Transcriptomic analysis of biomarkers of respiratory syncytial virus disease

RSV is one of the most prevalent causes of respiratory disease in children and the elderly worldwide. Despite this there is still a poor understanding of the role of host factors in severity of RSV disease. There is a significant unmet need to fully understand the relationship between host responses to viral pathogens and disease severity, both to enable more efficient stratification of patients and to assist in the development of novel treatments. This thesis had three primary aims. 1: To characterise the host transcriptomic response against RSV. 2: To link the host transcriptomic response to RSV disease severity in infants. 3: To develop a biomarker capable of robustly identifying patients with more severe disease. Microarrays were used to determine gene expression of RSV infected infants at the time of infection and in convalescence seven weeks later. Gene expression between these states were compared, associations between gene expression and severity were described using gene set enrichment analysis, and a panel of machine learning algorithms employed to identify the gene signature with the highest power to characterize severe RSV. Key findings of this thesis include a description of the immune response against RSV, and the refinement of a novel gene signature that can correctly classify individuals into disease severity groups. The immune response to RSV was characterised as being primarily defined by strong neutrophil and monocyte associated gene expression, alongside a downregulation of lymphocytes and a strong upregulation of interferon related genes. Genes with expression patterns that correlate most strongly with RSV also relate to neutrophils and monocytes, with higher expression of interferon, lymphocyte, B and plasma cell genes relating most strongly to milder disease. A novel feature selection technique was developed to identify a five gene signature (made up of ORF7, MMP8, IL1R2, ANXA3 and BCL6), that when used in conjunction with the age of a participant at the time of infection could correctly categorize severity of participants in this study at an AUC of .88, and in a completely independent validation cohort at an AUC of .6. The next challenge of this work is to be the refinement and validation of this biomarker, to assess the value that it can bring to physicians in decision making in the clinic