Depressive symptoms partially mediate the relationship between psychosocial factors and epigenetic age acceleration in a multi-racial/ethnic sample of older adults

Abstract

Psychosocial factors, including cumulative psychosocial stress and loneliness, have been linked to epigenetic aging in older adults. Further, depressive symptoms have established relationships with both psychosocial factors and epigenetic aging. However, it is not known whether depressive symptoms mediate the association between psychosocial factors and epigenetic aging.We conducted linear regression models to examine associations between psychosocial stress, loneliness, and depressive symptoms and five epigenetic age acceleration (AA) measures estimated by DNA methylation in a multi-racial/ethnic sample of 2681 older adults from the Health and Retirement Study (mean age: 70.4 years). For all identified associations, we tested for effect modification by sex and educational attainment and performed mediation analysis to characterize the role of depressive symptoms on these associations.Psychosocial stress, loneliness, and depressive symptoms were each associated with at least one measure of epigenetic AA (FDR q < 0.05). Further, we observed interactions between loneliness, psychosocial stress, and sex on DunedinPACE, as well as loneliness and educational attainment on GrimAA, PhenoAA, and DunedinPACE, with females and individuals without a college degree appearing more sensitive to the psychosocial effects on epigenetic aging. Depressive symptoms mediated between 24 % and 35 % of the relationships between psychosocial stress and HannumAA, GrimAA, and DunedinPACE, as well as 40 % and 37 % of the relationships between loneliness and both GrimAA and DunedinPACE, respectively.

Results: from this study may help elucidate the relationship between psychosocial factors and epigenetic aging, which is critical in understanding the biological mechanisms through which psychosocial factors may contribute to age-related disease.

Keywords: Depressive symptoms; Epigenetic aging; Epigenetic clocks; Loneliness; Psychosocial stress; Social epigenomics.

Fig. 1

Plots of significant interactions between psychosocial factors and sex on DunedinPACE (p < 0.05). Predicted DunedinPACE values for (A) cumulative psychosocial stress z-scores and (B) loneliness z-scores by sex are shown with 95 % confidence intervals, which intend to capture the true line in the population 95 % of the time across repeated samples. Interaction models: DunedinPACE ∼ psychosocial factor + sex + educational attainment + marital status + employment + has child + top 10 genetic ancestry PCs + WBC proportion + year of psychosocial battery + psychosocial factor∗sex.

Fig. 2

Plots of significant interactions between loneliness and educational attainment on epigenetic age acceleration measures (p < 0.05). Predicted (A) PhenoAA, (B) GrimAA, and (C) DunedinPACE values for loneliness z-scores by educational attainment (no college degree vs. college degree or higher) are shown with 95 % confidence intervals. Interaction models: EAA measure ∼ loneliness z_score + sex + educational attainment + marital status + employment + has child + top 10 genetic ancestry PCs + WBC proportion + year of psychosocial battery + loneliness z_score∗educational attainment.