Evaluation of the Roche cobas MTB and MTB-RIF/INH for detecting Mycobacterium tuberculosis complex and resistance to isoniazid and rifampicin: A prospective, multicenter diagnostic accuracy study

Objectives To evaluate the diagnostic performance of Roche cobas MTB and MTB-RIF/INH assays for detecting Mycobacterium tuberculosis complex (MTBC) and resistance to isoniazid (INH) and rifampicin (RIF). Methods This multicentre study was conducted in Taiwan between September 2023 and June 2024. Clinical specimens were collected from adult patients with presumptive tuberculosis (TB). All samples underwent acid-fast staining, culture, Xpert MTB/RIF Ultra assay and cobas MTB assays. The MTB-RIF/INH assay was conducted as a reflex test for specimens that tested positive with the cobas MTB assay. Thirty-seven MTBC control strains from the Taiwan Centers for Disease Control (TCDC) were also analysed to assess drug resistance detection. Whole genome sequencing was used to resolve discrepancies with phenotypic drug susceptibility testing (pDST). Results A total of 425 clinical samples from 378 adult patients were analysed. Among 392 respiratory samples, 53 were culture-positive for MTBC. cobas MTB showed 98.1% sensitivity (95% confidence interval [CI], 89.9%-100%) and 95.9% specificity (95% CI, 93.2%-97.7%) for MTBC detection. Drug resistance detection was evaluated using 37 TCDC stains. For RIF resistance, the cobas MTB-RIF/INH assay correctly identified all 11 phenotypically resistant strains. Among the 26 phenotypically susceptible strains, 25 were correctly identified, with 1 false-positive result (overall accuracy: 97.3%). For INH resistance (0.2 µg/mL), the assay identified 16 resistant and 19 susceptible strains, with 2 false negatives (accuracy: 94.6%). Conclusions Roche cobas MTB and MTB-RIF/INH assays demonstrated high accuracy for detecting MTBC and drug resistance, supporting their use as reliable diagnostic tools in clinical practice.