Improving the diagnosis of encephalitis through analysis of the host transcriptome, proteome and metabolome.

Background Encephalitis (inflammation of the brain) is a devastating neurological condition. It is most often caused by either acute viral infection or an autoimmune process. These can be difficult to distinguish clinically and require very different treatments. Encephalitis can also be mimicked by other neurological or systemic disease states. Improved biomarkers to distinguish encephalitis from its mimics, and autoimmune from viral encephalitis, could improve diagnosis and expedite effective treatment. Previous studies have identified signatures based on small numbers of proteins but none have investigated proteome, transcriptome or metabolome approaches to diagnosis using unbiased methodology. Aims 1. To establish whether multiomic methods can identify biomarkers to distinguish encephalitis from mimicking conditions, and autoimmune from viral encephalitis. 2. To investigate whether host responses in cerebrospinal fluid (CSF) and blood correlate with clinical outcome. 3. To explore how the pathogenesis of autoimmune and viral encephalitis differ from other mimicking conditions by pathway and network analysis. Methods 1. Using microarray, I investigated gene expression patterns in blood of patients with encephalitis and mimicking conditions. 2. Using liquid chromatography/mass spectrometry, I explored the CSF proteome of patients with encephalitis and mimics, validated by two separate cohorts. 3. Using 1H nuclear magnetic resonance (NMR) spectroscopy, I analysed metabolic profiles in CSF in patients with encephalitis and mimics. Results All three techniques revealed potential biomarkers to distinguish aetiological groups. Protein candidate biomarkers were the most effective and included adenosine deaminase 2 (ADA2) which distinguished encephalitis from mimics with a high degree of accuracy, including in patients without CSF pleocytosis. Encephalitis could also be distinguished from mimics by gene expression or metabolite profiles. Distinguishing viral from autoimmune encephalitis was more challenging, but could be achieved by combining small panels of genes or metabolites. Several markers correlated with clinical outcome in autoimmune and viral encephalitis, including proteins involved in leucocyte adhesion and cytokine signalling. Pathway analysis identified mRNA transcripts and protein networks which were enriched in encephalitis when compared with mimicking conditions, especially concerned with the innate and adaptive immune response and the complement and coagulation cascades. Conclusions Through analysis of transcriptome, proteome and metabolome, it was possible to identify candidate biomarkers to improve the aetiological diagnosis of encephalitis, to predict clinical outcome and to elucidate disease mechanisms. These findings now need to be validated through prospective clinical studies and should direct future mechanistic studies and the development of potential therapies