Rate of telomere shortening and metabolic and cardiovascular risk factors: A longitudinal study in the 1934–44 Helsinki Birth Cohort Study

Abstract

Introduction. Leucocyte telomere length (LTL) is associated with age-related health outcomes, but only few longitudinal studies have assessed changes in LTL in an ageing population.

Methods. A total of 1,082 subjects from the Helsinki Birth Cohort Study (born 1934–1944), undergoing two clinical visits ∼10 years apart, were studied. Relative LTL was measured twice by quantitative real-time PCR. Simple and multiple regressions were used to study associations between cardiometabolic risk factors and LTL.

Results. Telomere shortening was observed in 93.7%, and telomere elongation in 6.3% of the study participants. Telomere shortening was more rapid among males (–39.5% ± 1.1% versus –35.5% ± 1.0%, P < 0.01). In men a decrease in weight, waist circumference, BMI, and body fat percentage were all associated with telomere shortening during the follow-up (P < 0.05) independently of age and use of medication. Furthermore, higher body fat percentage and higher HDL-cholesterol level were associated with a slower rate of shortening in LTL (P < 0.05). Lower blood pressure levels were also associated with slower rate of telomere shortening in men (P < 0.05). No similar associations were observed among women.

Discussion. A decrease in adiposity was associated with telomere shortening, and higher body fat percentage and HDL-cholesterol were associated with a slower rate of shortening in telomere length in men.

Key words::

• Adiposity
• ageing
• blood pressure
• dyslipidaemia
• telomere shortening

Key messages

• Change in telomere length has been associated with ageing-related health outcomes, but only few longitudinal studies have been able to study this in an ageing population.

• A reduction in adiposity was associated with telomere shortening during the follow-up, while higher body fat percentage and HDL-cholesterol were associated with a slower rate of shortening in telomere length, but in men only.

• Cellular pathways regulating ageing differ during the ageing process, and they are sex-specific.

Introduction

The global population is rapidly ageing which is leading to a larger number of people surviving to very old age. The increase in longevity has not been accompanied by a parallel decline in the occurrence of chronic disorders, and unhealthy ageing is common (1,2). Traditionally when studying ageing most focus has been put on health-related behaviours including diet, smoking, and other lifestyle-related factors (3–6). Telomere length is considered a marker of cellular ageing and is inversely associated with age; the estimated shortening is 25 base pairs per year. Shorter telomere length has been associated with increased morbidity and mortality as well as with several risk factors for cardiovascular and metabolic health outcomes (7–13). One recent prospective study showed that rate of telomere shortening in midlife is related to vascular damage over a 10-year follow-up (14).

Obesity and elevated cholesterol levels are usually associated with health concerns. However, there seems to be a paradoxical association between ageing and cholesterol levels and adiposity (15–17). Higher cholesterol levels, as well as a higher body mass index in later life, are not necessarily associated with adverse health effects in all elderly people. The reasons behind this are not fully understood.

Most previous studies focusing upon telomere length have been cross-sectional (7–13). To our knowledge no previous study has focused upon the longitudinal association between telomere length, lipid metabolism, and adiposity-related outcomes in an elderly population. Furthermore, no previous study has assessed the rate of telomere shortening in later life and its relation to various cardiometabolic health outcomes using a longitudinal study setting.

Methods

The Helsinki Birth Cohort Study (HBCS) consists of 13,345 singletons born in Helsinki in 1934–1944 who were still alive in 1971 when all residents of Finland received a unique personal identification number; 8,760 of the individuals were born in Helsinki University Central Hospital. In order to reach for a target of 2,000 participants for a clinical examination, a total of 2,902 people were randomly selected from those 8,760 born at Helsinki University Central Hospital to participate in a clinical examination in the years 2001–2004. In total 2,003 individuals participated in the clinical study. The study design has been described in detail before (18,19). A follow-up study including 1,082 participants was conducted in the years 2011–2013. From the original clinical study cohort (n = 2,003) 1,311 people who were living within a 100-km distance from our study clinic were invited to participate in a new clinical follow-up; 1,082 participated. The participation rate was 83%. The participants in the present study were 67 to 77 years of age, mean age being 71 years. Those not participating declined mostly due to own or a family member's health conditions.

At the clinical visits height was measured with a KaWi stadiometer; weight with SECA (Brooklyn, NY, USA) alpha 770 scales. Height and weight were measured in light indoor clothing and without shoes. Height was measured to the nearest 0.1 cm and weight to the nearest 0.1 kg. Body composition was assessed by bio-impedance using the In Body (Seoul, Korea) 3.0 device.

Blood pressure was measured from the right arm while the subject was in the sitting position, and it was recorded as the mean of two successive readings from a mercury sphygmomanometer. All measurements were performed by trained study nurses. A questionnaire was used to obtain information on medical history, current medication, smoking, and leisure-time physical activity.

Blood samples were drawn for the assessment of glucose and lipids. Glucose was measured according to the hexokinase method. Serum cholesterol and triglyceride concentrations were measured with the use of standard enzymatic methods. LDL-cholesterol concentrations were calculated using the Friedewald formula (20). IL-6 and TNF-α were analysed with multiplex sandwich immunoassays (Milliplex Human Metabolic Hormone Panel; Darmstedt, Germany).

DNA extraction and telomere length

Relative telomere length was measured twice, at the time of the first visit in 2001–2004, and after 10 years (2011–2014 follow-up visit). DNA was extracted from peripheral whole blood using a commercially available kit according to the manufacturer's instruction (QIAamp blood Maxi kit and DNeasy blood and tissue kit, Qiagen s.r.l. (Venlo, The Netherlands) respectively). DNA concentration and purity were assessed by comparing ultraviolet absorbance at wavelengths of 260 nm to absorbance at 230 nm (260/230 ratio) for salts contamination, and to 280 nm (260/280 ratio) for other contaminants, including proteins. Samples with ratios ranging between 1.7 and 2.1 were considered pure and suitable for the following steps. DNA integrity was tested by electrophoresis in 0.8% agarose/0.5x TBE with 0.1 μL/mL Ethidium bromide at ∼100 V for 15–25 minutes.

At both time points relative telomere length was assessed by a real-time quantitative PCR method. At the first time point relative telomere length was determined as ratio of telomere DNA to haemoglobin beta single-copy gene signal intensities, as previously described (21,22). Briefly, based on O’Callaghan's method (23), a synthetic oligomer Sigma (St Louis, Missouri, USA) dilution series, hbg-120-mer and tel14x, was included in every plate to create reaction-specific standard curves, and plasmid DNA (pcDNA3.1) was added to each standard to maintain a constant 10 ng of total DNA concentration per reaction. Quality control (QC) was carried out with the Bio-Rad CFX Manager software v.1.6 9 (Bio-Rad Laboratories, Hercules, CA, USA). All plates included four genomic DNA control samples for the plate effect calibration and for monitoring the coefficient of variation (CV), which was 21.0% for the telomere reaction, 6.0% for the β- haemoglobin reaction, and 24.8% for their ratio (T/S). The plate effect was taken into account by normalizing the telomere signal and reference gene signal to the corresponding mean of four control samples that were analysed for every qPCR plate before taking the T/S ratio. Three outlier samples of T/S ratio were removed before statistical analyses.

The second relative telomere length measurement was performed using a multiplex quantitative real-time PCR method, as previously described by Cawthon (24) with minimum modifications. Briefly, DNA concentration was standardized to 4 ng/μL and combined with telomere primers pair 900 nM, beta-globin (as single-copy gene) primers pair 500 nM, and 2X master mix (IQ Sybr green supermix, Bio-Rad Laboratories). PCR reactions were set up in a 384-well plate (CFX384 Touch Real-Time PCR detection system, Bio-Rad Laboratories) and carried out in a final volume of 10 μL. The original thermal cycle (24) was used. A 1:3 serial dilution curve was run to assess the efficiency of the amplification. Threshold cycles (Ct) for both telomere and beta-globin amplification were analysed using a dedicated software (CFX Manager software, Bio-Rad Laboratories). The multiplex method provides a relative telomere length (T/S) expressed as ratio between the amplification of the telomere sequence (T) and that of a single copy gene (S), measured for each sample in the same well and PCR run, and normalized using a common reference DNA sample. Samples were run in triplicate; the mean coefficient of variation of each triplicate was 6.0%, and the mean inter-assay CV% was 6.2%.

Leucocyte telomere length (LTL) values obtained at the two time points correlated with each other (r = 0.325, P < 0.001). LTL change over 10 years was calculated adjusting for the baseline measurement (relative change in LTL = {[LTL at70] – [LTL at60]}/[LTL at60] × 100) in order to avoid mistakes due to different methodology used.

Ethical statement

The study was approved by the Ethics Committee of Hospital District of Helsinki and Uusimaa and conducted according to the guidelines in the Declaration of Helsinki. Written informed consent was obtained from all subjects.

Statistical methods

Data were examined in the whole cohort, and in women and men separately, by using SPSS for Mac OS X v.20 (Chicago, IL, USA). Results are expressed as mean ± SEM. Comparisons between groups were performed by ANOVA or ANCOVA when adjustments for covariates were introduced. Comparisons within group of repeated measurements were performed by paired t test or general linear models with adjustment for covariates (age and use of medication) when needed. For instance, analysis of change over time of blood pressure and lipid profile was performed both without and with adjustment for the use of medication for control of blood pressure and lipids respectively (both sets of information collected at the first visit in 2001–2004 and at the follow-up 10 years later were used in the models).

Since telomere length data were not normally distributed, Spearman's correlation coefficient was used to determine associations between them and clinical variables. Non-parametric tests were used to compare telomere length between men and women and their change over time. Logarithmic transformation of telomere length was also performed with no effect on the results of the statistical analysis. Regression analyses adjusted for age and use of medication were also performed. Pearson's chi-square test was used to assess whether observations on two categorical variables were independent of each other. A P value < 0.05 was considered statistically significant.

Results

Clinical characteristics

The study cohort (n = 1,082) included 610 women (56%) and 472 men (44%). Among the men 32.1% had been diagnosed with hypertension; the corresponding number among women was 31.2%. At follow-up the prevalence of hypertension was 49.2% among men and 49.0% among women. Coronary heart disease had been diagnosed in 6.8% of the men and in 6.6% of the women. The corresponding numbers at follow-up were 10.6% and 7.9%, respectively. Among the men 1.9% had a history of stroke at baseline and 3.4% at follow-up. The corresponding numbers among women were 0.5% and 2.3%, respectively. Prevalence of type 2 diabetes in men was 6.8% at baseline and 17.4% at follow-up. The corresponding numbers among women were 4.4% and 13.1%, respectively. Cancer had been diagnosed in 3.4% of the men at baseline and among 14.2% at follow-up. The corresponding values among the women were 6.1% and 16.1%, respectively. These increments in prevalence are significant in both men and women for all disease (P < 0.001), except for stroke in men and CHD in women.

Table I gives descriptive characteristics of the study population at both time points studied. Several changes in metabolic and anthropometric parameters occurred during the follow-up. There was a slight reduction in BMI among the men (P < 0.001), but no change among the women. Lean body mass decreased, body fat percentage increased, and systolic blood pressure increased during the follow-up in both men and women (all P values < 0.001) (Table I). After adjustment for use of medication affecting blood pressure, systolic blood pressure was still significantly higher in both men and women (P < 0.001) at follow-up.

Table I. Demographic, clinical, and biochemical parameters in 2001–2004 and in 2011–2013 among men and women.

	Males				Females
	2001–2004		2011–2013		2001–2004		2011–2013
	n	Mean ± SEM	n	Mean ± SEM	n	Mean ± SEM	n	Mean ± SEM
Age (years)	472	61.2 ± 0.1	472	70.8 ± 0.1**	610	61.3 ± 0.1	610	71.1 ± 0.1**
BMI (kg/m^2)	472	27.2 ± 0.2	472	26.8 ± 0.2**	610	27.2 ± 0.2	610	27.3 ± 0.2
SBP (mmHg)	472	144.5 ± 0.9	472	151.3 ± 0.9**	610	142.7 ± 0.8	610	151.8 ± 0.9**
DPB (mmHg)	472	89.7 ± 0.5	472	84.8 ± 0.5**	610	87.0 ± 0.4	610	83.2 ± 0.4**
Lean body mass (kg)	453	65.5 ± 0.3	460	63.2 ± 0.3**	592	47.7 ± 0.2	599	46.1 ± 0.2**
Fat percentage (%)	453	22.9 ± 0.3	460	23.6 ± 0.3**	592	33.2 ± 0.3	599	34.8 ± 0.3**
Glucose (mM)	472	6.0 ± 0.1	472	5.9 ± 0.1*	610	5.5 ± 0.1	607	5.6 ± 0.1*
Total cholesterol (mmol/L)	471	5.7 ± 0.1	472	5.1 ± 0.1**	610	6.0 ± 0.1	608	5.6 ± 0.1**
HDL-cholesterol (mmol/L)	471	1.47 ± 0.02	472	1.46 ± 0.01	610	1.76 ± 0.02	608	1.74 ± 0.01
LDL-cholesterol (mmol/L)	471	3.6 ± 0.1	472	3.1 ± 0.1**	610	3.6 ± 0.1	608	3.3 ± 0.1**
Triglyceride (mmol/L)	471	1.50 ± 0.03	472	1.29 ± 0.03**	610	1.37 ± 0.03	608	1.27 ± 0.02**

*P < 0.05 compared to baseline value in 2001–2004; **P < 0.001 compared to baseline value in 2001–2004.

Total cholesterol, LDL-cholesterol, and triglycerides concentrations decreased during follow-up (P < 0.001) in both men and women (Table I). Adjustments for the use of lipid-lowering medication did not change the findings in men. After adjustment only LDL-cholesterol was significantly lower at follow-up in women.

Telomere length was longer in women compared to men at both time points (1.42 ± 0.01 versus 1.38 ± 0.01, P < 0.05; and 0.89 ± 0.01 versus 0.81 ± 0.01, P < 0.01). Disease status was not associated with LTL at baseline in the whole population, nor in men and women separately. A negative correlation between number of chronic diseases and LTL was observed in women only (r = –0.0094, P = 0.020) at follow-up. Mean telomere length shortened significantly over time in both groups. Use of blood pressure-lowering medication was related to telomere length. Telomere length at 71 years was inversely correlated with the use of blood pressure medication both at baseline (r = –0.072, P = 0.018) and 10 years later at follow-up (r = –0.070, P = 0.022) in the whole study population in the age-adjusted model. No similar association was observed in relation to use of lipid-lowering medication.

Telomere shortening

Telomeres were on average 11% longer in women than in men (P < 0.001). Figure 1 shows change in telomere length separately for men and women during the follow-up. Telomere shortening was observed among 93.7% of the total study sample, while telomere elongation was observed in 6.3% of the study subjects. Telomere shortening during the follow-up was more rapid among males (P < 0.01). Adjustment for change in age did not affect the findings. Age, blood pressure, and glucose concentrations at baseline were not associated with change in telomere length over 10 years. Nor did disease state at baseline influence change in telomere length significantly. Furthermore, we assessed whether development of new chronic diseases would influence telomere length during follow-up; we did not observe any difference in change in LTL among those who developed at least one new chronic disease during follow-up compared to those who did not.

Figure 1. Changes in telomere length during the follow-up period in males (left) and women (right).

Clinical characteristics and telomere length

Baseline associations

Table II shows correlations between LTL and clinical characteristics. At baseline adiposity (body weight) (r = –0.072, P = 0.020), waist circumference (r = –0.069, P = 0.025), and lean body mass (r = –0.083, P = 0.008) were inversely associated with telomere length. Height was inversely associated with LTL (r = –0.080, P = 0.010). No significant associations were observed between glucose, lipid concentrations, or blood pressure and telomere length. Adjustment for age and use of lipid-lowering and anti-hypertensive medication only affected the findings little.

Table II. Correlations between leucocyte telomere length (LTL) and clinical characteristics in 2001–2004, unadjusted, adjusted for age, and adjusted for age and use of medication affecting blood pressure (SBP and DBP) and lipids (lipid parameters).

	All subjects		Males		Females
	r	P value	r	P value	r	P value
Weight (kg)	–0.072 –0.077	0.020 0.012	–0.082 –0.074	0.077 0.112	–0.026 –0.036	0.527 0.376
Waist (cm)	–0.069 –0.067	0.025 0.029	–0.103 –0.083	0.027 0.076	–0.014 –0.016	0.735 0.697
Systolic BP	–0.042 –0.034 –0.023	0.172 0.266 0.459	0.007 0.038 0.044	0.885 0.416 0.348	–0.068 –0.081 –0.066	0.095 0.049 0.109
Diastolic BP	–0.036 –0.043 –0.034	0.244 0.164 0.270	0.050 0.050 0.056	0.288 0.287 0.233	–0.079 –0.098 –0.088	0.054 0.017 0.032
Lean body mass (kg)	–0.083 –0.082	0.008 0.009	–0.061 –0.045	0.202 0.345	–0.040 –0.053	0.336 0.207
Fat percentage (%)	0.019 0.022	0.54 0.487	–0.085 –0.067	0.076 0.157	–0.007 –0.003	0.860 0.951
Total cholesterol (mmol/L)	0.038 0.048 0.051	0.213 0.119 0.095	0.023 0.016 0.016	0.627 0.729 0.740	0.041 0.056 0.063	0.318 0.172 0.125
HDL-cholesterol (mmol/L)	0.009 0.016 0.017	0.758 0.599 0.591	–0.030 –0.018 –0.018	0.521 0.706 0.701	–0.022 0.001 0.002	0.598 0.993 0.962
LDL-cholesterol (mmol/L)	0.037 0.036 0.039	0.230 0.244 0.207	0.010 0.001 –0.001	0.838 0.976 0.989	0.060 0.058 0.065	0.143 0.160 0.113
Triglycerides (mmol/L)	0.017 0.033 0.033	0.592 0.285 0.286	0.071 0.061 0.061	0.129 0.193 0.193	–0.015 0.019 0.018	0.713 0.636 0.669

Associations at follow-up

At follow-up, on average 10 years after the baseline visit, inverse associations were observed for body weight (r = –0.100, P < 0.001), waist circumference (r = –0.080, P = 0.009), and lean body mass (r = –0.128, P < 0.001) and telomere length (Table III). Height was inversely associated with LTL (r = –0.116, P < 0.001). Body fat percentage (r = 0.066, P = 0.032), total cholesterol (r = 0.072, P = 0.018), and HDL-cholesterol concentration (r = 0.074, P = 0.016) were all positively associated with telomere length at follow-up. These findings were little affected by adjustment for age and use of lipid-lowering medication.

Table III. Correlations between leucocyte telomere length (LTL) and clinical characteristics in 2011–2013, unadjusted, adjusted for age, and adjusted for age and use of medication affecting blood pressure (SBP and DBP) and lipids (lipid parameters).

	All subjects		Males		Females
	r	P value	r	P value	r	P value
Weight (kg)	–0.100 –0.102	< 0.001 0.001	0.007 0.022	0.881 0.638	–0.077 –0.098	0.057 0.017
Waist (cm)	–0.080 –0.087	0.009 0.005	0.010 0.024	0.825 0.614	–0.068 –0.088	0.093 0.032
Systolic BP	0.016 0.023 0.036	0.591 0.457 0.239	0.031 0.002 0.011	0.497 0.958 0.811	0.005 0.034 0.054	0.893 0.414 0.182
Diastolic BP	0.035 0.044 0.051	0.245 0.156 0.093	0.085 0.059 0.063	0.066 0.208 0.176	0.023 0.054 0.062	0.577 0.192 0.128
Lean body mass (kg)	–0.128 –0.136	< 0.001 < 0.001	0.020 0.013	0.675 0.778	–0.074 –0.095	0.070 0.021
Fat percentage (%)	0.066 0.070	0.032 0.024	–0.014 0.023	0.771 0.619	–0.054 –0.053	0.189 0.196
Total cholesterol (mmol/L)	0.072 0.062 0.063	0.018 0.047 0.038	0.068 0.036 0.009	0.143 0.440 0.848	0.023 0.021 0.012	0.579 0.605 0.769
HDL-cholesterol (mmol/L)	0.074 0.054 0.064	0.016 0.081 0.035	0.044 0.024 0.022	0.343 0.616 0.635	0.005 –0.006 –0.009	0.901 0.884 0.824
LDL-cholesterol (mmol/L)	0.054 0.047 0.029	0.076 0.127 0.336	0.059 0.035 0.007	0.200 0.453 0.886	0.018 0.024 0.013	0.659 0.569 0.759
Triglycerides (mmol/L)	0.006 0.001 0.001	0.850 0.995 0.987	0.003 –0.009 –0.016	0.956 0.850 0.736	0.011 0.012 0.015	0.788 0.766 0.711

Change in telomere length and clinical characteristics

In men a decrease in weight, waist circumference, BMI, and body fat percentage were all associated with telomere shortening during the follow-up (r = 0.116, 0.116, 0.127, and r = 0.113, all P < 0.02). Change in height was not associated with change in LTL during the follow-up. No similar associations between changes in anthropometric parameters and telomere length were observed in women (Table IV).

Table IV. Correlations between leucocyte telomere length (LTL) relative change (%) and relative change (%) in cardiometabolic risk factors, unadjusted, adjusted for age, and adjusted for age and use of medication affecting blood pressure (SBP and DBP) and lipids (lipid parameters).

	All subjects		Males		Females
	r	P value	r	P value	r	P value
Δ Weight (kg)	0.038 0.001	0.215 0.997	0.116 0.098	0.013 0.036	–0.040 –0.073	0.327 0.077
Δ Waist (cm)	0.058 0.032	0.060 0.299	0.116 0.096	0.013 0.040	–0.020 –0.031	0.628 0.458
Δ BMI (kg/m^2)	0.044 0.008	0.157 0.788	0.127 0.120	0.007 0.011	–0.043 –0.076	0.298 0.068
Δ Systolic BP (mmHg)	0.014 0.007 –0.001	0.661 0.831 0.978	0.038 0.029 0.019	0.411 0.542 0.678	–0.017 –0.013 –0.019	0.686 0.745 0.652
Δ Diastolic BP (mmHg)	0.041 0.039 0.031	0.180 0.202 0.315	0.100 0.078 0.063	0.032 0.097 0.180	–0.008 0.010 0.006	0.841 0.814 0.884
Δ Lean body mass (kg)	–0.013 –0.038	0.692 0.236	0.078 0.071	0.107 0.139	–0.076 –0.105	0.072 0.012
Δ Fat percentage (%)	0.060 0.061	0.059 0.053	0.113 0.110	0.019 0.022	–0.008 0.017	0.846 0.682
Δ Total cholesterol (mmol/L)	0.059 0.048 0.036	0.056 0.117 0.241	0.043 0.045 0.035	0.364 0.340 0.461	0.052 0.034 0.021	0.206 0.411 0.618
Δ HDL-cholesterol (mmol/L)	–0.014 0.011 0.012	0.643 0.734 0.708	–0.051 –0.006 –0.005	0.276 0.902 0.914	0.013 0.016 0.017	0.745 0.704 0.684
Δ LDL-cholesterol (mmol/L)	0.056 0.022 0.012	0.070 0.484 0.700	0.017 –0.001 –0.008	0.710 0.979 0.861	0.064 0.046 0.034	0.118 0.263 0.403
Δ Triglycerides (mmol/L)	0.046 0.017 0.013	0.133 0.588 0.675	0.119 0.068 0.066	0.011 0.146 0.162	–0.034 –0.034 –0.038	0.414 0.414 0.356

We further divided the study participants according to change in telomere length around the median value. Males had a faster rate of telomere shortening compared with women (P < 0.001). Moreover, a higher body fat percentage and higher HDL-cholesterol level were both associated with a slower rate of change in telomere length (P < 0.05). Similarly, among men lower systolic and diastolic blood pressure levels were associated with slower rate of telomere shortening (P < 0.05). No similar associations were observed among women.

Clinical characteristics predicting increase in telomere length

We then divided the participants according to whether their telomere length had increased or decreased during the follow-up period. Those individuals with an increase in telomere length were characterized by significantly lower IL-6 concentrations (102 ± 27 pg/mL versus 228 ± 30 pg/mL, P = 0.002) and lower TNF-α concentration (9.2 ± 1.4 pg/mL versus 31.4 ± 5.7 pg/mL, P < 0.0001) at baseline. None of the other parameters studied was associated with an increase in telomere length.

Discussion

In this longitudinal study based upon findings in participants born 1934–1944 from the Helsinki Birth Cohort Study we describe that telomere length shortens with increasing age, as expected, and much faster among males than among females. Our findings further suggest a protective role of adipose tissue as well as higher total and HDL-cholesterol concentrations especially in elderly men in relation to telomere shortening. Also lower blood pressure levels were associated with slower rate of telomere shortening in men. Lower levels of inflammatory markers predicted slower rate of telomere length shortening during follow-up.

Telomere length was associated with several clinical characteristics both at baseline and at follow-up. At baseline at a mean age of 61 years adiposity was inversely associated with telomere length. Neither lipids nor blood pressure were associated with telomere length at baseline. The association between adiposity and telomere length has been described previously, although it has not been consistent (7,25,26). Interestingly the association between telomere length and some of the clinical characteristics was different at follow-up at a mean age of 71 years. The previous inconsistent findings between adiposity and telomere length could perhaps be explained by the difference observed in relation to ageing (26). Total body fat percentage was positively associated with telomere length at follow-up. Especially among men we noted that both higher body mass index and higher body fat percentage were positively associated with telomere length.

Interestingly at follow-up total cholesterol levels were positively associated with telomere length. These findings suggest that the association between risk factors and telomere length change with increasing age in some individuals. This could also explain the paradoxical findings that obesity and higher cholesterol levels are not necessarily associated with the same negative health outcomes in elderly individuals as often seen in younger age groups (15–17).

A higher body fat percentage, higher HDL-cholesterol concentration, and lower blood pressure were all associated with a slower rate of reduction in telomere length among men. Oxidative stress has been implicated in affecting telomere length (27,28). In support of this we observed that those individuals with an increase in telomere length had the lowest levels of inflammatory markers. However, there was not a large proportion of men with an increase in telomere length. Several studies have suggested that oxidative stress is a potential mediator of the association between shorter telomere length and increased cardiovascular disease risk. Oxidative stress is also believed to be one of the main drivers of atherosclerosis (29–31). Therefore, exposure to increased levels of oxidative stress may explain the association between inflammatory markers and change in telomere length observed in the present study. Elevated blood pressure is a well-known risk factor for many cardiovascular diseases, many of which have been related to short telomere length. We report an association between blood pressure levels and telomere length, especially among men, supporting the need to control blood pressure and potentially prevent premature telomere shortening. Obviously there are several other health outcomes supporting treatment of elevated blood pressure levels.

We observed a reduction in telomere length among the majority of the study participants (94%). This is a much higher number than reported by Masi et al., who reported that among 74% of the participants there was telomere shortening over the follow-up time. However, the participants in the present study were about 10 years older than those in the study of Masi et al. (14). This probably explains the difference. Similarly, 26% of the subjects in the UK study had an increase in telomere length (14). The corresponding number in the present study was 6%.

There are several strengths in our study. This study was longitudinal and included more than 1,000 well-characterized subjects from the Helsinki Birth Cohort Study, reflecting a general urban population through a period of 10 years in later life when biological involution sets in. Furthermore, telomere length was measured at two different time points from leucocytes. There are also limitations to our study. The HBCS is restricted to people who were both born 1934–1944 in Helsinki, and who attended child welfare clinics in the city of Helsinki. Most children and their parents attended these clinics, which were free. The people in our study may not be representative of all people living in Helsinki in those days. However, at birth the distribution of fathers’ occupations was similar to that in the city as a whole. Body composition was assessed with bio-impedance. This is not the gold-standard for assessment of body composition. However, bio-impedance offers accurate estimates of body composition within a wide range of adiposity. In the present study we assessed relative telomere length. Using the real-time quantitative PCR method, the telomere signal is normalized to the signal for a single copy gene to generate a T/S ratio which is proportional to the absolute quantification of LTL in base pairs. Although this is perhaps not the gold-standard for assessment of telomere length, it is the most commonly used method in larger samples, but has also been used in studies with smaller sample sizes (32,33). Furthermore, several other studies have used relative telomere length in longitudinal settings. Also, the relative change in LTL over 10 years in the present study is of the same magnitude as reported by others (31). Telomere length has been associated with several different disease outcomes; however, the power of the present study does not allow us to focus upon this subject in greater detail. Furthermore, the associations reported in this study are observational, and therefore no conclusions can be made regarding causality.

We believe that these findings might have important implications. In some elderly individuals adiposity as well as elevated cholesterol levels seem to be associated with longer telomeres and therefore ‘aggressive’ treatment of overweight and obesity in elderly people does not seem warranted. Another noteworthy result of the present study was that there were marked sex-related differences in the relationship linking telomere length and risk factors for metabolic and cardiovascular disease outcome.

In conclusion, we analysed the relationship between a marker of cellular ageing, leucocyte telomere length, and various measures of cardiovascular and metabolic risk factors. Our findings support the hypothesis that cellular pathways regulating ageing differ during the ageing process and they are gender-specific.

Funding: HBCS has been supported by grants from Finska Läkaresällskapet, the Finnish Special Governmental Subsidy for Health Sciences, Academy of Finland, Samfundet Folkhälsan, Liv och Hälsa, the Signe and Ane Gyllenberg Foundation, and EU FP7 (DORIAN) project number 278603.

Declaration of interest: The authors report no conflicts of interest.