Title: Additive interaction and mediation-interaction decomposition: DNA methylation age acceleration, education, and cognitive impairment in the Health and Retirement Study
Ware EB, Higgins C, Harris S, Fisher JD, Bakulski KM. Additive interaction and mediation-interaction decomposition: DNA methylation age acceleration, education, and cognitive impairment in the Health and Retirement Study. In review.
This Github repository contains the data management and analytic scripts to produce the following manuscript:DNA methylation age acceleration, education, and cognitive impairment in the Health and Retirement Study
Dementia represents a significant and increasing public health burden. DNA methylation age acceleration may be associated with dementia and dementia risk factors, such as education, but investigating their impact on dementia is necessary. Objective: To evaluate the association of educational attainment on dementia and cognitive impairment through DNA methylation age acceleration, while accommodating exposure-mediator interaction effects.
In the 2016 Health and Retirement Study wave, we evaluated six epigenetic clocks, including GrimAge, with Langa-Weir classified dementia, cognitive impairment non-dementia, and normal cognition. Age acceleration was the residual between chronological age of participants and DNA methylation clock, dichotomized at zero. To understand the joint association of low education (≤12 years) and DNA methylation age acceleration in relation to cognitive impairment, we used weighted logistic regression and calculated interaction on the additive scale adjusting for chronological age, sex, race/ethnicity, and cell type composition. We performed four-way mediation and interaction decomposition analysis to estimate the: 1) controlled direct effect of education on cognition, 2) interaction reference, 3) interaction mediation, and 4) the pure indirect effect of DNA methylation age acceleration on cognition.
Analysis was conducted on a subsample of Health and Retirement Study participants in the 2016 Venous Blood Study (N=3,724).
Both GrimAge acceleration (OR=1.6 95%CI 1.3 – 2.1) and low educational attainment (OR= 2.4 95%CI 1.9 – 3.0) were associated with higher odds of cognitive impairment, non-dementia in a mutually adjusted logistic model. We found additive interaction associations between low education and GrimAge acceleration on dementia. We observed that 6-8% of the association of education on dementia was mediated through GrimAge acceleration. While mediation effects were small, the portion of the association of education due to additive interaction with GrimAge acceleration was between 23.6 and 29.2%.
Accelerated DNA methylation age was associated with increased odds of cognitive impairment and we observed more than additive interaction effects between education and age acceleration on dementia. These results support the interplay of social disadvantage and biological aging processes on impaired cognition.
This work was supported by grants from the National Institute on Aging (R01 AG055406, R01 AG067592, P30 AG072931, R01 AG055654, R01 AG067592-01S1, R01AG06759201, R25AG053227); and the National Institute of Minority Health and Disparities (R01 MD013299); and the National Institute of Environmental Health Sciences (P42ES017198, P30ES017885); and the National Center for Advancing Translational Sciences (UL1TR002240).
This repo contains the STATA script used for data management, and data analysis of the manuscript. We used STATA for our main models and analysis of measures of interaction in the additive scale, and mediation-interaction decomposition analysis. We built our own code in STATA to calculate measures of interaction in the additive scale and the mediation-interaction decomposition analysis. We decided not rely on existing packages (i.e., STATA med4way: https://pubmed.ncbi.nlm.nih.gov/30452641/) because this package does not accommodate for survey weights.
Health and Retirement Study survey data are publicly available (https://hrs.isr.umich.edu/data-products), and genetic data are available through dbGaP (https://dbgap.ncbi.nlm.nih.gov; phs000428.v2.p2) Original Article
A preprint version of this article can be found at medRxiv (https://www.medrxiv.org/content/10.1101/2022.12.23.22283919v1)
Export File Tables: including the tables related to this project
Stata Script: including the STATA .do file which contains code on creating epigenetic clock age acceleration variables and tables