In the study, the research team engineered viruses to express 78 distinct hemagglutinin proteins, representing the diversity of H3N2 flu viruses circulating in 2023, as well as recent vaccine strains. Each of these 78 variants was tagged with a unique genetic "barcode." By mixing these barcoded viruses with human sera—the antibody-rich component of blood—and using Illumina sequencing, the researchers could measure exactly how well each individual’s antibodies neutralized every one of the 78 variants.
The scale of the data generated was unprecedented. The team analyzed 150 serum samples collected in 2023 from a diverse group of children and adults in the United States. In total, they produced over 11,000 individual titer measurements. This massive dataset created a detailed "snapshot" of the state of human immunity at the threshold of the 2023–2024 flu season.
The results revealed a staggering level of variation in how individuals respond to the flu. The researchers found that immunity is highly personalized; two people of the same age living in the same city might have radically different levels of protection against the same strain of flu based on which specific viruses they caught years earlier. The study found that some children possessed antibodies that could strongly neutralize almost every strain tested. In contrast, other children had much weaker responses across the board. While adults generally exhibited more consistent levels of immunity, they still showed significant individual variation.
One of the most striking observations was that the highest rates of neutralization were found in a subset of children. The researchers suggested two possible reasons for this. First, it aligns with the concept that immune responses are often most robust against strains encountered during the first decades of life—a phenomenon sometimes referred to as "original antigenic sin" or primary exposure. Second, children are generally more susceptible to frequent respiratory infections, meaning they may have had more recent "immunological boosting" from natural infections compared to adults.
To determine how this individual variation actually influences the path of viral evolution, the researchers utilized a statistical model known as multinomial logistic regression. They compared their 11,000 titer measurements with the actual growth rates of different H3N2 strains observed during the 2023 flu season. The goal was to see if the immunity profile of the 150 individuals could predict which viral strains would successfully spread through the wider population.
The analysis revealed a clear and significant correlation: the flu strains that spread most successfully were those that managed to "escape" neutralization in a larger fraction of the individual serum samples. Crucially, the researchers found that a strain’s success was best predicted by the percentage of the population that fell below a specific threshold of immunity. If a high percentage of individuals had weak or no protection against a specific variant, that variant was much more likely to surge in frequency.
