Assessing Protective Efficacy of a Vaccine using Immunological Surrogate Endpoints

Gebre M and Alemu G

Published on: 2023-10-18

Abstract

The ability to evaluate the protective efficacy of a vaccine by determining the percentage of vaccines who produce a particular immune response, without having to measure clinical outcomes, has significant advantages. Hence, the substitute endpoints, immunological correlates of protection, are frequently used to predict a vaccine efficacy on the clinical outcome. Knowing this, this study was conducted in order to investigate the variability of the predicted vaccine efficacy of a Herpes Zoster disease as compared to placebo using continuous correlate of protection. This is clinical study in which simulated data from 10000 subjects (5000 vaccinated and 5000 unvaccinated) were used. The clinical outcome, occurrence or non-occurrence of disease was considered as the response, whereas the continuous correlate of protection, log antibody titer value of the subjects was considered as a predictor variable. Different statistical methods, such as exploratory data analysis, logistic regression and scaled logit models were employed to predict the vaccine efficacy. Besides, different estimation methods, delta, bootstrap and simulation methods were used in order to scrutinize the variability of the vaccine efficacy. Findings from the methods implied that the variability is higher at the lower values of the vaccine efficacy as compared to that of at the higher values in all methods. In addition to that, these predicted confidence intervals are closer to the observed confidence intervals at the higher values of the vaccine efficacies as compared to that of at the lower values. However, the confidence intervals of the delta method is wider than that the other methods confidence intervals, and far from the observed one at the lower values of the VE while they are very similar at the higher values of the VE. At the lower values of VE, the delta method overestimates the VE. In conclusion, at the lower values of the vaccine efficacy, the confidence intervals are less precise, especially for delta method as the variability is high and not close to the observed CIs comparing with at the higher values of VE. From these three methods, the bootstrap method is better method as its results are clinically precise and close to the true results, whereas the delta method is worse relatively. But, the bootstrap method is computationally more difficult while the simulation method easier. At the higher values of VE, delta and simulation methods slightly underestimate the VE.