Abstract
HIV infection promotes the progression of latent infection of Mtb to the active disease with the primary challenge of diagnosis being the development of efficient and sensitive methods to detect latent TB in HIV infected individuals. Previous studies have identified transcriptional signatures for active TB along with signatures predicting the risk of active TB disease in latent TB infected individuals or those with other diseases. Existing studies have also identified characteristic genes for active TB in HIV infected patients. However, no studies have identified predictive transcriptional signatures that discriminate latent TB from active TB disease in HIV positive persons as well epigenetic mechanisms associated with latent TB/HIV coinfection. The aim of this study was to develop a computational pipeline using statistical modelling and machine learning (ML) methods to identify a transcriptomic signature associated with latent TB in HIV positive patients and to identify candidate epigenetic modifications for future studies. A novel pipeline, that leverages statistical differential expression analyses (OPLS-DA) and supervised ML and feature selection methods, was applied to an existing transcriptomic dataset (NCBI GEO repository accession number GSE37250) and the outcome of the two methodologies were integrated to define a gene signature characterising the progression of latent to active TB in HIV infected patients. Enrichment analysis was performed on the transcriptomic panel of genes to predict candidate epigenetic marks in latent TB/HIV coinfection. An 11-gene minimal signature was identified of which the expression levels discriminate between latent TB and active TB in HIV positive patients. A broader analysis of DEGs identified by the ML and OPLS-DA revealed enrichment of pathways related to T- and B-cell receptor signalling, metabolic processes, insulin signalling, endocrine resistance and ATP-binding. Candidate epigenetic alterations associated with latent TB in the HIV positive cohort were identified using transcription factor (TF), histone modification (HM) and miRNA enrichment analyses. This novel integrative approach to identify a discriminative latent TB gene signature provided new insights into the response mechanism of HIV co-infection with Mtb, and pathways that merit further investigation was identified. The genes of interest identified may provide novel diagnostic and therapeutic targets for latent TB in patients who are HIV positive.
M.Sc. (Biochemistry)