Statistical tools for immune correlates analysis of vaccine clinical trial data

Abstract

In vaccine research, it is important to identify biomarkers that can reliably predict vaccine efficacy relative to a clinical endpoint. Such biomarkers are known as immune correlates of protection (CoP) and can serve as surrogate endpoints in vaccine efficacy trials to accelerate the approval process. CoPs must be rigorously validated, and one method of doing so is through the controlled vaccine efficacy (CVE) curve, a function that represents the causal effect of the biomarker on population-level vaccine efficacy. In this dissertation, we propose and study two methods to estimate the CVE curve and construct pointwise confidence bands. The first method assumes a Cox proportional hazards model, allowing for possible nonlinearity in the additive predictor. The second method assumes monotonicity of the CVE curve and leverages modern tools from shape-constrained inference and nonparametric efficiency theory. We also develop a nonparametric test of the null hypothesis that the CVE curve is flat, and discuss extensions of these methods that lead to improved finite sample performance. Finally, we describe open-source software that we published to implement these methods, and apply the methods to data from several vaccine efficacy trials.