New insights into DNA methylation profiles associated with insulin resistance in children and adolescents with obesity

Background: The role of DNA methylation (DNAm) in metabolic dysregulation is emerging, but few studies have investigated the early epigenetic modifications associated with insulin resistance (IR) in children and adolescents with obesity. We aimed to investigate i) whether DNAm in blood was associated with early IR status in children and adolescents with obesity; ii) whether DNAm at some of the identified IR-associated sites partially mediated the effect of birthweight, an early metabolic risk factor, on IR status in childhood using causal mediation analysis. Methods: We performed an epigenomic case-control analysis on peripheral blood using Infinium MethylationEPIC BeadChip array. About 100 versus 100 children and adolescents with obesity, corresponding, respectively, to the lowest and the highest decile of Matsuda index (i.e., dynamic index of insulin sensitivity) adjusted for age, puberty, sex, and BMI, were selected from a cohort of ~1000 children and adolescents with obesity (aged 8-16 years). Gene pathway analysis, literature search, and overlap with previous epigenome-wide association studies (EWAS) were performed to validate our results and assess their biological relevance. A causal mediation analysis was used to investigate if DNAm at any of the identified sites may mediate the association between birthweight and childhood IR. Replication in peripheral blood on an independent cohort of children and adolescents with obesity by Infinium MethylationEPIC v2.0 BeadChip is ongoing. Results: We identified 127 differentially methylated positions (DMPs), annotated to 89 unique genes, between cases (IR+) and controls (IR-) according to Bonferroni adjusted P-value <0.05 and 16,834 DMPs with FDR <1%. Most DMPs were hypermethylated in cases compared to controls. The 16,834 DMPs with FDR <1% were annotated to genes overrepresented in biological processes potentially involved in the development of IR, such as AGE-RAGE signaling pathway, MAPK signaling pathway, cytoskeleton organization, vesicle-mediated transport, and phosphatidyl-inositol signaling. A significant enrichment of previously identified DMPs in blood from EWAS on IR and incident type 2 diabetes (T2D) (e.g., LGALS3BP and MAN2A2) was found in our study. Furthermore, a consistent overlap was observed with the results of previous EWAS in relevant tissues for IR such as adipose tissue (e.g., PTPN22) and pancreatic islets (e.g., CTNNA2 and SLC8A1). Next, we identified an association between birthweight Z-score and IR in our cohort, independent of age, sex, and BMI (β= 0.262, P= 0.043), and a causal mediation analysis highlighted that DNAm at 25 DMPs (e.g., USP53 and DTNB) partially mediated the association between BW Z-score and IR (Proportion mediated average effect = 45%, P=0.034). Conclusion: These findings contribute to understanding the biological pathways implicated in early IR, confirming the plausible role of epigenetic programming of IR in early life. Further studies are needed to validate our findings in tissues relevant for IR and disclose the functional involvement of identified pathways in the pathogenesis of T2D. This could provide early blood-based epigenetic biomarkers of IR and novel potential therapeutic targets for early personalized prevention and treatment of IR and thus T2D from a precision medicine perspective.