Blood-based epigenome–wide analyses of chronic low–grade inflammation across diverse population cohorts

Authors: Robert F. Hillary, Hong Kiat Ng, Daniel L. McCartney, Hannah R. Elliott, Rosie M. Walker, Archie Campbell, Felicia Huang, Kenan Direk, Paul Welsh, Naveed Sattar, Janie Corley, Caroline Hayward, Andrew M. McIntosh, Cathie Sudlow, Kathryn L. Evans, Simon R. Cox, John C. Chambers, Marie Loh, Caroline L. Relton, Riccardo E. Marioni, Paul D. Yousefi, Matthew Suderman

Epigenome-wide association studies were performed on log-transformed blood C-reactive protein levels in Generation Scotland. Log-transformed CRP was the outcome. CpG beta-values were the exposure and were measured using the EPIC array (n=752,722 sites). 

A basic model adjusted for age, sex, experimental batch and Houseman-estimated white blood cell proportions. A fully-adjusted model further adjusted for alcohol consumption (units/week), log-transformed body mass index (kg/m2), education (11-category ordinal variable), deprivation (Scottish Index of Multiple Deprivation), EpiSmokEr (a methylation-based surrogate score for cigarette smoking) and 20 genetic principal components (to control for population structure).