BACKGROUND:Latin America has been experiencing an Oropouche virus (OROV) outbreak of unprecedented magnitude and spread since 2023-24 for unknown reasons. We aimed to identify risk predictors of and areas at risk for OROV transmission. METHODS:In this multidisciplinary, laboratory-based, modelling study, we retrospectively tested anonymised serum samples collected between 2001 and 2022 for studies on virus epidemiology and medical diagnostics in Bolivia, Brazil, Colombia, Costa Rica, Ecuador, and Peru with nucleoprotein-based commercial ELISAs for OROV-specific IgG and IgM antibodies. Serum samples positive for IgG from different ecological regions and sampling years were tested against Guaroa virus and two OROV glycoprotein reassortants (Iquitos virus and Madre de Dios virus) via plaque reduction neutralisation testing (PRNT) to validate IgG ELISA specificity and support antigenic cartography. Three OROV strains were included in the neutralisation testing, a Cuban OROV isolate from the 2023-24 outbreak, a contemporary Peruvian OROV isolate taken from a patient in 2020, and a historical OROV isolate from Brazil. We analysed the serological data alongside age, sex, cohort, and geographical residence data for the serum samples; reported OROV incidence data; and vector occurrence data to explore OROV transmission in ecologically different regions of Latin America. We used the MaxEnt machine learning methodology to spatially analyse and predict OROV infection risk across Latin America, fitting one model with presence-absence serological data (seropositive results were recorded as presence and seronegative results were recorded as absence) and one model with presence-only, reported incidence data from 2024. We computed marginal dependency plots, variable contribution, and permutation metrics to analyse the impact of socioecological predictors and fitted a generalised linear mixed-effects model with logit link and binary error structure to analyse the potential effects of age, sex, or cohort type bias and interactions between age or sex and cohort type in our serological data. We conducted antigenic cartography and evolutionary characterisations of all available genomic sequences for all three OROV genome segments from the National Center for Biotechnology Information, including branch-specific selection pressure analysis and the construction of OROV phylogenetic trees. FINDINGS:In total, 9420 serum samples were included in this study, representing 76 provinces in the six Latin American countries previously mentioned. The sex distribution across the combined cohorts was 48% female (4237 of 8910 samples with available data) and 52% male (4673 of 8910 samples) and the mean age was 29·5 years (range 0-95 years). The samples were collected from census-based cohorts, cohorts of healthy individuals, and cohorts of febrile patients receiving routine health care. The average OROV IgG antibody detection rate was 6·3% (95% CI 5·8-6·8), with substantial regional heterogeneity. The presence-absence, serology-based model predicted high-risk areas for OROV transmission in the Amazon River basin, around the coastal and southern areas of Brazil, and in parts of central America and the Caribbean islands, consistent with case data from the 2023-24 outbreak reported by the Pan American Health Organization. Areas with a high predicted risk of OROV transmission with the serology-based model showed a statistically significant positive correlation with state-level incidence rates per 100 000 people in 2024 (generalised linear model, p=0·0003). The area under the curve estimates were 0·79 (95% CI 0·78-0·80) for the serology-based model and 0·66 (95% CI 0·65-0·66) for the presence-only incidence-based model. Longitudinal diagnostic testing of serum samples from cohorts of febrile patients suggested constant circulation of OROV in endemic regions at varying intensity. Climate variables accounted for more than 60% of variable contribution in both the serology-based and incidence-based models. Antigenic cartography, evolutionary analyses, and in-vitro growth comparisons showed clear differentiation between OROV and its glycoprotein reassortants, but not between the three different OROV strains. PRNT titres of OROV-neutralising serum samples were strongly correlated between all three tested OROV isolates (r>0·83; p<0·0001) but were not correlated with the two glycoprotein reassortants. INTERPRETATION:Our data suggest that climatic factors are major drivers of OROV spread and were potentially exacerbated during 2024 by extreme weather events. OROV glycoprotein reassortants, but not individual OROV strains, probably have distinct antigenicity. Preparedness for OROV outbreaks requires enhanced diagnostics, surveillance, and vector control in current and future endemic areas, which could all be informed by the risk predictions presented in this Article. FUNDING:European Union. TRANSLATIONS:For the Spanish and Portuguese translations of the abstract see Supplementary Materials section.
更多
原文链接
分享
引用
