Understanding the factors that influence emergence of novel, zoonotic pathogens is a key challenge in disease ecology. Bottom-up control is a factor that can regulate emergence if existing population immunity to related pathogens interferes with invasion by the novel pathogen. Distinct influenza subtypes are not typically thought to cross-react, so population-level immunity is not thought to exist for influenza subtypes novel to humans. However, new evidence shows that cross-immunity between emerging and established subtypes is possible. We investigated influenza A subtypes H7N9 and H5N1, two avian influenza viruses of great concern for pandemic emergence. H5N1 and H7N9 repeatedly spill over into humans, where cases show an unusual and unexplained age distribution. These unusual age distributions could be signatures of existing, age-structured immunity in the human population. We compiled and analyzed data from all known human cases of H5N1 and H7N9 to test whether existing, age-structured influenza immunity explains observed case age distributions. We developed a set of multinomial models and used model selection to test our immune history hypothesis alongside all previously proposed explanations. We used maximum likelihood estimation to quantify the effect of each tested factor on subtype-specific infection risk.
We found strong evidence that an individual’s first influenza A infection leads to lifelong immune imprinting against the hemagglutinin antigen, whereby an individual’s primary infection confers lifelong, partial protection against subsequent exposures to novel hemagglutinin subtypes (e.g. H7 or H5) in the same phylogenetic group as the primary exposure. Model selection showed that the hemagglutinin imprinting effect was the most important predictor of case age distribution for both H5N1 and H7N9. Individuals with protective imprinting were estimated to have a 75% reduced risk of severe infection. This implies that susceptibility to specific emerging influenza viruses is predictable based on a person’s birth year. Our results challenge the current paradigm, in which novel influenza viruses invade on an immunological blank slate, and pandemic emergence is facilitated by an absence of bottom-up control. More generally, our results suggest it should be possible to forecast the changing invasion potential of novel influenza viruses and other zoonotic pathogens by understanding how population immunity to related human pathogens exerts bottom-up control on emergence.