Clinical value of targeted next-generation sequencing and metagenomic next-generation sequencing in the diagnosis of pulmonary tuberculosis: a real-world propensity score matching study

Targeted next-generation sequencing (tNGS) and metagenomic next-generation sequencing (mNGS) are widely used for diagnosing pulmonary infections. However, their comparative clinical value and impact on healthcare resource utilization remain unclear. This retrospective study enrolled patients with confirmed pulmonary infections who underwent either tNGS or mNGS. Propensity score matching (PSM) was performed to balance the baseline covariates. The diagnostic performance for tuberculosis (TB) and clinical outcomes was compared between the two groups. After PSM, 403 patients were included (304 in tNGS and 99 in mNGS). No significant differences were observed between tNGS and mNGS in the final TB diagnosis rate (17.4% versus 23.2%; p = 0.257) or Mycobacterium tuberculosis complex positivity rate (20.4% versus 28.3%, p = 0.134). The area under the curve for TB diagnosis was 0.936 (95% confidence interval [CI]: 0.898–0.975) for tNGS and 0.967 (95% CI: 0.939–0.995) for mNGS, with no statistically significant difference ( p = 1.000). In the matched cohort (mNGS group, n = 99), mNGS achieved 100% sensitivity and 100% negative predictive value (NPV); however, in the unmatched full cohort ( n = 104), the sensitivity and NPV were 95.8% and 98.7%, respectively. Nevertheless, the mNGS group incurred substantially higher healthcare costs. The previously noted extended hospital stay lost its statistical significance after adjusting for the calendar period, indicating that the variation in length of stay may be attributed in part to temporal changes in clinical practice rather than to the test modality itself. tNGS and mNGS did not show statistically significant differences in diagnostic efficacy for pulmonary TB based on the AUC comparison in our study. In post-PSM subset-specific AUC analyses, MTBC-NGS outperformed IGRA, TB-DNA, and acid-fast staining, and also showed a smaller but statistically significant advantage over Xpert MTB/RIF after Bonferroni correction. mNGS showed a very high negative predictive value in our matched cohort, suggesting potential utility for ruling out TB in challenging cases; however, this finding requires validation in larger studies, and it consumes more healthcare resources, whereas tNGS was associated with lower healthcare costs in this cohort. These two methods should be considered complementary. Not applicable.