Graphdiyne-assisted LDI-MS for rapid, non-invasive urine metabolomic profiling in tuberculosis screening

Abstract

Tuberculosis (TB) remains a major global health burden. Here we report a rapid, noninvasive urine-based metabolomics approach using graphdiyne (GD)-assisted laser desorption ionization mass spectrometry (LDI-MS) combined with machine learning. We applied GD-assisted LDI-TOF MS to urine samples from healthy controls (HC) and active TB patients, generating rich metabolite fingerprints. Supervised classifiers trained on the GD-assisted LDI-TOF MS spectral features achieved excellent discrimination, consistent with previous reports of Matrix-Assisted Laser Desorption Ionization-Time-of-Flight Mass Spectrometry (MALDI-TOF MS) with machine learning as a powerful screening tool. Key discriminatory urine metabolites and pathways were putatively annotated and included markers of altered energy, nucleotide, and amino-acid metabolism in TB patients, reflecting a shift in cellular energy handling and immune-related nucleotide turnover. These biochemical insights underscore TB-associated inflammatory and energetic perturbations. Overall, the GD-assisted LDI-TOF MS platform enables fast, high-throughput metabolite profiling and, when coupled with machine learning, offers a patient-friendly, noninvasive screening strategy for early TB detection and monitoring.