All Times EDT
An objective of phase 2b or 3 preventive vaccine efficacy trials is to assess post-vaccination immunological biomarkers as various types of “immune correlates of protection,” with objectives including to identify correlates of disease risk (alternatively, predictors/signatures of disease risk), modifiers of vaccine efficacy, and mediators of vaccine efficacy, and ultimately to define surrogate/replacement endpoints that can constitute the basis for traditional or accelerated approval, as well as for making bridging predictions of vaccine efficacy to new settings. These objectives can be assessed statistically using regression/machine learning modeling, principal stratification causal inference, natural direct and indirect effects causal inference, and a variety of approaches for quantifying the quality of a biomarker(s) as a surrogate endpoint, respectively. In this talk I will describe some recent work on statistical methods for assessing correlates of risk and modifiers of vaccine efficacy, with application to dengue and HIV vaccine efficacy trial data sets. The correlates of risk methods (Sun, Qi, Heng, Gilbert, 2020, JRSS-C) assess how biomarker correlates of risk depend on the amino acid sequence distance of the pathogen to the pathogen strain(s) represented in the vaccine construct, helping define the spectrum of pathogen strain diversity for which the biomarker may be a correlate of protection. These methods use a mark-specific proportional hazards model with nonparametric kernel smoothing in the sequence distance, augmented inverse probability weighting for missing biomarkers, and multiple imputation for missing pathogen sequences. The modifiers of vaccine efficacy methods (Gilbert, Blette, Shepherd, Hudgens, 2020, JCI) extend Survivor Average Causal Effect (SACE) methods to assess binary biomarker vaccine efficacy moderation under minimal assumptions without requiring baseline immunogenicity predictors or close-out placebo vaccination.