Beyond the mean: quantile regression to explore the air pollution effect on gene-specific DNA methylation in the Normative Aging Study

摘要

BACKGROUND: Air pollution has been related to mean changes in outcomes, including DNA methylation. Mean regression analyses may not capture associations that occur primarily in the tails of the outcome distribution, and can miss associations primarily in susceptible groups. OBJECTIVES: This study examines whether air pollution affects the DNA methylation distribution differently across quantiles. We focus on methylation on genes related to coagulation and inflammation: coagulation factor Ⅲ (F3), intercellular adhesion molecule 1 (ICAM-1), toll-like receptor 2 (TRL-2), interferon gamma (IFN-γ), interleukin 6 (IL-6). METHODS: We measured gene-specific blood DNA methylation in 777 elderly men participating in the Normative Aging Study (1999-2009). We fit quantile regressions for longitudinal data to investigate whether the intermediate-term air pollution effect (4-weeks moving average) on DNA methylation (expressed in %5mC) varied across quantiles of the DNA methylation distribution. RESULTS: We observed significant effects of PM2.5, particle number, and ozone on the 10th to 50th percentiles of ICAM-1, IFN-γ, and IL-6 DNA methylation, while there was no significant air pollution effect on the 70th to 90th quantiles. An interquartile range increase in PM2.5 was associated with a 1.28%5mC (95%CI: [0.42; 2.14]) decrease on the 20th quantile of IFN-γ methylation, and was not significantly related to the 80th quantile (Estimate: 0.03%5mC, 95%CI: [-0.48; 0.54]). The effect of ultrafine particles captured by particle number was stronger on the 50th to 90th quantiles of F3 methylation, compared to the lower quantiles. CONCLUSIONS: We found that primary and secondary air pollutant effects were different across quantiles of the DNA methylation distribution, which would not have been detected by mean regression. These findings may suggest that ambient air pollution impacts systemic inflammation and links to disease exacerbation by low ICAM-1, IFN-γ, and IL-6 methylation, as well by high F3 methylation.