Performance of host blood transcriptomic signatures for TB diagnosis and monitoring TB treatment

Muwanga, Vanessa

Performance of host blood transcriptomic signatures for TB diagnosis and monitoring TB treatment

Thesis / Dissertation

2025

Publisher

Universiy of Cape Town

Department

Department of Pathology

Faculty

Faculty of Health Sciences

Abstract

Non-sputum-based diagnostic tests are necessary to enable tuberculosis (TB) diagnosis and monitoring of TB treatment. This work aimed to identify which published blood transcriptomic TB signatures have the best diagnostic performance to distinguish between TB cases and other respiratory diseases (ORDs) in symptomatic adults. These signatures were also evaluated to monitor treatment in young children, to identify signatures that might indicate when treatment could be stopped. Diagnostic performance of signatures was evaluated in adults presenting for care with symptoms associated with TB, who were recruited from primary healthcare clinics in six African countries. To identify signatures that can monitor treatment response, the same set of signatures was evaluated in young children randomised to 4- or 6-month TB treatment in the SHINE trial. Twenty transcriptomic signatures were selected and measured in whole blood using multiplex, microfluidic RT-qPCR. In the adult diagnostic cohort, nine signatures achieved equivalent performance for differentiating patients with ORDs from all TB cases. Factors associated with signature scores included HIV and country. With sensitivity benchmarked at 90%, these nine signatures achieved specificities between 44%-54%. In pooled analyses, none of the signatures met the minimal target product profile criteria for a TB triage test. Country-specific analyses, however, showed that several signatures met the minimal criteria in some countries. In children from the SHINE trial, baseline scores for all signatures were highest in children with confirmed, relative to unconfirmed and unlikely TB. Baseline scores were also higher among children with disease classified as more severe on chest x-ray. Scores declined progressively during TB treatment in the confirmed and unconfirmed TB groups; no changes in scores were observed for most signatures in the unlikely TB group. Scores were higher at the end of treatment in the 4-month than the 6-month treatment arm in the confirmed TB group; no differences were observed between treatment arms in the unconfirmed and unlikely groups. These results suggest that host blood transcriptomic signatures have potential to monitor TB treatment response in children. This work supports further development of transcriptomic signatures as TB triage tests and treatment monitoring tools.

Keywords

TB diagnosis

TB treatment

Reference:

Collections

PhD / Doctoral

Full item page