Longitudinal cerebrospinal fluid metabolomics changes during treatment of tuberculous meningitis: A prospective...

Tuberculous meningitis (TBM) is the most severe form of extrapulmonary tuberculosis and is associated with delayed diagnosis, high morbidity, and substantial mortality despite appropriate treatment. As TBM directly involves the meninges and brain parenchyma, disease-associated metabolic perturbations arising from neuroinflammation, neuronal injury, altered energy metabolism, and host-pathogen interactions are more accurately reflected in cerebrospinal fluid (CSF) than in peripheral biofluids. However, conventional CSF biochemical parameters lack sufficient specificity and sensitivity for reliable diagnosis and are inadequate for monitoring therapeutic response. In this study, we employed high-field (800 MHz) ¹H nuclear magnetic resonance (NMR)-based metabolomics to longitudinally characterize CSF metabolic alterations in confirmed TBM patients at baseline and after three months of anti-tubercular therapy. Multivariate analyses, including PCA, OPLS-DA, and Random Forest classification, revealed clear discrimination between baseline and follow-up TBM samples, which was further improved after feature pruning. A total of 52 CSF metabolites were identified, encompassing amino acids, organic acids, energy metabolites, ketone bodies, membrane-related metabolites, and aromatic compounds. Baseline TBM was characterized by marked perturbations in energy metabolism (elevated lactate and reduced glucose), amino acid turnover, and redox-related pathways, along with the presence of metabolites such as gluconate, 1,5-anhydro-D-sorbitol, acetoin, and cyclohexanone that were absent or markedly reduced at follow-up. Several metabolites exhibited partial or complete normalization with treatment, reflecting metabolic recovery and clinical improvement. Overall, this longitudinal NMR-based CSF metabolomics study provides mechanistic insight into TBM-associated neuro-metabolic reprogramming and identifies treatment-responsive metabolic signatures with potential utility for disease monitoring and improved understanding of TBM pathophysiology.