Discovery and validation of host transcriptomic and proteomic biomarkers for paediatric tuberculosis
File(s)
Author(s)
Vito, Ortensia
Type
Thesis
Abstract
Although being a curable disease, Tuberculosis (TB) is still one of the leading causes of mortality and morbidity worldwide. Most of the global cases reside in Low- and Middle-Income Countries, where the resources are limited and a high number of patients go undiagnosed or untreated. The existing diagnostic tests for TB are mainly sputum-based, with lengthy turnaround time, low sensitivity, and high cost. Good-quality sputum samples are difficult to collect in children, where invasive procedures, such as induced sputum or gastric aspiration, are needed. Moreover, sputum-based diagnostic tests are suboptimal in patients with Extra-Pulmonary Tuberculosis (EPTB).
The presence of pathogens (i.e., bacteria or viruses) causes changes in gene expression and/or protein abundance detectable in the host’s blood. Sparse combinations of biological molecules, which are known as host diagnostic signatures, can be used to distinguish between disease groups of interest and developed into a Point-of-care Test (POCT) for rapid and accurate disease diagnosis. A host signature for TB, especially childhood TB, might be the way towards a low-cost, easy-to-use, fast and blood-based POCT.
In this thesis, I have used host transcriptomic and proteomic profiles to identify sparse diagnostic signatures for paediatric TB. Firstly, I have described a 6-gene signature that can accurately discriminate between children with microbiologically confirmed TB and unlikely TB, with or without Human Immunodeficiency Virus (HIV) infection, with an Area Under the ROC Curve (AUC) of 89.3% [CI95% 82.5%-96.1%], sensitivity of 88.5% [CI95% 78.9%-96.2%] and specificity of 86.0% [CI95% 75.4%-94.7%] in an independent validation set. Secondly, I have described a 4-protein signature that achieved an AUC of 88.7% [CI95% 85.0%-92.4%], sensitivity of 78.5% [CI95% 71.1%-85.6%] and specificity of 86.1% [CI95% 81.3%-90.4%] in an independent validation set. Lastly, I have described a novel multi-omics approach where I have integrated transcriptomic and proteomic datasets to identify a multi-omics signature...
The presence of pathogens (i.e., bacteria or viruses) causes changes in gene expression and/or protein abundance detectable in the host’s blood. Sparse combinations of biological molecules, which are known as host diagnostic signatures, can be used to distinguish between disease groups of interest and developed into a Point-of-care Test (POCT) for rapid and accurate disease diagnosis. A host signature for TB, especially childhood TB, might be the way towards a low-cost, easy-to-use, fast and blood-based POCT.
In this thesis, I have used host transcriptomic and proteomic profiles to identify sparse diagnostic signatures for paediatric TB. Firstly, I have described a 6-gene signature that can accurately discriminate between children with microbiologically confirmed TB and unlikely TB, with or without Human Immunodeficiency Virus (HIV) infection, with an Area Under the ROC Curve (AUC) of 89.3% [CI95% 82.5%-96.1%], sensitivity of 88.5% [CI95% 78.9%-96.2%] and specificity of 86.0% [CI95% 75.4%-94.7%] in an independent validation set. Secondly, I have described a 4-protein signature that achieved an AUC of 88.7% [CI95% 85.0%-92.4%], sensitivity of 78.5% [CI95% 71.1%-85.6%] and specificity of 86.1% [CI95% 81.3%-90.4%] in an independent validation set. Lastly, I have described a novel multi-omics approach where I have integrated transcriptomic and proteomic datasets to identify a multi-omics signature...
Version
Open Access
Date Issued
2023-04-03
Date Awarded
01/10/2023
License URL
Advisor
Kaforou, Myrsini
Levin, Michael
Publisher Department
Department of Medicine
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)