Xpert MTB/RIF Ultra assay for the detection of Mycobacterium tuberculosis in people with negative conventional Xpert MTB/RIF but chest imaging suggestive of tuberculosis in Dhaka, Bangladesh

BACKGROUND: The World Health Organization is considering substituting Xpert MTB/RIF (Xpert) with Xpert MTB/RIF Ultra (Ultra) for tuberculosis (TB) diagnosis, but supportive evidence is scarce, particularly among people more likely (presumptive) to have paucibacillary pulmonary TB (PTB). METHODS: During January-July 2018, presumptive PTB patients visiting TB Screening and Treatment Centres of Dhaka for routine chest X-ray (CXR) and conventional Xpert were enrolled. Sputum specimens were additionally tested with microscopy, culture, and Ultra. Specimens with "Trace call" by Ultra (Ultra-trace) were retested. Yield and diagnostic accuracy using various approaches to Ultra-trace and concordance of Ultra with bacteriological-positive PTB were assessed. RESULTS: Altogether, 1,083 participants (104 'Xpert-positive'; 979 'Xpert-negative and CXR-suggestive') were enrolled. All Xpert-positives and 900 (92%) Xpert-negatives were concordant with Ultra, however, seventy-nine (8.1%) Xpert-negative specimens tested positive with Ultra; 37 (46.8%) were categorically positives, and 42 (53.2%) were Ultra-trace. Sixteen of the 42 were retested, of whom eight (50.1%) Ultra-trace turned categorically positive, leading to 45 (4.6%) additionally detected by Ultra. Ultra sensitivity and specificity were 93.9% and 94.6%, and it additionally detected 5.4% more TB patients with a concordance of 94.6% (kappa, □=0.78) compared to any bacteriologically positive specimen (microscopy, culture, or Xpert). CONCLUSION: Ultra exhibited improved detection and accuracy among Xpert-negatives in a cohort with a high likelihood of PTB.