Background Accurate diagnosis of tuberculosis (TB) and drug-resistant (DR-TB) is crucial for effective treatment, transmission control, and TB management. We evaluated the effectiveness of Xpert MTB/RIF Ultra (Xpert Ultra) in reducing false-positive rifampicin resistance (RIF-R) and indeterminate (RIF-Ind) results from Xpert MTB/RIF (Xpert Classic). Methods Participant with RIF-R and RIF-Ind results identified by Xpert Classic were enrolled from TB Screening and Treatment Centers in Dhaka, Bangladesh, between January and December 2023. The specimens were then retested using Xpert Ultra. Phenotypic drug susceptibility testing (pDST), line probe assay (LPA), and targeted next-generation sequencing (tNGS) were performed only on isolates classified as rifampicin-resistant by Xpert Ultra, and a composite reference standard (CRS) was applied to determine the final rifampicin resistance status in these cases. Results A total of 62 RIF-R and 54 RIF-Ind cases were enrolled based on Xpert Classic. Upon retesting with Xpert Ultra, out of 62 RIF-R, Xpert Ultra re-categorize 50 (50/62; 80.6%) as Rifampicin sensitive (RIF-S), 6 (6/62; 9.7%) RIF-R, 3 (3/62; 4.8%) RIF-Ind, and 3 (3/62; 4.8%) MTB not detected. Similarly, among 54 RIF-Ind cases, Xpert Ultra re-categorize 40 (40/54;74%) as RIF-S, 2 (2/54;3.7%) RIF-R, 9 (9/54;16.7%) Trace positive, and 3 (3/54;5.5%) MTB not detected. False-positive RIF resistance results were predominantly 72.6% (45/62) observed in specimens with low bacterial load. Compared to pDST, Xpert Ultra showed 100% sensitivity and 96.4% specificity, with a strong agreement (Kappa, 0.65). Out of eight RIF-R isolates by Xpert Ultra, three remained RIF-R, two were RIF-S in pDST and three were culture negative. According to the composite reference standard, six isolates were confirmed as RIF-R, while two were showed discordant results or RIF-S. Conclusions Xpert Ultra demonstrated higher specificity in detecting RIF resistance compared to Xpert Classic. Integrating Xpert Ultra into routine TB diagnostics could improve RIF resistance detection, especially in samples with low bacterial loads.
Uddin et al. (Fri,) studied this question.