Tuberculosis (TB) remains a major global cause of morbidity and mortality. Current tools for monitoring treatment response rely on sputum-based microscopy and culture, which are often insensitive, time-consuming, and impractical in extrapulmonary or pediatric TB and in individuals unable to produce sputum. Metabolomics has emerged as a promising approach for identifying host-derived biomarkers that reflect treatment-associated immunometabolic changes; however, the available evidence remains heterogeneous and has not been comprehensively synthesized. We conducted a comprehensive literature review of human studies evaluating metabolomic biomarkers in relation to TB treatment response or outcomes. PubMed, Scopus, and EMBASE were searched for human studies evaluating targeted or untargeted metabolomics (NMR, LC-MS, GC-MS, CE-MS) in relation to treatment response or outcomes. Two reviewers independently screened studies, extracted data, and assessed risk of bias using QUIPS and PROBAST. Findings were synthesized using a structured framework organized across treatment stages and outcomes. Of 218 records identified, 139 titles and abstracts were screened and 42 full texts assessed; 15 studies met the inclusion criteria. Recurrent treatment-associated signals involved amino acid metabolism, particularly the tryptophan–kynurenine pathway, as well as vitamin and cofactor metabolites (pyridoxate, nicotinamide, trigonelline). Plasma studies frequently reported lipid remodeling and bile acid perturbations, whereas urine studies highlighted polyamine metabolism (e.g., N1,N12-diacetylspermine) and fatty acid β-oxidation markers. Common limitations included inadequate adjustment for confounders and, in prediction models, small sample sizes and limited external validation. Metabolomics reveals reproducible but heterogeneous immunometabolic changes during TB therapy. Key pathways include tryptophan–kynurenine metabolism, vitamin and cofactor metabolism, lipid remodeling, and urine polyamine pathways. Standardization and prospective multicenter validation are needed for clinical translation.
Nguyen et al. (Thu,) studied this question.