Cross-cohort meta-analysis of respiratory microbiota predicts metabolic adaptations of microbes in pulmonary tuberculosis

Background: Understanding the microbial species and their metabolic pathways associated with pulmonary tuberculosis (PTB) has significant potential to improve current treatments through the identification of novel biomarkers and therapeutic targets. Recent metagenomic studies have enhanced our understanding of PTB associated microbes; however, their findings are often inconsistent due to small, geographically restricted cohorts. Therefore, a systematic cross-cohort meta-analysis of these studies is necessary to obtain a more generalized picture. Methods: Seven publicly available 16S rRNA amplicon sequencing datasets derived from sputum, bronchioalveolar lavage, lung biopsy and stool samples were re-analyzed to generate a homogeneous taxonomic abundance dataset. Microbial metabolic pathways were inferred using PICRUSt2. Differential abundance analysis between MTB+ and MTB- samples was conducted to identify PTB associated microbial taxa and metabolic pathways. In addition, three machine learning (ML) models were trained to assess the diagnostic potential of sputum microbiota. Findings: PTB-associated microbial dysbiosis was marked by reduced alpha diversity, particularly in sputum samples. Differential abundance analysis of 832 sputum samples identified 18 taxa and 39 metabolic pathways. Commensal Prevotella species (P. intermedia, P. nigrescens, P. jejuni, P. melaninogenica, and P. oris) were depleted, while opportunistic pathogens (Schaalia odontolytica, TM7x, Fusobacterium periodonticum, etc.) increased. Metabolic pathway abundance estimates favoring Mycobacterium tuberculosis included enhanced formaldehyde assimilation and reduced L-rhamnose degradation. ML models trained on sputum microbiota efficiently classified disease status, with the random forest classifier achieving the highest specificity (average recall = 0.81).Interpretation: This large-scale cross-cohort meta-analysis of the PTB respiratory microbiome reveals its strong diagnostic potential and provides insights into possible microbial metabolic adaptations. Funding: National Supercomputing Mission, Government of India.