Telomere length in circulating leukocytes is associated with lung function and disease

Introduction

An increasing life expectancy is associated with an increased susceptibility for prevailing chronic diseases, such as cardiovascular and neurological diseases, type 2 diabetes and cancer, suggesting that common, age-related processes are involved in these diseases [1–4]. The lung is known to undergo age-related processes associated with a continuous decline in lung function [5, 6]. Chronic obstructive pulmonary disease (COPD), lung cancer and idiopathic pulmonary fibrosis are lung diseases with an increase in incidence by age. In COPD, chronic inflammation and oxidative stress are among the generally accepted pathophysiological mechanisms but, as discussed for other degenerative diseases, accelerated ageing resulting in premature cell senescence may contribute to the pathobiology [7–9]. Although clinical and epidemiological studies provide some evidence for this hypothesis, the data base is still limited [8], so our study aimed to provide further clues on this.

Telomeres, made up of the simple repetitive sequence TTAGGG in humans [10], are specialised DNA structures that, in concert with telomere-specific proteins, protect the ends of chromosomes and maintain the integrity of the genome [3]. They progressively shorten with each cell division [11] and, after reaching a critical length, either cellular senescence or apoptosis is induced [3]. Telomeres are therefore considered a marker of biological age. Their length is affected by various factors, such as chronic inflammation or oxidative stress [3], but has also been shown to have a heritable component [12]. Recent data suggest that shorter telomere length is associated with an increased risk of total and cancer mortality in COPD patients [13]. Shortened telomeres were found in pulmonary vascular endothelial cells, alveolar epithelial cells and circulating leukocytes of COPD patients [14–18], supporting the notion of accelerated ageing in COPD. Rode et al. [18] reported multivariable-adjusted odds ratios of 1.15 (95% CI 1.06–1.25) for shortest versus longest telomere quartiles for COPD. Smoking, a major cause of COPD, is associated with different signs of premature ageing, which is particularly obvious in the skin [8], suggesting that environmental exposure may play an important role in triggering processes related to biological ageing.

Since the lung is continuously exposed to environmental hazards, lung function per se may be a surrogate marker for biological age in light of the large interindividual variability observed [19]. Only a few case–control studies [14, 20] have investigated the relationship between telomere length and spirometric indices. One large population-based study was recently published by Rode et al. [18] reporting a modest correlation between telomere length and lung function in terms of the spirometric indices forced expiratory volume in 1 s (FEV₁), forced vital capacity (FVC) and their ratio, FEV₁/FVC.

In a meta-analysis of 14 European studies embedded in the ENGAGE (European Network for Genetic and Genomic Epidemiology) consortium, we studied the relationship between telomere length, lung function and lung disease to address the question of whether accelerated ageing, indicated by telomere length, may be involved in the pathogenesis of COPD and asthma, and whether intrinsic age-related processes may contribute to the interindividual variability of lung function. For this purpose, we explored the association between circulating leukocyte telomere length and COPD or asthma. Although these are different obstructive disease entities, they share chronic inflammation as a hallmark of disease, which may result in accelerated ageing and telomere shortening. We further investigated the association between telomere length and lung function using the spirometric indices FEV₁, FVC and FEV₁/FVC.

Results

COPD

By combining nine studies including 934 COPD cases and 15 846 controls according to the GLI criteria (table 1), we observed a significant negative association between COPD and telomere length (β= −0.0982, p=0.001 with fixed effects; β= −0.0953, p=0.012 with random effects, I²=27.7%, p-value for heterogeneity (p_het)=0.198). With COPD defined by the GOLD criteria, there were more COPD cases (n=1189) with an increase in older and a reduction in younger individuals. On the other hand, the association with telomere length using GOLD criteria was more heterogeneous across studies and less significant (β= −0.0676, p=0.018 with fixed effects; β= −0.0629, p=0.117 with random effects; I²=42.3%, p_het=0.085). Figure 1 shows study-specific and combined effect estimates of a fixed- and random-effects model with COPD defined by the GLI criteria. The association was strong in the female subsample (n=10 435; β= −0.1306, p=3.6×10⁻⁴ with fixed effects; β= −0.1312, p=6.7×10⁻⁴ with random effects; I²=5.8%, p_het=0.385) whereas in the male subsample, no significant relationship was detectable (n=6345; β= −0.0242, p=0.663 with fixed effects; β= −0.0329, p=0.622 with random effects; I²=23.5%, p_het=0.250). Online supplementary figure E1 shows the meta-analysis stratified for sex. The two studies showing positive effect estimates, EX-ATHLETES and NAG-FIN, are selective in their study design: NAG-FIN comprises families of heavy ever-smokers, and the EX-ATHLETES are male former elite athletes and matched controls with a mean age of 72.7 years. Since COPD shows an age-related increase in incidence, we conducted sensitivity analyses to investigate whether the observed association is affected by age. Exclusion of subjects >70 or >60 years of age showed no significant differences for the effect estimates in subgroups of both sexes (fig. E2), confirming that the association is not driven by older age groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Associations between telomere length and chronic obstructive pulmonary disease (COPD). Forest plot comparing effects between studies, and showing combined effects in a fixed-effects model as well as in a random-effects model. Sex-stratified results are based on random-effects models. 95% confidence intervals are given for all estimates. COPD was defined by the Global Lungs Initiative criteria. FITSA: Finnish Twin Study on Ageing; KORA: Cooperative Health Research in the Region Augsburg; NAG-FIN: Nicotine Addiction Genetics Finland; NFBC: Northern Finland Birth Cohort Study.

Asthma

By combining 13 studies with 2834 asthma cases and 28 195 controls (table 2 and fig. 2), we observed a negative association between telomere length and asthma. This was significant in a fixed-effects model (β= −0.0452, p=0.024) and suggestive in a random-effects model (β= −0.0446, p=0.093; I²=30.4%, p_het=0.140). The effect was driven by the female subsample (n=18 724, β= −0.0764, p=0.002 with fixed effects; β = −0.0809, p=0.005 with random effects; I²=15.8%, p_het=0.289) and not seen in the male subsample (n=12 305; β=0.0182, p=0.594 with fixed effects; β=0.0139, p=0.771 with random effects; I²=40.3%, p_het=0.080; fig. E3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Association between telomere length and asthma. Forest plot comparing effects between studies, and showing combined effects in a fixed-effects model as well as in a random-effects model. Sex-stratified results are based on random-effects models. 95% confidence intervals are given for all estimates. ERF: Erasmus Rucphen Family; FITSA: Finnish Twin Study on Ageing; KORA: Cooperative Health Research in the Region Augsburg; NAG-FIN: Nicotine Addiction Genetics Finland; NFBC: Northern Finland Birth Cohort Study; NTR: Netherlands Twin Register; QIMR: Queensland Institute of Medical Research.

Lung function

By combining the data of seven studies, totalling 12 595 individuals (table 3 and fig. 3), we observed highly significant positive associations between telomere length and spirometric indices measured as FEV₁ (β=0.0455, p=1.07×10⁻⁷ with fixed and random effects; I²=0%), FVC (β=0.0401, p=2.07×10⁻⁵ with fixed and random effects; I²=0%) and FEV₁/FVC (β=0.0238, p=5.27×10⁻³ with fixed and random effects; I²=0%). The effect for FEV₁ was higher in females than in males (β=0.0548 versus β=0.286 with random effects). The meta-analysis stratified by sex is shown in figure E4.

View this table:

• View inline
• View popup

Table 3– Characteristics of studies in lung function index analysis

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Telomere length is associated with a) forced expiratory volume in 1 s (FEV₁), b) forced vital capacity (FVC) and c) FEV₁/FVC. Forest plots comparing effects between studies, and showing combined effects in fixed-effects models as well as in random-effects models. Sex-stratified results are based on random-effects models. 95% confidence intervals are given for all estimates. FITSA: Finnish Twin Study on Ageing; KORA: Cooperative Health Research in the Region Augsburg; NFBC: Northern Finland Birth Cohort Study.

Figure 4 shows the association results stratified by lung health status. Compared with “healthy” controls, we found somewhat stronger effects in the disease subgroups of COPD and asthma cases. Comparisons of the effect estimates stratified by health status and by sex are presented in figure E5. The stronger effects found in COPD were predominantly seen in males and were less obvious in females. The opposite was observed for asthma, where the effects were strong in females. With regard to smoking, within the healthy subgroups, the observed effects were stronger in “healthy never-smokers” than in “healthy ever-smokers” for FEV₁ and FVC (fig. E6). To further explore the contribution of smoking, we stratified the group of healthy smokers (current or ex-smokers) by lung function status (lower 25% or upper 25% within this subgroup). The results suggest that there was no difference between those two groups for FEV₁. In FVC, an effect was only seen in individuals with poor lung function (lower 25%). Although differences between subgroups were noticeable, they did not differ significantly.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4–

Associations between telomere length and a) forced expiratory volume in 1 s (FEV₁), b) forced vital capacity (FVC) and c) FEV₁/FVC, stratified by health status. Forest plots showing effects in the overall sample as well as in “apparently lung-healthy” individuals, asthmatics and patients with chronic obstructive pulmonary disease (COPD) only to test the impact of lung disease on the effect. Effects are based on random-effects models. 95% confidence intervals are given for all estimates.

Discussion

We conducted a large study investigating the relationship between telomere length and lung function and disease, assessed in multiple European cohorts unselected for lung disease. We found highly significant positive associations between telomere length from peripheral leukocytes and spirometric indices measured as FEV₁, FVC and FEV₁/FVC in a meta-analysis of seven studies with 12 595 individuals. When stratifying by lung health status, larger effects were seen in the subgroups of COPD patients or asthmatics. This suggests that lung function decline, in part, reflects biological ageing due to intrinsic processes, which are likely to be aggravated in asthma or COPD due to mechanisms promoting cell senescence.

Tobacco smoke is a well-known environmental factor that promotes ageing processes and cellular senescence [8]. The recent large, population-based study by Rode et al. [18] revealed that having ever smoked and the cumulative pack-years smoked were associated with telomere length in 46 396 adults. This is in line with previous population-based findings [26]. However, in clinical studies, these effects were not seen [14, 17, 20]. When stratifying by smoking status, we observed no evident associations between telomere length and lung function indices in ever-smokers. In addition, stratifying by lung function within the group of healthy smokers (current and ex-smokers) did not clarify the issue, although for FVC, an effect could be detected in smokers of the lower quartile. The underlying reason for not detecting an effect may be found in the complex interactions between smoking and genetic and/or lifestyle factors, such as dietary antioxidants or physical activity, affecting biological age and its markers [1, 3, 27–32], but also in the heterogeneity of smoking exposure, and its amount and duration.

Up to now, a limited number of studies have addressed the association between lung function and telomere length. Mui et al. [20] observed a positive association in a sample of COPD patients and Tsuji et al. [16] reported that airflow limitation and alveolar cell senescence were positively correlated with each other. Other studies failed to find significant associations between spirometric parameters and telomere length [14, 17, 33], although one study found that parameters of gas exchange (arterial oxygen and carbon dioxide pressures) were associated with telomere length in COPD patients [14]. The large population-based study by Rode et al. [18] found weak but statistically significant positive correlations between telomere length and spirometric indices, which is confirmed by our results. Furthermore, we observed the effects to be somewhat stronger in COPD or asthma patients compared with healthy individuals, suggesting that intrinsic age-related processes may be aggravated by chronic inflammatory processes occurring either locally in the lung or systemically.

In both COPD and asthma patients, we observed shorter telomeres compared with healthy controls. In COPD, the association was stronger when defining cases by the age-dependent LLN for FEV₁/FVC according to the GLI criteria [19] compared with the fixed cut-off given by GOLD [21]. This suggests that the GLI approach better suits the transition from health to disease. For COPD and asthma, effect estimates appeared to be larger in females than in males, although sex differences were not statistically significant. With respect to COPD, this might be consistent with the observation that females appear to be more susceptible to COPD than males [34] but the causality certainly needs to be proved. Whereas our findings for asthma are novel, the results for COPD are in line with previous findings in case–control [14–17] and population-based studies [18]. This suggests that, as already shown for other chronic diseases [1, 3, 35], accelerated ageing may contribute to the pathogenesis of asthma and COPD.

Our epidemiological finding is supported by results from clinical studies showing shorter telomeres in alveolar epithelial and endothelial cells from patients with emphysema [16]. Amsellem et al. [15] reported a higher percentage of pulmonary vascular endothelial cells stained for senescence markers in COPD patients than in lung healthy controls. This is in line with senescent-type growth characteristics of lung fibroblasts [36]. Furthermore, cellular senescence of pulmonary smooth muscle cells appears to be involved in the pathogenesis of pulmonary hypertension in COPD [37]. Several mechanisms have been proposed for abnormal ageing in COPD, including chronic inflammation, oxidative stress, mitochondrial dysfunction, altered energy metabolism, abnormal regulation of ageing processes, cellular and immunosenescence, and telomere dysfunction [3, 7–9, 15]. Analogous mechanisms in asthma have not been explored in detail but, due to the inflammatory nature of both diseases, some of the aforementioned factors are also likely to play a role in asthma. Possible consequences for both diseases are very difficult to predict but one might speculate that the ageing processes may be reflected in shorter life expectancy and an increased risk of cognitive impairment and dementia as reported for asthma and COPD [38–40].

The observed associations between telomere length and both diseases were not as strong as one might have initially expected, considering the large sample size. In the case of COPD, in the population-based approach of Rode et al. [18], a much weaker association was also observed than previous case–control studies had suggested [14]. In our case, this may be related to the fact that mostly mild-to-moderate disease stages were included in the study samples. For example, for COPD, in the KORA cohorts, 94.3% of COPD cases were in GOLD stages I and II. Rode et al. [18] also observed weaker effects for stages I (OR 1.09, 95% CI 0.97–1.23) and II (OR 1.17, 95% CI 1.04–1.31) than for more advanced GOLD, stages after multiple adjustment. In the case of asthma, the weakness of association could be due to the different methods of ascertaining asthma cases or phenotypic heterogeneity.

Moreover, as in many other studies, telomere length was measured in circulating leukocytes and not in the cells of the primarily affected organ, i.e. the lung. Studies addressing the comparability of telomere length between different tissues found strong correlations in healthy individuals, but somewhat blunted ones in disease [41–44]. We cannot rule out that this may be a factor that affects the association of telomere length with lung diseases. Leukocyte telomere length is a complex trait that is shaped by many factors including genetic, epigenetic, lifestyle and environmental determinants, and their complex interaction is hardly understood. In the present analysis, we therefore adjusted only for well accepted determinants. Certainly, the population-based epidemiological nature of the studies included comes along with distinct limitations. Population-based samples typically only scarcely include severe cases of lung diseases; therefore, early disease stages predominate, in which age-related mechanisms may not be as significant as in more advanced stages that are primarily included in clinical cohorts [14–17]. Furthermore, our study design did not allow us to take into account the known heterogeneity of COPD and asthma phenotypes. These entities may have different impacts on telomere length or may trigger premature ageing processes via telomere-independent mechanisms, as reported for fibroblasts of patients with lung emphysema [45]. Unlike in clinical studies, diagnosis of COPD and asthma had to rely mainly on self-reported physician diagnosis, which certainly adds some potential for misclassification error. However, we employed the recent GLI reference values [19] for COPD diagnosis wherever possible. In addition, confounding effects, such as smoking or medication, could not be accounted for in the same way as in a matched case–control study design.

As this analysis combines a number of studies that are heterogeneous in their design (online supplementary material), we report results of random-effects models in addition to the results based on fixed effects, in order to account for this heterogeneity. Naturally, confidence intervals are larger with random effects. While both models reach statistical significance in COPD, statistical significance was only reached with the fixed-effects model for asthma. The observed effect point estimates, however, were comparable between the fixed- and the random-effects model, supporting the biological significance of our findings. In COPD, our sensitivity analysis did not show an age dependency (fig. E2). Two studies showing positive effect estimates for COPD were selective in their study design; in the case of the EX-ATHLETES study, the positive effect of lifetime physical exercise on biological age may have played a role [3, 27, 46, 47].