Biomarkers are important tools to improve the early detection of patients at high risk for developing diabetes as well as the stratification of diabetic patients towards risks of complications. In addition to clinical variables, we analyzed 155 metabolic parameters in plasma samples of 51 healthy volunteers and 66 patients with diabetes using nuclear magnetic resonance (NMR) spectrometry. Upon elastic net analysis with lasso regression, we confirmed the independent associations of diabetes with branched-chain amino acids and lactate (both positive) as well as linoleic acid in plasma and HDL diameter (both inverse). In addition, we found the presence of diabetes independently associated with lower concentrations of free cholesterol in plasma but higher concentrations of free cholesterol in small HDL. Compared to plasmas of non-diabetic controls, plasmas of diabetic subjects contained lower absolute and relative concentrations of free cholesterol in all LDL and HDL subclasses except small HDL but higher absolute and relative concentrations of free cholesterol in all VLDL subclasses (except very small VLDL). These disbalances may reflect disturbances in the transfer of free cholesterol from VLDL to HDL during lipolysis and in the transfer of cell-derived cholesterol from small HDL via larger HDL to LDL.
Keywords: biomarker; diabetes mellitus type 2; free cholesterol; lipoprotein subclasses; nuclear magnetic resonance spectroscopy.