Untargeted plasma metabolome identifies biomarkers in patients with extracranial arteriovenous malformations

Objective: This study aimed to investigate the plasma metabolic profile of patients with extracranial arteriovenous malformations (AVM). Method: Plasma samples were collected from 32 AVM patients and 30 healthy controls (HC). Ultra-high performance liquid chromatography-mass spectrometry (UHPLC-MS) was employed to analyze the metabolic profiles of both groups. Metabolic pathway enrichment analysis was performed through Kyoto Encyclopedia of Genes and Genomes (KEGG) database and MetaboAnalyst. Additionally, machine learning algorithms such as Least Absolute Shrinkage and Selection Operator (LASSO) and random forest (RF) were conducted to screen characteristic metabolites. The effectiveness of the serum biomarkers for AVM was evaluated using a receiver-operating characteristics (ROC) curve. Result: In total, 184 differential metabolites were screened in this study, with 110 metabolites in positive ion mode and 74 metabolites in negative mode. Lipids and lipid-like molecules were the predominant metabolites detected in both positive and negative ion modes. Several significant metabolic pathways were enriched in AVMs, including lipid metabolism, amino acid metabolism, carbohydrate metabolism, and protein translation. Through machine learning algorithms, nine metabolites were identify as characteristic metabolites, including hydroxy-proline, L-2-Amino-4-methylenepentanedioic acid, piperettine, 20-hydroxy-PGF2a, 2,2,4,4-tetramethyl-6-(1-oxobutyl)-1,3,5-cyclohexanetrione, DL-tryptophan, 9-oxoODE, alpha-Linolenic acid, and dihydrojasmonic acid. Conclusion: Patients with extracranial AVMs exhibited significantly altered metabolic patterns compared to healthy controls, which could be identified using plasma metabolomics. These findings suggest that metabolomic profiling can aid in the understanding of AVM pathophysiology and potentially inform clinical diagnosis and treatment.