A method to estimate probability of disease and vaccine efficacy from clinical trial immunogenicity data

Abstract

AbstractVaccine efficacy is often assessed by counting disease cases in a clinical trial. A new quantitative framework proposed here (“PoDBAY,” Probability of Disease Bayesian Analysis), estimates vaccine efficacy (and confidence interval) using immune response biomarker data collected shortly after vaccination. Given a biomarker associated with protection, PoDBAY describes the relationship between biomarker and probability of disease as a sigmoid probability of disease (“PoD”) curve. The PoDBAY framework is illustrated using clinical trial simulations and with data for influenza, zoster, and dengue virus vaccines. The simulations demonstrate that PoDBAY efficacy estimation (which integrates the PoD and biomarker data), can be accurate and more precise than the standard (case-count) estimation, contributing to more sensitive and specific decisions than threshold-based correlate of protection or case-count-based methods. For all three vaccine examples, the PoD fit indicates a substantial association between the biomarkers and protection, and efficacy estimated by PoDBAY from relatively little immunogenicity data is predictive of the standard estimate of efficacy, demonstrating how PoDBAY can provide early assessments of vaccine efficacy. Methods like PoDBAY can help accelerate and economize vaccine development using an immunological predictor of protection. For example, in the current effort against the COVID-19 pandemic it might provide information to help prioritize (rank) candidates both earlier in a trial and earlier in development.