Blood-based cepheid 3-gene host response test for TB disease: diagnostic performance at a specialized respiratory hospital in China

OBJECTIVE: To evaluate the diagnostic performance of the Cepheid 3-gene Host Response Test (Xpert-MTB-HR) for tuberculosis disease (TBD). METHODS: Among 1,128 participants enrolled in this study, 740 were diagnosed with TBD, and the remaining 388 were classified as Non-TB participants, including 218 cases of other pulmonary diseases (OPD), 48 cases of TB infection (TBI), and 122 healthy controls (HC). TB scores were obtained for all participants using the Xpert-MTB-HR assay, and diagnostic performance was evaluated by ROC curves and AUC. RESULTS: The corresponding AUCs of the TB score for distinguishing TBD from non-TB, HC, OPD, and TBI were 0.717 (95% CI, 0.686-0.748; sensitivity 63.6%, specificity 72.7%), 0.846 (95% CI, 0.814-0.878; sensitivity 70.9%, specificity 88.5%), 0.625 (95% CI, 0.583-0.667; sensitivity 54.9%, specificity 67.9%), and 0.807 (95% CI, 0.761-0.853; sensitivity 60.6%, specificity 93.8%), respectively. TB scores were negatively correlated with bacterial load and C-reactive protein levels (both p < 0.001). Differences in TB scores were observed across treatment duration groups in this cross-sectional analysis. CONCLUSION: Xpert MTB-HR did not fully meet WHO performance targets in the overall population, with performance influenced by pathogen load, particularly in low bacterial load or post-treatment patients. TB score showed patterns consistent with treatment response, warranting longitudinal validation. CLINICAL TRIAL: Not applicable.