|Home | About | Journals | Submit | Contact Us | Français|
Leukocyte telomere length (LTL) decreases over the adult life course due to the cumulative burden of oxidative stress, inflammation, and exposure to vascular risk factors. Left ventricular (LV) mass is a biomarker of long-standing exposure to cardiovascular disease risk factors. We hypothesized that LTL is related inversely to LV mass.
We related LTL (measured by Southern blot analysis) to echocardiographic LV mass and its components (LV diastolic dimension, LVDD; LV wall thickness, LVWT) in 850 Framingham Heart Study participants (mean age 58 years, 58% women) using multivariable linear regression adjusting for age, sex, height, weight, systolic blood pressure, hypertension treatment, and smoking. Overall, multivariable-adjusted LTL was positively related to LV mass (beta coefficient per SD increase [β]=0.072; p= 0.001), LVWT (β=0.053; p= 0.01), and LVDD (β=0.035; p= 0.09). We observed effect modification by hypertension status (p for interaction =0.02 for LV mass); LTL was more strongly associated with LV mass and LVWT in individuals with hypertension (β per SD increment of 0.10 and 0.08, respectively; p<0.01 for both). Participants with hypertension who were in the top quartile of LV mass had LTL that was 250 base pairs longer relative to those in the lowest quartile (p for trend across quartiles 0.009).
In contrast to our expectation, in our community-based sample, LTL was positively associated with LV mass and wall thickness, especially so in participants with hypertension. These data are consistent with the hypothesis that longer LTL may be a marker of propensity to LV hypertrophy.
Telomere dynamics (telomere length and its age-dependent shortening) provide valuable insights into the pathogenesis of aging-related diseases in general, and atherosclerosis, in particular. Age-dependent shortening of leukocyte telomere length (LTL) registers the accruing burden of oxidative stress and inflammation, two key contributors to both aging1-4 and the development and progression of atherosclerosis.5,6 Accordingly, LTL has been reported to be relatively shorter in individuals with atherosclerotic cardiovascular disease (CVD).7-12 LTL also correlates inversely with cardiovascular disease risk factors, including cigarette smoking,13-15 obesity,14-16 sedentary lifestyle,17 and with an unhealthy life style in general.18 Of note, LTL shortening is assumed to parallel telomere shortening in the hematopoietic stem cells.19
Based on the association of shorter LTL with manifest CVD and with several of its major risk factors, we hypothesized that LTL also would be related inversely with manifestations of cardiovascular target organ damage. Left ventricular (LV) mass and LV hypertrophy are key indicators of such target organ damage, representing the time-averaged exposure to blood pressure and other vascular risk factors (such as age,20-23 obesity,20-25 and diabetes21,22,24,26) over the life course of individuals.27,28 Thus, we postulated that LTL would vary inversely with LV mass. We tested this hypothesis in the community-based Framingham Offspring cohort, in which measurements of LTL and echocardiographic LV mass were routinely obtained in a subsample of individuals.
The design and selection criteria of the Offspring cohort of the Framingham Study have been described previously.29 Briefly, 5124 individuals (who were children of the Original Cohort, and their spouses) were enrolled in 1971 and these participants are examined approximately every 4 years at the Heart Study. Offspring cohort participants were eligible for the present investigation if they attended the sixth examination cycle (1995-1998), and had available data on LTL measurements and echocardiographic LV mass. Of 3532 attendees at that examination, LTL measurements were obtained on 1589 unrelated individuals; no LTL measurements were made on the remaining 1943 attendees, a group that consisted of individuals who were biologically related, being part of one or more families, as noted previously.15 We excluded 739 participants from the present investigation for the following reasons: inadequate quality or amount of DNA for LTL measurements (n=345); previous myocardial infarction (MI) or heart failure (n=57); missing or unavailable data on LV mass (n=262), or presence of LV systolic dysfunction as evidenced by a fractional shortening of <0.29 (n=73), or missing covariates (n=2). We excluded participants with MI because the condition can render inaccurate the echocardiographic assessment of LV mass with M-mode, and people with heart failure or LV systolic dysfunction because of prior reports indicating shorter LTL in people with these conditions.12 After these exclusions, 850 participants (511 non-hypertensive and 399 hypertensive individuals) remained eligible for the present investigation. We have previously reported that participants undergoing LTL measurements were not systematically different from the entire sample of eligible attendees at the sixth examination cycle.15 The study protocol was approved by the Institutional Review Board at Boston University Medical Center, and all participants provided written informed consent.
At the sixth examination cycle, all attendees underwent a routine standardized medical history and physical examination (including blood pressure measurement), anthropometry and laboratory assessment of cardiovascular risk factors. Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or use of anti-hypertensive medications.30 Diabetes mellitus was defined as fasting glucose ≥126 mg/dl, or use of insulin or hypoglycemic medications.31 Participants were considered current cigarette smokers if they reported having smoked cigarettes regularly during the previous year.
All attendees at the sixth examination cycle underwent routine M-mode and two-dimensional echocardiography with Doppler color flow imaging on a Sonos 1000 Hewlett-Packard machine. Digitized images were stored on optical disks and measured using an off-line analysis system by a certified reader (sonographer or cardiologist), who was blinded to all clinical information, including LTL measurements. The end-diastolic left ventricular internal dimensions (LVDD) and the thicknesses of the interventricular septum (IVST) and the LV posterior wall (PWT) were obtained by averaging digital M-mode measurements in at least 3 cardiac cycles using the leading edge technique, according to the American Society of Echocardiography guidelines.32 End-diastolic LVWT was calculated as the sum of IVST and PWT. LVM was calculated as 0.8[1.04(LVDD+LVWT)3-(LVDD)3]+0.6.33 The reproducibility (inter-reader and intra-reader) of echocardiographic measurements was very good, as reported previously.34
LTL was derived from the mean length of the terminal restriction fragments, measured by Southern blot analysis, as described previously.15 Personnel of the laboratory that measured LTL were blinded to the clinical and echocardiographic characteristics of the participants. The CV of the LTL measurements in duplicate DNA samples resolved on different gels was 2.4%.
We used multivariable linear regression to relate LTL measurements (dependent variable) to echocardiographic LV mass (independent variable), adjusting for potential confounders. We also related LTL to each of the two components of LV mass, i.e., LVDD and LVWT, to gain insights into whether any potential association of LV mass with LTL was mediated by a relation of LTL to LVDD versus LVWT (or to both). We did not observe effect modification by sex (p>0.75 for interaction terms of LTL with LV mass, LVDD and LVWT in multivariable models). So, all analyses were sex-pooled in order to maximize statistical power. Given the known differences in the distributions of these echocardiographic measurements between men and women,35 we standardized each echocardiographic variable within sex, and then combined the z-scores for pooled-sex analyses.
We evaluated two sets of multivariable models: initially, adjusting for demographic and anthropometric variables that are key correlates of LTL and of echocardiographic measurements, i.e., age, sex and height and weight; next, additionally adjusting for smoking, systolic blood pressure and hypertension treatment. We chose to evaluate these models sequentially because blood pressure is a key correlate of LV mass,20 and smoking is strongly related to LTL,13,14 and so these variables may confound and/or mediate any potential association of LV mass with LTL. We evaluated several measures of goodness of fit, including the coefficient of determination of the model (R square), normality of residuals, presence of outlier values and excessive leverage by influential observations. Plots of residuals against predicted telomere length confirmed lack of outliers, and the appropriateness of using linear models. We tested for effect modification by age, hypertension status, obesity, and weight; if the corresponding interaction terms were statistically significant, we performed additional stratifying by these variables.
We did not adjust for diabetes status in the multivariable analyses noted above. Because diabetes has been associated with greater LV mass in epidemiological studies,36 we repeated our analyses additionally adjusting for diabetes. Also, urine albumin excretion, an indicator of target organ damage, has also been associated with LV mass in epidemiological studies.37 Therefore, we repeated all analyses adjusting additionally for the urine albumin-creatinine ratio (UACR) in a subsample of individuals (n =758; 89% of study sample).
In our primary analyses, echocardiographic LV measures were modeled as the independent variables and LTL was the dependent variable. We chose to do so because we postulated that echocardiographic LV mass would chronicle the long-standing exposure to multiple risk factors, all of which are associated with shorter LTL. However, in retrospect based on our findings, LTL dynamics may be causally related to the occurrence of cardiovascular target organ damage. Therefore, we also repeated our analyses with LTL as the independent variable and sex-standardized echocardiographic LV measures as the dependent variables. For these analyses, we considered two sets of models that incorporated the same set of covariates as in the primary analyses.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
Table 1 displays the baseline characteristics of our study sample overall, and stratified by hypertension status. About 40% of the sample was hypertensive.
Table 2 displays the results of multivariable models relating echocardiographic measures to LTL. LV mass and wall thickness were positively associated with LTL in both sets of models. A SD increment in LV mass (29 gm in women and 43 gm in men) was associated with an increment in LTL of 69-72 basepairs (models 1 and 2, respectively). LVDD was also positively related to LTL, but the relations were borderline statistically significant.
We observed effect modification by hypertension status (p for interactions of hypertension with: LV mass =0.02; LVWT =0.02; LVDD =0.24) but not by age, obesity or weight (p for interactions exceeded 0.13). So, additional analyses were stratified by hypertension status only (Table 3). The positive association of LV mass with LTL was stronger (both in terms of effect size and statistical significance) in participants with hypertension. Thus, in non-hypertensive individuals, a SD increment in LV mass was marginally associated with an increment in LTL of 49 basepairs (p=0.09), whereas in hypertensive individuals, a SD increment in LV mass was associated with an increment in LTL of 100 basepairs (p=0.003; comparing results for model 2 in the 2 groups). Figure 1 demonstrates the adjusted mean LTL across sex-specific quartiles of LV mass in the whole sample (Panel A), in participants with hypertension (Panel B) and in those without hypertension (Panel C). These data are consistent with the analyses using LV mass as a continuous variable.
We observed a positive association of LVWT with LTL in hypertensive participants but not in the non-hypertensive group. LVDD was not associated with LTL in either group.
In additional analyses, adjustment for diabetes (in addition to other covariates) did not alter the findings (data not shown); the positive association of LTL with echocardiographic LV measures was maintained. In a subsample of individuals with available measures of UACR, the association of LTL with LV mass remained robust after additional adjustment for UACR (data not shown).
Table 3 displays the results of analyses in which LTL was the independent variable and echocardiographic LV measures were treated as the dependent variables. In these analyses, results consistent with the primary analyses were observed. Figure 2 displays the increment in adjusted mean LV mass across quartiles of LTL in the entire sample, and in the nonhypertensive and hypertensive subsets.
The interest in LTL has had a remarkable ascendancy during the last decade primarily because of the hypothesis that age-dependent LTL shortening registers the cumulative burden of systemic oxidative stress and inflammation.7,11,13,38 If this premise is proven valid, sequential LTL measurements over time may constitute traits that characterize several aging phenotypes, including those pertaining to the cardiovascular system. It is therefore crucial to systematically decipher the links between LTL dynamics and indices of cardiovascular aging. Several previous studies have documented that key cardiovascular traits that increase with age, i.e., coronary atherosclerosis,9,10 arterial stiffening,39 increased carotid artery intima/media thickness8,40,41 and clinically overt CVD events such as myocardial infarction9,10 and stroke,40 are all associated with shortened LTL. Likewise, vascular risk factors such as smoking,13-15 body mass index,14-16 and hypertension42 are inversely associated with LTL.
LV mass represents time-averaged exposure to cardiovascular disease risk factors over the entire life course and is an indicator of target organ damage.27 Higher LV mass also has been reported to correlate with greater atherosclerotic burden and with increased risk of clinical CVD.28,43,44 Each of the key positive correlates of LV mass, notably age, high blood pressure, insulin resistance, and obesity, is inversely related to LTL.11,13,14,16,42 We postulated, therefore, that LV mass would be inversely related to LTL. Yet, to our surprise, we observed a positive association of LV mass with LTL. This presumably counterintuitive observation warrants an explanation.
We postulate that select biological determinants of LV mass must be positively related to LTL in order to explain the observed positive relations between the two. It follows that such ‘explanatory’ correlates must be different from the standard set of clinical variables (age, high blood pressure, insulin resistance, obesity) that are inversely related to LTL11,13,14,16,42 and positively to LV mass.20,27,28 What might these biological determinants of LV mass be?
The association of LTL with LV mass and LVWT, but not with LVDD may provide some clues. In epidemiological studies, key determinants of LVDD include blood volume, stroke volume, heart rate, and other hemodynamic factors.45 On the other hand, LVWT varies more with systolic blood pressure load,45 likely reflecting a compensatory increase in myocardial cell mass due to cardiac remodeling. The stronger association noted in hypertensive individuals also raises the possibility that a positive relation of LTL to myocardial growth may underlie the observed association.
We offer two potential explanations that may contribute to the observed positive association between LTL and both LVM and LVWT. One explanation is related to the role of resident cardiac stem cells/progenitor cells46 and circulating endothelial progenitor cells (EPCs, derived primarily from the hematopoietic stem cell pool) in LV hypertrophy and in LTL dynamics. It has been postulated that cardiac hypertrophy is angiogenesis-dependent.47-49 The increased muscle mass that is a sine quo non of LV hypertrophy requires a concomitant expansion of a microvascular network in order to meet the increased metabolic demands associated with greater cardiac mass. EPCs are essential for angiogenesis50 and their number increases with LV hypertrophy.51 It follows that adequate hematopoietic stem cell reserves (which sustain the numbers and functions of EPCs) may be necessary for the development of LV hypertrophy. Consistent with this hypothesis, investigators have reported that telomere length may reflect the replicative capacity of hematopoietic stems cells,19,52 and so LTL may serve as a marker of the angiogenic potential needed for LV hypertrophy. It is also of interest, therefore, that EPC reserves might depend on telomere length.53,54
A second explanation may be related to complex role of insulin-like growth factor 1 (IGF-1) in both LV hypertrophy and LTL dynamics. Higher circulating IGF-1 concentrations and up-regulation of the IGF-1 pathway within the heart are observed with LV hypertrophy, whether its etiology is hypertension or acromegaly.55-57 Indeed, IGF-1 has been reported to contribute to a substantial proportion of interindividual variation in LV mass in some reports.58 In addition, systemic administration of IGF-1 causes LV hypertrophy in experimental animals.59 IGF-1 also stimulates vascular endothelial nitric oxide synthase,60 exerting a powerful anti-inflammatory and anti-oxidant effect in concert with increasing numbers of EPCs and retarding atherosclerosis.61 Thus, higher IGF-1 levels may be the critical link between increased LV mass on the one hand, and longer LTL on the other hand (via its anti-inflammatory and anti-oxidant effects).
The moderate-sized community-based sample, and the assessments of LV dimensions and LTL blinded to the results of each other strength our study. Several limitations must be acknowledged. We did not measureIGF-1 levels or EPCs at the index examination to elucidate the mechanisms underlying the observed association. We acknowledge that our hypothesized explanations are speculative and hypothesis generating. The cross-sectional, observational design also precludes any causal inferences and definitive mechanistic insights regarding the association. Also, our sample is comprised of white participants of European ancestry; the generalizability of our observations to other ethnicities is uncertain. It would be of great interest to evaluate the association in African-Americans, who have longer LTL62 and also a greater burden of LV hypertrophy.63
We observed a seemingly paradoxical positive association of LTL with LV mass in our moderate-sized community-based sample. These findings should be replicated in other studies and the mechanisms that might explain this association should be elucidated in experimental studies.
This work was supported by the National Heart, Lung and Blood Institute's Framingham Heart Study (Contract No. N01-HC-25195), 2 K24 HL04334, RO1HL080124 (RSV), 6R01-NS 17950; and R01AG021593 (AA).
Disclosures: None (for any of the authors).