Uncovering dengue serotype-specific transmission, cross-reactivity, and immune profiles from cross-sectional serosurveys

Dengue remains a growing global health concern, with four co-circulating serotypes (DENV-1 to DENV-4). Cross-sectional serosurveys are essential for inferring past transmission and population immunity, but serotype-specific interpretation is constrained by antibody cross-reactivity across serotypes. Here, we developed a novel modelling framework to jointly infer serotype-specific force of infection (FOI) and cross-reactivity from cross-sectional serosurveys. The model is flexible to data granularity, functioning with binary serostatus alone or combined with quantitative titres. In simulations, the model effectively recovered FOI and cross-reactivity across varying endemicity, sampling designs, and sample age coverage. Applied to annual cross-sectional serosurveys in Vietnam (2013-2017), the model estimated serotype-specific FOI patterns comparable in overall magnitude to analysis requiring supplementary longitudinal post-infection antibody measurements, which are difficult to collect. We further identified higher DENV-3 transmission intensity than previously estimated. Our estimates revealed asymmetric cross-reactivity between infecting and heterologous antibody-response serotypes after primary infections, and high cross-reactivity after post-primary infections. We reconstructed population immune profiles by age, time, serotype, and past infection number, resolving single-serotype and multi-serotype exposure histories not directly distinguishable from observed seroprevalence. Our estimates showed that susceptibility to secondary infection in Vietnam peaked at approximately age 10 for each serotype, and averaged ∼25% for DENV-1 to DENV-3 and ∼30% for DENV-4 among individuals aged 1-30 years. Additionally, incorporating titres enabled individual-level infection-history inference and revealed exposure-history signals in titre profiles despite extensive cross-reactivity. These findings show that improved modelling can substantially expand the information recoverable from cross-sectional serology, strengthening the public-health utility of serosurveys for risk assessment, burden estimation, and vaccination planning.