Modeling the interferon response to avian influenza

Highly pathogenic avian influenza viruses, such as H5N1, cause severe disease in humans and pose a persistent pandemic threat, yet the mechanism by which the type I interferon response shapes their within-host dynamics remains unclear. Here we fit three ordinary differential equation models, in which interferon reduces the viral infection rate, reduces viral production, or drives cells into a refractory state, to paired viral load and IFN- \\(\\alpha \\) 2 time courses from human nasal and tracheal respiratory epithelium infected with a seasonal H3N2 strain and with 2005 and 2022 clades of H5N1. Although the refractory-state model attains the lowest residual error in most conditions, the small-sample-corrected Akaike information criterion does not justify its additional parameter, and the two alternative mechanisms, reduction of infection rate or reduction of production rate, describe the data comparably well. While a structural identifiability analysis suggests all parameters are globally identifiable when both viral titer and interferon are observed, with only ten observations per condition the data cannot single out one mechanism of interferon action. The bootstrapped parameter distributions are largely similar across strains when compared with a Mann-Whitney U-test, differing mainly in the rate of interferon clearance, with only modest differences between the nasal and tracheal epithelium. These results quantify the interferon response to avian influenza in human respiratory tissue and underscore the need for more frequently sampled data to resolve the dominant mechanism of interferon action in vivo.