Changes in serum metabolites with the stage of chronic kidney disease: Comparison of diabetes and non-diabetes

Background The renal dysfunction of chronic kidney disease (CKD) alters serum metabolite levels, but it is not clear how diabetes mellitus (DM) affects the metabolic changes in CKD. Methods Serum metabolites from pre-dialysis CKD patients (n = 291) with or without DM and from healthy controls (n = 56) was measured using nuclear magnetic resonance. Results Initial principal components analysis and partial least squares-discriminant analysis score plots segregated the CKD patients according to CKD stage and separated DM from non-DM patients. In the CKD patients, associations were seen with clinical characteristics, hyperglycemia, altered amino acid metabolism, accumulated uremic toxins, and dyslipidemia. Of interest, diabetes more strongly affected the metabolic signature during early stage CKD. Furthermore, serum metabolite profiles were successfully applied to the PLS regression model to predict the estimated glomerular filtration rate. The R² values from the PLS models for CKD patients with DM were higher than those for CKD without DM. Conclusions Metabolomics is useful clinically for providing a metabolic signature that is associated with the CKD phenotype and diabetes more seriously affects patients with early stage CKD compared to those with advanced CKD.