Abstract
A traditional phase 3 clinical efficacy study for a Zika vaccine may be unfeasible because of the current low transmission of Zika virus (ZIKV). An alternative clinical development approach to evaluate Zika vaccine efficacy (VE) is therefore required, delineated in the US FDA's Accelerated Approval Program for licensure, which utilizes an anti-Zika neutralizing antibody (Zika NAb) titer correlated with non-human primate (NHP) protection as a surrogate endpoint. In this accelerated approval approach, the estimation of VE would be inferred from the percentage of phase 3 trial participants achieving the established surrogate endpoint. We provide a statistical framework to predict the probability of protection for human participants vaccinated with a purified inactivated ZIKV vaccine (TAK-426), in the absence of VE measurements, using NHP data under a single-correlate model. Based on a logistic regression (LR) with bias-reduction model, a probability of 90% protection in humans is expected with a ZIKV NAb geometric mean titer (GMT) ≥ 3.38 log(10) half-maximal effective concentration (EC(50)). The predicted probability of protection of TAK-426 against ZIKV infection was determined using the two-parameter LR model that fit the calculated VE in rhesus macaques and the flavivirus-naïve phase 1 trial participants' ZIKV NAb GMTs log(10) EC(50), measured by a ZIKV reporter virus particle assay, at 1 month post dose 2. The TAK-426 10 µg dose predicted a probability of protection from infection of 98% among flavivirus-naïve phase 1 trial participants.