Insights into microbiome and ARGs diversity in patients with upper and lower respiratory tract infections by targeted next-generation sequencing.

Respiratory tract infections (RTIs) cause substantial global morbidity and mortality, with antimicrobial resistance presenting an increasing challenge to the effective management. Characterizing the differences in microbiome composition and antimicrobial resistance genes (ARGs) between upper respiratory tract infections (URTIs) and lower respiratory tract infections (LRTIs) may inform site-specific diagnostic and therapeutic strategies. We retrospectively analyzed 1,340 URTIs samples (nasopharyngeal swab) and 699 LRTIs samples (bronchoalveolar lavage fluid) admitted to a single medical center to characterize the epidemiology of the respiratory microbes and ARGs using targeted next-generation sequencing (tNGS). Microbiome diversity, ARGs profiles, and coinfection patterns were compared between LRTIs and URTIs groups. Random forest machine learning was employed to identify discriminating species. LRTIs patients exhibited significantly higher microbiome abundance and ARGs diversity than URTIs patients (P < 0.001. Beta-lactam, multidrug, phenicol, and fluoroquinolone resistance genes were significantly more abundant in LRTIs (P < 0.01). Bacteria-virus coinfections predominated in both LRTIs (39.3%) and URTIs (54.6%). Thirty species were identified as potential discriminators between LRTIs and URTIs, with an Area Under Curve (AUC) of 0.852 in the training set. These findings reveal distinct microbial and ARGs profiles between URTIs and LRTIs patients, and provide a foundation for understanding site-specific microbial ecology in RTIs for clinical diagnosis and antimicrobial stewardship.