Insulin clearance has emerged as a potential key player in the deterioration of glycaemic control. Associations between circulating metabolites and insulin clearance are unexplored, especially in longitudinal terms. Here, we investigated significant associations between temporal changes in OGTT-assessed insulin clearance (ICL) and changes in circulating targeted metabolites over the same period, in individuals at risk of developing type 2 diabetes (T2D). In the IMI DIRECT cohort of white North European individuals at risk of developing T2D (N=1983), targeted metabolomics were assessed using the AbsoluteIDQ Biocrates p180kit and OGTTs were conducted at months 0, 18, and 48. Metabolite concentrations were normalized to reference plasma samples and adjusted for study center. Longitudinal slopes for 128 metabolites and for ICL were computed over the 48 months via linear mixed effects modeling. ICL slopes were adjusted for insulin sensitivity (as OGIS index) slopes. The relationships between slopes for ICL and for metabolites were investigated through nested cross-validated LASSO multivariate analysis, including as potential confounders mean insulin secretion and BMI slopes, sex, age, and smoker status. LASSO selected 8 metabolites slopes and three confounders as important variables for ICL slope multivariate linear regression (figure; cross-validated R2=0.13). Adjusted R2 from multivariate linear regression without LASSO penalization improved from 0.05 (confounders only) to 0.15 with the addition of the selected features. The highest importance was reported for slopes of proline (Pro), with negative association, and acetyl-L-carnitine (C2) and octadecenoylcarnitine (C18:1), with positive association. We identified a set of metabolites longitudinally associated to OGTT-assessed insulin clearance, adjusted for insulin sensitivity. Further investigations may demonstrate the utility of these metabolites as biomarkers of insulin clearance, or point to a mechanistic role of some of these metabolites in its evolution.
