Analysis of Coronavirus Prevention Network COVID-19 vaccine efficacy trials: Do vaccine side effects predict better efficacy to prevent COVID-19?

Abstract

Many previous studies have investigated the predictiveness of antibody markers on COVID-19 and there are also mixed results indicating the association between immunogenicity and reactogenicity induced by vaccines. However, little is known about the direct association of reactogenicity and clinically relevant endpoints such as symptomatic COVID-19. Among the Moderna trial, the Novavax trial, and the Sanofi Stage 1 and Stage 2 trials, we assessed the associations of reactogenicity variables, specifically the occurrence of any AR, any local AR, any systemic AR, and fever with future risk of COVID-19 among vaccine and placebo recipients using marginal variable importance measures (VIM) analysis, and we examined the vaccine efficacy in the subgroups defined by reactogenicity outcome if assigned vaccine using principal stratification (PS) analysis. By the VIM analysis, we do not find a significant increase in predictiveness for reactogenicity variables across all vaccine trials. By the PS analysis, the results in the Moderna trial suggest improved vaccine efficacy among the subgroup with vaccine-caused reactogenicity compared to the subgroup without reactogenicity, while there’s an opposite direction of vaccine efficacy suggested by the Sanofi trials. Although many results are not significant, the opposite pattern we found between mRNA and recombinant protein vaccines can contribute to the design of future vaccines.