A cartridge-based assay for improved detection of multidrug-resistant Mycobacterium tuberculosis directly from sputum

Tuberculosis (TB) is a global health threat complicated by increasing INH resistance (INH-R) and rifampicin resistance (RIF-R), including RIF-R due to rpoB Ile491Phe (I491F) mutations, which are not detected by current molecular resistance tests. We developed a low-complexity automated nucleic acid amplification test MDRmDx, designed to operate on the 10-color GeneXpert instrument, which detects Mycobacterium tuberculosis (MTB), and RIF-R in both the rifampicin resistance determining region (RRDR) and the I491F allele, and INH-R due to mutations in katG, fabG1, and the inhA promoter directly from sputum. In an analytic study, the limit of detection (LoD) for detecting MTB in sputum was 21.6 CFU/mL (95% confidence interval [CI]: 14.9-28.3) for the MDRmDx assay, versus 19.1 CFU/mL (95% CI: 14.0-24.1) for the Xpert MTB/RIF Ultra assay (Ultra). The LoD for rifampicin susceptibility (RIF-S) detection was 108.8 CFU/mL (95% CI: 77.4-140.2) versus 100.9 CFU/mL (95% CI: 67.3-134.5) for MDRmDx versus Ultra. The LoD for INH susceptibility detection was 92.2 CFU/mL (95% CI: 65.3-119.1). MDRmDx identified all rpoB mutations with a ≥0.5% global prevalence, including several not detected by Ultra. MDRmDx was 100% specific. Testing frozen sputum enriched for smear-negative, RIF-R, and INH-R samples, MDRmDx sensitivity was 91.3% (95% CI: 85.6-94.8) for MTB detection, 100% (95% CI: 95.1-100) for RIF resistance, and 98.7% (95% CI: 93.0-99.8) for INH resistance, with a specificity of 98.0% (95% CI: 89.3-99.6) for MTB detection, 93.8% (95% CI: 83.2-97.9) for RIF resistance, performance comparable to Ultra. MDRmDx specificity for INH-R was 100% (95% CI: 90.8-100). Importance: The Xpert MTB/RIF Ultra assay (Ultra) is a gold standard test for rapidly diagnosing tuberculosis (TB) as well as resistance to rifampicin directly from sputum samples. However, the Ultra assay does not detect certain rifampicin resistance mutations, which are observed clinically and cannot detect resistance to isoniazid, another key drug against TB. We modified the Ultra assay, creating a prototype MDRmDx assay designed for 10-color instruments that detects isoniazid resistance, which can be a precursor of potential rifampicin resistance and has improved ability to detect rifampicin resistance. In both analytic and retrospectively collected clinical sputum samples, we found that the MDRmDx assay performed at least as well as the Ultra assay with the additional advantage of detecting isoniazid resistance and rpoB I491F mutations with high sensitivity and specificity. The MDRmDx assay's improved performance should improve detection of drug-resistant TB and facilitate the selection of effective treatment.