Using a multiplex serological assay to estimate time since SARS-CoV-2 infection and past clinical presentation in Malagasy patients

Background: The world is facing a 2019 coronavirus (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). In this context, efficient serological assays are needed to accurately describe the humoral responses against the virus. These tools could potentially provide temporal and clinical characteristics and are thus paramount in developing-countries lacking sufficient ongoing COVID-19 epidemic descriptions.Methods: We developed and validated a Luminex xMAP® multiplex serological assay targeting specific IgM and IgG antibodies against the SARS-CoV-2 Spike subunit 1 (S1), Spike subunit 2 (S2), Spike Receptor Binding Domain (RBD) and the Nucleocapsid Protein (NP). Blood samples collected periodically for 12 months from 43 patients diagnosed with COVID-19 in Madagascar were tested for these antibodies. A random forest algorithm was used to build a predictive model of time since infection and symptom presentation.Findings: The performance of the multiplex serological assay was evaluated for the detection of SARS-CoV-2 anti-IgG and anti-IgM antibodies. Both sensitivity and specificity were equal to 100% (89.85-100) for S1, RBD and NP (S2 had a lower specificity = 95%) for IgG at day 14 after enrolment. This multiplex assay compared with two commercialized ELISA kits, showed a higher sensitivity. Principal Component Analysis was performed on serologic data to group patients according to time of sample collection and clinical presentations. The random forest algorithm built by this approach predicted symptom presentation and time since infection with an accuracy of 87.1% (95% CI = 70.17-96.37, p-value = 0.0016), and 80% (95% CI = 61.43-92.29, p-value = 0.0001) respectively. Interpretation: This study demonstrates that the statistical model predicts time since infection and previous symptom presentation using IgM and IgG response to SARS-CoV2. This tool may be useful for global surveillance, discriminating recent- and past- SARS-CoV-2 infection, and assessing disease severity.