Diet, lifestyle and telomere length: using Copula Graphical Models on NHANES data

Abstract

Telomere length has been related to human health and ageing in multiple studies. However, these studies have analyzed a small set of variables, according to pre-formulated hypotheses. We used data from NHANES 1999-2002 to perform a preregistered cross-sectional analysis. From these four years we selected the participants with available leukocyte telomere length measure and with plausible daily energy intake, leading to a total study population of 7096 participants. Then, we divided the participants in three groups according to age: Young 20-39 (n = 2623), Middle 40-59 (n = 2210), Old 60-84 (n = 2263). On each group we performed Copula Graphical Modelling (CGM) to capture the links between the variables of interest, and we conducted certainty and sensitivity analyses to understand the robustness of the results. Blood levels of C-reactive protein and γ-tocopherol, and intake of caffeine and fibers are inversely related to telomere length across the age strata. Sex, race, smoking, physical activity and indicators of socioeconomic status have almost no direct connection with telomeres; however, they are directly linked to C-reactive protein, which in turn is connected to leukocyte telomere length. C-reactive protein is therefore a possible central mediator of the effect of these factors on telomeres.