Candidate DNA methylation markers for gestational diabetes: findings from technical and biological validation in a South African population

Abstract

Gestational diabetes mellitus (GDM) increases risk of adverse maternal and fetal outcomes. Using Genome-wide Beadchip arrays, we previously reported differential methylation of 1046 CpG sites in women with GDM. This study aimed to validate three candidate CpG sites in an independent cohort of South African women with and without GDM. DNA methylation was assessed in maternal peripheral blood collected at < 28 weeks of gestation, at three top-most significant CpG sites within Calmodulin Binding Transcription Activator 1 (CAMTA1), Kelch domain-containing protein 3/ Male-Enhanced Antigen 1 (KLHDC3/MEA1), and Collectin Subfamily member 10 (COLECT10), using bisulfite pyrosequencing, in a technical validation group (12 GDM and 12 non-GDM), and an independent biological validation group (46 GDM and 135 non-GDM). Associations between differentially methylated CpGs and maternal clinical characteristics were assessed. Technical validation confirmed decreased methylation at cg23643951 (CAMTA1; -1.53-fold; p = 0.04) and cg21910650 (KLHDC3/MEA1; -1.57-fold; p = 0.004), and increased methylation at cg16306629 (COLECT10; 4.5-fold; p = 0.003) in women with GDM, consistent with the original array findings. No significant differences were observed in the biological validation group. Within the technical validation group, cg21910650 methylation was inversely correlated with fasting glucose (rho=-0.625; p = 0.015) and 1 h OGTT (rho=-0.555; p = 0.034), while cg16306629 was positively correlated with fasting glucose (rho = 0.709; p = 0.006). Technical validation confirmed differential methylation at three CpG sites, though these findings were not reproduced in the larger biological validation group, highlighting the need for validation in independent cohorts. Associations between CpG methylation with glucose concentrations suggest a potential association with glucose regulation during pregnancy, providing a foundation for future validation of these CpG sites as biomarkers for GDM. Technical validation confirmed decreased methylation at cg23643951 within CAMTA1 and cg21910650 within KLHDC3/MEA1, and increased methylation at cg16306629 nearest to COLECT10 in women with GDM in this population, consistent with original array findings Two of these CpG sites were associated with glucose concentrations, suggesting a possible role in glucose tolerance during pregnancy. These findings were not reproduced in the larger independent biological validation cohort, highlighting challenges in translating methylation findings into clinically useful biomarkers. This study provides a foundation for future research to validate the clinical utility of these three CpG sites as biomarkers for GDM. Technical validation confirmed decreased methylation at cg23643951 within CAMTA1 and cg21910650 within KLHDC3/MEA1, and increased methylation at cg16306629 nearest to COLECT10 in women with GDM in this population, consistent with original array findings Two of these CpG sites were associated with glucose concentrations, suggesting a possible role in glucose tolerance during pregnancy. These findings were not reproduced in the larger independent biological validation cohort, highlighting challenges in translating methylation findings into clinically useful biomarkers. This study provides a foundation for future research to validate the clinical utility of these three CpG sites as biomarkers for GDM.