Tissue Tropism and Transmission Ecology Predict Virulence of Human RNA Viruses

Liam Brierley; Amy B. Pedersen; Mark E. J. Woolhouse

The value of predictive modelling as an inexpensive and rapid tool for risk assessments The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/581512 doi: bioRxiv preprint 14 for one virus rated 'nonsevere' from literature protocols, Usutu virus, potentially suggesting 2 6 5 the capability for more severe disease to be recognised in future. However, our models have restricted function in predicting the virulence of a newly identified 2 6 8 virus. Although taxonomy is easily accessible and applicable to give simple virulence 2 6 9 estimates, the most informative non-taxonomic predictor, tissue tropism, is not likely to be More widely, our analysis brings a novel focus that complements comparative models 2 8 2 predicting other aspects of the emergence process, such as zoonotic transmission certain virus families (e.g. Hantaviridae) or tissues (e.g. neural tissue) could contribute to a 2 9 0 viable strategy to detect future virulent zoonoses. This work adds to the comparative and predictive modelling efforts surrounding emerging 2 9 4 infectious diseases. Here, we contribute a novel focus in ecological predictors of virulence of 2 9 5 human RNA viruses, which can be combined in holistic frameworks with other models such 2 9 6 as those predicting emergence dynamics. As a predictive model, the featured random forest infections. We propose that future predictive studies and preparedness initiatives with respect to emerging diseases should carefully consider potential for human virulence. and 4 were carried out only when fatality or tropism data respectively were not already found virus species. Therefore, virulence was rated using a simple two-category measure of severity viruses as 'severe' if they had frequent reports of hospitalisation, were associated with 3 2 0 significant morbidity from certain conditions (haemorrhagic fever, seizures/coma, cirrhosis, The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/581512 doi: bioRxiv preprint described as "severe" or "causing severe disease" (S1 Table, S2 Table) . We rated viruses as 3 2 3 'nonsevere' if none of these conditions were met. Note that this led to 'nonsevere' ratings for 3 2 4 some viruses with clinically severe, but rare syndromes, e.g. Dengue virus can cause 3 2 5 haemorrhagic dengue fever, though this is much rarer than typical acute dengue fever vulnerable individuals (defined as age 16 and below or 60 and above, immunosuppressed, 3 2 8 having co-morbidities, or otherwise cited as being 'at-risk' by sources for specific viruses) and in healthy adults, and whether any 'nonsevere' virus has atypically severe strains (for disease; however, poliovirus, which causes severe paralytic disease, is also classified under 3 3 2 this species). These were examined both individually and within a composite six-rank system 3 3 3 (S5 Table) . The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/581512 doi: bioRxiv preprint vascular, or 'systemic' (primary tropism within multiple organ systems). We accepted isolation accept generalised symptoms such as inflammation. However, many human viruses were 3 4 8 isolated from blood with no further evidence of any specific tissue tropisms (n = 69).