BackgroundThe antibody response generated following infection with SARS-CoV-2 is expected to decline over time. This may cause individuals with confirmed SARS-CoV-2 infection to test negative according to serological diagnostic tests in the months and years following symptom onset.
MethodsA multiplex serological assay was developed to measure IgG and IgM antibody responses to four SARS-CoV-2 Spike (S) antigens: spike trimeric ectodomain (S tri ), its receptor-binding domain (RBD), spike subunit 1 (S1), and spike subunit 2 (S2).Antibody responses were measured in serum samples from patients in French hospitals with RT-qPCR confirmed infection (n = 259), and negative control serum samples collected before the start of the SARS-CoV-2 epidemic in 2019 (n = 335).The multiplex antibody data was used to train a random forests algorithm for classifying individuals with previous SARS-CoV-2 infection. A mathematical model of antibody kinetics informed by prior information from other coronaviruses was used to estimate time-varying antibody responses and assess the potential sensitivity and classification performance of serological diagnostics during the first year following symptom onset.
ResultsIgG antibody responses to one S antigen identified individuals with previous RT-qPCR confirmed SARS-CoV-2 infection with 90.3% sensitivity (95% confidence interval (CI); 86.1%, 93.4%) and 99.1% specificity (95% CI; 97.4%, 99.7%). Using a serological signature of IgG to four antigens, it was possible to identify infected individuals with 96.1% sensitivity (95% CI; 93.0%, 97.9%) and 99.1% specificity (95% CI; 97.4%, 99.7%). Antibody responses to SARS-CoV-2 increase rapidly 1-2 weeks after symptom onset, with antibody responses predicted to peak within 2-4 weeks. Informed by prior data from other coronaviruses, one year following symptom onset antibody responses are predicted to decay by approximately 60% from the peak response. Depending on the selection of sero-positivity cutoff, we estimate that the sensitivity of serological diagnostics may reduce to 56% -97% after six months, and to 49% -93% after one year.
ConclusionSerological signatures based on antibody responses to multiple antigens can provide more accurate and robust serological classification of individuals with previous SARS-CoV-2 infection. Changes in antibody levels over time may cause reductions in the sensitivity of serological diagnostics leading to an underestimation of sero-prevalence. It is essential that data continue to be collected to evaluate this potential risk.