Blood-Derived DNA Methylation Biomarkers Predict Diabetic Kidney Disease and Map to Epigenomic Remodeling in Renal Cells

Abstract

Diabetic kidney disease (DKD) is a leading cause of end-stage kidney disease in type 1 diabetes (T1D); however, albuminuria and estimated glomerular filtration rate (eGFR) lack the precision necessary for early prognostication. We sought blood-derived DNA methylation biomarkers that predict DKD progression and reflect renal epigenomic remodeling. In the PROFIL cohort, we performed high-depth genome-wide methylation sequencing of leukocyte DNA from 101 adults with T1D and 20 control participants without diabetes, stratified by Kidney Disease: Improving Global Outcomes (KDIGO) risk and observed for a median of 5.6 years. Differentially methylated regions were identified across baseline KDIGO strata and prioritized based on hypomethylation, four or more CpGs per region, and discrimination between KDIGO categories; their associations with eGFR decline and rising albumin excretion were then evaluated. From these candidates, we derived a four-locus methylation risk score (MRS; SLC4A4, FSTL4, ICA1, and PTK2), validated by targeted assays in an independent subset of 330 participants with T1D with low or moderate KDIGO risk at baseline. Those with T1D showed global loss of 5-methylcytosine, with hypomethylated regions enriched at regulatory sites. In the discovery cohort, the combined MRS plus clinical model discriminated baseline KDIGO moderate- and high/very-high-risk strata (areas under the curve 0.91 and 0.86, respectively). In the independent validation cohort, the four-locus MRS predicted DKD progression within baseline low- and moderate-risk categories and improved risk stratification beyond clinical covariates, including eGFR and albumin excretion. In human podocytes, high glucose induced locus-specific hypomethylation, altered chromatin binding, and increased expression of MRS genes. These findings support a four-locus MRS for earlier, mechanistically grounded DKD risk stratification in T1D. Article Highlights We undertook this study because albuminuria and estimated glomerular filtration rate often identify diabetic kidney disease risk only after substantial kidney injury in type 1 diabetes. We asked whether blood DNA methylation patterns could predict future DKD progression and improve clinical risk stratification. Genome-wide leukocyte methylation sequencing identified hypomethylated regions and a four-locus methylation risk score (SLC4A4, FSTL4, ICA1, and PTK2) that improved prediction beyond clinical risk models and was validated in 330 participants. Podocyte multiomics showed concordant hyperglycemia-induced hypomethylation and gene activation, supporting the biological plausibility of the blood-derived risk score.