Current diabetes risk scores often fail to detect early metabolic changes. Acylcarnitines species (ACs), intermediates of mitochondrial fatty acid oxidation, may indicate early dysregulation in diabetes, but prospective evidence is limited. We investigated the associations between circulating ACs and 10-year incidence of diabetes, and whether AC improve the predictive performance of established diabetes risk scores. Data from 1,976 apparently healthy adults in a population-based cohort. Incident diabetes was defined as HbA1c ≥ 48 mmol/mol or initiation of glucose-lowering medication at follow-up. Associations were analysed using bivariate and multivariate models, and predictive value was assessed by adding ACs individually to FINDRISC, SDA, Kahn clinic, Kahn clinic+biologic, Wilson, Balkau and Griffin scores. Over 10 years, 35 participants developed diabetes. Higher baseline levels of Propionyl-, Isovaleryl-, Hexanoyl-, and Octenoylcarnitine were associated with incident diabetes in bivariate analysis whereas only Propionylcarnitine remained associated in multivariate analyses, mean ± sem 376.3 ± 2.7 vs. 419.1 ± 20.4 nmol/L in non-Diabetes and incident diabetes, respectively, p < 0.05. Sensitivity analyses using both diabetes and prediabetes incidence showed similar pattern of associations compared to the primary analysis. In predictive analyses, Propionyl- and Isovalerylcarnitine remained independently associated with incident diabetes after adjustment for each established diabetes risk scores. Adding Isovalerylcarnitine to the risk scores improved discrimination in three out of seven risk scores: AUC 0.785 (0.710–0.859), 0.841 (0.789–0.892), and 0.841 (0.789–0.892) for Balkau, SDA and FINRISK, respectively. Specific AC species, particularly Propionyl- and Isovalerylcarnitine, predicted 10-year risk of diabetes and improved predictive performance beyond established risk scores, suggesting their potential as early metabolic markers.
Maung et al. (Sun,) studied this question.