Science Spotlight

Community virus surveillance can double the data of vaccine trials

From Mr. Craig Magaret, Vaccine and Infectious Disease Division

Emerging viruses constantly challenge researchers to make vaccines that protect people from disease. And the difficulties do not always dissipate after an initial vaccine is produced, as seen for influenza virus (flu). Over time, viruses adapt or evolve to produce slightly different “flavors” from the original virus. For influenza virus, this means yearly adaptations in the compositions of the flu vaccine are made to account for predicted changes to the circulating strains. The current SARS-CoV-2 pandemic represents a different challenge, as new variants have emerged several times throughout the year, and it is unclear how well the vaccine protects against each emerging virus variant. Mr. Craig Magaret, Associate Director of Bioinformatics with Coronavirus Prevention Network and the HIV Vaccine Trials Network, as part of the Vaccine and Infectious Disease Division at Fred Hutchinson Cancer Center, collaborated with Drs. Dean Follmann and Michael Fay from NIAID to complete the computational “leg work” necessary to build the modeling infrastructure that will predict vaccine efficacy for emerging virus variants. This new approach was published last month in the journal Statistics in Medicine.

A vaccine in clinical trials is given to a randomly selected subset of the healthy participants, while the unvaccinated or placebo group acts as a control. Following the trial, the placebo group may be given the opportunity to receive the vaccine, representing a delayed, secondary time point of vaccination termed “deferred vaccination”. "One of the major challenges with fighting the ongoing SARS-CoV-2 epidemic is how to evaluate the efficacy of COVID vaccines to new variants that emerge after the placebo group has been vaccinated,” stated Magaret. “Such information can help guide public health agencies and vaccine developers with their strategies to curtail the epidemic by understanding how much efficacy is lost with emerging variants." To learn how well a vaccine holds up against emerging virus variants, the researchers aimed to extend what is learned from a vaccine trial by using the placebo group vaccination as a second vaccination time point and adapting surveillance data of circulating variants to deduce control rates of infection for this delayed vaccination group. Magaret described this approach further: "The method described in our paper uses data from the vaccine efficacy trial, in concert with lineage assignments and other clinical metadata from publicly available sequence surveillance repositories, to estimate the efficacy of a vaccine to a variant that arose following the vaccination of the trial's placebo group. The key requirements are that the distribution of surveillance strain lineages accurately reflects the distribution of lineages for the trial's placebo group following vaccination, and that at least one lineage is present before and after placebo vaccination."

Vaccine efficacy for ancestral and variant viruses is estimated from period 1 and period 2 by comparing infection ratios for ancestral virus and virus variants in the vaccinated and unvaccinated populations. The unvaccinated population for period 2 is inferred from the ancestral vaccine effect in period 1 combined with community surveillance data of the ratio of infected individuals with either ancestral or variant viruses.
Vaccine efficacy for ancestral and variant viruses is estimated from period 1 and period 2 by comparing infection ratios for ancestral virus and virus variants in the vaccinated and unvaccinated populations. The unvaccinated population for period 2 is inferred from the ancestral vaccine effect in period 1 combined with community surveillance data of the ratio of infected individuals with either ancestral or variant viruses. Figure taken from primary publication

This approach requires accurate community surveillance of emerging virus variants over the timespan of the vaccine trial, which is available for SARS-CoV-2 variants. Another critical aspect of this approach is that a new virus variant must co-circulate with the ancestral virus in period 2. If period 1 is all ancestral virus and period 2 all variant virus, the method breaks down. While this requirement relies on a bit of luck for timing, such overlap has occurred for multiple SARS-CoV-2 vaccine trials. Current methods of estimating vaccine efficacy assume that the efficiency is constant over time because they only evaluate one point in time. A two-period approach enriches our understanding of changes in vaccine efficacy over a short duration of time and may provide essential insights into changes in vaccine efficacy during the current SARS-CoV-2 pandemic.

The application of this modeling approach to SARS-CoV-2 variants is ongoing. Dr. Follmann specified that "These methods are currently being applied to the Novavax trial to assess [vaccine efficacy] (VE) against Delta [SARS-CoV-2] which emerged after the Novavax trial vaccinated placebo recipients." While these approaches do not determine the vaccine efficacy to variants in real-time, studying vaccine efficacies across multiple trials over time may inform on how frequently adaptations to the vaccine targets are needed to maintain high vaccine efficacy for a particular circulating virus type.

The spotlighted research was funded by the National Institute of Allergy and Infectious Disease (NIAID) and National Institutes of Health Department of Health and Human Services (HHS).

Follmann D, Fay M, Magaret C. 2022. Estimation of vaccine efficacy for variants that emerge after the placebo group is vaccinated. Stat Med. 41(16):3076-3089.