Correlates of protection for booster doses of the BNT162b2 vaccine

Variants of concern (VOC) of SARS-CoV2 and waning immunity pose a serious global problem. Overall, vaccination and prior infection appear to provide significant protection to the majority of individuals, but some remain susceptible to infection and severe disease. Rigorously identifying a broad spectrum of correlates of protection (COP) is necessary to identify these susceptible populations. The extent to which additional booster doses provide protection is also poorly understood. To address this need, we conducted a multicenter prospective study assessing the association between serological profiles and the risk for SARS-CoV-2 infection, comparing those vaccinated with three to four doses of Pfizer BNT162b2 vaccine. Of 608 healthy adults, 365 received three doses and 243 received four doses. During the first 90 days of followup, 239 (39%) were infected, of whom 165/365 (45%) received 3 doses and 74/243 (30%) four doses. We found that the fourth dose elicited a significant rise in antibody binding and neutralizing titers against multiple variants, and reduced the risk of symptomatic infection by 37% [95% I, 15% - 54%]. We identified several parameters based on IgG and IgA binding that were COPs. The strongest association with infection risk was reduced IgG levels to RBD mutants and IgA levels to VOCs, which was a COP in the three-dose group (HR=6.34, p=0.008) and in the four-dose group (HR=8.14, p=0.018). A combination of two commercially available ELISA assays were also associated with protection in both groups (HR = 1.84, p = 0.002; HR = 2.01, p = 0.025, respectively). Most importantly, we identified a subset of individuals with low antibody levels after three doses of vaccine that responded with a significant boost in neutralizing antibody titers after a fourth dose, but were still at significantly increased susceptibility to infection when compared to those who had pre-existing high levels of neutralizing antibodies. Thus, we identify a highly susceptible population that remains susceptible despite apparent responsiveness to vaccines. Further, we develop several specific and sensitive COPs that show dramatic effect sizes and may be utilized to identify individuals most at risk from future exposures.