Cobas tests accurately detect TB and drug resistance in South Africa

Tuberculosis (TB) remains a leading global health problem where fast, accurate diagnosis and identification of drug resistance are vital. In high-burden settings, delays or missed detection of Mycobacterium tuberculosis complex (MTBC) and resistance to key drugs can lead to worse outcomes and ongoing transmission. To address this, a team led by corresponding author Anura David evaluated two moderate-complexity nucleic acid amplification tests, the Cobas® MTB assay and the Cobas MTB/RIF-INH assay, as tools to detect MTBC and resistance to rifampicin (RIF) and isoniazid (INH). The study took place in South Africa and enrolled 354 symptomatic adults, a group with 56% HIV prevalence. Researchers compared results from the Cobas tests on sputum with a liquid culture reference standard and used Xpert MTB/RIF Ultra (Xpert Ultra) as a comparator test. They also explored whether tongue swabs (TS) could be a usable alternative specimen for the Cobas® MTB assay, testing different sample pre-processing conditions. The aim was practical: to see whether these assays could reliably identify TB and drug resistance where accurate and scalable testing is needed most.

The Cobas® MTB and Cobas MTB/RIF-INH assays are molecular tests that detect MTBC and, in the case of the MTB/RIF-INH assay, profile resistance to RIF and INH. Using liquid culture as the reference standard, the Cobas MTB assay on sputum showed an overall sensitivity of 93.8% (95% CI: 84.8-98.3) and specificity of 100% (95% CI: 98.7-100). Agreement between Cobas MTB and Xpert Ultra was very high (Cohen’s kappa = 0.904). Among HIV-positive participants the reported sensitivity was 88.2% (95% CI: 97.8-100). RIF resistance calls made by the Cobas MTB/RIF-INH assay were fully concordant with culture and Xpert Ultra results, while three INH-resistant cases were not detected by the Cobas assay—an issue the authors attribute to likely genotypic-phenotypic discordance. The study also tested the Cobas® MTB assay on tongue swabs (TS); although the number of TS specimens was small, diagnostic accuracy improved when a diluted (66%) Microbial Inactivation Solution was used during processing.

These results suggest the Cobas® MTB and Cobas MTB/RIF-INH assays could serve as reliable alternatives for TB detection and resistance profiling in clinical settings that demand rapid and accurate results. The assays’ strong performance on sputum and near-perfect agreement with Xpert Ultra position them as useful options where liquid culture is not practical because culture takes longer and requires specialized infrastructure. A particular advantage of the Cobas MTB/RIF-INH assay is its ability to detect INH resistance in addition to RIF, and the authors note scalability through high-throughput platforms as a practical benefit for busy laboratories. The preliminary tongue swab findings point to the possibility of expanding testing to an easier specimen type, provided pre-processing is optimized (for example, using a 66% Microbial Inactivation Solution). At the same time, missed INH-resistant cases underscore limits of molecular tests when genotypic results do not match phenotypic resistance, and they highlight the continued need for complementary methods and careful interpretation. Overall, the study supports incorporating Cobas assays into diversified TB testing strategies to improve detection and management in high-burden settings.