Abstract
The MTBDRplus line probe assay (LPA) and Xpert MTB/RIF have been endorsed by the World Health Organization for the rapid diagnosis of drug-resistant tuberculosis. However, there is no clarity regarding the superiority of one over the other. In a double-blinded prospective study, we evaluated the efficacy of the Xpert MTB/RIF on samples that were first tested by LPA under the revised national tuberculosis control program of India. A total of 405 sputum samples from suspected drug-resistant tuberculosis patients were included. Of these, 285 smear-positive samples were subjected to LPA. Seventy-two (25.8%) samples showed multidrug resistance, 62 (22.2%) showed rifampin monoresistance, 29 (10.3%) showed isoniazid monoresistance, and 116 (41.5%) were pan-susceptible. Six (2.1%) of the samples gave invalid results. Of the 62 rifampin-monoresistant samples by LPA, 38 (61.4%) showed rifampin resistance, while 21 (33.8%) were found susceptible to rifampin by Xpert MTB/RIF using cartridge version G4. Three (4.8%) samples gave an error. Of the 116 pan-susceptible samples, only 83 were available for Xpert MTB/RIF testing; 4 (5.1%) were rifampin resistant, 74 (94.8%) were susceptible, and 5 (6.0%) showed an error. The 25 discrepant samples were further subjected to MGIT960 drug susceptibility testing. The MGIT960 results showed 100% agreement with LPA results but only 64.4% agreement with Xpert MTB/RIF results. Sequencing analysis of discrepant samples showed 91.3% concordance with LPA but only 8.7% concordance with the Xpert MTB/RIF assay. These findings indicate that by using Xpert MTB/RIF testing we might be underestimating the burden of drug-resistant tuberculosis and indicate that country-specific probes need to be designed to increase the sensitivity of the Xpert MTB/RIF.