Modifiable traits, healthy behaviours, and leukocyte telomere length: a population-based study in UK Biobank

Abstract

Background: Telomere length is associated with risk of several age-related diseases and cancers. We aimed to investigate the extent to which telomere length might be modifiable through lifestyle and behaviour, and whether such modification has any clinical consequences.

Methods: In this population-based study, we included participants from UK Biobank who had leukocyte telomere length (LTL) measurement, ethnicity, and white blood cell count data. We investigated associations of LTL with 117 potentially modifiable traits, as well as two indices of healthy behaviours incorporating between them smoking, physical activity, diet, maintenance of a healthy bodyweight, and alcohol intake, using both available and imputed data. To help interpretation, associations were summarised as the number of equivalent years of age-related change in LTL by dividing the trait β coefficients with the age β coefficient. We used mendelian randomisation to test causality of selected associations. We investigated whether the associations of LTL with 22 diseases were modified by the number of healthy behaviours and the extent to which the associations of more healthy behaviours with greater life expectancy and lower risk of coronary artery disease might be mediated through LTL.

Findings: 422 797 participants were available for the analysis (227 620 [53·8%] were women and 400 036 [94·6%] were White). 71 traits showed significant (p<4·27 × 10^-4) associations with LTL but most were modest, equivalent to less than 1 year of age-related change in LTL. In multivariable analyses of 17 traits with stronger associations (equivalent to ≥2 years of age-related change in LTL), oily fish intake, educational attainment, and general health status retained a significant association of this magnitude, with walking pace and current smoking being additionally significant at this level of association in the imputed models. Mendelian randomisation analysis suggested that educational attainment and smoking behaviour causally affect LTL. Both indices of healthy behaviour were positively and linearly associated with LTL, with those with the most healthy behaviours having longer LTL equivalent to about 3·5 years of age-related change in LTL than those with the least heathy behaviours (p<0·001). However, healthy behaviours explained less than 0·2% of the total variation in LTL and did not significantly modify the association of LTL with risk of any of the diseases studied. Neither the association of more healthy behaviours on greater life expectancy or lower risk of coronary artery disease were substantially mediated through LTL.

Interpretation: Although several potentially modifiable traits and healthy behaviours have a quantifiable association with LTL, at least some of which are likely to be causal, these effects are not of a sufficient magnitude to substantially alter the association between LTL and various diseases or life expectancy. Attempts to change telomere length through lifestyle or behavioural changes might not confer substantial clinical benefit.

Funding: UK Medical Research Council, UK Biotechnology and Biological Sciences Research Council, and British Heart Foundation.

Figure 1

A circular plot showing traits nominally associated with leukocyte telomere length For each trait, the p value shown is from a multivariable linear regression model adjusted for age, sex, ethnicity, and white blood cell count. For categorical traits, the global p value from a likelihood ratio test is shown. Bonferroni significant traits (p<4·27 × 10^–4) are in purple text. Nominally significant traits (p<0·05) are in black text. Non-significant traits (p>0·05) are shown in the appendix (pp 14–20). eGFR=estimated glomerular filtration rate. HbA_1c=glycated haemoglobin. MET=metabolic equivalents.

Figure 2

Individual traits most strongly associated with LTL The traits are shown on the y-axis and the β coefficients for the associations LTL on the x-axis. Error bars indicate 95% CIs. Effect in years is the ratio of the trait β coefficient and the absolute value of the age β coefficient (–0·023). Continuous traits are estimated for a 1 SD increase in the trait. Binary traits compare yes with no. Poor overall health rating is compared with excellent. Current smoking is compared with never. Always added salt is compared with never or rarely. Soft (tub) margarine is compared with olive-oil spread. Always standing job or usually manual job is compared with never or rarely. Processed meat, cheese, or oily fish intake (five or more times per week) is compared with never. Usually on a shift is compared with never. Muesli cereal intake is compared with other types. Wholemeal bread intake is compared with white bread. Brisk walking pace is compared with slow. University or college degree is compared with no qualifications. LTL=leukocyte telomere length.

Figure 3

Association between the number of components of the primary healthy behaviour index and leukocyte telomere length In the base model adjustments were made for age, sex, ethnicity, and white blood cell count, and in the final model additional adjustments were made for self-reported diagnosed by doctor of chronic medical conditions (diabetes, cancer, hypertension, and vascular disease), insomnia, fed-up feelings, LDL cholesterol, C-reactive protein, estimated glomerular filtration rate, and educational attainment. Error bars represent the 95% CIs.

Figure 4

Mendelian randomisation analysis The plots show the results of the bidirectional mendelian randomisation analysis of the associations between LTL and years spent in education; initiation of regular smoking; and smoking intensity. Regression coefficients were derived from the inverse-weighted variance mendelian randomisation method. Error bars represent the 95% CIs. LTL=leukocyte telomere length.