Blood DNA methylation profiling identifies candidate diabetic kidney disease risk signatures in the PROPHECY Aboriginal Diabetes Study

Abstract Background Diabetic kidney disease is a leading cause of kidney failure, yet molecular predictors of risk remain poorly defined, particularly among Indigenous Australians who experience a disproportionate burden. We aimed to identify blood-derived candidate DNA methylation signatures associated with Kidney Disease: Improving Global Outcomes (KDIGO) risk categories in the PROPHECY Aboriginal Diabetes Study. Methods We applied methyl-CpG binding domain sequencing to blood-derived DNA from 95 participants stratified as non-diabetic (n = 23), low-risk (n = 21), moderate-risk (n = 26), high-risk (n = 14) or very-high-risk (n = 11). Differential MBD-enriched methylated regions were assessed using covariate-adjusted models incorporating clinical, leukocyte-marker and CpG-density covariates. Candidate regions were prioritised using a nominal discovery threshold and explored using annotation, random forest modelling and human podocyte experiments. Results For the primary diabetic kidney disease risk contrast, 1,317 regions met the nominal discovery threshold of P < 0.001, but no region reached FDR < 0.10. Across 16 predefined analysis terms - 11 pairwise/grouped contrasts and 5 single-group terms - 22,392 nominal DMR entries met P < 0.001, corresponding to 13,024 unique candidate regions. Exploratory random forest models suggested that methylation features may contribute discriminatory information beyond clinical covariates, although estimates require independent validation. Podocyte analyses provided locus-level functional context but not genome-wide cross-tissue replication. Conclusions This cross-sectional discovery study nominates research-grade candidate methylation biomarkers and pathways for future longitudinal validation in PROPHECY and external cohorts.