Genetic influences on the association between DNA methylation and obesity measures: insights from a twin study design

Abstract

Background: Both obesity and DNA methylation (DNAm) are influenced by genetic factors. Despite more than a thousand of obesity-related DNAm sites (CpGs) being identified, studies that account for genetic influences in these associations are limited. Results: Using data from 1,074 twins in the Chinese National Twin Registry and bivariate structural equation models (SEMs), we investigated the phenotypic (Rph), genetic (Ra), and environmental (Re) correlations between genome-wide DNAm and three obesity indices: BMI, waist circumference (WC), and waist-to-hip ratio (WHR). Genome-wide, correlations between DNAm and obesity were small (Rph = 0.04, Ra = 0.08–0.09, Re = 0.02–0.03). For CpGs with high phenotypic correlation (Rph > 0.1), the mean genetic and environmental correlations were 0.23–0.24 and 0.03–0.05, respectively, indicating significant genetic influence on the DNAm-obesity associations. To further investigate the role of genetic influences, we then categorized the CpGs into different groups: high phenotypic correlation (Rph ≥ 0.2); high phenotypic and genetic correlations (Rph > 0.1 and Ra > 0.5); high phenotypic and low genetic correlations (Rph > 0.1 and Ra < 0.5). Association studies were conducted in the full population and in the monozygotic (MZ) twin-paired design, where genetic influences were controlled. For CpGs with Rph ≥ 0.2, 9, 8, and 22 were associated with BMI, WC, and WHR in the full population, but only 6, 1, and 1 CpGs remained significant after controlling for genetic effects in MZ twin-pair analyses. For CpGs with Rph > 0.1 and Ra > 0.5, genetic factors predominantly drove the association, and none of the 155/155/189 CpGs associated with BMI/WC/WHR in the full population were significant in MZ-paired analyses. For CpGs with Rph > 0.1 and Ra < 0.1, genetic effects were minimal or confounding, with 89, 4, and 17 significant in both full population and MZ-paired analyses. Conclusions: Our results highlight the significant genetic influences on the DNAm-obesity relationships, which may explain the low replicability of obesity-related DNAm markers. This indicates that genetic influences should be carefully considered in DNAm-related studies.