Untargeted Metabolomics Identifies N‐Lactoyl‐Amino Acids as Dose‐Responsive Plasma Biomarkers of Metformin Adherence in Type 2 Diabetes

Metformin is the most widely prescribed antidiabetic drug, yet adherence remains difficult to objectively assess. Using untargeted metabolomics and lipidomics, we analyzed plasma from 637 patients with type 2 diabetes (T2D) with confirmed metformin use and 143 nondiabetic controls, annotating 614 metabolites. Patients were stratified by plasma metformin into sub-therapeutic, therapeutic, and supra-therapeutic groups, and associations were evaluated by multiple linear regression and composite metabolite ranking. Five previously unannotated features were structurally identified as N-lactoyl-amino acids, whose levels correlated strongly with plasma metformin (ρ = 0.42-0.55, P < 0.0001) and increased up to 7.2-fold in the supra-therapeutic group (> 2000 ng/mL). While N-lactoyl-amino acids were consistently detected in the nanomolar range, they still displayed robust and dose-dependent associations with metformin. Broader metabolic changes in T2D included elevated lactate, organic acids, and branched-chain amino acids, together with reduced urea cycle metabolites. Lipidomics showed increases in saturated triacylglycerols and diacylglycerols and decreases in cholesteryl esters, sphingomyelins, and phospholipids. These findings establish N-lactoyl-amino acids as robust, dose-responsive plasma biomarkers of metformin exposure. Despite being up to four orders of magnitude less abundant than their amino acid precursors, they sensitively reflect mitochondrial lactate overflow and pharmacodynamic adaptation, offering objective assessment of adherence.