Editors' ChoiceInfectious Disease

A high five for TB screening

See allHide authors and affiliations

Science Translational Medicine  05 Dec 2018:
Vol. 10, Issue 470, eaaw0519
DOI: 10.1126/scitranslmed.aaw0519

Abstract

Whole-blood transcriptional profiling identifies a five-gene signature for tuberculosis screening among people living with HIV.

Mycobacterium tuberculosis (TB) causes more deaths worldwide each year than any other single pathogen. Systematic clinical screening of high-risk populations followed by targeted diagnostic testing is a strategic approach for combating the TB pandemic, but it is limited by a lack of screening tests with adequate performance characteristics. Host transcriptional profiling has emerged as an attractive alternative to traditional pathogen–based TB diagnostics but has not previously been evaluated in the context of TB screening.

Rajan and colleagues addressed the unmet need for a better TB screening test by studying whole-blood host gene expression profiles from a cohort of HIV-positive Ugandan adults initiating antiretroviral therapy with or without active TB. They began by identifying nested sets of one to five transcript signatures that could distinguish TB-positive patients from matched controls with a sensitivity and specificity greater than the target thresholds of ≥ 90% and ≥ 70%, respectively, proposed by the World Health Organization. Out of 417 candidate signatures, a single 5-gene signature met these criteria with an area under the curve of 0.87 when tested in an independent validation cohort, performing better than any previously identified signature in the context of TB screening. One of the five genes, GPB6, which has a still undefined role in immune response, was also identified in two previously reported TB signatures. Surprisingly, when comparatively assessed, the C-reactive protein (CRP), a century-old biomarker of host inflammation, also performed better any previously identified TB gene signatures.

Although the idea of using host transcriptional profiling for TB diagnosis has been around for over a decade, Rajan and colleagues extend the earlier work to specifically address the need for better tests in a TB screening setting. Importantly, with only five targets, their signature has the potential to be incorporated into a rapid PCR assay for point-of-care diagnosis. Next steps that could build on these findings include determining whether assay performance can be improved by incorporating clinical metrics, such as fever, night sweats, weight loss, chest x-ray findings, or CRP. Additional work integrating metagenomic sequencing of both host and microbial transcripts could be an intriguing complementary approach for minimally invasive confirmatory testing, and it might also have utility in identifying gene signatures that predict disease severity or treatment outcome.

Highlighted Article

View Abstract

Navigate This Article