Evaluating the Cepheid Xpert TB/LTBI research-use-only assay for detection of active and latent Mycobacterium tuberculosis infection

Background Accurate non-sputum tests are essential for improving tuberculosis (TB) detection. We report the first evaluation of the Xpert TB/LTBI assay (research-use-only, Cepheid, USA), which detects nine Mycobacterium tuberculosis antigen-stimulated mRNA targets from blood. Methods We enrolled individuals ≥12 years with presumptive TB from clinics in Uganda and Vietnam. All participants underwent sputum- (liquid culture; Xpert MTB/RIF Ultra) and blood-based (Xpert TB/LTBI; QuantiFERON-TB Gold Plus [QFT-Plus]) testing. Participants were classified as having active TB (positive sputum results), latent TB infection (LTBI; positive QFT-Plus with negative sputum results), or no TB infection. Diagnostic accuracy for active TB (primary analysis) and of a two-step algorithm that first predicted TB infection (active TB or LTBI) and then distinguished active TB from LTBI was assessed using logistic regression and receiver operating characteristic analysis. Results Among 214 participants included, 56.5% were male, 9.3% were living with HIV, 29.0% had active TB, and 31.8% had LTBI. Xpert TB/LTBI area under the curve (AUC) was 0.92 (95% CI 0.88-0.96) for identifying active TB, and sensitivity was 93.5% (95% CI 84.6-97.5) and specificity 76.3% (95% CI 69.0-82.4) at a cut-point that achieved ≥90% sensitivity. In the two-step algorithm, Xpert TB/LTBI had high accuracy for identifying TB infection in the first step, but accuracy was lower for distinguishing active TB from LTBI in the second step. Conclusions Xpert TB/LTBI exceeded World Health Organization (WHO)-recommended minimum accuracy targets for a TB screening test. However, further refinement is needed to improve its ability to distinguish active TB from LTBI.