Blood lipid–linked DNA methylation patterns replicate in Korean adults and tie to epigenetic aging

An epigenome-wide association study in Korean adults has confirmed that the best-established links between DNA methylation and blood lipid levels hold outside the European-ancestry populations where they were first described, and it adds a generational twist: the relationship between lipid traits and epigenetic aging measures differed across five generations of epigenetic clocks. The work, published in Clinical Epigenetics, analyzed peripheral whole-blood leukocyte DNA from 1,527 Korean adults, with replication in two non-overlapping subsets of the Korean Genome and Epidemiology Study health examinees (HEXA) cohort, comprising 389 and 417 individuals, all profiled on the Illumina EPIC array.

The study’s primary aim was to determine whether lipid–methylation loci established in earlier, predominantly European-ancestry research transfer to a Korean population. Its secondary aim was to characterize how lipid-specific epigenetic signatures relate to biological aging across clock generations. The researchers examined total cholesterol, log-transformed triglycerides, high-density lipoprotein (HDL), low-density lipoprotein, and hypertriglyceridemia, and evaluated associations with five epigenetic clocks.

The headline result concerns triglycerides. The team identified 1,081 differentially methylated positions associated with log-transformed triglycerides at a Bacon-corrected false discovery rate below 0.05, and seven positions associated with HDL. Critically, all five CHARGE consortium CpGs—located at or near CPT1A, TXNIP, ABCG1, SREBF1, and DHCR24—were confirmed in both replication subsets by meta-analysis, with P values ranging from 2.2 × 10−3 to 2.6 × 10−17.

Replication was not limited to those five sentinel sites. Across the wider discovery set, 89.8 percent of CpGs showed directional concordance in the primary replication subset, a binomial P value of 3.4 × 10−170, although only 24 reached individual meta-analysis significance. The authors also built a five-CpG methylation risk score, which explained 23.6 percent of additional triglyceride variance in the discovery set, 16.1 percent in HEXA-I, and 5.9 percent in HEXA-II—a pattern of shrinking explanatory power across subsets that the published results report directly.

On the aging side, the study observed a generational gradient in the associations between age acceleration and triglycerides, meaning that the strength of the link varied depending on which generation of epigenetic clock was used. Among the five clocks evaluated, DunedinPACE showed the broadest associations across lipid traits, with a log-triglyceride association at P = 4.0 × 10−7.

In mediation analysis comparing two alternative causal orderings linking epigenetic aging, methylation, and lipids, ABCG1 methylation accounted for 63 percent of the DunedinPACE–triglyceride association. Notably, that estimate was essentially unchanged when the causal ordering was reversed, so the analysis does not by itself resolve whether methylation drives the aging–lipid link or reflects it.

Methodological rigor received explicit attention. The authors assessed robustness through sensitivity analyses including surrogate variable adjustment, fixed-effect meta-analysis, and methylation risk scores. The abstract also states plainly that the study is an early-release, peer-reviewed accepted version subject to further edits before the final Version of Record.

The conclusions carry clear caveats. Because both replication subsets came from the same cohort program, and because methylation was measured in blood, the authors state that generalization to other populations and tissues requires further study. The findings support the conservation of core lipid–methylation loci across ancestries, implicate ABCG1 on the axis linking accelerated epigenetic aging to dyslipidemia, and describe a generational gradient in clock–lipid associations—but as an observational EWAS in blood, the work maps associations rather than demonstrating causal mechanisms in humans.

For the field, the study addresses a long-standing gap: East Asian data on lipid–methylation associations have been sparse, and the predominantly European-ancestry evidence base left open whether the same CpGs would appear in other populations. This Korean EWAS indicates that the core loci do transfer, while the mediation results position ABCG1 methylation as a candidate node connecting epigenetic aging measures to triglyceride levels—a hypothesis that further work in diverse cohorts and tissues will need to test.

Subject of Research: Epigenome-wide association study of blood lipid traits and epigenetic aging in a Korean population

Article Title: Epigenetic aging and lipid traits with a generational gradient in a Korean population with independent replication

Article References: Lee, J., Hwang, J.-T., Choi, H.-K., & Park, J.-H. (2026). Epigenetic aging and lipid traits with a generational gradient in a Korean population with independent replication. Clinical Epigenetics. https://doi.org/10.1186/s13148-026-02258-y

Image Credits: AI Generated

DOI: 10.1186/s13148-026-02258-y

Keywords: DNA methylation, epigenetic clocks, triglycerides, HDL, ABCG1, EWAS, Korean population, DunedinPACE, methylation risk score, hypertriglyceridemia, Clinical Epigenetics, replication