Multiparametric Prediction Models for Coronavirus Disease 2019 Vaccine Selection: Results of a Comparative Population-Based Cohort Study

Abstract

Background. An understanding vaccine-dependent effects on protective and sustained humoral immune response is crucial to planning future vaccination strategies against coronavirus disease 2019 (COVID-19). Methods. In this multicenter, population-based, cohort study including 4601 individuals after primary vaccination against COVID-19 ≥ 4 months earlier we compared factors associated with residual antibody levels against severe acute respiratory syndrome coronavirus-2 receptor-binding domain (RBD) across different vaccination strategies (BNT162b2, mRNA-1273, or ChAdOx1). Results. Our main model including 3787 individuals (2 × BNT162b2, n = 2271; 2 × mRNA-1273, n = 251; 2 × ChAdOx1, n = 1265), predicted significantly lower levels of anti-RBD antibodies after 6 months in individuals vaccinated with ChAdOx1 (392.7 binding antibody units per milliliter [BAU/mL]) compared with those vaccinated with BNT162b2 (1179.5 BAU/mL) or mRNA-1273 (2098.2 BAU/mL). Vaccine-dependent association of antibody levels was found for age with a significant predicted difference in BAU/ml per year for BNT162b2 (−21.5; 95% confidence interval [CI], −24.7 to −18.3) and no significant association for mRNA-1273 (−4.0; 95% CI, −20.0 to 12.1) or ChAdOx1 (1.7; 95% CI, .2 to 3.1). The predicted decrease over time since full immunization was highest in mRNA-1273 (−23.4; 95% CI, −31.4 to −15.4) compared with BNT162b2 (−5.9; 95% CI, −7 to −4.8). Conclusions. Our study revealed population-based evidence of vaccine-dependent effects of age and time since full immunization on humoral immune response. Findings underline the importance of individualized vaccine selection, especially in elderly individuals.