Telomere length and aging-related outcomes in humans: A Mendelian randomization study in 261,000 older participants. Chia-Ling Kuo Luke C. Pilling George A. Kuchel Luigi Ferrucci David Melzer. Aging Cell, August 24 2019. https://doi.org/10.1111/acel.13017
Abstract: Inherited genetic variation influencing leukocyte telomere length provides a natural experiment for testing associations with health outcomes, more robust to confounding and reverse causation than observational studies. We tested associations between genetically determined telomere length and aging‐related health outcomes in a large European ancestry older cohort. Data were from n = 379,758 UK Biobank participants aged 40–70, followed up for mean of 7.5 years (n = 261,837 participants aged 60 and older by end of follow‐up). Thirteen variants strongly associated with longer telomere length in peripheral white blood cells were analyzed using Mendelian randomization methods with Egger plots to assess pleiotropy. Variants in TERC, TERT, NAF1, OBFC1, and RTEL1 were included, and estimates were per 250 base pairs increase in telomere length, approximately equivalent to the average change over a decade in the general white population. We highlighted associations with false discovery rate‐adjusted p‐values smaller than .05. Genetically determined longer telomere length was associated with lowered risk of coronary heart disease (CHD; OR = 0.95, 95% CI: 0.92–0.98) but raised risk of cancer (OR = 1.11, 95% CI: 1.06–1.16). Little evidence for associations were found with parental lifespan, centenarian status of parents, cognitive function, grip strength, sarcopenia, or falls. The results for those aged 60 and older were similar in younger or all participants. Genetically determined telomere length was associated with increased risk of cancer and reduced risk of CHD but little change in other age‐related health outcomes. Telomere lengthening may offer little gain in later‐life health status and face increasing cancer risks.
1 INTRODUCTION
Telomeres are end fragments of chromosomes consisting of thousands of repeats of the noncoding sequence TTAGGG. Telomeres function to protect chromosome ends against genomic instability. Telomeres shorten with each cell cycle and contribute to replicative senescence when reaching the Hayflick limit (Hayflick & Moorhead, 1961). Telomerase is a ribonucleoprotein complex, which replenishes telomere loss during replication. Telomerase is active at early developmental stages but almost completely inactive in somatic tissues of adults (Collins and Mitchell, 2002). Telomerase activation may treat aging‐related diseases and prolong human lifespan (de Jesus & Blasco, 2013). Previous studies on adult or old mice have shown successes from improving physical function and lifespan without increasing incidence of cancer, but the translation from mice to humans is unknown (de Jesus & Blasco, 2013).
Telomere length is often approximated using leukocyte telomere length, which is easy to extract from blood and highly correlated with telomere length in other tissues (Daniali et al., (2013)). Measured telomere length has been associated with mortality and aging‐related outcomes in humans (Mather, Jorm, Parslow, & Christensen 2011; Sanders & Newman, 2013; Brown, Zhang, Mitchel, & Ailshire, 2018), including cancer (Zhang et al., 2017), cardiovascular disease (Haycock et al., 2014), cognitive function, physical performance such as grip strength, sarcopenia, and frailty (Lorenzi et al., 2018; Zhou et al., 2018), plus biomarkers of lung function, blood pressure, bone mineral density, cholesterol, interleukin 6, and C‐reactive protein. Observational associations cannot be consistently replicated likely due to study populations, measurement methods, and statistical modelling (Sanders & Newman, 2013). In addition, a number of factors may confound observational associations such as sex and race/ethnicity, paternal age at birth, smoking, psychological stress, and other psychosocial, environmental, and behavioral factors (Blackburn, Epel, & Lin, 2015; Starkweather et al., 2014).
Telomere length has a strong inherited genetic component in humans (heritability estimates ranging from 34% to 82% (Broer, Codd, & Nyholt 2013). Mendelian randomization (MR) is a powerful statistical method to evaluate the causal relationship between an exposure and an outcome, under certain assumptions (Davey Smith & Hemani, 2014). Analogous to randomized clinical trials, MR creates groups determined by genotypes, which are inherited at random and are independent of confounding factors. In theory, if the groups are associated with the outcome, the association is independent of confounders and is via the exposure, assuming no pleiotropy is present. MR studies are more robust than observational studies to confounding effects, measurement errors or bias, and reverse causation (i.e., free of downstream effects appearing to be causes).
By applying MR, we were able to study the effect of telomere length on aging, with robustness to confounding effects. To date, 16 inherited genetic variants from genome‐wide association studies (GWAS) have been shown to be strongly associated with human leukocyte telomere length using European‐descent population samples (Haycock et al., 2017). Many of these loci harbor telomerase and telomere‐protective protein genes, including TERC, TERT, NAF1, OBFC1, and RTEL1 (Codd et al., 2013; Haycock et al., 2017). These variants have been used to perform MR, but the focus was on diseases (Haycock et al., 2017; Zhan et al., 2015). Additionally, previous studies tend to be underpowered due to an insufficiently large sample size for a small percent of variance (2%–3%) explained by the genetic variants (Haycock et al., 2017). The small percent of variance affects the power but not validity of the causal inference, if the genetic variants meet the Mendelian randomization assumptions: (a) associated with telomere length, (b) independent of all confounders for the association between telomere length and the outcome, and (c) independent of the outcome conditional on telomere length and all the confounders (Haycock et al., 2017).
In this study, we investigated causal relationships between telomere length and aging‐related outcomes with the focus on common measures of human aging such as grip strength, frailty, and cognitive function. We analyzed European‐descent participants from UK Biobank, with a wealth of genetic and phenotypic data. This study was not designed to analyze every aging trait in UK Biobank. Instead, we selected traits to cover different aspects of aging, using inputs from senior investigators in the team. Cancer, coronary heart disease, hypertension, and pneumonia were selected as they were common in older adults, but we did not attempt to include every individual disease. Disease‐specific MR associations were reported elsewhere (Haycock et al., 2017). Our project is focused on aging traits and is not powered for diseases that require a longer time to accumulate sufficient cases.