Development and evaluation of the phenotypic 2G test to detect drug-resistant TB

TL;DR

A new test for diagnosing TB and drug resistance was developed and tested in Mozambique, showing high accuracy in the lab but lower sensitivity in real-world settings.

Contribution

A novel colorimetric, thin-layer agar-based test (2G test) was developed and evaluated for TB diagnosis and drug susceptibility testing.

Findings

The 2G test showed 100% accuracy in detecting drug-resistant Mycobacterium tuberculosis strains in the lab.

In clinical settings, the 2G test had 45.9% sensitivity compared to Xpert and 45.2% compared to MGIT culture using fresh sputa.

Sensitivity dropped to 23.1% when using frozen decontaminated sputum samples.

Abstract

Early diagnosis of TB with drug susceptibility testing (DST) is critical to achieve successful treatment outcomes. We aimed to develop and test a novel colorimetric, 12-well, thin-layer agar-based test to assess its accuracy for TB diagnosis and DST in a clinical setting in Southern Mozambique.

Development of the first prototype of the second generation (2G) test in the laboratory setting followed by a cross-sectional diagnostic accuracy study with consecutive recruitment of subjects with microbiologically confirmed TB using GeneXpert MTB/RIF Ultra.

In the laboratory setting, the 2G test showed 100% accuracy in detecting resistance of genotypically characterised drug-resistant Mycobacterium tuberculosis strains. In the clinical setting, the sensitivity of the 2G test to detect M.tb complex versus Xpert and Mycobacteria Growth Indicator Tube (MGIT) culture using fresh sputa was 45.9% and 45.2%, respectively. The 2G test sensitivity versus MGIT decreased to 23.1% when using frozen decontaminated sputum samples.

In the clinical setting, the 2G test showed a low sensitivity versus Xpert and MGIT. The 2G test sensitivity was lower when frozen instead of fresh sputa was used. Despite these results, important information was collected to further improve this 2G test prototype and its implementation in resource-constrained settings.

Figures