AUTHOR=Xiao Yu , Shi Caifeng , Qin Songyan , He Aiqin , Wu Xiaomei , Dai Chunsun , Zhou Yang 

TITLE=Urinary lipid metabolites and progression of kidney disease in individuals with type 2 diabetes

JOURNAL=Frontiers in Endocrinology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2025.1650498

DOI=10.3389/fendo.2025.1650498

ISSN=1664-2392

ABSTRACT=ObjectiveA substantial proportion of individuals with type 2 diabetes (T2D) experience a fast decline (FD) in kidney function, a high-risk phenotype not reliably identified by current clinical markers. This study aimed to evaluate the potential of urinary lipid metabolites as novel predictors for the rapid progression of diabetic kidney disease (DKD).MethodsThis investigation employed a dual-phase design comprising cross-sectional screening and longitudinal validation. In the initial phase, targeted lipidomic profiling of urine samples from 152 patients with T2D and DKD and 152 age- and sex-matched individuals with uncomplicated diabetes revealed distinct metabolite patterns. The subsequent validation phase utilized an independent cohort of 248 T2D patients, in which rapid kidney function decline was defined as the highest quartile of annual estimated glomerular filtration rate (eGFR) reduction. Feature selection was performed using machine learning algorithms (random forest and Boruta) to identify potential biomarkers from the differentially expressed metabolites. The prognostic value of these lipid markers for predicting future renal function decline was assessed against clinical variables using receiver operating characteristic (ROC) analysis.ResultsThe analysis of fasting spot urine specimens quantified 104 lipid metabolites out of 508 targeted species, with all concentrations normalized to urinary creatinine. The comparative analysis identified 21 lipid metabolites that were significantly upregulated in the DKD group. Feature selection algorithms isolated nine (Boruta) and eight (random forest) candidate biomarkers from this pool. During a median follow-up period of 33 months (IQR 17–47), 62 participants showing the most rapid eGFR decline were classified as the FD group. These individuals exhibited significantly elevated baseline levels of the identified lipid metabolites. The lipid panel demonstrated superior predictive performance for future kidney function decline compared with traditional clinical predictors, including baseline eGFR, hemoglobin A1c, and albuminuria.ConclusionsOur findings reveal a strong association between urinary lipid metabolites and DKD progression. Specifically, urinary lipid profiling shows promise as a non-invasive tool to identify T2D patients at a high risk for rapid kidney function decline, outperforming the current clinical standard of albuminuria and eGFR.