Abstract
Multiplex immunoassays can facilitate the parallel measurement of antibody responses against multiple antigenically related pathogens, generating a wealth of high-dimensional data that depict complex antibody-antigen relationships. In this study, we developed a generalizable analytical framework to maximize inferences from multipathogen serological studies. We fit the model to measurements of immunoglobulin antibody binding against 10 arboviral pathogens from a cross-sectional study in northwest Bangladesh with 1453 participants. We used our framework to jointly infer the prevalence of each pathogen by location and age as well as between-pathogen antibody cross-reactivity. Reconstructing immunological profiles, we found evidence of endemic transmission of Japanese encephalitis virus and recent outbreaks of dengue and chikungunya viruses in this district. Our estimates of antibody cross-reactivity were highly correlated with phylogenetic distances inferred from genetic data [correlation coefficient (r) = 0.94], demonstrating how antigenic landscapes can be inferred from population-level serological studies. Furthermore, we showed how our framework could be used to identify the presence of antigenically related pathogens that were not directly tested for, representing a potential opportunity for the detection of emerging pathogens. The presented analytical framework offers a tool that can be applied to a growing number of multipathogen studies and will help support the integration of serological testing into disease surveillance platforms.