Immunogenicity Assays for SARS-CoV-2 Vaccine

A proportion of serum samples from vaccine recipients at PB28 were tested on 3 different assays with 4 assay readouts. All NAAT+ cases were tested if sample volume allowed, and a proportion of noncases were tested. Samples were tested blinded to case status. The data from noncases were obtained first, and consisted mainly of the samples processed for the initial application for emergency use which needed 15% of samples included in the efficacy cohort to be processed on validated assays. Subsequent to this NAAT+ cases were sent for testing as they occurred, if not already including the 15%. We assume the mechanism of missingness for samples that were not tested to be missing at random^{39 (link)}. To account for the missing data, factors associated with sample availability were controlled as weights in the analysis (see ‘Correlates of risk’ and ‘Inverse probability weighting’ below).
Anti-SARS-CoV-2 Spike and RBD IgG were measured by a multiplex immunoassay on the MSD platform at PPD Laboratories. The assay sequences were based on the ancestral sequences from Wuhan, China. Antigen information and sequence information are provided in Supplementary Table 1. Assay validation included precision and ruggedness, dilutional linearity, selectivity, and relative accuracy for each SARS-CoV-2 antigens. Post-validation studies for stability and for conversion to the WHO standard, as well as the establishment of a cut-point, were performed. The lower limit of quantifications (LLOQs) for anti-spike and anti-RBD are 33 and 204 AU/ml, respectively.
Antibody neutralization was measured with a lentivirus-based pseudovirus particle expressing the D614 SARS-CoV-2 spike protein. The pseudovirus neutralizing antibody assay was validated at Monogram Biosciences. Validation included accuracy, repeatability, intermediate precision, linearity, specificity/selectivity, sensitivity, and stability utilizing pooled sera from high-titer, intermediate-titer, and low-titer pooled convalescent SARS-CoV-2 sera, as well as historical negative samples collected in the year 2017 (prior to SARS-CoV-2 circulation). The LLOQ for pseudovirus neutralizing antibody is 40 (ID₅₀).
Antibody neutralization was also measured by a live microneutralization assay using the Victoria/01/2020 strain of the virus (Public Health England). Qualification of the assay included assessment of specificity, parallelism, dilutional linearity, repeatability, intermediate precision, and assessment of the assay range. A formal validation has since been completed (after the testing of clinical study samples in this manuscript). Normalized values (NF₅₀) were used for the main analyses, as the normalization process removes the plate-to-plate variability and normalized values are more highly correlated with binding antibody and pseudovirus neutralization assays. However, normalized values cannot be converted into WHO standard units. A sensitivity analysis is provided in Supplementary Table 4 using non-normalized values (ND₅₀), which are also presented as IU/ml using the WHO standard, but are less highly correlated with other assays. The LLOQ of the assay is 58 (ND₅₀) and 8.6 (NF₅₀).
Due to the limitations of laboratory capacity, fewer samples were tested for virus neutralization than were tested using the quicker multiplex assay.