Abstract
This study investigated the diagnostic potential of mNGS for detecting MTB in pulmonary tuberculosis patients. We analyzed pulmonary microbiome data to assess its impact on mNGS diagnostic accuracy and explored the association between microbiome profiles and clinical diagnosis. Bronchoalveolar lavage fluid samples were collected from 236 patients with pulmonary infections, and the diagnostic performance of mNGS was compared with traditional methods in detecting MTB. Furthermore, the incidence of false negatives and false positives, as well as the characteristics of the lung microbiota in TB patients, was analyzed to improve the diagnostic precision of mNGS. We observed that among all detection methods, mNGS showed the highest sensitivity (73.33%), followed by X-pert (60.00%), culture (53.33%), RT-PCR (53.33%), and sputum smear (23.33%). Notably, mNGS produced 3 false positive results in 236 samples, yielding a specificity of 98.54%. Analysis of the pulmonary microbiome revealed significant differences in both α-diversity and β-diversity between patients with TB and uninfected controls (P<0.05). Shannon index and Chao1 index were identified as significant predictors associated with MTB infection. ROC curve analysis demonstrated an AUC of 0.765, indicating good discriminatory performance. This study suggested that integrating wet-laboratory techniques with bioinformatics analysis can further enhance the diagnostic accuracy of mNGS for TB. Furthermore, microbiome analysis holds significant potential for the diagnosis of MTB infection. IMPORTANCE: This study focuses on the application of next-generation sequencing (NGS) technology in detecting Mycobacterium tuberculosis in bronchoalveolar lavage fluid and explores the impact of M. tuberculosis infection on the pulmonary microbiome. By optimizing the methods and conducting microbial analyses, the accuracy of metagenomic NGS for detecting M. tuberculosis has been improved.