Factors affecting diagnostic and prognostic performance of a transcriptomic signature of risk of tuberculosis in HIV-uninfected South African adults

Mulenga, Humphrey

Factors affecting diagnostic and prognostic performance of a transcriptomic signature of risk of tuberculosis in HIV-uninfected South African adults

Doctoral Thesis

2022

Abstract

Background Host blood transcriptomic signatures, such as RISK11, have potential as tests for diagnosing and predicting tuberculosis. This thesis aimed to review the literature, evaluate host and non-host factors associated with variability of the RISK11 signature and impact on discriminatory performance and evaluate RISK11 performance in combination with tests of Mycobacterium tuberculosis sensitization. Methods A systematic review of discriminatory performance of transcriptomic signatures for tuberculosis was conducted. RISK11, QuantiFERON-TB Gold-Plus and host factors were analysed in a prospective cohort, in which a cross-sectional study of upper respiratory organisms was nested. Effects on RISK11 were quantified using multivariable generalised regression. Discriminatory performance of RISK11, and RISK11/QuantiFERON combinations, were quantified by area under the curve and/or sensitivity and specificity. Results In the literature, one signature (90% sensitivity; 74% specificity) met the minimal criteria for a triage test; one signature (86% sensitivity; 84% specificity) met the minimal criteria for a predictive test. In the prospective cohort, RISK11 scores were higher among individuals with prevalent tuberculosis (+18.90%), night sweats (+14.65%) and incident tuberculosis (+7.29%). Cough was associated with 72.55% higher RISK11 score in prevalent tuberculosis cases. Stratification by cough improved diagnostic performance from area under curve of 0.74 overall, to 0.97 in cough-positive participants. Adjustment for host factors affecting controls did not change RISK11 discriminatory performance. In the cross-sectional study, RISK11 scores were higher by +16.7%, +67.8% and +13.5% in participants with coronavirus, influenza and rhinovirus, respectively, such that RISK11 could not differentiate prevalent tuberculosis from upper respiratory viruses. Compared to RISK11, the Either-Positive test combination decreased diagnostic negative likelihood ratio from 0.7 to 0.3, and prognostic negative likelihood ratio from 0.9 to 0.3, but did not improve upon QuantiFERON alone. Compared to QuantiFERON, the Both-Positive test combination increased diagnostic positive likelihood ratio from 1.3 to 4.7, and prognostic positive likelihood ratio from 1.4 to 2.8, but did not improve upon RISK11 alone. Conclusion RISK11 holds promise as a triage test for tuberculosis. Further optimisation, or development of new signatures is needed to improve discrimination of subclinical tuberculosis, without cough, and to mitigate the impact of viral co-infection. RISK11/QuantiFERON combination testing is not recommended.

Keywords

Pathology

Reference:

Collections

PhD / Doctoral

Full item page