Abstract

Background: The metabolic dysregulation underlying knee osteoarthritis (KOA) is not fully elucidated, particularly the specific molecular shifts occurring within diseased cartilage compared to adjacent healthy tissue. This study aimed to define the distinct metabolomic signature of KOA-affected cartilage using a patient-matched control design, leveraging machine learning to identify key discriminatory metabolites and altered metabolic pathways.
Methods: Articular cartilage samples from both lesional (KOA) and visually healthy areas were collected from 17 patients undergoing total knee arthroplasty. A targeted, broad-spectrum metabolomic analysis was performed using liquid chromatography-mass spectrometry (LC-MS). To distinguish between the two tissue types and identify significant metabolic features, the resulting data were subjected to multivariate statistical analyses, including Partial Least Squares-Discriminant Analysis (PLS-DA) and Orthogonal PLS-DA (OPLS-DA), as well as a Random Forest (RF) machine learning model. Pathway analysis was conducted using the KEGG and SMP databases.
Results: A clear and significant metabolic distinction was found between KOA and healthy cartilage, demonstrating profound metabolic reprogramming in the diseased tissue. The PLS-DA and OPLS-DA models successfully separated the two groups. Pathway analysis revealed that the most significantly disrupted pathways in KOA cartilage included arginine biosynthesis, glutathione metabolism, arginine and proline metabolism, glycerophospholipid metabolism, and alanine, aspartate, and glutamate metabolism. The Random Forest model identified L-kynurenine, lipoic acid, and phosphatidylcholine (PC) (36:1) as top-ranking metabolites for discriminating between healthy and osteoarthritic cartilage. Conclusion: Utilizing a robust patient-matched design, this study reveals significant metabolic reprogramming in KOA cartilage, characterized by major disruptions in amino acid, antioxidant, and lipid metabolism. The metabolites L-kynurenine, lipoic acid, and PC (36:1) were identified as novel and promising biomarker candidates for KOA. These findings enhance our molecular understanding of KOA pathogenesis and pinpoint key metabolic nodes that could serve as future therapeutic targets. Further validation is necessary to confirm the diagnostic potential of these biomarkers and to explore the functional consequences of these metabolic shifts in disease progression.