Identification of metabolomic signatures of diabetic kidney disease in urine and plasma using untargeted gas chromatography-mass spectrometry

Diabetic kidney disease (DKD) is a complication of diabetes and the leading cause of kidney failure among diabetic patients. Unfortunately, it is typically diagnosed after normal kidney function is already significantly impaired. We used gas chromatography-mass spectrometry (GC-MS) of urine and plasma to explore whether metabolites can serve as potential biomarkers for the early diagnosis of DKD. Urine and plasma samples were obtained from 41 individuals [11 healthy controls, 20 patients with diabetes (diabetes mellitus or DM) and microalbuminuria, 10 patients with DKD]. A total of 342 metabolites were identified in urine samples and 252 metabolites were identified in plasma samples, with 182 metabolites overlapping between the two sample types. Among these, 58 metabolites from urine and 22 metabolites from plasma showed suggestive evidence (P-value < .05) of differences between samples from patients with DKD and samples from healthy control individuals. Sparse partial least squares discriminant analysis (sPLS-DA) was applied to identify metabolomics profiles (in urine, plasma, or both) that maximize separation between DKD patients and controls. A set of four metabolites in plasma, including tyrosine and threo-hydroxyaspartic acid, shows promise as an early-stage biomarker signature that will need to be validated in a larger study. Using this set of metabolites, we estimated the probability of DKD for each of the DM patients, based on their metabolomic profiles, and found a significant correlation with DM designation (low, medium, and high) and estimated glomerular filtration rate. Further validation in larger cohorts is needed to confirm their clinical utility and the ability to predict individuals at risk for DKD.