Diagnostic performance of Xpert MTB/RIF Ultra assay with pulmonary and extrapulmonary specimens: a retrospective evaluation in a low-incidence setting in Finland

Objective The aim was to evaluate the sensitivity and specificity of the Xpert MTB/RIF Ultra (Xpert Ultra) assay in the detection of extrapulmonary tuberculosis (TB) in comparison to pulmonary TB in a low-incidence setting in the Helsinki capital area of Finland. Methods The retrospective analysis included results from 1112 pulmonary and 705 extrapulmonary samples collected between 2018 and 2023, of which 193 and 136 were culture-positive for Mycobacterium tuberculosis (MTB), respectively. Xpert Ultra results were compared with mycobacterial culture. PCR-positive and culture-negative cases were separately compared with available clinical data (composite reference standard, CRS). Results Compared with culture, Xpert Ultra demonstrated 95.3% sensitivity (95% CI, 91.3%-97.7%) and 94.5% specificity (95% CI, 92.8%-95.8%) with pulmonary samples, 47.1% (95% CI, 26.2%-69.0%) and 96.7% (95% CI, 93.8%-98.4%) with pleural fluid, 100% (95% CI, 86.9%-100%) and 81.8% (95% CI, 72.4%-88.6%) with tissue, 96.6% (95% CI, 81.4%-100%) and 75.0% (95% CI, 62.2%-84.6%) with pus, and 95.1% (95% CI, 83.0%-99.5%) and 67.5% (95% CI, 51.9%-80.0%) with lymph node samples, respectively. Other, less common sample types were also included. When CRS was also considered, specificity exceeded 93% for all sample types. Sensitivity was 100% with both smear-positive pulmonary and smear-positive extrapulmonary samples. Neither false rifampicin susceptibility testing results nor cross-reactivity with nontuberculous mycobacteria was detected. Discussion Xpert Ultra detected MTB in lymph node, tissue, and pus samples with high-accuracy comparable with the analysis of pulmonary samples while reducing the time to diagnosis by up to several weeks compared with mycobacterial culture.