Use this URL to cite or link to this record in EThOS:
Title: Computational investigation of systemic pathway responses in severe pneumonia among the Gambian children and infants
Author: Jafali, James
ISNI:       0000 0004 7969 1886
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Pneumonia remains the leading cause of infectious mortality in under-five children, and the burden is highest in sub-Saharan Africa. To mitigate this burden, further knowledge is required to accelerate the development of innovative and cost-effective approaches. To gain a deeper insight into the pathogenesis of pneumonia, I investigated the central hypothesis that systemic pathway (cellular and molecular) responses underpin the development of severe pneumonia outcomes. Mainly, I compared whole blood transcriptomes between severe pneumonia cases (clinically stratified as mild, severe and very severe) and non-pneumonia community controls (prospectively matched by age and sex). In total, 803 whole blood RNA samples were collected from Gambian children (aged 2-59 months) between 2007 and 2010, of which, 518 passed laboratory quality control criteria for the microarray analysis. After data cleaning, the final database reduced to 503 samples including the training (n=345) and independent validation (n=158) data sets. To investigate the cellular responses, I applied computational deconvolution analysis to assess the variations of immune cell type proportions with pneumonia severity. To further enhance the computational performance, I applied a data fusion approach on 3,475 immune marker genes from different resources to derive an optimal and integrated blood marker list (IBML, m=277) for Neutrophils, Monocytes, NK, Dendritic, B and T cell types; which robustly performed better than the existing individual resources. Using the IBML resource, pneumonia severity was significantly associated with the depletion of B, T, Dendritic and NK cell types, and the elevation of Monocytes and neutrophil proportions (P-value < 0.001). At the molecular level, pneumonia severity was associated (false discovery rate < 0.05) with a battery of systemic pathway (innate, adaptive and metabolic) responses in a range of biomedical databases. While the up-regulation of inflammatory innate responses was also observed in mild cases, severe pneumonia cases were predominantly associated with the co-inhibition of the cells of the adaptive immune response (B and T) and Natural killer cells, and the up-regulation of fatty acid and lipid metabolism. While most of these findings were anticipated, the involvement of NK cells was unexpected, and potentially presents a novel immune-modulation target for mitigating the burden of pneumonia. Together, the cellular and molecular pathways responses consistently support the central hypothesis that systemic pathway responses contribute significantly to the development of severe pneumonia outcomes. Clinically, the identification and appropriate treatment of patients at the higher risk of developing severe pneumonia outcomes remains the major challenge. To address that, I applied supervised machine-learning approaches on cellular pathway based transcriptomic features; and derived a 33-gene classifier (representing the NK, T, and neutrophils cell types), which accurately detected severe pneumonia cases in both the training (leave-one-out cross-validated accuracy=99%) and independent validation (accuracy=98%) datasets. Independently, similar performance (98% in each dataset) was associated with a subset (m=18) of the validated 52-gene neonatal sepsis classifier. Conversely, at least 75% of the cellular biomarkers were differentially expressed (false discovery rate < 0.05) in bacterial neonatal sepsis. Further, very severe pneumonia cases were predominantly associated with antibacterial responses; and mild pneumonia cases with blood-culture-confirmed positivity were also associated with an increased frequency of differentially expressed genes. These findings suggest the significant contribution of bacterial septicaemia in the development of serious pneumonia outcomes. Together, this study highlights the future potential of host-derived systemic biomarkers for early identification and novel treatment modalities of high-risk cases presenting at a resource-constrained clinic with mild pneumonia. However, further validation studies are required.
Supervisor: Ghazal, Peter ; Forster, Thorsten Sponsor: Medical Research Council (MRC)
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: pneumonia ; transcriptomics ; Gambia ; computational deconvolution analysis ; clinical identification ; high-risk