Plasma metabolomic and lipidomic signatures characteristic of treatment non-response in rheumatoid arthritis

ObjectivesThis study aimed to investigate plasma metabolomic and lipidomic profiles in rheumatoid arthritis (RA) to identify potential biomarkers for distinguishing treatment responses.MethodsPlasma samples were collected from 106 RA patients and 10 healthy controls, with 30 RA samples selected based on predefined inclusion criteria. Using liquid chromatography–mass spectrometry (LC-MS)-based untargeted metabolomics and lipidomics analysis, a total of 2,279 metabolites and 2,987 lipids were tentatively annotated.ResultsThrough rigorous statistical evaluation (variable importance in the projection (VIP) > 1 and false discovery rate (FDR) < 0.05), 22 metabolites and lipids were found to be positively associated with RA risk based on logistic regression analysis.

These were further refined into 12 core features using least absolute shrinkage and selection operator (LASSO) regression. The 12 core features identified in the treatment non-response group included lipids Cer(d18:1/16:0), PS(16:0/20:0), and Palmitic acid; metabolites Mimosine and D-Xylose; drug-related metabolites Dihydralazine, Tridihexethyl, Acyclovir monophosphate, Indoline, and Melleolide; and pollution-related metabolites Norcotinine and 2-Chloro-1-(chloromethyl)ethyl carbamate.