Identifying viral infections through metagenomic Next Generation Sequencing of undiagnosed respiratory fevers in Madagascar (2014-2019)

Background Respiratory illness contributes to substantial global morbidity and mortality. In Madagascar, an island nation off the southeastern coast of the African continent, hospital-based public health surveillance for respiratory pathogens screens for common respiratory viruses. However, many cases remain undiagnosed. Methods We conducted metagenomic Next Generation Sequencing (mNGS) to identify the pathogen profile of 102 undiagnosed febrile patients who presented to public hospitals with respiratory symptoms and screened negative on a 14-virus multiplex RT-qPCR. We analyzed the diversity of the respiratory microbiome of each patient from mNGS data and identified viral infections potentially linked to undiagnosed fever. We assembled whole genome consensus sequences of viruses with sufficient read depth and coverage, characterized each phylogenetically, and identified any discrepancies with the primers used in the multiplex RT-qPCR panel. Finally, we compared all whole genome sequences against publicly available global databases in a phylogenetic analysis. Results We identified evidence of infection by a wide range of known human viruses in approximately two thirds (64.7%) of study participants from nine different families of viruses and generated 30 complete or nearly complete consensus sequences of known respiratory viruses including orthopneumoviruses, metapneumoviruses, rhinoviruses, coronaviruses, parainfluenza virus, and bocaparvovirus. mNGS-attributed evidence of infection was predominantly due to orthopneumovirus (also called respiratory syncytial virus [RSV]; n = 24) and enterovirus (n = 20) detections, despite previous negative RT-qPCR results for these and other common respiratory viruses for most patients investigated. Finally, phylogenetic analysis identified two distinct phylogenetic clusters of RSV subtype A, suggesting local transmission following distinct international introductions for this virus. Conclusion mNGS provides a sensitive pan-pathogenic tool for virus detection. We demonstrate the diversity of viruses driving respiratory fevers in Madagascar, emphasize the importance and relevance of the existing respiratory surveillance in the country, and highlight the interconnectedness of regional respiratory infection dynamics with global networks of respiratory pathogen transmission.