Methods

Top Methods Discussion Author & Article Info References

The MEDLINE and Current Contents databases available at the Biomedical Library of the University of Pennsylvania were searched. The MEDLINE searches were done with Ovid software and were limited to the English-language literature published between 1966 and 1997. Textword searches for electron beam computed, electron beam CT, EBCT, ultrafast computed, and ultrafast CT were combined by a Boolean or. In addition, an exploded keyword search for myocardial ischemia was combined by a Boolean or with keyword and textword searches for calcium; this set was combined with keyword and textword searches for plaque, atherosclerosis, fluoroscopy, and digital subtraction by a Boolean and. The Current Contents searches used the keywords coronary artery and calcium. These searches yielded several hundred references. These references were reviewed by the author, and articles relating to coronary artery calcium and its detection by fluoroscopy or electron-beam CT were selected on the basis of title and abstract. The reference lists of selected articles were reviewed for additional relevant references. Articles were excluded from review if they failed to present data sufficient for the calculation of sensitivity, specificity, and positive and negative predictive value. All prospective studies of electron-beam CT were included. Abstracts, which are not peer reviewed, were excluded from review.

Pathogenesis of Coronary Artery Disease

Understanding the mechanisms of atherosclerosis is fundamental to assessing coronary artery calcium as an indicator of coronary artery disease. The pathogenesis of atherosclerosis has been the subject of several recent, comprehensive reviews [1-5]. The fundamental mechanism of atherogenesis seems to be chronic mild endothelial injury that, depending on certain local and systemic factors, results in areas of coronary vessels that show macrophage infiltration and increased uptake and retention of low-density lipoprotein (LDL) cholesterol. The predominant local factor seems to be regional blood flow, which results in altered shear stress on the vessel wall and lining endothelium, causing changes in endothelial structure, function, and binding of lipoproteins. Plaque configuration and inflammation are local factors thought to contribute to plaque rupture and thrombosis. Other, more recently and more tentatively identified local factors include infectious agents, such as cytomegalovirus, Chlamydia pneumoniae, and Helicobacter pylori [6, 7]. Systemic factors resulting in endothelial injury or increased risk for plaque formation and rupture include elevated serum LDL cholesterol levels (particularly oxidized LDL cholesterol levels) and perhaps other lipoproteins, advanced glycosylation end products found in patients with diabetes mellitus, circulating catecholamines, toxins from cigarette smoke, inflammatory states, platelet and coagulation factors, and other circulating factors. Lipid uptake into endothelial cells depends on both active receptor-mediated mechanisms and passive, endothelial damage-dependent mechanisms. Macrophages are recruited by chemotaxis and accumulate through the binding of upregulated cell adhesion molecules; both effects are mediated by oxidized LDL cholesterol. Finally, smooth-muscle cells proliferate and elaborate extracellular matrix molecules in response to wall stress and locally secreted growth factors.

Early in their development, atherosclerotic lesions are largely composed of lipids and grow mainly by lipid accumulation [3, 4]. Prelesions located at areas of altered shear stress show adaptive smooth-muscle thickening. Type I through type IV lesions show the progressive accumulation of intracellular and then extracellular lipids and of macrophage foamy cells (Table 1). From the fourth decade of life, type IV (atheromatous) lesions may evolve in multiple directions. Type V (fibroatheromatous) lesions may have a lipid core and a fibrotic layer (or many such cores and layers) or may be primarily calcific or primarily fibrotic. Type IV and, more commonly, type V lesions may progress to type VI (complicated) lesions with surface defect, hematoma, hemorrhage, or thrombus formation. Severe lesions (>75% cross-sectional lumen narrowing) consist almost exclusively of advanced fibrolipid or purely fibrotic plaques without ulceration or thrombosis [8].

Smaller lesions with a lipid core are thought to be at risk for rupture, especially if there has been little deposition of matrix or other lesion-stabilizing material [9-11]. A recent study of acute coronary thrombi in patients with sudden death showed that 69% of these thrombi were due to rupture of a vulnerable plaque; the rest were due to the erosion of fibrous plaque [12]. It is the exposure of the lipid core to blood that is thought to result in acute thrombus formation. Changes in shear stress, coronary pressure and tone, and plaque configuration all contribute to rupture. Local inflammatory processes also play a role by causing the release of matrix-degrading enzymes and weakening plaque structure. Creation of a surface defect and subsequent thrombus formation is not always catastrophic, however, and may result in formation of a clinically silent thrombus that subsequently undergoes organization and incorporation. Certain plaque characteristics, such as the lesion components, degree of stenosis, surface roughness, and presence of residual thrombus, influence the degree of thrombosis and clinical presentation, which is also affected by vasospasm and the degree of preexisting collateralization.

The Role of Calcium in Plaque Biology

Calcium has long been known to be a component of atherosclerotic plaques and has been associated with advanced coronary artery disease [5, 13-16]. In atherosclerotic plaques, calcium is not found as amorphous calcium phosphate but rather as hydroxyapatite [17], the form of calcium found in bone; such deposits may appear identical to lamellar bone [18]. Studies of atherosclerotic calcification have shown local expression of multiple genes related to calcium and bone metabolism, including osteopontin, osteonectin, osteocalcin [19, 20], bone morphogenic protein type II [21], and the presence of so-called calcifying vascular cells that deposit hydroxyapatite [22]. Calcified lesions may be stiffer and less likely to rupture than uncalcified lesions [23]. However, the role of calcium in plaque rupture is unclear; it is thought that plaques are less susceptible to rupture only after extensive calcification, whereas partial calcification may increase susceptibility [5]. The interface of hard calcium deposits with plaque components weakened by inflammation may represent an unstable structure that is more likely to rupture [24]. Furthermore, advanced, calcified plaques are associated with the presence of other lipid-rich, uncalcified plaques that may be considerably more unstable.

The first clinical studies of coronary artery calcium used routine fluoroscopy and digital subtraction fluoroscopy [25] to measure coronary artery calcium [5, 26]. There is a strong correlation between the amount of calcium detected by digital subtraction fluoroscopy and the true calcium mass in excised coronary artery segments [27], and the technique has good interobserver reproducibility [28]. Coronary calcium detected by fluoroscopy is associated with increased age, male sex, family history of coronary artery disease, and diabetes [29, 30]. Sensitivity for the detection of significant lesions (>50% stenosis) ranges from 40% to 92%; specificity ranges from 52% to 95% [5, 26]. In symptomatic patients, fluoroscopically detected calcium is associated with coronary events [31]. Detrano and colleagues [32, 33] have shown that coronary artery calcium detected by digital subtraction fluoroscopy seems to have prognostic significance in asymptomatic, high-risk patients. At 1 year, fluoroscopically detected calcium was associated with various cardiac end points, including angina, myocardial infarction, revascularization, and cardiac death (risk ratio, 2.7) and was an independent predictor of cardiac end points [32]. After 55 months of follow-up, myocardial infarction or cardiac death was 2.2 times more likely to occur in patients with coronary artery calcium detected in more than one vessel; the relative risk increased by 1.4 for each vessel involved [33]. However, of the 53 patients who reached cardiac end points after 1 year, 5 had no calcium detected by fluoroscopy. Of the 15 reported deaths and nonfatal myocardial infarctions, 5 were in patients without coronary artery calcium [33]. Moreover, compared with white and Asian-American patients, asymptomatic, high-risk, African-Americans had a decreased prevalence of coronary artery calcium but a significantly higher incidence of angina, myocardial infarction, and cardiac death after an average of 20 months of follow-up [34].

Electron-Beam Computed Tomography and Coronary Artery Calcium

Electron-beam CT is a relatively recent innovation in radiology that has been applied to the detection of coronary artery calcium only since the 1980s (Figure 1) [35-38]. In conventional CT, the x-ray source is rotated around the patient; in electron-beam CT, an electron beam is directed toward four tungsten targets to generate x-rays that pass through the patient to detectors. The absence of a moving x-ray source allows for 100-ms scans; when triggered by electrocardiography to capture during late diastole, a full series of accurate, sharp, 3-mm slices can be obtained in less than 1 minute. The entire procedure lasts approximately 10 minutes, requires minimal patient effort (only breath holding), and requires no contrast agents. The estimated cost is equal to or less than that of an electrocardiography exercise stress test or a radionucleotide cardiac imaging study.

View larger version (162K): [in this window] [in a new window]   Figure 1. Sample electron-beam computed tomographic scan showing calcification of the left anterior descending coronary artery (thick arrow) and the aortic root (thin arrow).

Coronary artery calcium is measured in an automated manner by quantitating the number of pixels with a density greater than 130 Housefield units [39], a density more than twice that of blood [40]. The area of each detected region is multiplied by a factor related to the peak density for each (1, 130 to 199 Hounsfield units; 2, 200 to 299 units; 3, 300 to 399 units; and 4, 400 units), and these values are summed, giving the total coronary artery calcium score. The number of adjacent pixels with a density greater than 130 Hounsfield units required for detection and quantitation varies from study to study (Table 2). Requiring only a single pixel of the requisite density can result in misidentification of tomographic noise as calcium, whereas requiring too many adjacent pixels may cause loss of sensitivity for small lesions. A threshold of eight adjacent pixels was found to give 50% repeated detection of lesions on a second scan without excessive loss of sensitivity [40]. Measures of coronary artery calcium that use number of plaques, calcific area, average density, coronary artery calcium score, and coronary artery calcium product (the product of density and area) are well correlated with each other [46, 53].

In addition to being used to quantify calcification, electron-beam CT images can be used to create three-dimensional reconstructions of the coronary vasculature [54]. With the use of intravenous contrast agents, coronary artery lumen size can be assessed [55, 56]. The study of coronary arteries harvested at autopsy shows a strong correlation of calcium detection by electron-beam CT with both histologic calcium detection [57, 58] and luminal narrowing [59, 60]. In autopsy specimens, coronary artery calcium scores can predict coronary artery narrowing [58, 59], and coronary artery calcium area is highly correlated with histologically defined coronary plaque area [61].

Several methodologic issues have been raised about the measurement of coronary artery calcium by electron-beam CT. Detrano and coworkers [62] attempted to quantify known amounts of hydroxyapatite placed in model coronary vessels and found that the standard quantitation algorithm described above resulted in significant variation in the quantitation of hydroxyapatite, depending on position, angle, and particle size. They describe an alternative method for the calculation of coronary artery calcium that markedly decreases variation, although this method has yet to become standard. Wang and colleagues [63] have observed significant test-to-test variability in the measurement of coronary artery calcium, suggesting that the quantitation is not sufficiently reproducible to allow serial evaluation of patients over intervals of less then 2 years; others have found better reproducibility [64]. An increase in slice thickness to 6 mm decreases the test-to-test variability [63]; clinically, 3-mm and 6-mm slices give equivalent predictive ability [53].

Pathologic studies in which coronary arteries from autopsied hearts are examined by electron-beam CT and light microscopy show that coronary artery calcium, as measured by electron-beam CT, does not precisely correlate with actual luminal narrowing [60] and that some high-grade lesions and many smaller lesions lack detectable calcium [57, 61]; the total area of calcification determined by electron-beam CT is roughly 20% of the total plaque area [61]. Although the absence of calcium strongly suggests the absence of an obstructive lesion, it does not rule out the presence of atherosclerotic disease in an arterial segment. Morphologic evaluation of calcium detected by electron-beam CT correlates with the structure of angiographic lesions and improves the ability to detect stenoses on a site-by-site basis [51]. Used in conjunction with intracoronary ultrasonography, electron-beam CT detects calcium in almost all hard plaques, 47% of soft plaques, and 25% of segments that are plaque-free on ultrasonography [52]. However, only 3 of 15 soft plaques causing obstructions greater than 50% had coronary artery calcium detectable on electron-beam CT in this study.

Since it was first introduced as a method of detecting coronary artery calcium [41], electron-beam CT has been used to relate the presence and extent of calcification both to risk factors for coronary artery disease and to disease as determined by angiography. Several studies have linked risk factors to coronary artery calcification. Goel and colleagues [65] noted an increased prevalence of coronary artery calcium associated with diabetes, hypertension, smoking, and a history of chest pain or previous myocardial infarction in men and associated with hypercholesterolemia and smoking in women. Overall, age, male sex, smoking, hypertension, and number of risk factors were associated with an increased prevalence of calcification. In asymptomatic adults, Wong and coworkers [66] found a higher prevalence of coronary artery calcium among men with hypertension, diabetes, hypercholesterolemia, smoking, infrequent exercise, and obesity. Among women, however, the prevalence was higher only in those with hypercholesterolemia.

Maher and colleagues [67] noted that the extent of coronary calcification in 740 asymptomatic adults was significantly associated with sex, age, body size, blood pressure, ratio of total cholesterol to high-density lipoprotein (HDL) cholesterol, and smoking, although these risk factors accounted for less than 40% of the variation in the quantity of coronary artery calcium. Elevated serum triglyceride levels [68] and fibrinogen levels [69] are also associated with coronary artery calcification. In a study of 11 children and young adults with familial hypercholesterolemia (mean LDL cholesterol level ±SD, 671 ± 198 mg/dL [17.35 ± 5.12 mmol/L]), 7 patients had significant lesions on coronary angiography. All 7 (excluding 2 patients younger than 15 years of age) had coronary artery calcium detected; 1 patient with a normal angiogram also had coronary artery calcium [70]. Coronary artery calcium scores in this population correlated best with age and cholesterol-years and were independent of serum lipoprotein and lipid levels.

Mahoney and coworkers [71] evaluated risk factors in 384 boys and girls at an average age of 15 years and again at age 27 to 33 years; the second evaluation included electron-beam CT. Calcification was seen in 31% of the men and 10% of the women, and the presence of coronary artery calcium was associated most strongly with increased body mass index, elevated blood pressure, and decreased HDL cholesterol levels at the time of the second evaluation. Childhood weight, body mass index, and triglyceride levels were most strongly associated with the development of coronary artery calcium 15 years later. Higher coronary artery calcium scores and increased prevalence of coronary artery calcium are found in patients with a history of documented coronary artery disease compared with patients who have risk factors alone [72].

Tanenbaum and colleagues [41] were the first to use electron-beam CT to evaluate patients undergoing cardiac catheterization and coronary angiography. Soon thereafter, Agatston and coworkers [39] defined the standard method of quantitating coronary artery calcification. Several features emerge as common to all or most of these studies (Table 2): 1) The amount of detectable coronary artery calcium is strongly correlated with the extent of coronary artery disease [39, 41, 42, 44-4952, 73, 74]; 2) coronary artery calcium scores and prevalence increase with age [39, 41, 44-4765, 66, 74, 75]; 3) sensitivity for hemodynamically significant lesions is very high and specificity is poor to moderate, whereas sensitivity is worse and specificity is better for detecting any angiographically evident disease [40, 42-44, 46-48]; and 4) coronary artery calcification occurs later and to a lesser extent in women than in men, although the sensitivity and specificity of coronary artery calcium scores for disease is similar for men and women [44, 47, 50, 65]. Increasing the threshold for coronary artery calcium score decreases sensitivity and improves specificity [46, 49, 50]; thus, this can be done to compensate for the increase in coronary artery calcium with age [39]. Adjustment of the threshold for detection of coronary artery calcium can also be used to tailor the sensitivity and specificity of electron-beam CT to specific clinical uses (for example, as a very sensitive test for evaluation of patients in the acute setting compared with a more specific test for the evaluation of stable, asymptomatic patients [76]).

Electron-beam CT compares favorably with exercise stress testing both with and without radionucleotide imaging [45]. Its sensitivity is lower and its specificity higher in distinguishing mildly diseased coronary arteries from normal arteries than in distinguishing significantly diseased arteries from normal arteries [47, 50]. When electron-beam CT studies are repeated after 1 year or more, lesions have progressed [54, 77]. The number of calcified lesions and the total area of calcified plaque correlate with the number of coronary artery segments with significant obstruction on angiography [13]. Coronary artery calcium scores are up to fivefold higher in patients undergoing dialysis than in patients not undergoing this treatment [77]. In a small study [78], the amount of calcium detected by electron-beam CT at sites of angioplasty was higher in patients who had restenosis.

Fewer studies have considered the role of electron-beam CT in the evaluation of asymptomatic patients, and few patients with coronary artery calcium detected on electron-beam CT (symptomatic or asymptomatic) have been prospectively examined for clinical events. Janowitz and coworkers [75] recruited 1898 asymptomatic persons 14 to 88 years of age, including patients with known risk factors who were referred by physicians. They saw an increasing prevalence of coronary artery calcium with age; it ranged from 11% for men younger than 30 years of age to 100% for men older than 80 years of age and from 6% for women younger than 30 years of age to 100% for women older than 80 years of age. The increase in coronary artery calcium scores in women paralleled that of men but with a lag time of about 10 years.

Kaufmann and coworkers [74] studied 772 asymptomatic adults 20 to 59 years of age. Using a stricter threshold for calcium detection than Janowitz and colleagues did [75] (eight adjacent pixels were required rather than only two) and avoiding patients referred by physicians, they found an increasing incidence of coronary artery calcium with age and an overall incidence of 20%; 7.4% of women and 32.6% of men had evidence of calcification [74]. Eight percent of their patients had a low-risk profile but had coronary artery calcium scores above the 75th percentile. In a sample of 875 referred adults 22 to 85 years of age with known risk factors, Wong and coworkers [66] found that 57% of men and 44% of women had coronary calcification (one-pixel threshold); the risk for calcification was associated with age and other risk factors.

In studies of asymptomatic French men with risk factors who were 30 to 70 years of age (four-pixel threshold), Megnien and colleagues [68] found that 65% of 111 patients had coronary artery calcium, and Simon and coworkers [79] identified calcium in 63% of 618 patients. Guerci and coworkers [80] did coronary angiography in 18 asymptomatic patients with elevated coronary artery calcium scores and 18 patients with low coronary artery calcium scores. The first group had an average coronary artery calcium score of 573 ± 504 and an average worst stenosis of 45% ± 16%; the second group had an average coronary artery calcium score of 27 ± 38 and an average worst stenosis of 7% ± 14%. The worst stenosis was strongly correlated with the square root of the coronary artery calcium score.

Despite the extent of these studies, few data are available on how coronary artery calcium detected by electron-beam CT affects prognosis. After up to 45 months of follow-up, a study of 422 symptomatic patients who had undergone angiography and electron-beam CT showed significantly higher event-free survival for patients with initial calcium scores less than 100; the coronary artery calcium score, but not the number of angiographically diseased vessels, could predict cardiac events (cardiac death or hospital admission for chest pain or suspected myocardial infarction) [49]. Events were more likely in the persons with the highest coronary artery calcium scores. In a sample of 1173 asymptomatic men and women who were followed for an average of 19 months, Arad and coworkers [81] found that patients who had cardiac events (including myocardial infarction, cardiac death, revascularization, and non-hemorrhagic stroke) had significantly higher coronary artery calcium scores than did those who had no such events. Increasing the coronary artery calcium score threshold from 100 to 680 decreased sensitivity (from 89% to 50%) and increased specificity (from 77% to 95%). The positive predictive value of electron-beam CT was poor (6% to 14%), but the negative predictive value was greater than 99% for all thresholds. Separate data for hard events (infarction and cardiac death) were not provided.

In a recent study by Secci and colleagues [53]. 326 asymptomatic, high-risk patients were followed for 2 years before electron-beam CT was done and for an average of 32 months afterward. Twelve events (5 myocardial infarctions and 7 revascularizations) occurred in 11 patients before electron-beam CT, and 23 events (5 cardiac deaths, 6 infarctions, and 12 revascularizations) occurred in 18 patients after electron-beam CT. Events (before and after scanning) were three times more frequent in patients with coronary artery calcium scores above the median. However, no significant association was seen between calcium detected on electron-beam CT and hard events (infarction or cardiac death) occurring before or after scanning, and only electrocardiographic evidence of left ventricular hypertrophy, not coronary artery calcium scores or mass, predicted hard events over the follow-up period. Revascularization was predicted by coronary artery calcium. Furthermore, hard events occurred in several patients who had little or no detectable coronary calcium. In patients receiving heart transplants, detection of coronary artery calcium predicted cardiac events over a 4-year follow-up period [82].

Discussion

Top Methods Discussion Author & Article Info References

In a short time, a large amount of data on the use of electron-beam CT for the evaluation of coronary artery calcification has accumulated. Calcium detected in coronary arteries by electron-beam CT is a sensitive but not a specific indicator of angiographically detected atherosclerotic lesions; the sensitivity is better and the specificity is worse for the detection of lesions causing a 50% or greater reduction in coronary artery diameter. Unfortunately, the data linking calcification to angiographic lesions have not yet been correlated with specific risk for ischemic events. This is partly because the necessary studies have not yet been done and partly because our understanding of the biology of atherosclerosis, plaque rupture, and acute myocardial syndromes is incomplete. Because we do not fully understand which lesions are most likely to cause cardiac events, interpretation of angiography (the gold standard for detecting coronary artery disease in electron-beam CT studies) is compromised.

Electron-beam CT has the greatest sensitivity for the detection of angiographically significant lesions, so stenoses of less than 50% that may be at high risk for rupture can be missed. Thus, the absence of coronary artery calcium, although it suggests that angiographically significant disease is absent, does not necessarily imply a low risk for acute events. Similarly, the high sensitivity of coronary artery calcium for significant disease cannot be interpreted as high sensitivity for disease at risk to progress catastrophically, because a calcified lesion detected by electron-beam CT may be a stable plaque unlikely to rupture. However, the presence of such lesions may imply the presence of other, less stable, higher-risk lesions.

Given these reservations, the current role for electron-beam CT in clinical practice is unclear. It has been suggested that this technique be used to screen asymptomatic patients with or without significant risk factors and to assess patients with atypical or acute chest pain. It has great advantages: It is fast, is noninvasive, and does not require exercise, and its reported cost is equal to or less than that of other forms of cardiac evaluation. However, few data are available on outcomes in symptomatic and asymptomatic patients. In the single study that has addressed symptomatic patients [49], a positive test result conferred an increased risk for a cardiac event; however, the clinical significance of this increased risk will be unclear until the differential management of symptomatic patients with positive scans and negative scans is studied. In asymptomatic patients, data are conflicting. Fluoroscopically detected coronary artery calcium in asymptomatic patients confers increased risk but produces too many false-negative results and paradoxically detects less calcium in African-American patients, who go on to have more events [34]. Only one of two prospective studies using electron-beam CT has shown an increased number of all cardiac events (including hard events and revascularizations) in patients with elevated coronary artery calcium scores [81], whereas a significantly increased incidence of hard events (infarction or cardiac death) has not been associated with increased scores [53].

Given these findings, the appropriate management of a patient with a positive result is unclear. For patients who have already had angiography, or even stress testing, electron-beam CT adds little information; although it may predict risk better than angiography in symptomatic patients, it is unclear how this altered risk should affect management. The use of electron-beam CT in symptomatic patients who have not been studied invasively has not been examined, although a markedly positive result on electron-beam CT probably indicates that angiographically significant disease is present. This result might be used to support an aggressive approach to the management of symptoms and risk factors, an approach that may include angiography and angioplasty. Given the potential clinical significance of uncalcified lesions, the absence of coronary artery calcium may not be reassuring in some cases and would not eliminate the need to manage modifiable risk factors. For a patient who requires an exercise tolerance test or angiography but cannot exercise or has renal failure, electron-beam CT might be useful. Given conflicting reports, the use of electron-beam CT for screening asymptomatic patients currently has little support, although the limited data from both electron-beam CT and fluoroscopy studies, in conjunction with the high correlation between coronary artery calcium scores and angiographically significant disease, suggest that electron-beam CT may become an effective screening tool as more prospective studies accumulate.

Although electron-beam CT shows great promise as a tool in the evaluation of patients who have or are at risk for coronary artery disease, further study in several areas is needed before this promise is realized. Despite the strong association between coronary artery calcium and angiographic evidence of atherosclerosis, more prospective data on outcomes in asymptomatic, high-risk patients and symptomatic patients are needed, addressing both patients with very low calcium scores and patients with high calcium scores. The use of electron-beam CT in acute risk stratification, as part of the emergency department evaluation of patients with chest pain, must also be examined. The extent to which outcomes in patients with coronary artery calcium can be affected by aggressive medical and risk factor management (and by revascularization when indicated) must be assessed. Continued molecular, cellular, and physiologic studies on the role of calcium in plaque stability and rupture will provide the pathophysiologic framework for future clinical investigations.