DNA Methylation and 28-Year Incidence of Kidney Disease in Type 1 Diabetes

Abstract

Key Points:

1. DNA methylation may be an early biomarker for diabetic kidney disease, but prior studies in type 1 diabetes focused on advanced disease.

2. Cytosine phosphate guanine sites in histone deacetylase 11 and cysteine-rich epidermal growth factor-like domain 2 were associated with microalbuminuria incidence in type 1 diabetes suggesting roles in early kidney damage.

3. Epigenetic regulation of those loci warrants further study as a potential target for prevention or treatment of diabetic kidney disease.

Background: Prior epigenome-wide association studies (EWAS) of DNA methylation (DNAm) and diabetic kidney disease (DKD) in type 1 diabetes (T1D) focused on advanced disease. We hypothesized DNAm is a biomarker of earlier manifestations of DKD, microalbuminuria (MA), and overt nephropathy (ON), independent of traditional risk factors in T1D. We thus performed separate EWAS of 28-year MA and ON in the Pittsburgh Epidemiology of Diabetes Complications T1D cohort.

Methods: We analyzed baseline whole-blood DNAm (total n=405 participants; 683,597 cytosine phosphate guanines [CpGs]) and time to event in Epidemiology of Diabetes Complications participants free of each respective complication at baseline (n=224 eligible for MA analysis, n=306 eligible for ON analysis). We also identified differentially methylated regions (DMRs), assessed associations between DKD-associated CpGs and established clinical risk factors, and performed in silico functional analyses. In secondary analyses, we performed EWAS of ESKD in participants with existing DKD at baseline (ON or eGFR 15–59 ml/min per 1.73 m², n=99) for comparison with prior studies.

Results: DNAm at cg06568490 (histone deacetylase 11) and cg13618568 (cysteine-rich epidermal growth factor-like domain 2) was associated with MA incidence (false discovery rate [FDR] <0.05). Associations between both CpGs and MA remained similar after risk factor adjustment. We also identified 48 significant DMRs (Šidák value <0.05) for MA. For ON, no individual CpGs had FDR <0.05, but there were 35 significant DMRs. Pathways in Signal Transduction were the most significantly enriched for both end points. In EWAS of ESKD, no CpGs were significant in minimally adjusted models, while cg09867934 (ETFDH) was significantly associated with ESKD after adjusting for clinical risk factors (FDR=0.002), but there was little overlap with previously published studies of DNAm and ESKD in T1D.

Conclusions: Epigenetic modification of loci, including histone deacetylase 11, which has been associated with kidney fibrosis, and cysteine-rich epidermal growth factor-like domain 2, which is involved in angiogenesis, may contribute to early kidney damage in T1D, thus warranting further study as potential biomarkers of future risk.

Keywords: ESKD; albuminuria; biomarkers; cohort studies; diabetes; diabetic nephropathy; epidemiology and outcomes; human genetics; microalbuminuria; nephropathy.

Graphical abstract

Figure 1

Pathway analysis revealed enrichment for 28 subpathways for microalbuminuria and 32 for overt nephropathy highlighting potential mechanisms underlying DKD development in type 1 diabetes. Dot plots summarizing Reactome pathway enrichment among loci with suggestive CpGs (FDR <0.10) or significantly DMRs (Šidák value <0.05) for MA (A) and ON (B). Significance (−log P value) is indicated by the x axis; dot area represents the proportion of total proteins in the Reactome set present in the network derived using results of the EDC EWAS. CpGs, cytosine phosphate guanines; COPI, coat protein I; DKD, diabetic kidney disease; DMR, differentially methylated region; EDC, Epidemiology of Diabetes Complications; ER, endoplasmic reticulum; ESR, estrogen receptor; EWAS, epigenome-wide association studies; FDR, false discovery rate; HOX, homeobox; MA, microalbuminuria; NTR, neurotrophin receptor; ON, overt nephropathy; PDGF, platelet-derived growth factor; TCR, T-cell receptor.