# Associations of DNA methylation estimators of protein abundance with concurrent and future physical health risk factors

## TL;DR

This study explores how DNA methylation-based protein abundance estimates relate to physical health traits in children and young adults, finding potential associations.

## Contribution

The study introduces episcores as a novel tool for identifying protein-phenotype associations in populations without direct protein measurements.

## Key findings

- 9 cross-sectional and 11 prospective phenotype-episcore associations were identified.
- CHIT1 and SEMA3E episcores showed multiple associations with physical health traits.
- No causal effects were found using Mendelian randomisation between proteins and phenotypes.

## Abstract

DNA methylation (DNAm) is an epigenetic modification which plays a role in gene regulation and has genetic and environmental influences. Recently, DNAm-based models of protein abundance (termed episcores) have been developed and were found to be associated with incident disease in older adults. Here, we ask if these episcores are associated with latent physical health phenotypes in children and young adults in the Avon Longitudinal Study of Parents and Children (ALSPAC) cohort. Episcores were projected in ALSPAC participants who had DNAm measurements in cord blood, and peripheral blood at ages seven, nine, 17, and 24 (n = 192–2857). We analysed cross-sectional associations between 108 episcores and 17 physical health phenotypes, followed by an examination of prospective associations between episcores and the same phenotypes measured 2 + years after the blood samples used for episcore calculation. Two-sample Mendelian randomisation (2SMR) was then used to evaluate evidence for causal relationships between the underlying proteins and any associated physical health phenotypes. Of the associations tested between 17 physical health phenotypes and 108 episcores at multiple timepoints, 9 cross-sectional (CHIT1 is associated with 8 of these) and 11 prospective (SEMA3E is associated with 7 of these) phenotype-episcore associations were discovered. Of these, no 2SMR analyses suggested a causal effect of a protein on its related phenotype. We find evidence to suggest that episcores may be useful for discovering protein-phenotype associations in populations lacking direct measurements of protein abundance.

The online version contains supplementary material available at 10.1038/s41598-025-31843-z.