similar testing programmes.⁶