Abstract
BACKGROUND: Tuberculosis (TB) diagnosis in the UK is impacted by delay and suboptimal culture-based microbiological confirmation rates due to the high prevalence of paucibacillary disease. We examine the real-world clinical utility of Xpert MTB/RIF Ultra (Xpert-Ultra) as a diagnostic test and biomarker of transmissible infection in a UK TB service. METHODS: Clinical specimens from suspected TB cases triple tested (smear microscopy, mycobacterial culture and Xpert-Ultra) at University Hospitals of Leicester NHS Trust (1 March 2018-28 February 2019) were retrospectively analysed. Diagnostic sensitivity and specificity were calculated using positive MTB culture and clinical TB diagnosis as reference standards. The QuantiFERON (QFT) positive proportion of pulmonary TB (PTB) contacts was used as a metric of transmitted infection to evaluate Xpert-Ultra and smear grade as markers of infectiousness. RESULTS: 251 samples (188 respiratory) from 231 patients (86 TB) were analysed. Compared with microscopy, Xpert-Ultra had higher diagnostic sensitivity (24.7% vs 78.7%, p<0.001) and comparable specificity (97.5% vs 99.4%). Xpert-Ultra and culture had comparable sensitivity (78.7% vs 71.9%) and specificity (99.4% vs 100.0%). Incorporating Xpert-Ultra with culture increased microbiologically verified diagnosis to 91.7% for PTB and 75.9% for extrapulmonary TB, compared with 85.0% and 44.8%, using culture alone. In PTB, both smear and Xpert-Ultra grade were positively associated with the proportion of contacts testing QFT positive. However, Xpert-Ultra had a higher negative predictive value than smear (QFT-positive contacts 6.7% vs 17.7%). CONCLUSION: In low-TB-burden settings, systematic adoption of Xpert-Ultra for clinical assessment of suspected TB can improve the proportion of microbiologically verified diagnoses and improve the stratification of transmission risk.