Metabolomics and Lipidomics of Plasma Biomarkers for Tuberculosis Diagnostics using UHPLC-HRMS

Gaofeng Sun¹Quan Wang²Xinjie Shan³Maierheba Kuerbanjiang⁴Ruiying Ma²Wensi Zhou²Lin Sun^5*Qifeng Li^3*

• ¹Xinjiang Medical University, Ürümqi, Xinjiang Uyghur Region, China
• ²The Infectious Disease Hospital of Xinjiang Uygur Autonomous Region, Urumqi, China
• ³Children's Hospital of Xinjiang Uygur Autonomous Region, Urumqi, Xinjiang Uyghur Region, China
• ⁴Department of Nutrition and Food Hygiene, School of Public Health, Xinjiang Medical University, Urumqi 830017, China., Urumqi, China
• ⁵Beijing Key Laboratory of Respiratory Infectious Diseases Research in Children, Beijing Children’s Hospital, Capital Medical University, Beijing, China

Introduction: Determining metabolic profiles during host-pathogen interactions is crucial for developing novel diagnostic tests and exploring the mechanisms underlying infectious diseases. However, the characteristics of the circulating metabolites and their functions after Mycobacterium tuberculosis infection have not been fully elucidated. Therefore, this study aimed to identify the differential metabolites in tuberculosis (TB) patients and explore the diagnostic value of these metabolites as potential biomarkers. Methods: Seventy-two TB patients and 78 healthy controls (HCs) were recruited as the training set, while 30 TB patients and 30 HCs were enrolled as the independent validation set. Metabolites in plasma samples were analyzed by high-resolution mass spectrometry. Differential metabolites were screened using principal component analysis and machine learning algorithms including LASSO, Random Forest, and XGBoost. The diagnostic accuracy of the core differential metabolites was evaluated. Pearson correlation analysis was performed.Result: The metabolic profiling of TB patients showed significant separation from that of the HCs. In the training set, 282 metabolites were identified as differentially expressed in TB patients, with 214 metabolites validated in the independent validation cohort. KEGG pathway enrichment analysis showed that the differential metabolites were mainly enriched in lipid metabolism. Seven core differential metabolites were identified by the three machine learning algorithms. Receiver operating characteristic analysis revealed that Angiotensin IV had high accuracy in diagnosing TB. Conclusion: These newly identified plasma metabolites are expected to serve as potentially valuable biomarkers for TB, potentially facilitating the diagnosis of the disease and enhancing the understanding of its underlying mechanisms.