Leukocyte Telomere Length and Coronary Artery Calcium

Abstract

Patients with a history of myocardial infarction display shortened leukocyte telomere length (LTL), but conflicting findings have been reported on the relation between LTL and subclinical coronary artery atherosclerosis, as expressed in coronary artery calcium (CAC). We examined the relation between LTL, measured by Southern blots, with CAC in 3,169 participants in the NHLBI Family Heart Study. Participants consisted of 2,556 whites, 613 blacks, 1,790 women and 1,379 men. The odds of having CAC≥100 for the shortest LTL tertile versus the longest LTL tertile were 1.95 (1.28, 3.16) in white males and 1.76 (1.18, 2.45) in white females, after adjusting for multiple covariates of CAC. The corresponding odd ratios for blacks were 1.53 (0.67, 3.50) and 0.87 (0.37, 2.00). Tests for trend across LTL tertiles were p=0.002 in white males, p=0.006 in white females, p=0.32 in black males and p=0.74 in black females. The associations, or lack of associations, were independent of C-reactive protein levels and other risk factors for CAC. As previously shown in other studies, whites displayed a shorter LTL than blacks (p<0.0001). In conclusion, the higher the coronary artery atherosclerotic burden in whites, the shorter the LTL. This LTL-atherosclerosis connection is not found in blacks. The mechanisms for the racial difference in LTL, CAC and their inter-relations do not seem to be related to inflammation and merit further research.

Both cardiovascular disease (CVD), principally in the form of atherosclerosis, and leukocyte telomere length (LTL) are associated with aging and ostensibly with each other.¹ A recent meta-analysis, although confirming that individuals who have suffered myocardial infarction have a shorter LTL than their peers, failed to show a significant association of LTL with coronary artery disease (CAD), the root cause of most cases of myocardial infarction.² Coronary artery calcium (CAC) is a highly sensitive and specific index of CAD risk, since arterial calcium is pathognomonic of atherosclerosis and also a sensitive indicator of atherosclerotic plaque burden in the coronary arteries.³ A body of research indicates that independent of traditional risk factors, CAC predicts myocardial infarction.^4–7 Even in the presence of multiple cardiac risk factors, persons with no evidence of CAC have low near-term risk of having myocardial infarction.⁵ Here we present the relation of LTL with CAC based on a cohort comprising 3,169 participants in the National Heart, Lung and Blood Institute (NHLBI)-Family Heart Study. This work has also examined whether gender, ethnicity (black or white) and high-sensitivity serum CRP affect the association of LTL with CAC.

Methods

The NHLBI Family Heart Study is a multi-center investigation of the genetic and epidemiological basis of cardiovascular disease.⁸ Between January 2002 and January 2004, 3,359 subjects were examined (2,737 whites, 622 blacks). These subjects had either participated in a prior clinic visit between February 1994 and March 1996 or were newly recruited blacks from the Hypertension Genetic Epidemiology Network (HyperGEN) field center in Birmingham Alabama.⁹ The study protocol was approved by Institutional Review Boards at each of the participating centers and each participant gave written informed consent.

EDTA blood was stored at −70 °C until shipment to the Center of Human Development and Aging at the New Jersey Medical School, Rutgers, for LTL measurement. DNA was extracted by Gentra Puregene Blood Kit (Qiagen, Valencia, CA) and samples were subjected to DNA integrity tests. LTL was derived from the mean length of the terminal restriction fragments, determined by Southern blot analysis.¹⁰ The inter-assay coefficient of variation for duplicate samples, which were resolved on different gels on different occasions, was 2.4%.

CAC measurements were obtained from 5 field centers using 4-slice multi-detector CT systems without contrast (General Electric Health Systems LightSpeed Plus & LightSpeed Ultra, Siemens Volume Zoom, or Philips MX 8000). The protocol and scoring (Agatston score) at a central reading lab have been previously described, and was consistent across centers.⁴

Measurements of glucose and lipid levels were performed on fasting blood samples. Thirty-six subjects who were fasting for <6 hours were excluded from all analyses. High-sensitivity serum CRP was measured at the Laboratory for Clinical Biochemistry Research, University of Vermont by an enzyme-linked immunosorbent assay method calibrated with World Health Organization reference material. ¹¹ The New Jersey and Vermont laboratories were blinded to the identity of the subjects.

Packs-years of smoking were obtained by a questionnaire. An automated blood pressure device (Dinamap Monitor Pro 100) was used to record 3 measurements of blood pressure, 1 minute apart. Hypertension was defined as currently on antihypertensive medications or untreated blood pressure of ≥140/90 mmHg. Medication use for hypertension or dyslipidemia was obtained from each subject and coded as current or not for this analysis.

Subject characteristics were compared across gender and race categories using generalized estimating equations and an exchangeable correlation matrix to model the relatedness of the subjects (PROC GENMOD, SAS Institute, Cary, NC). To examine the mean LTL by 4 different CAC categories (0, 1–100, 101–400, and ≥400 Agatston units), we first adjusted LTL for age and analyzed the means of the residuals after the overall mean was added back to each residual. A test for a linear trend across the categories was performed using contrasts in this same model. Tertiles of LTL were analyzed for the primary association tests so that a linear relationship between LTL and CAC did not need to be assumed.

To predict the risk of developing significant coronary atherosclerotic calcium, CAC was modeled as a dichotomous dependent variable (CAC≥100 vs. CAC<100) in a logistic general linear model that incorporated the effects of LTL nested within gender and race categories as the independent variable, the relatedness of the sample, and multiple covariates that have been shown to be related to the development of CAC. Generalized estimating equations with an exchangeable correlation matrix were used to adjust for relatedness of the sample. Two sets of covariates were used for additional statistical adjustment. First, adjustment was performed using age and BMI only as covariates. The second set of covariates included age, BMI, pack-years of smoking, glucose, LDL-C, HDL-C, triglycerides, mean arterial blood pressure, blood pressure medication status, and CRP. Gender-combined models included an adjustment for gender.

Results

A total of 3,169 of 3,359 subjects had an LTL measurement. The 190 subjects without LTL were those who did not provide a blood sample to get DNA, had too little DNA for LTL measurement, or whose samples failed the DNA integrity tests. Table 1 shows the characteristics of the subjects. Women were slightly older, had lower diastolic blood pressure, higher CRP and HDL-cholesterol but lower fasting glucose and triglycerides than men. Blacks were younger and displayed higher BMI, systolic and diastolic blood pressures, higher CRP, fasting glucose and HDL-C, and lower triglycerides than whites. Fewer women than men were smokers. Compared with whites, more blacks were current smokers and on anti-hypertensive medications and fewer were on lipid-lowering drugs.

Table 1.

Subject Characteristics in the NHLBI Family Heart Study

Variable	Black		White		p-value*
	Men (N=215)	Women (N=398)	Men (N=1164)	Women (N=1392)	Gender Effect*	Race Effect*
Age (year)	52±0.7	54±0.6	57±0.4	58±0.4	0.01	<0.0001
Body mass index (kg/m2)	30.4±0.4	33.8±0.4	29.3±0.2	28.4±0.2	0.47	<0.0001
Systolic blood pressure (mmHg)	132±1.4	133±1.2	122±0.5	119±0.6	<0.0001	<0.0001
Diastolic blood pressure (mmHg)	79±0.8	74±0.6	73±0.3	67±0.3	<0.0001	<0.0001
Coronary arterial calcium (Agatston)	342±44	166±19	521±29	125±13	<0.0001	0.09
Leukocyte telomere length (kb)	6.85±0.04	7.11±0.03	6.73±0.02	6.87±0.02	<0.0001	<0.0001
C-reactive protein (mg/l)	4.6±0.4	7.6±0.5	2.8±0.1	4.1±0.2	<0.0001	<0.0001
Glucose (mg/dl)	116±3.0	111±2.2	103±0.8	97±0.6	<0.0001	<0.0001
Low density lipoprotein cholesterol (mg/dl)	113±2.6	116±1.8	109±1.0	112±1.1	0.01	0.03
High density lipoprotein (mg/dl)	47±1.1	57+0.7	42±0.4	54±0.5	<0.0001	<0.0001
Triglycerides (mg/dl)	117±5.7	110±3.9	152±3.2	136±2.4	<0.0001	<0.0001
Pack-years of smoking	15.4±1.4	7.0±0.6	15.9±0.8	7.6±0.5	<0.0001	0.49
Antihypertensive drugs	56%	70%	33%	30%	0.37	<0.0001
Lipid lowering drugs	20%	15%	26%	21%	<0.0001	0.16

^*p-value from two-way ANOVA test for gender and race, after adjusting for age.

Data are presented as mean ± SEM

Age-adjusted LTL was longer in women than in men (6.98±0.02 vs. 6.81±0.02, p<0.0001) and in blacks than in whites (7.01±0.03 vs. 6.78±0.02, p<0.0001). Shorter LTL was significantly related to increased CAC category, but only in whites (Table 2). Blacks with either CAC=0 or CAC≥400 had longer LTL than the 2 intermediate CAC categories, resulting in more of a quadratic relationship between the 2 variables. When age-adjusted LTL is divided by tertiles, a shorter LTL predicted an increased probability that CAC≥100 in both white men and women after adjustment for age and BMI (Figure 1A). Odd ratios for CAC≥100 for the shorter LTL tertile versus the longer LTL tertile were 2.08 (1.38, 3.13) for white males and 1.55 (1.06, 2.28) for white females after age and BMI adjustment. After including other covariates that have been related to CAC in the model (mean arterial pressure, LDL-C, HDL-C, triglycerides, pack years of smoking, C reactive protein), the previous odds ratios were only slightly reduced and remained significant (1.95 (1.28, 3.16) for white males and 1.76 (1.18, 2.45) for white females). The tests for linear trend across the 3 LTL tertiles were p=0.002 for white males and p=0.006 for white females. The odds ratios for blacks (Figure 1B) were not significant (1.53 (0.67,3.50) for black males and 0.87 (0.37, 2.00) for black females) in the fully adjusted model. The tests for linear trends across the LTL tertiles in blacks were p=0.32 for males and p=0.74 for females.

Table 2.

Age-Adjusted Leukocyte Telomere Length (kb) by Coronary Arterial Calcium (CAC, Agatston score) Category, Sex, and Race

CAC	Black			White
	Men	Women	Both	Men	Women	Both
0	6.98±0.08 (82)	7.20±0.04 (205)	7.08±0.04 (287)	6.78±0.04 (300)	6.85±0.03 (721)	6.80±0.02 (1021)
1–100	6.92±0.07 (71)	7.16±0.06 (127)	7.04±0.05 (198)	6.78±0.03 (371)	6.88±0.03 (380)	6.81±0.02 (751)
101–400	6.90±0.10 (29)	7.15±0.09 (36)	7.02±0.07 (65)	6.57±0.04 (154)	6.77±0.05 (147)	6.69±0.04 (301)
≥400	6.97±0.10 (33)	7.23±0.10 (30)	7.10±0.07 (63)	6.61±0.04 (339)	6.74±0.06 (144)	6.70±0.03 (483)
p*	0.91	0.80	0.90	0.0003	0.036	0.002

Sample size is given for each category in parentheses. Mean leukocyte telomere length adjusted for age and sex when combining men and women. Data are presented as mean ± SEM.

^*Test for linear trend across CAC categories

Figure 1.

Figures 1A and 1B. Odds ratios of coronary calcium (≥100 versus <100 Agatston units) by tertile of leukocyte telomere length (LTL): (1A) white males and females and (1B) black males and females. The longest LTL tertile is used as the referent group. Odds ratios and 95% confidence intervals are shown after adjustment for either age and BMI or adjustment for age, BMI, mean arterial pressure, LDL-C, HDL-C, triglycerides, pack years of smoking, C reactive protein and use of antihypertensive or lipid lowering medications.

CRP was not significantly associated with CAC in the fully adjusted models for either race, and in addition CRP was also not related to LTL (p=0.65 in whites and p=0.40 in blacks, after adjusting for gender, age, BMI and pack years). Age and pack-years of smoking were significantly related to CAC for both genders in the whites, with BMI also associated in white females and mean arterial pressure associated in white males.

Discussion

A longer LTL in women than men¹² and in blacks than in whites^13,14 has been consistently observed when LTL measurements were performed by Southern blots. In the present study, these sex and race differences in LTL were confirmed. Furthermore, we observed an inverse relation between LTL and CAC for only whites, whose age-adjusted LTL between individuals (both men and women) with CAC scores <100 Agatston units vs. those with CAC scores ≥100 Agatston units amounted to about 110 base pairs (Table 2). This compares with average differences of 170 base pairs between men and women and 230 base pairs between whites and blacks. Thus, factors that affect LTL dynamics (LTL at birth and age-dependent shortening afterward) and their relation with CAC may differ in nature and magnitude between blacks and whites. However, the nature of these factors is poorly understood.

A recent meta-analysis has confirmed that subjects who experienced myocardial infarction were likely to have a shorter LTL than their peers.² However, the meta-analysis failed to show a significant association of short LTL with CAD, the principal cause of myocardial infarction. In addition, 2 studies^15,16 observed an inverse relation of LTL with CAC, while a third study¹⁷ did not. One of the studies examined both whites and blacks.¹⁶ It observed a lower strength of the association between LTL and CAC in blacks than in whites; in blacks this association was not statistically significant. These studies comprised relatively small sample sizes. The findings of the present study clearly establish that in whites LTL is inversely associated with CAC, a finding that is not observed in blacks.

CAC scores are typically lower in women than in men and in blacks than for whites.^18–21 Previous studies have also found that CRP levels were lower in men than in women and in whites than in blacks.^22–24 These sex and race-related differences in CAC and CRP were observed in the present work. The ultimate trigger of myocardial infarction is often the disruption of the atherosclerotic plaque, a process that is prompted by inflammation and thrombosis^25,26 Accordingly, CRP has been the focus of studies that have examined the ability of inflammatory biomarkers to predict this eventuality, but its value for clinical practice is a matter of debate.^26–28 CRP appears to predict myocardial infarction,²⁸ but it is a weak index of coronary atherosclerosis, especially as represented by CAC.²⁹ In the present study we found no relations between LTL and CRP and between CRP and CAC.

The large sample size of the present study and the use of the more precise Southern blot method to measure LTL,³⁰ we suggest, provide confidence in the validity of the findings. Nevertheless, we would like to underscore the study limitations. First, this is a cross-sectional analysis. A longitudinal study that monitors LTL and CAC score over many years might be more informative. Second, there were fewer blacks than whites and, on average, blacks were younger than whites. Therefore, there was less chronological time for LTL to shorten and CAC to develop in blacks. That being said, the sample size for the blacks was much larger than previously published studies and the odds ratios were much lower and more inconsistent across LTL tertiles than for whites. Nevertheless, we cannot rule out smaller risks in blacks compared with whites of high CAC levels with shorter telomeres.