Author + information
- Received June 7, 2010
- Revision received August 11, 2010
- Accepted October 12, 2010
- Published online January 1, 2011.
- Kamakki Banks, MD, MPH†,‡,
- Divijani Puttagunta, MD†,
- Sabina Murphy, MS⁎,
- Monica Lo, MD†,‡,
- Darren K. McGuire, MD, MHSc⁎,†,‡,
- James A. de Lemos, MD⁎,†,‡,
- Alice Y. Chang, MD, MSCS⁎,†,§,
- Scott M. Grundy, MD, PhD⁎∥ and
- Amit Khera, MD, MSc⁎,†,‡,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Amit Khera, Division of Cardiology, University of Texas Southwestern Medical Center, 5909 Harry Hines Boulevard, Room HA9.133, Dallas, Texas 75390-9047
Objectives We sought to evaluate the relationship between angina and coronary artery calcium (CAC) in women, and among women without CAC, the associations between angina and clinical, vascular, and inflammatory factors.
Background Angina in women without coronary atherosclerosis is associated with significant morbidity, yet its determinants are poorly understood.
Methods Women ages 30 to 65 years from the Dallas Heart Study, a multiethnic probability sample of Dallas County residents, who completed a Rose angina questionnaire and had complete data for CAC by computed tomography were selected for this analysis. Soluble intercellular adhesion molecule-1, soluble vascular cell adhesion molecule-1, high sensitivity C-reactive protein, monocyte chemoattractant protein-1 and aortic compliance by magnetic resonance imaging were measured.
Results Among the 1,480 women in this cohort (mean age 45 years, 49% African-American), angina was present in 6.9% but was not associated with CAC (19% CAC prevalence with angina vs. 15% without, p = 0.2). Among women without CAC, angina was related to variables reflecting obesity and insulin resistance and was independently associated with African-American ethnicity, premature family history of myocardial infarction, and waist circumference (all p < 0.05). Such women with angina also had higher levels of soluble intercellular adhesion molecule-1 (668 vs. 592 ng/ml, p = 0.02) and soluble vascular cell adhesion molecule-1 (1,106 vs. 968 ng/ml, p = 0.01) and reduced aortic compliance (mean 22 vs. 26 ml/mm Hg, p = 0.007) than such women without angina. Conversely, there was no difference in C-reactive protein or monocyte chemoattractant protein-1 levels for women with and without angina (p = not significant, each).
Conclusions Angina among women in the general population is common and is not associated with subclinical atherosclerosis. Additionally, angina in the absence of subclinical atherosclerosis is not related to many traditional atherosclerotic risk factors but is associated with clinical, inflammatory, and vascular factors that reflect endothelial dysfunction and vascular stiffness, suggesting a distinct vascular etiology and alternative potential therapeutic targets.
- aortic compliance
- coronary artery calcium
- soluble intracellular adhesion molecule
- soluble intravascular adhesion molecule
Angina in women is a morbid condition for which the determinants are poorly characterized. More than 50% of women with angina may have normal or near-normal coronary arteries visualized at coronary angiography, a rate much higher than is observed among men (1). Coronary angiography, however, is an insensitive measure of atherosclerosis burden (2), and some have hypothesized that subclinical atherosclerosis may promote microvascular dysfunction and ischemia in women with angina in the absence of obstructive epicardial coronary disease (3). Presently, there are few data regarding the relationship between angina and subclinical atherosclerosis in women to evaluate this hypothesis.
In women with angina but without angiographic obstructive coronary artery disease (CAD), recent studies have revealed the considerable morbidity suffered by these women despite low overall mortality. In the Women's Ischemia Syndrome Evaluation study, 46% of women with chest pain and negative coronary angiograms had continued anginal symptoms at 5-year follow-up, and 20% had symptoms severe enough for rehospitalization. Additionally, lifetime angina-related healthcare costs for a woman in this population were estimated to be >$700,000 (1). Despite the significant morbidity accompanying this illness, the determinants of angina in women without obstructive CAD are still poorly defined (4). Most studies evaluating this entity are limited by small sample sizes and referral bias inherent in studying cohorts of patients referred for angiography on clinical grounds. Given the high negative predictive value of normal coronary artery calcium (CAC) scans for obstructive CAD (5), coupling CAC scanning and angina data in a population-based cohort can provide novel insights into this entity.
The objectives of this study are twofold: 1) to assess the relationship between angina and subclinical atherosclerosis in women; and 2) to evaluate the clinical and biological factors associated with angina in women without CAC in a multiethnic, population-based sample.
The Dallas Heart Study is a multiethnic, population-based, probability sample of subjects in Dallas County designed to examine cardiovascular disease. Details of study design and variable definitions have been published previously (6). African-American subjects were deliberately oversampled to constitute 50% of the study population. All participants provided written informed consent to participate in the study, and the protocol was approved by the institutional review board of the University of Texas Southwestern Medical Center at Dallas. Briefly, the first in-home visit comprised 6,101 participants ages 18 to 65 years who completed a detailed interview with attainment of demographic and medical history. Subjects between the ages of 30 years and 65 years who completed the initial visit were invited to participate in a second visit, which included 3,398 who provided fasting venous blood and urine samples. Subsequently, 2,971 participants returned for various imaging procedures including cardiac electron-beam computed tomography (EBCT) scanning and magnetic resonance imaging during a third visit. There were no significant differences in demographics and clinical variables between subjects participating in the 3 visits (6). Sampling weights, reflecting the different probabilities of selection for participants and sample attrition between visits, were constructed to generate unbiased estimates of population frequencies (6). The present analyses were restricted to Caucasian, African-American, and Hispanic women who completed the Rose angina questionnaire (n = 1,866) and had interpretable EBCT scans (n = 1,502). Subjects with self-described ethnicity as “other” (n = 22) were excluded from the analyses as the validity of the Rose angina questionnaire in these groups is uncertain.
Rose angina questionnaire
Angina was determined using the complete World Health Organization Rose angina questionnaire (7). Definite Rose angina was defined per standard criteria as having all of the following: 1) pain or discomfort in the chest; 2) onset with walking fast or uphill, or walking at an ordinary pace on a level surface; 3) causing the participant to stop or slow down; 4) relieved with resting; 5) within 10 min or less; and 6) located along the sternum (any level) or the left anterior chest and left arm. Participants providing a negative response to any of these items were categorized as not having angina per the Rose definition.
Cardiac EBCT scans were performed in duplicate to determine CAC scores as previously described (8). The CAC score was reported as the mean score in Agatston units of 2 scans done in immediate succession, and prevalent CAC was classified as an average CAC score above 10, which was the data-derived threshold that maximized signal-to-noise ratio and reproducibility between the 2 EBCT scans (8).
Abdominal magnetic resonance imaging was performed using a 1.5-T whole-body system (Intera, Philips Medical Systems, Best, the Netherlands). Six transverse slices of the infrarenal abdominal aorta were obtained using a free-breathing, electrocardiography-gated, T2-weighted turbo spin-echo (black-blood) sequence as described previously (9). Abdominal aortic plaque was identified as hyperintense signal volume that protruded ≥1 mm from the endoluminal surface of the aortic wall as previously defined.
Cardiac magnetic resonance was performed on the same scanner using a body coil with electrocardiography gating. Aortic cross-sectional area was then measured on the axial images at the level of the pulmonary bifurcation using the QFLOW software package (version 4.1.6, Medis Medical Imaging Systems, Leiden, the Netherlands), and aortic slice volume was calculated as the aortic cross-sectional area multiplied by the aortic slice thickness as previously described (10). Arm blood pressure measurements were obtained at the time of the scan using an automated Welch-Allyn sphygmomanometer, and pulse pressure was then calculated as systolic blood pressure minus diastolic blood pressure. Aortic compliance (AC) (ml/mm Hg) was calculated using the following formula: (maximum aortic slice volume minus minimum aortic slice volume) divided by pulse pressure.
Measurement of biomarkers
Venous blood samples were obtained after an overnight fast in ethylenediaminetetraacetic acid tubes and were stored for ≤4 h at 4°C before processing. Plasma aliquots were frozen at −80°C until assays were performed. Soluble intercellular adhesion molecule (sICAM)-1, soluble vascular cell adhesion molecule (sVCAM)-1 (11), high-sensitivity C-reactive protein (CRP) (12), and monocyte chemoattractant protein (MCP)-1 (13) were measured as previously described.
Other Variable definitions
Variable definitions for traditional cardiovascular disease risk factors (13) and left ventricular hypertrophy (14) have been previously described. Premature family history of myocardial infarction (FHMI) was defined using a more restrictive definition to ensure the sensitivity of our premature family history variable, and included a myocardial infarction in a first-degree relative <50 years of age for men and <55 years of age for women. Metabolic syndrome was defined according to the National Cholesterol Education Program, Adult Treatment Panel III guidelines, as the presence of ≥3 of 5 standard criteria. Menopause was determined by self-reported history of menopause, ≥1 year from last menstrual cycle, or history of bilateral oopherectomy/ovariectomy. Sedentary lifestyle was defined as persons with leisure time physical activity of 0 metabolic equivalents per week. Low education was defined as less than completion of high school. Ethnicity was determined by self-report. Insulin resistance was calculated according to the homeostasis model assessment-insulin resistance (15).
Categorical data are presented as proportions, and continuous data as mean values with standard deviations or median values with interquartile ranges. Sample weight adjustment was used to estimate the population prevalence of angina in women in Dallas County. Statistical comparisons of the prevalence of CAC and abdominal aortic plaque in participants with and without angina were made using the chi-square test. Participants were categorized into 4 groups according to the presence or absence of angina and the presence or absence of CAC: 1) normal group (angina [−], CAC [−]); 2) angina group (angina [+], CAC [−]); 3) CAC group (angina [−], CAC [+]); and 4) angina/CAC group (angina [+], CAC [+]). Pairwise comparisons were performed using the chi-square test for categorical variables and the t test for continuous variables. All variables associated with angina in those without CAC (angina [+], CAC [−]) compared with the normal group (angina [−], CAC [−]) in univariable analyses with a p value < 0.1 were tested in multivariable logistic regression models to determine the independence of association. As body mass index, waist circumference (WC), and waist-to-hip ratio are collinear, WC alone was chosen for these models because of its higher Wald chi-square value. Separate models were created using either WC and hypertension, or metabolic syndrome, given the redundancy in these variables. The relationship between angina and number of metabolic syndrome components was assessed using the chi-square trend test. The association of sICAM-1, sVCAM-1, CRP, MCP-1, and AC with angina was assessed using the Wilcoxon rank-sum test and also in multivariable logistic regression models controlling for variables independently associated with angina from the models above.
The overall study cohort of 1,480 women (mean age 45 years, range 30 to 65 years; 49% African-American, 30% Caucasian, 20% Hispanic) had a high prevalence of traditional CAD risk factors (25% smokers, 11% diabetes mellitus, 30% hypertension, 13% hypercholesterolemia) consistent with a multiethnic, urban population. The sample weight-adjusted prevalence of angina in the entire cohort was 6.9%. In univariable analyses, angina was not associated with either CAC (p = 0.2) or abdominal aortic plaque (p = 0.9) (Fig. 1). These relationships remained unchanged after age and ethnicity adjustment.
Among women without CAC, the sample-weight adjusted prevalence of angina was 5%. Compared with normal subjects (angina [−], CAC [−]), those with angina (angina [+], CAC [−]) were more commonly African-American and post-menopausal, and had a higher prevalence of hypertension, metabolic syndrome, FHMI, as well as higher values for body mass index, waist-to-hip ratio, WC, and insulin resistance (Table 1). We also observed a stepwise increase in the prevalence of angina with increasing number of metabolic syndrome factors (p trend = 0.003) (Fig. 2). However, other traditional risk factors for atherosclerosis such as age, dyslipidemia, diabetes, and smoking were not associated with angina in the absence of CAC (angina [+], CAC [−]). In contrast, among subjects with CAC who did not have angina (angina [−], CAC [+]), most traditional atherosclerotic risk factors were more frequent in this group compared with normal subjects (Table 1). The few subjects with both angina and CAC (n = 22) were predominantly African-American, had the highest prevalence of many traditional risk factors, and had larger values for adiposity measures.
In multivariable analyses, only African-American ethnicity, FHMI, and WC were independently associated with angina in women without CAC (Table 2). When metabolic syndrome was substituted for WC and hypertension in these models, there was a trend toward an association with angina (odds ratio: 1.54, 95% confidence interval: 0.98 to 2.41).
Subjects with angina in the absence of CAC (angina [+], CAC [−]) had higher levels of sICAM-1 and sVCAM-1 compared with normal subjects (Table 3). In contrast, CRP and MCP-1 had no independent relationship with angina. Angina in the absence of CAC was also associated with lower mean AC (22 ml/mm Hg vs. 26 ml/mm Hg, p = 0.007). The associations of angina with sICAM-1, sVCAM-1, and AC remained statistically significant when controlling for variables independently associated with angina. Applying appropriate statistical transformations of these variables to create normal distributions did not materially change the results. In subjects with CAC, angina had no statistical association with sICAM-1, sVCAM-1, or AC (Table 4).
In this large multiethnic population-based sample of women, angina was reported by approximately 7% of women ages 30 to 65 years but was not associated with subclinical atherosclerosis. The variables related to angina in women without CAC differed from the traditional spectrum of risk factors associated with atherosclerosis, and favored obesity and insulin resistance factors as well as African-American ethnicity, FHMI, measures of endothelial dysfunction, and large artery stiffness. These findings suggest that angina in absence of atherosclerosis may be a distinct entity from atherosclerotic cardiovascular disease, and that vascular dysfunction may play a more prominent role in the underlying pathophysiology of this syndrome.
Angina and coronary and peripheral subclinical atherosclerosis
We found no association between angina and CAC in our cohort of relatively young women. Our results are somewhat discordant with those from the Rotterdam Study, the only other large epidemiological study to investigate the association between angina and CAC. In this population-based cohort study performed in the Netherlands, women with angina had a 4.8 increased odds of CAC >1,000 Agatston units compared with women without angina (16). These discrepant results are likely due in part to the much older age, higher overall prevalence and severity of CAC, and evaluation of the amount of calcium rather than the presence of calcium in the Rotterdam study.
Our null results with CAC do not exclude the possibility that angina is associated with “soft,” noncalcified atherosclerotic plaque in women that may be present in approximately 6% of those without CAC (17). Accordingly, we evaluated the relationship between angina and magnetic resonance imaging measured abdominal aortic plaque, a noncalcific measure of atherosclerosis in a second vascular territory, and also failed to observe any significant associations with angina prevalence. These findings suggest that subclinical atherosclerosis, whether calcified or noncalcified, is not likely a major contributor to angina among relatively young women in the general population.
Clinical determinants of angina in women without obstructive CAD
Investigating the determinants of angina in women without obstructive CAD is challenging, and prior studies in this area have reported inconsistent results. While some studies reveal associations with traditional cardiovascular disease risk factors, menopausal status, and ethnicity, others do not. Despite these inconsistencies, most studies suggest a prominent role of endothelial dysfunction in the underlying pathophysiology (18).
Our study used a novel model to define angina without obstructive CAD in a population-based cohort, Rose angina without CAC. This algorithm applies a validated measure of angina to a population free from referral bias, and one with negligible coronary atherosclerosis. We found that most traditional risk factors for atherosclerosis were not associated with angina in women without CAC; notably, there was no association with hypercholesterolemia, smoking status, or age. In contrast, several variables reflecting insulin resistance and obesity were related to this entity, including an independent association with WC as well as a stepwise association with a greater number of metabolic syndrome factors. Increased WC and metabolic syndrome both correlate with higher amounts of visceral adiposity, which in turn has been linked to impaired endothelial function as measured by decreased brachial artery flow-mediated dilation (19).
Our findings regarding the association of African-American ethnicity, FHMI, and WC with angina are consistent with a recent report from the Women's Ischemia Syndrome Evaluation study investigators (20), although that study derived from a referral population that may have had higher prevalence of cardiovascular risk factors and did not employ multivariable adjustment. As with WC, African-American ethnicity and FHMI have also been correlated with endothelial dysfunction. African-Americans have decreased flow-mediated dilation (21) and increased levels of endothelin-1, a potent vasoconstrictor released by the endothelium, compared with Caucasians (22). Further, the protective effect of female sex on endothelial function is less evident in African-Americans compared with Caucasians (23). A study of offspring from patients with FHMI also revealed decreased flow-mediated dilation compared with healthy controls (24).
Biomarkers of endothelial dysfunction and aortic compliance
Both sICAM-1 and sVCAM-1 are inflammatory biomarkers that are indicative of endothelial dysfunction. Turhan et al. (25) evaluated sICAM-1 and sVCAM-1 levels in subjects with no coronary atherosclerosis and either normal or abnormal coronary blood flow and found that sICAM-1 and sVCAM-1 levels were higher in those with lower flow. These results suggest that increased levels of soluble adhesion molecules and concomitant endothelial dysfunction may result in slow coronary flow, a known precipitant of ischemia and angina in those with normal coronary arteries (26). In contrast to sICAM-1 and sVCAM-1, CRP and MCP-1 are nonspecific inflammatory markers, and the lack of association of CRP and MCP-1 with angina in our study indicates that the findings with sICAM-1 and sVCAM-1 are not just reflective of generalized information. However, the null findings with CRP could be the result of reverse confounding by higher statin medication use in patients with angina.
Reduced AC may be related to angina in women without CAC either because of a shared pathophysiology with endothelial dysfunction or because of a direct causal association. Several studies have demonstrated correlations between measures of large artery stiffness and endothelial dysfunction (27). Some researchers have hypothesized that endothelial dysfunction in the vasovasorum may lead to ischemia in parts of the aorta resulting in stiffening. Aortic stiffening, in turn, could lead to abnormal ascending aortic flow, reduced coronary artery flow, and ischemia (28). Alternatively, reduced AC may be associated with increased afterload causing a widened pulse pressure, decreased diastolic coronary perfusion pressure and angina in this population.
Our study findings suggest that angina in women is unrelated to subclinical atherosclerosis. Moreover, many traditional risk factors for atherosclerosis do not associate with angina in women without CAC. Thus, attempting to target standard atherosclerosis risk factors may not be an effective strategy for reducing the angina-related morbidity experienced by these women. In contrast, increased levels of sICAM-1 and sVCAM-1, and lower AC independently associated with angina in this population, and each of these factors could plausibly play a direct causal or contributory role in this syndrome. Therefore, therapies that favorably impact endothelial dysfunction and vascular stiffness such as L-arginine (29), angiotensin-converting enzyme inhibitors, and statins may prove to be efficacious. The association with abdominal obesity in our analyses suggests that weight loss could also have salutary effects on angina in this population. Finally, these findings should heighten awareness of the increased prevalence of the syndrome of angina without obstructive CAD among African-Americans and among persons with a FHMI, a possible treatable diagnosis in those with chest pain complaints.
The present study used the Rose angina questionnaire rather than a clinician evaluation or diagnostic studies to define angina. However, the Rose angina questionnaire has been extensively validated and used in numerous epidemiologic studies of angina (30). In addition, physician-diagnosed angina may not be reflective of the true burden and nature of angina in the general population. Indeed, our definition of angina may be a more “pure” form of angina that is free from selection biases or confounding by subclinical atherosclerosis. The small numbers of participants with angina defined by the Rose angina questionnaire (n = 112) yields little power to rigorously assess associations, potentially yielding type 2 error. Despite this limitation of power, we found consistent associations between Rose angina and patterns of clinical and biological factors, suggesting that the Rose angina questionnaire is a valid measure that is quantifying a true clinical entity.
Typical exertional angina affects 1 of every 14 women between the ages of 30 and 65 years in Dallas County. Interestingly, angina was not associated with subclinical atherosclerosis among women. Angina that occurs in the absence of subclinical atherosclerosis was not related to many traditional atherosclerotic risk factors; however, it was associated with measures of vascular stiffness and endothelial dysfunction, suggesting a distinct vascular etiology and alternate potential therapeutic targets for this entity compared with atherosclerosis.
The Dallas Heart Study was funded by the Donald W. Reynolds Foundation (Las Vegas, Nevada) and was partially supported by USPHS GCRC Grant M01-RR00633 from NIH/NCRR-CR. The authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- aortic compliance
- coronary artery calcium
- coronary artery disease
- C-reactive protein
- electron-beam computed tomography
- family history of myocardial infarction
- monocyte chemoattractant protein
- soluble intercellular adhesion molecule
- soluble vascular cell adhesion molecule
- waist circumference
- Received June 7, 2010.
- Revision received August 11, 2010.
- Accepted October 12, 2010.
- American College of Cardiology Foundation
- Shaw L.J.,
- Merz C.N.,
- Pepine C.J.,
- et al.
- Schuijf J.D.,
- van Werkhoven J.M.,
- Pundziute G.,
- et al.
- Cannon R.O. III.
- Rumberger J.A.,
- Sheedy P.F. III.,
- Breen J.F.,
- Schwartz R.S.
- Jain T.,
- Peshock R.,
- McGuire D.K.,
- et al.
- Khera A.,
- de Lemos J.A.,
- Peshock R.M.,
- et al.
- Rohatgi A.,
- Owens A.W.,
- Khera A.,
- et al.
- Khera A.,
- McGuire D.K.,
- Murphy S.A.,
- et al.
- Deo R.,
- Khera A.,
- McGuire D.K.,
- et al.
- Hausleiter J.,
- Meyer T.,
- Hadamitzky M.,
- Kastrati A.,
- Martinoff S.,
- Schomig A.
- Opherk D.,
- Mall G.,
- Zebe H.,
- et al.
- Piatti P.,
- Fragasso G.,
- Monti L.D.,
- et al.
- Hemingway H.,
- Langenberg C.,
- Damant J.,
- Frost C.,
- Pyorala K.,
- Barrett-Connor E.