OUP user menu

Telomere length is associated with left ventricular function in the oldest old: the Newcastle 85+ study

Joanna Collerton, Carmen Martin-Ruiz, Antoinette Kenny, Karen Barrass, Thomas von Zglinicki, Tom Kirkwood, Bernard Keavney
DOI: http://dx.doi.org/10.1093/eurheartj/ehl437 172-176 First published online: 10 January 2007


Aims Heart failure is a condition increasingly prevalent at older ages; however, mechanisms by which the ageing process affects cardiac function are largely unknown. Telomere length is a biomarker of ageing that has been suggested to be associated with a variety of diseases of late onset, but its relationship with cardiac function has not previously been studied. We measured telomere length in peripheral blood mononuclear cells and carried out echocardiography in a group of 85-year old subjects recruited from the community as part of the Newcastle 85 + Study.

Methods and results Eighty-nine subjects were recruited through local family practitioners. They were visited in their homes for clinical assessment and echocardiography, which was performed using a handheld device. Telomere length was measured by a real-time PCR method. High sensitivity C-reactive protein was measured using ELISA. Echocardiographic M-mode ejection fraction (EF) was strongly associated with telomere length (P = 0.006) in subjects without evidence of previous MI. Sex and telomere length were significant predictors of EF while current smoking, blood pressure, plasma high sensitivity C-reactive protein, and use of cardiovascular medications were not. One standard deviation longer telomeres were associated with a 5% higher EF. Telomere length accounted for 12% of the observed variability in EF.

Conclusion These data show influences of the ageing process on myocardial function in the oldest old, apparently independent of other specific disease processes. This may be of importance in the aetiology of heart failure in this age group.

  • Telomeres
  • Ageing
  • Left ventricular function


The age group comprising ‘the oldest old’, those over 85 years of age, is expected to increase in number faster than any other age group in all developed countries. The incidence of heart failure increases sharply at older ages, and the anticipated demographic shift has led to predictions of a possible ‘epidemic’ of heart failure as the population ages.1 There are significant differences in the epidemiology and clinical pathophysiology of heart failure in the very old when compared with younger subjects. Left ventricular (LV) systolic dysfunction due to myocardial infarction is a less dominant cause of heart failure than at younger ages. Indeed, a large proportion of older patients with clinical heart failure have preserved LV ejection fraction (EF).2 The contribution of the ageing process itself, independent of the incidence of specific cardiovascular diseases, to heart failure susceptibility in the elderly is contentious, but there is some evidence in favour of such a contribution. For example, Chimenti et al.3 presented evidence from human cardiac biopsy specimens in favour of a specific age-related cardiomyopathy characterized by severe telomeric shortening, cellular senescence, and cell death. Current paradigms of cellular ageing involve progressive accumulation of molecular faults leading to deficient responses to intrinsic and extrinsic stresses resulting ultimately in cellular dysfunction and death.4 The importance of cellular ageing processes for determining myocardial function in later life is presently little studied. Structural and functional myocardial changes with ageing (such as LV wall thickening and decreased LV compliance) are well described; however, no studies thus far have related biological markers of ageing to clinically relevant measures of cardiac function in very old people.

Telomeres are specialized structures consisting of DNA and protein which are located at the ends of chromosomes. They function to maintain chromosomal integrity by preventing the ends of the chromosomes being recognized as double-stranded DNA breaks. Telomeres are maintained by the activity of the telomerase ribonucleoprotein complex, the activity of which is high in germline cells, but low or undetectable in most somatic cells. Telomeres progressively shorten in somatic cells with age (due to numerous factors including oxidative stress), and with increasing number of cell divisions. When telomere length falls below a critical level, replicative senescence (permanent growth arrest of dividing cell types, such as fibroblasts) is triggered. Both genetic and environmental factors are involved in the control of human telomere length, and the intra-individual correlation between telomere lengths in different tissues is high.5 Many studies have shown that average telomere length in white blood cells shortens with age, although there is high variability between individuals of the same age. It has been suggested that blood cell telomere length can serve as a biomarker of an individual's cumulative exposure to oxidative stress.6 Cardiac tissue from animal models of ageing shows substantial loss of telomere length with age, and telomerase knockout mice develop heart failure, suggesting a possible role for telomere shortening in the development of human heart failure.79

Recently, associations between blood cell telomere length and a variety of age-related diseases have been shown in humans. With respect to the cardiovascular system, shorter white blood cell telomeres have been associated with a variety of conditions related to the presence and extent of atherosclerosis including premature MI, vascular dementia, a higher mortality from heart disease in individuals over 60 years of age, increased pulse pressure, and increased carotid atherosclerosis in hypertensive subjects.1015 Since atherosclerosis is characterized by increased levels of oxidative stress and chronic systemic inflammation, such an association between shorter white blood cell telomeres and more extensive atherosclerosis is highly biologically plausible. Perhaps surprisingly given these findings, a relationship between WBC telomere length and mortality has not been consistently observed. Such an association was found in a study of 143 individuals over 60 years of age,16 but two subsequent larger studies in older individuals (including a total of 1410 subjects aged 73–100 years) found no evidence of association.17,18 One of these studies found a borderline significant association between shorter telomeres in peripheral blood mononuclear cells (PBMC) and a previous history of myocardial infarction in individuals over the age of 85 years.17 We have performed the first study to examine the relationship between telomere length and echocardiographically determined LV function. To maximize the power to detect an independent effect of the ageing process on LV function, we have studied ‘the oldest old’. Such individuals constitute a ‘survivor cohort’ unlikely to have suffered from severe premature atherosclerosis, and thus represent the most powerful resource to detect an effect of telomere length on LV function independent of the association of telomere length with atherosclerosis (and other inflammatory conditions). Both telomere length and the rate of telomere shortening over time vary considerably between individuals; therefore, we studied subjects who were all close to 85 years of age at recruitment.


Community-dwelling individuals whose 85th birthday was between 1 January 2003 and 31 December 2003 were recruited from four family practices in Newcastle upon Tyne, UK. Subjects were visited at their place of residence by a nurse, and a questionnaire detailing aspects of their general health, functional status, and past medical history was administered. Clinical measurements including weight, demi-span, and blood pressure were made. An ECG was recorded and electronically transmitted to a reference centre for automated Minnesota coding. Blood was drawn for plasma and DNA extraction. The medical record was examined, and previous diagnoses and present medications recorded. Subjects consenting to an echocardiogram were subsequently visited at their place of residence by a trained echocardiographer and underwent two-dimensional, M-mode, and Doppler echocardiography which was performed in the left lateral decubitus position using a SonoHeart Elite handheld instrument (Sonosite, Bothel, WA) and C15 broadbandwidth (4-2Mz) array transducer with a 15 mm footprint. Measurements were made according to American Society of Echocardiography guidelines. EF was calculated, by the standard Teicholz formula, from M-mode measurements obtained from left parasternal long axis views taken at the level of the chordae tendineae. To assess diastolic function, the ratio of the mitral inflow E and A waves was calculated, and peak systolic and diastolic flows in the pulmonary veins were measured. We defined normal diastolic function as E/A ratio 0.75 to 1.5 and peak pulmonary venous systolic forward flow (S) greater than or equal to peak pulmonary venous diastolic forward flow (D). Grades of diastolic dysfunction were defined based on the E/A ratio and relationship between pulmonary vein S and D peak flows as in Redfield et al.1 Subjects with significant aortic valve disease, mitral valve disease, or pericardial disease evident on echo were excluded from analysis. All patients from one family practice (N = 26) were approached to undergo a second echocardiogram about one month after their initial echocardiogram to establish the reproducibility of these domiciliary studies; 19 consented.

High molecular weight DNA was isolated from PBMCs. DNA was obtained from 84 subjects (the other five subjects declined a blood sample). Telomere length was measured as abundance of telomeric template vs. a single gene by quantitative real-time PCR with modifications as described.17 Measurements were performed in quadruplicate. Three DNA samples with known telomere lengths were run as internal standards in each batch of samples. High-sensitivity C-reactive protein was measured using a commercial ELISA assay.

We assessed the reproducibility of domiciliary echocardiography by calculating the correlation coefficient, coefficient of variation, and repeatability coefficient.19 We used linear regression to establish whether sex, current smoking, previous MI, hypertension, diabetes, previous stroke, use of regular cardiovascular medications, measured blood pressure, and high sensitivity C-reactive protein were significant covariates of telomere length. We used linear regression to establish whether these same variables were significant covariates of EF. The relationship between the covariate-adjusted M-mode LVEF and telomere length (our primary hypothesis) was then examined using linear regression. Similar subsidiary analyses examined the relationship between covariate-adjusted LV mass or mitral E/A ratio and telomere length. Because of the previously reported associations between telomere length and MI, we removed two individuals with pathological Q-waves on the ECG and three individuals without ECG data available from the principal analyses relating telomere length to EF. Subsidiary analyses to check our result was not due to confounding were carried out including only those individuals who had no history of MI in the family practice record (N = 50), and including only those individuals with EF > 50% (N = 54). All statistical tests were two-sided; P < 0.05 was chosen as the threshold of significance for our primary hypothesis (association with LVEF) and P < 0.01 chosen as the threshold for our subsidiary hypotheses (association with LV mass and E/A ratio) to make some allowance for multiple testing. Analyses were performed using MINITAB v14.


134 people born in 1918 registered in the participating family practices were considered eligible to participate in this study. Nine were excluded because of death, terminal illness, or because they could not be contacted. 125 people were approached to participate of whom 89 consented. Of these, three died after the initial nurse visit and before the echo visit could be made, 16 declined an echo or withdrew from the study before the echo visit, and three echo studies were not performed for logistical reasons. Thus, echocardiograms were performed on 67 participants. There was no statistically significant difference in telomere length between subjects who did, and who did not have an echo (P = 0.167). Of 67 echo examinations, satisfactory quality M-mode data were available in 64. The reproducibility of M-mode EF measured in the home was similar to previously published hospital-based data in mixed populations (r = 0.85; CV = 12.5%; repeatability coefficient 16.62%). With respect to diastolic function, pulmonary venous flows could not be accurately determined in 19 subjects.

Characteristics of the study subjects are shown in Table 1. The mean age of participants at blood sampling was 85.2 years (range 84.9–85.7). LV systolic function was generally good: mean LVEF was 63.9%, and only 9 of 64 (14%) individuals had an EF below 50%. By contrast, 57% of those in whom diastolic function could be quantified had evidence of at least mild diastolic dysfunction. There was no association between telomere length and sex, current smoking, history of MI, hypertension, diabetes, stroke, use of regular cardiovascular medications, measured blood pressure (systolic, diastolic or pulse pressure), or high sensitivity C-reactive protein (all P > 0.1). There was a significant association between M-mode EF and sex: EF was 9.9% (95% CI 3.0–16.8) higher in women than in men (P = 0.004); there were no other significant covariates of EF. There was significant association between sex-adjusted EF and telomere length. One standard deviation difference in telomere length was associated with a 5% higher sex-adjusted EF, and telomere length accounted for 12.3% of the observed variability of EF (β = 4.954; SE(β) = 1.751; P = 0.006; Figure 1). We performed some subsidiary analyses to check the robustness of our result. The association remained statistically significant (P = 0.019) when the most influential participant (telomere length 7900 bp) was excluded from the analysis. We used the ECG to define inclusion in the principal analyses because of concern about the reliability of a previous history of MI in this age group, particularly if the supposed MI occurred in the distant past. Indeed, of nine participants with a history of MI and both echo and ECG data available, normal EFs and normal or only borderline significant ECG abnormalities were observed in six. Therefore, to provide additional assurance that the result was not confounded by association of telomere length with prior MI, we repeated the analyses additionally excluding those participants who had a previous history of MI from the family practice record, regardless of whether that history was supported by abnormalities on the ECG or reduced EF on echo. In those analyses, which included 50 subjects, the association between sex-adjusted EF and telomere length remained significant (β = 3.900; SE(β) = 1.845; P = 0.04). To rule out the possibility that the few individuals with low EF were disproportionately influencing the result, we carried out analyses restricted to those 54 subjects who had EF > 50%. In that subset of participants, the result remained significant (P = 0.025).

Figure 1

Scatterplot of sex-adjusted residuals for M-mode EF(y-axis) against telomere length in bp (x-axis), with superimposed regression line.

View this table:
Table 1

Characteristics of study participants

CharacteristicNumber (%)
Male sex40 (45%)
Age in years (range)85.2 (84.9–85.7)
History of
 MI13 (14.6%)
 Heart failure21 (23.6%)
 Hypertension57 (64%)
 Stroke11 (12.4%)
 Diabetes7 (7.9%)
 Current smoking9 (10.5%)
 Cardiovascular medications63 (70.8%)
Pathological Q-wave on ECG3 (4%)
EF < 50%9 (14%)
Diastolic function (normal/mild/moderate/severe)20/18/5/5
Mean (SD)
Height (m)1.62 (0.08)
Weight (Kg)65.50 (13.12)
Systolic BP (mmHg)151.90 (20.35)
Diastolic BP (mmHg)72.97 (9.06)
Pulse pressure (mmHg)78.91 (17.84)
LVIDs (cm)2.90 (0.92)
LVIDd (cm)4.66 (0.74)
LVPWd (cm)1.10 (0.22)
IVSd (cm)1.22 (0.21)
Mitral inflow E wave velocity (m/s)0.79 (0.24)
Mitral inflow A wave velocity (m/s)0.86 (0.24)
E/A ratio0.98 (0.56)
M-mode EF(%)63.90 (14.96)
LV mass index (g/m2)119.35 (34.88)
High sensitivity C-reactive protein (g/L)6.09 (9.44)
Telomere length (Kb)3.935 (0.880)
  • ECG was available on 75/89. Blood pressure was available on 79/89. Echo measurements were available on 64/89. C-reactive protein and Telomere length were available on 84/89.

There was no association between the sex-adjusted LV mass index and telomere length (P > 0.1). There was no association between the mitral E/A ratio, or the grade of diastolic dysfunction (graded either absent/mild/moderate/severe or absent/present) and telomere length (P > 0.1).


We have shown a highly significant association between telomere length and LVEF in a group of 85-year old subjects. This is the first study to show association between telomere length and any aspect of LV function. The association between telomere length and LV function in the present study was independent of MI or its risk factors and was also observed in analyses restricted to that subset of participants with conventionally normal EFs. We also observed, independently of telomere length, a 10% higher EF in 85-year old women than in men; this confirms results from other studies of subjects over 75 years of age.20 Our results suggest that cumulative age-related biological stress may affect myocardial function in the oldest old independently of specific cardiovascular conditions (particularly, myocardial infarction and hypertension) that are common at older ages. If so, the responsible mechanisms could be important contributors to the high incidence of heart failure in elderly people. We found no association between mitral inflow E/A ratio, graded severity of diastolic dysfunction, or LV mass index and telomere length.

Previous studies have shown association between shorter telomere length and the presence of angiographic coronary artery disease (CAD) or premature MI;12,13 the severity of carotid artery atheroma in hypertensive subjects;10 and increased arterial stiffness.11,15 These studies have led to the hypothesis that altered rates of biological ageing (consequent upon an individual's genetically and environmentally determined degree of exposure and resistance to damage-inducing stress) may be important in the development of atherosclerosis, the major cause of heart failure. A history of MI was relatively uncommon in our cohort, and our result retained statistical significance when participants with such a history were excluded from analysis. Thus, it seems unlikely that our findings are confounded by any association of telomere length with MI. We observed no association between previous MI and telomere length in this study, but numbers of events were few and power to detect such an association correspondingly low. The relative risk for MI of a particular difference in telomere length may be smaller at older ages, in keeping with other ‘classical’ risk factors for MI (for example smoking);21 the borderline (P = 0.038) statistical significance of association between telomere length and a previous history of MI observed in a large (N = 598) sample of participants whose mean age was around 90 years would support this notion.17

Previous studies in humans and animals have suggested mechanisms whereby ageing could affect myocardial function directly. Age-associated telomere shortening can be prevented in transgenic mice by cardiac-restricted expression of telomerase reverse transcriptase. The myocardium of such mice is to an extent protected from various types of injury, suggesting that myocardial telomere length preservation may render cardiomyocytes more robust to age-related stress.22 Chimenti and colleagues presented evidence from biopsy specimens for a specific age-related cardiomyopathy in man.3 Our results provide evidence that this entity may represent an extreme manifestation of a generally occurring process as the heart ages. By contrast, in a study of telomere length in the human heart, Takubo et al.5 found similar telomere length in cardiac autopsy samples from 168 individuals in the age range 0–104 years. Although that observation casts some doubt on the importance of telomere-mediated cellular senescence in the myocardium, animal models have shown cardiac myocyte subpopulations in the adult heart that are particularly susceptible to telomere shortening in response to stress.23 It is not currently known whether such cellular heterogeneity exists in the human heart. We studied telomere length in PBMC's, which is a biomarker of systemic cumulative oxidative and inflammatory stress. Thus, it is possible that, even if myocardial telomeres were preserved, other stress-related myocardial changes (such as increased mitochondrial inefficiency) that are correlated with PBMC telomere length may account for our observations. Further studies involving tissue samples will be necessary to investigate these mechanistic hypotheses.

This study could have important clinical implications. Our regression model indicated that telomere length accounted for 12% of the observed variability in EF, a substantial proportion. This suggests that molecular pathways involved in the response to age-associated stress may be particularly interesting to investigate for novel therapeutic agents for heart failure. Telomere length was associated with EF in participants within the ‘normal range’ of EF in this cross-sectional study. It is possible that those participants with currently normal EF but shorter telomeres are predisposed to the development of heart failure over time, and that telomere length may be a suitable biomarker to guide presymptomatic intervention to prevent heart failure in elderly people. We are presently conducting prospective studies to test this hypothesis.

A particular strength of this study is that we have enrolled community-dwelling individuals and studied them in their places of residence, which should maximize the representativeness of the cohort. Around 50% of these often frail elderly subjects stated that they would not have participated in this study had it required a hospital visit, emphasising the potential for selection bias in hospital-based studies of persons in this age group. To our knowledge, this is the only study thus far conducted using domiciliary echocardiography in elderly people. The technical quality of the echocardiograms using the handheld instrument, with respect to the proportion of technically adequate studies, the reproducibility of measurements of M-mode EF and the proportion of participants scorable for pulmonary venous flow was comparable to that achieved in hospital-based studies. Another important strength, given the high degree of inter-individual variation in telomere attrition rates, is the homogeneity of the ages of the participants. However, this study has certain limitations. Our capacity to score diastolic dysfunction was limited to 75% of the sample by the capacity of the handheld instrument then current; recent developments incorporating tissue Doppler imaging in handheld instruments will improve scoring of diastolic function in future studies of this type. We showed no association between measures of diastolic dysfunction and telomere length; since diastolic dysfunction is a commoner cause of clinical heart failure in the elderly than is systolic dysfunction, this is perhaps surprising but is likely to reflect the greater imprecision of measurement of diastolic function. Although our findings are highly statistically significant, our sample size is relatively modest and further studies to confirm or refute these novel findings, ideally in this age group, are warranted.

In conclusion, we have shown that telomere length, a biomarker of ageing, is strongly and independently associated with LVEF in the oldest old. This suggests the existence of novel mechanisms whereby the biological stress of ageing directly affects myocardial function.


The principal acknowledgement is to the participants in the study. We also thank Tyne and Wear Contractor Services Agency, Newcastle Primary Care Trust, Professor Peter Macfarlane, University of Glasgow for Minnesota coding, the participating family practices and the study nurses. The study was funded by the British Heart Foundation, Dunhill Medical Trust, Unilever PLC, Research into Ageing, and Newcastle Primary Care NHS Trust. None of the funding agencies had any role in the collection and interpretation of the data, or in the preparation of the manuscript.

Conflict of interest: none declared.


  • Other members of the Newcastle 85+ Study Core Team were John Bond, Martin Eccles, Carol Jagger, Oliver James, and Louise Robinson.


View Abstract