A Comparison of Risk Factors for Calcified Atherosclerotic Plaque in the Coronary, Carotid, and Abdominal Aortic Arteries

Abstract

The extent of shared risk factors for calcified atherosclerotic plaque (CAP) of the coronary, carotid, and abdominal aortic arteries is unknown. CAP was measured by computed tomography in 1,125 individuals in families affected with diabetes. Statistical methods adjusted for the lack of independence between observations. CAP scores were standardized, and tests of interaction were conducted to compare risk factor relations across vascular beds. The average age of the cohort was 61 years, and 84% had diabetes. The correlation in CAP scores across vascular beds ranged from 0.59 to 0.72. Age, albumin/creatinine ratio, hemoglobin A1c, diabetes, hypertension, and lipid-lowering therapy were correlated with quantity of CAP in all vascular beds (all p < 0.05); no differences in the strength of these relations were noted. In contrast, other significant correlates differed in the strength of their relations with CAP. The risk factor pack-years of smoking was most strongly correlated with CAP in the abdominal aorta (p < 0.005). Male gender, previous myocardial infarction, and coronary revascularization were most strongly correlated with CAP in the coronary arteries (p < 0.0001). In summary, CAPs of the coronary, carotid, and abdominal aortic arteries generally share common risk factors, even though several of these factors have a greater impact on CAP in one vascular bed than another.

Calcified atherosclerotic plaque (CAP) of the coronary arteries as measured by computed tomography is associated with prevalence and incidence of cardiovascular disease (1–4), even among subgroups at high risk such as persons with type 2 diabetes mellitus (1,5,6). Its correlates have been described, generally establishing the role of traditional cardiovascular disease risk factors in the etiology of CAP of the coronary arteries (7–9).

Very little is known regarding the distribution and determinants of CAP of the carotid arteries apart from a single study (10). In contrast, CAP of the abdominal aorta has been more extensively evaluated. Four epidemiologic studies have reported associations with cardiovascular disease risk factors (1, 10–12). Furthermore, CAP of the abdominal aorta as quantified by lateral lumbar radiographs is an independent predictor of incident coronary heart disease (13) and congestive heart failure (14).

Atherosclerosis is a systemic disorder, occurring throughout the arterial tree (15). Despite this, the quantity of calcification in atherosclerotic plaques is only moderately correlated across vascular beds, ranging from 0.28 to 0.59 (10–12). This leaves a large portion of the variance of CAP unexplained. Differences in hemodynamic forces across the vasculature are thought to be responsible for differences in responses to systemic risk factors across the vascular beds (16). However, information is limited on the selective atherosclerotic response of different vascular beds to risk factors. Noninvasive methods such as computed tomography provide the opportunity for an assessment of global atherosclerotic risk.

The literature regarding the epidemiology of CAP is lacking in two specific areas. First, there are few reports of risk factor relations with CAP across multiple vascular beds in a single cohort. Second, few data are available that describe the correlates of CAP of the carotid arteries. The present study attempts to overcome these gaps. CAP was quantified in three vascular beds (coronary, carotid, and abdomen) by use of computed tomography, along with an extensive risk factor assessment in a sibling study of 1,125 individuals with and without type 2 diabetes mellitus. The statistical approach used provides a valid comparison of risk factor associations across the three vascular beds.

MATERIALS AND METHODS

The Diabetes Heart Study was conducted in Forsyth County, North Carolina, to study the genetic and epidemiologic origins of cardiovascular disease in families affected with type 2 diabetes mellitus. Siblings concordant for diabetes were recruited from medical clinics and community advertising. Type 2 diabetes mellitus was defined as a physician diagnosis of diabetes after the age of 34 years, in the absence of historical evidence of ketoacidosis. Unaffected siblings, similar in age to the siblings with type 2 diabetes mellitus, were also invited to participate. Individuals with health conditions that would restrict their ability to participate in a 4-hour examination were not invited to participate. Recruitment was based upon family structure, and there were no exclusions based upon prior or current evidence of cardiovascular disease. All protocols were approved by the institutional review board of Wake Forest University School of Medicine, and participants gave informed consent.

The examinations were conducted in the General Clinical Research Center of Wake Forest University School of Medicine between 1999 and 2005. The examinations included interviews for medical history and health behaviors, anthropometric measures, resting blood pressure, fasting blood draw, and spot urine collection. Laboratory assays included urine albumin and creatinine, total cholesterol, high density lipoprotein cholesterol, triglycerides, glucose, and blood chemistries. Low density lipoprotein cholesterol was calculated by use of the Freidewald equation (17); the calculation was considered valid for subjects in whom triglycerides were less than 796 mg/dl (18). C-reactive protein in serum was measured with a high-sensitivity enzyme-linked immunosorbent assay. Hemoglobin A1c was measured by automated ion exchange on high-performance liquid chromatography. Pack-years were quantified among all participants as the average number of packs per day × the total number of years smoked. Never smokers (those who reported having smoked less than 100 cigarettes in their lifetime) were coded as having zero pack-years. Hypertension was defined as current use of antihypertensive medication or having systolic blood pressure greater than 140 mmHg or diastolic blood pressure greater than 90 mmHg. Coronary revascularization included coronary artery bypass graft surgery or angioplasty.

CAP was measured under a standardized protocol (19). Over the course of this study, three versions of the General Electric computed tomography system (CTi, LightSpeed QXi, and Pro16; General Electric Medical Systems, Waukesha, Wisconsin) capable of 520-, 520-, and 244-millisecond temporal resolutions, respectively, when operating in the cardiac mode have been used. The robustness of the CAP score using various computed tomography systems with different temporal resolutions in this range has been established in standardized protocols used at this and other institutions (19–21). In brief, CAP was measured in each of the epicardial coronary arteries (right, left main, left anterior descending and posterior descending) and summed to create the total coronary CAP burden. Two cardiac scans were performed sequentially, and the average of the two measurements was used. For the carotid examination, an unenhanced computed tomography scan was performed through the right and left carotid bifurcations, and CAP was measured in four vascular segments: common, bulb, internal, and external. Specifically, the carotid bifurcation level was identified, and the 15 mm of internal and external carotid above the bifurcation and 30 mm of carotid bulb and common carotid artery below the bifurcation were measured. The entire abdominal aorta and iliac arteries were measured. The proximal segment (suprarenal abdominal aorta) was defined as starting 25 mm proximal to the origin of the superior mesenteric artery. The entire juxtarenal and infrarenal aorta and a 25-mm length of the common iliac arteries below the aortic bifurcation were measured. Scans of the abdominal aorta were initiated 2 years into the study, resulting in a reduced sample size. Computed tomographic examinations were analyzed by experienced analysts producing an Agatston score corrected for slice thickness on a GE Advantage Windows Workstation using the Smart-Scores software package (General Electric Medical Systems). The slice thickness was 2.5–3.0 mm for all scans, and the number of adjacent pixels used to define a calcified plaque was one, resulting in a minimum lesion size of 1 mm². The reproducibility of CAP scores in the coronary and carotid arteries was assessed by obtaining duplicate scans. The inter- and intraobserver variability exceeded 0.96.

For statistical analysis, CAP scores were log transformed after adding a constant and then standardized to a mean of zero and a variance of one to accommodate comparisons across vascular beds. Conditional on the covariates described below, the distribution of the residuals from the models was consistent with conditional normality and did not exhibit significant heterogeneity of variance. To account for the correlations among the observations (i.e., family members, single or multiple observations within an individual), we based tests of significance on generalized estimating equations (GEEs) assuming exchangeable correlation and using the robust estimator of the variance (22). The sandwich estimator estimates the correlations among single or multiple observations of members of the pedigree and is an asymptotically unbiased (i.e., consistent) estimator of the variance used in the tests of significance. The consistency of the sandwich estimator holds even if the correlation structure is misspecified. Thus, the modeling approach is a repeated-measures approach with the three CAP scores within an individual treated as a cluster. An important advantage of using GEEs in the analysis of CAP is that it does not assume conditional normality of the outcome. Standard regression diagnostics for collinearity and influence were computed (23).

To test whether a putative risk factor had differential effects dependent upon the specific vascular bed, a multivariable GEE linear model was computed. Specifically, the standardized CAP score at each vascular bed was entered in the model as an outcome similar to a repeated-measures model. A binary covariate was included in the model as an indicator variable denoting the vascular bed. An interaction between the indicator variable and the risk factor was then computed to test whether the risk factor’s influence on CAP score differed by vascular bed. Thus, a significant interaction indicates that the magnitude of the effect differs by vascular bed. This interpretation is possible since the vascular calcium measures have been standardized. Thus, the regression coefficients can be compared directly. Dichotomous risk factors are summarized as a difference in means of the standardized variable for the two strata (e.g., men vs. women).

Our group and others have reported lower CAP scores in African Americans compared with European Americans (24–26). Furthermore, it is plausible that there are ethnic-specific risk factors that influence CAP. Therefore, analyses are stratified on ethnicity, and tests for ethnic interactions have been conducted as a hypothesis-generating exercise (alpha < 0.10), recognizing that the small sample of African Americans provides only modest power for detecting interactions.

RESULTS

The average age was 62 years in European Americans and 59 years in African Americans (table 1). Nearly all participants in the Diabetes Heart Study had type 2 diabetes mellitus (84 percent overall), and more than 50 percent were current or past smokers. Hypertension was highly prevalent (exceeding 80 percent). Previous myocardial infarction and coronary vascular procedures were reported by two times more European-American than African-American participants.

TABLE 1.

Distribution of cardiovascular risk factors and prevalent cardiovascular disease, Diabetes Heart Study, Forsyth County, NC, 1999–2005^*

	European Americans (n = 950) (mean (SD†) or %)	African Americans (n = 175) (mean (SD) or %)
Age (years)	61.8 (9.4)	58.6 (9.0)
Women (%)	53	68
Body mass index (kg/m2)	31.8 (6.6)	33.7 (7.2)
Systolic blood pressure (mmHg)	138.8 (19.3)	142.9 (20.8)
Diastolic blood pressure (mmHg)	72.6 (10.3)	76.8 (11.6)
Current smoker (%)	16.3	25.4
Past smoker (%)	41.3	37.6
Pack-years	20.9 (28.6)	12.9 (19.8)
Lipid-lowering medications (%)	43	35
Diabetes (%)	83	89
Measures restricted to persons with diabetes
Duration of diabetes (years)	10.5 (7.2)	10.8 (8.0)
Insulin therapy (%)	27	42
Oral hypoglycemic medications (%)	77	67
Laboratory measures
High density lipoprotein cholesterol (mg/dl)	43.4 (12.5)	50.6 (14.8)
Low density lipoprotein cholesterol (mg/dl)	105.1 (33.0)	113.5 (33.3)
Triglycerides (mg/dl)	203.2 (128.2)	133.4 (92.4)
Hemoglobin A1c (%)	7.3 (1.7)	8.5 (2.7)
Fasting glucose (mg/dl)	140.9 (57.7)	153.2 (80.1)
Serum creatinine (mg/dl)	1.1 (0.28)	1.2 (0.44)
Albumin/creatinine ratio (mg/g), median	11.7	16.4
Albuminuria (albumin/creatinine ratio: >30 mg/g) (%)	28.7	36.5
C-reactive protein (mg/dl)	0.62 (0.95)	0.69 (0.76)
Self-reported cardiovascular conditions and procedures
Hypertension (%)	81.6	85.1
Stroke (%)	9.4	6.4
Myocardial infarction (%)	19.8	8.7
Coronary artery bypass graft (%)	14.5	5.1
Coronary angioplasty (%)	15.5	7.5
Carotid endarterectomy (%)	2.2	0

^*Sample sizes are reduced for European Americans by no more than 21 except for C-reactive protein (n = 826) and for African Americans by no more than eight except for C-reactive protein (n = 144).

^† SD, standard deviation.

The prevalence of CAP (score: >0) was highest for the coronary arteries and abdominal aorta (approximately 90 percent for women and 92–98 percent for men) (table 2). Prevalence was similar for European and African Americans. The prevalence of CAP in the carotid arteries was lower overall (65–71 percent for women and 57–80 percent for men), being somewhat lower in African than in European Americans. Median scores of CAP were higher in European- than in African-American participants for all vascular beds.

TABLE 2.

Distribution of calcified atherosclerotic plaque scores in the coronary arteries, carotid artery, and abdominal aorta, Diabetes Heart Study, Forsyth County, NC, 1999–2005

	European Americans		African Americans
	Men	Women	Men	Women
Coronary arteries
No.	445	502	56	119
Mean (SD*)	2,982 (4,352)	820 (1,505)	1,233 (2,583)	682 (1,490)
Median	1,335	104	166	57
% >0	98	90	95	91
Carotid artery
No.	446	493	56	117
Mean (SD)	464 (763)	242 (607)	195 (585)	184 (618)
Median	170	24	15	13
% >0	80	71	57	65
Abdominal aorta
No.	330	343	50	93
Mean (SD)	4,854 (5,474)	2,524 (3,410)	1,717 (3,834)	1,742 (2,731)
Median	2,773	1,105	547	488
% >0	96	90	92	81

^*SD, standard deviation.

CAP scores were significantly correlated across vascular beds (p < 0.0001). Spearman’s correlation coefficients were highest between the abdominal and carotid arteries (0.72 and 0.67 in European Americans and African Americans, respectively). Similarly, the correlations between the abdominal and coronary arteries were 0.67 and 0.61, being lower between the carotid and the coronary arteries (r = 0.59 and 0.53).

Among European Americans, age, pack-years, albumin/creatinine ratio, and hemoglobin A1c were the strongest and most consistent correlates of the CAP scores across the three vascular beds (all p < 0.01; r ranging from 0.13 to 0.53) (table 3). Duration of diabetes, triglycerides, high density lipoprotein cholesterol, C-reactive protein, and body mass index were correlated modestly but significantly with CAP in the coronary and carotid arteries (all p < 0.05; r ranging from 0.08 to 0.13). Low density lipoprotein cholesterol was inversely correlated with CAP in the coronary arteries (p < 0.05). Among the dichotomous variables, male gender, diabetes, hypertension, myocardial infarction, revascularization, and lipid-lowering therapy were all associated with increased CAP scores in all vascular beds (table 4). For example, CAP in the coronary arteries was, on average, 2.1 standard deviations greater in men than in women (p < 0.0001).

TABLE 3.

Age- and gender-adjusted Spearman’s correlation coefficients between risk factors and calcified atherosclerotic plaque scores, Diabetes Heart Study, Forsyth County, NC, 1999–2005

	European Americans			African Americans
	Coronary arteries (n = 947)	Carotid artery (n = 939)	Abdominal aorta (n = 673)	Coronary arteries (n = 175)	Carotid artery (n = 173)	Abdominal aorta (n = 143)
Age	0.38***	0.46***	0.53***	0.33***	0.45***	0.48***
Duration of type 2 diabetes mellitus (among type 2 diabetics only)	0.13**	0.12**	0.07	0.17*	0.20*	0.19*
Pack-years	0.15***	0.16***	0.35***,†,‡	0.13	0.09§	0.33***
Triglycerides	0.09*	0.08*	0.07	0.04	0.03	0.08
High density lipoprotein cholesterol	−0.10*	−0.08*	−0.07	−0.10	0.00§	−0.19*
Low density lipoprotein cholesterol	−0.08*	−0.02	−0.01	0.01§	−0.01	0.03
Albumin/creatinine ratio	0.20***	0.23***	0.19***	0.15	0.21*	0.13
Hemoglobin A1c	0.22***	0.16***	0.13**	0.05§	0.15*	0.08
C-reactive protein	0.10*	0.09*	0.07	−0.06	−0.15§	−0.11
Body mass index	0.13***	0.11**	0.01	0.14	0.09	−0.09†

^*p < 0.05;

^**p < 0.001;

^***p < 0.0001 (tests of significance for difference from zero).

^† Correlation differs between aorta and coronary arteries, within ethnic group (pack-years in European Americans, p = 0.005; body mass index in African Americans, p = 0.04) (tests of significance for interaction between vascular bed and risk factor within ethnic group).

^‡ Correlation differs between aorta and carotid artery, within ethnic group (pack-years in European Americans, p < 0.0001) (tests of significance for interaction between vascular bed and risk factor within ethnic group).

^§ Correlation differs between European Americans and African Americans, within vascular bed, p < 0.10 (tests of significance for interaction between ethnicity and risk factor within vascular bed).

TABLE 4.

Difference between the means of standardized coefficients for dichotomous variables and age-and gender-adjusted tests of significance, Diabetes Heart Study, Forsyth County, NC, 1999–2005

	European Americans			African Americans
	Coronary arteries (n = 947)	Carotid artery (n = 939)	Abdominal aorta (n = 673)	Coronary arteries (n = 175)	Carotid artery (n = 173)	Abdominal aorta (n = 143)
Gender (men vs. women)	2.1***,†,‡	1.1***	1.0***	0.8*,§	0.0§	0.0§
Diabetes	2.0***	1.4***	1.0***	1.5*,‡	1.4*	0.3
Hypertension	1.5***	1.7***	0.9***	1.1*	1.6**	1.2*
Myocardial infarction	2.6***,†,‡	1.7**	1.5***	2.3*	1.3	1.6*
Coronary revascularization	3.0***,†,‡	1.7***	1.5***	2.6***	2.1**	2.4***,§
Lipid-lowering medications	1.3***	0.9***	0.9***	1.2*	1.5**	1.2**

^*p < 0.05;

^**p < 0.001;

^***p < 0.0001 (tests of significance for difference from zero).

^†Difference between the means of standardized coefficients differs between coronary arteries and carotid artery, within ethnic group (gender, myocardial infarction, and coronary revascularization in European Americans, p < 0.001) (tests of significance for interaction between vascular bed and risk factor within ethnic group).

^‡Difference between the means of standardized coefficients differs between coronary arteries and aorta, within ethnic group (gender, myocardial infarction, and coronary revascularization in European Americans, p < 0.001; diabetes in African Americans, p = 0.03) (tests of significance for interaction between vascular bed and risk factor within ethnic group).

^§ Difference between the means of standardized coefficients differs between European Americans and African Americans, within vascular bed, p < 0.05 (tests of significance for interaction between ethnicity and risk factor within vascular bed).

Among European Americans, there were few instances where the risk factor relations differed across vascular beds. Pack-years was a stronger correlate of CAP in the abdominal aorta than in either the coronary or carotid artery (p < 0.005) (table 3). Male gender, myocardial infarction, and revascularization were stronger correlates of CAP in the coronary arteries than in either the carotid artery or the abdominal aorta (p < 0.001) (table 4).

Because of a considerably smaller sample size, there were few statistically significant correlations between risk factors and CAP score among African Americans (table 3). Age and duration of diabetes were significantly and consistently associated with vascular calcium in all vascular beds (p < 0.05). Pack-years and high density lipoprotein cholesterol (inverse) were associated with CAP in the abdominal aorta (p < 0.05), and albumin/creatinine ratio and hemoglobin A1c were associated with CAP in the carotid artery. Among the dichotomous variables, hypertension, revascularization, and lipid-lowering therapy were associated with increased CAP in all vascular beds (table 4).

Among African Americans, there were two instances where the risk factor relations differed statistically across vascular beds. The correlation between body mass index and CAP in the coronary arteries was different from that between body mass index and CAP in the abdominal aorta (table 3) (r = 0.14 vs. r = −0.09; p = 0.04). Diabetes was more strongly related to CAP in the coronary arteries than in the abdominal aorta (table 4) (p < 0.05).

A test for ethnic interactions indicated that the risk factor–CAP score relation differed between European Americans and African Americans for pack-years and CAP in the carotid artery (r = 0.16 vs. 0.09), high density lipoprotein cholesterol and CAP in the carotid artery (r = −0.08 vs. 0.00), low density lipoprotein cholesterol and CAP in the coronary arteries (r = −0.08 vs. 0.01), and hemoglobin A1c and CAP in the coronary arteries (r = 0.22 vs. 0.05) (table 3). Notably, correlations between C-reactive protein and CAP scores were all positive in European Americans but were all inverse in African Americans, being significantly different from one another only in the carotid artery (r = 0.09 vs. −0.15). Among the dichotomous variables, male gender was strongly associated with CAP in all vascular beds in European Americans but not in African Americans (table 4). Also, the revascularization effect in the abdominal aorta was stronger in African Americans.

DISCUSSION

The research question examined in this report was whether risk factors for CAP are shared across the three different vascular beds. Our motivation was that, despite being a systemic process, the atherosclerosis burden is only moderately concordant across the vascular beds as evidenced by reports that correlation coefficients between CAP scores range from 0.28 to 0.59. This suggests that the variation in CAP score in one vascular bed explains only up to 35 percent of the variation in another vascular bed. It has also been hypothesized that regional differences in hemodynamic profiles lead the endothelium to respond distinctly to systemic risk factors such as hypercholesterolemia and smoking (16). However, upon evaluation of this question, we observed few instances where the relation between risk factors and CAP differed across the vascular beds. For the most part, risk factors were associated with the extent of CAP in a consistent, systemic manner. These risk factors included age, albumin/creatinine ratio, hemoglobin A1c, diabetes, hypertension, and lipid-lowering therapy. There were several exceptions. The risk factors that had significantly different relations across the vascular beds in European Americans were smoking, gender, myocardial infarction, and revascularization. Smoking was a significant correlate of increased CAP score in all three vascular beds. However, the relation between smoking and CAP was significantly stronger in the abdominal aorta than in the carotid or coronary arteries. This finding extends the one other report of systematic calcified atherosclerosis in which smoking was most strongly associated with calcium in the distal aorta and iliac arteries (10). An autopsy-based study in young adults also observed that smoking selectively increased the raised lesions of the abdominal aorta while not influencing the atherosclerotic burden of the coronary artery (27). Thus, smoking has a differential impact on atherosclerosis burden across the vasculature, now confirmed using a computed tomography measure of CAP.

This is the first report to formally contrast risk factor profiles of CAP across multiple vascular beds using rigorous statistical testing. To make comparisons across the vascular beds on the same scale, the CAP scores were first transformed to better approximate conditional normality and then standardized to a mean of zero and a variance of one. Consequently, we are able to compare correlation coefficients (for continuous variables) or differences in means (for dichotomous variables) directly across the vascular beds. Only one previous report presented the patterns and risk factors of calcified atherosclerosis in multiple vascular beds (10).

Our report adds to the limited literature on the extent of and risk factors for CAP in the carotid artery (10). We observed a lower prevalence of CAP in the carotid artery compared with the coronary arteries and abdominal aorta. Furthermore, we observed that several traditional, as well as novel, risk factors were significant correlates of CAP in the carotid artery in European Americans, specifically male gender, age, duration of diabetes, smoking, triglycerides, reduced high density lipoprotein cholesterol, albumin/creatinine ratio, hemoglobin A1c, C-reactive protein, body mass index, hypertension, myocardial infarction, revascularization, and lipid-lowering therapy. Low density lipoprotein cholesterol was not associated with CAP in the carotid artery. In African Americans, point estimates were of similar magnitude with the exception of pack-years and high density lipoprotein cholesterol, in which the point estimates were significantly weaker and in which no gender difference in CAP was observed. Furthermore, in our formal analysis in European Americans, there were three risk factors for which significantly weaker associations were observed for CAP in the carotid artery compared with CAP in the coronary artery: male gender, myocardial infarction, and revascularization. Apart from these findings, the risk factor profiles are generally similar between the carotid and coronary arteries.

This report also adds to a growing literature regarding the epidemiology of CAP in the aorta as assessed by computed tomography (1, 10–12). Similar to CAP in the coronary arteries, the prevalence of CAP in the abdominal aorta is quite high, generally exceeding 90 percent. CAP in the abdominal aorta of European Americans shared many risk factors with the other vascular beds. The unique contribution made by the current report is its formal statistical comparison of risk factor relations across vascular beds. Notably, the few differences that were observed comparing the strength of the risk factor relations across beds often identified CAP in the abdominal aorta as one that differed. For example, among European Americans, smoking was a stronger correlate of CAP in the abdominal aorta than of CAP in either the carotid or coronary arteries. Furthermore, gender, myocardial infarction, and revascularization were associated with smaller increases in CAP in the abdominal aorta than in the coronary arteries. In African Americans, body mass index and diabetes had different relations with CAP in the coronary arteries than in the abdominal aorta. These findings suggest that the epidemiology of CAP in the abdominal aorta differs somewhat from that of other vascular beds.

Racial differences in coronary heart disease risk have been studied extensively. In one report in which African Americans and European Americans were followed for nearly 6 years, African Americans suffered more coronary heart disease events despite a lower prevalence of coronary calcium (28). This and other observations led to our hypothesis that the relation between risk factors and CAP would differ between European Americans and African Americans. Because of the smaller sample size of African Americans, there was limited power to detect interactions. Despite this, several statistically significant interactions were observed. Pack-years of smoking and high density lipoprotein cholesterol were weaker correlates of CAP in the carotid artery in African Americans compared with European Americans. Hemoglobin A1c and low density lipoprotein cholesterol were weaker correlates of CAP in the coronary arteries. Male gender was not associated with CAP in the carotid or abdominal aorta in African Americans, in contrast to the highly significant association in European Americans. Interestingly, C-reactive protein was inversely correlated with CAP in all vascular beds in African Americans but positively correlated in European Americans, being significantly different between races only in the carotid artery. These findings are suggestive of a pattern of differential risk factor relations with CAP, which requires further exploration. Only one other study has contrasted risk factor relations for vascular calcium across ethnic groups (29). Age, smoking, hypertension, and diabetes were consistently associated with coronary artery calcium in African Americans and European Americans, while the relations among body mass index, low density lipoprotein cholesterol, and high density lipoprotein cholesterol were weaker in African Americans. Limitations of that report are the lack of formal statistical testing for interactions and an analysis that was restricted to presence/absence of calcium. Together, these findings may be valuable to understanding the lower prevalence of CAP and the weaker link between CAP and subsequent cardiovascular disease events in African Americans relative to European Americans.

A cohort comprised primarily of persons with type 2 diabetes mellitus provides both strengths and limitations. The high prevalence and quantity of CAP in all vascular beds provide an advantage to the statistical approaches used herein, allowing an analysis of the quantity of CAP. Many studies are limited to an analysis of a dichotomous variable of presence of CAP. Similarly, the wide variance of hemoglobin A1c and the albumin/creatinine ratio has allowed a more rigorous evaluation of these factors.

A cohort composed primarily of persons with type 2 diabetes mellitus is also a limitation. In particular, the unique nature of the cohort may have led to spurious findings. First, the homogeneity of the sample, as evidenced by the higher correlation of CAP scores across vascular beds than has previously been reported, may have limited our ability to detect a differential impact of risk factors on CAP across the vascular beds. Second, the inverse relation between CAP in the coronary arteries and low density lipoprotein cholesterol may be a spurious finding. Studies of samples not selected for diabetes report a significant positive relation between CAP of the coronary arteries and low density lipoprotein cholesterol (30), yet studies of samples selected for diabetes observe the inverse association as we have (1, 31). In the latter study, as in ours, statin therapy was positively associated with CAP score. Thus, the likely explanation for the inverse relation between CAP and low density lipoprotein cholesterol is the pharmacologic treatment with lipid-lowering medications among persons with type 2 diabetes mellitus. Thus, cohorts with type 2 diabetes mellitus provide valuable information regarding the etiology of vascular calcification particularly due to the extent of disease, but they may lead to spurious findings due to concomitant conditions.

In summary, we have conducted a systematic analysis of risk factors for CAP in three vascular beds in a large epidemiologic cohort. There were few instances in which the risk factor relations differed across the vascular beds; the differences generally identified CAP of the abdominal aorta as unique. In addition, we presented the distribution and determinants of CAP in the carotid artery, adding to an extremely small literature on this topic. Finally, although statistical power was limited, we observed several instances wherein risk factor relations with CAP differed between African Americans and European Americans.

Abbreviations

CAP

calcified atherosclerotic plaque

GEE

generalized estimating equation