Catalog Home Page

Multi-omics analysis of immune-metabolic blood responses of very preterm infants for improving diagnosis of late-onset sepsis

Ng, Sherrianne Qin Yin (2019) Multi-omics analysis of immune-metabolic blood responses of very preterm infants for improving diagnosis of late-onset sepsis. PhD thesis, Murdoch University.

[img]
Preview
PDF - Whole Thesis
Download (18MB) | Preview

Abstract

Very preterm infants are at highest risk of developing late-onset sepsis (LOS) through nosocomial infection, but our understanding of the associated transcriptional and metabolic responses remains very limited. Interrogating the neonatal immunometabolome could provide novel insights into the complex and dynamic pathophysiology and has potential to identify new signatures/biomarkers and improve LOS diagnosis. We hypothesised that neonatal sepsis will trigger characteristic changes in transcriptional regulation of critical innate immune response genes that will relate to their blood metabolome. Additionally, that the pathophysiological changes can be used to identify infected infants using a minimum metabolic and transcriptional signature specific for LOS.

The blood transcriptome and plasma metabolome of very preterm infants with suspected LOS were profiled using RNA sequencing (RNA-seq) (n=18) and non-targeted liquid chromatography-mass spectrometry (LC-MS; n=20). Transcriptional profiling included bioinformatic and statistical analyses consisting of differential gene expression analysis, over-representation analysis, and protein-protein interaction (PPI) network and pathway enrichment analysis. Metabolomic analyses consisted of univariate and multivariate statistical methods including principal component analysis (PCA), partial least squaresdiscriminant analysis (PLS-DA) and orthogonal partial least squares-discriminant analysis (OPLS-DA). The metabolite features potentially important for discriminating infants with confirmed LOS, possible LOS and without LOS were identified using stringent criteria of: p<0.05, fold-change (FC)>3 and OPLS-DA variable importance in projection (VIP)>1. Separately, we used a multi-dimensional approach by integrating the transcriptomic, metabolomic, demographic and laboratory data of matching very preterm infants with suspected LOS using multiple factor analysis (MFA; n=17) to assess patient classification.

Our transcriptomic findings show that blood leukocyte transcriptional responses of infants with confirmed LOS are significantly altered compared to infants with no LOS. The transcriptional aberrations of infants with confirmed LOS were associated with pathogen recognition (mainly toll-like receptor (TLR) pathways), cytokine signalling (both pro-inflammatory and inhibitory responses), cellular functions (related to homeostasis), immune and haematological regulation (including cell death pathways), and metabolism (altered cholesterol biosynthesis). Immunoassays confirmed that interleukin (IL)-1α, IL-6 and IL-10 responses, predicted from the pathway analysis were elevated in sepsis. Separately, metabolomic data revealed that the majority of infants with confirmed LOS had similar profiles based on 36 metabolite features that consisted of biologically relevant putative metabolites including pyruvate and fatty acids. Additionally, we identified 20 metabolite features including urea and fatty acids that could clearly distinguish infants with possible LOS from those without LOS. The integration of transcriptomic (308 distinct genes) and metabolomic (36 metabolite features), demographic and laboratory data resulted in stratification of all infants with confirmed LOS into individual clusters that were separate from infants with possible or no LOS. The sub-groups within infants with confirmed LOS were associated with infecting pathogen type.

This thesis is the first to characterise both the transcriptome and metabolome from very preterm infants with LOS. The blood transcriptome of very preterm infants with confirmed LOS was characterised by skewed host immune responses that are reflective of unbalanced immunometabolic homeostasis. Non-targeted analysis of the plasma metabolome revealed 36 metabolite features that were significantly different in infants with confirmed LOS compared to those with no LOS. The integration of 308 distinct genes and 36 metabolite features associated with confirmed LOS with demographic and laboratory data resulted in superior identification of confirmed LOS and highlights the potential for integration of multi-dimensional data. Further characterisation of transcriptional signatures and/or metabolite biomarkers for the diagnosis of neonatal sepsis is clearly warranted.

Item Type: Thesis (PhD)
Murdoch Affiliation: Medical, Molecular and Forensic Sciences
United Nations SDGs: Goal 3: Good Health and Well-Being
Supervisor(s): Currie, Andrew, Strunk, Tobias, Maker, Garth, Reinke, Stacey and Decuypere, S.
URI: http://researchrepository.murdoch.edu.au/id/eprint/46552
Item Control Page Item Control Page

Downloads

Downloads per month over past year