Atherosclerotic calcification is an organized, regulated process similar to bone formation that occurs only when other aspects of atherosclerosis are also present. Nonhepatic Gla-containing proteins like osteocalcin, which are actively involved in the transport of calcium out of vessel walls, are suspected to have key roles in the pathogenesis of coronary calcification. Osteopontin and its mRNA, known to be involved in bone mineralization, have been identified in calcified atherosclerotic lesions. Calcified human atherosclerotic plaque also contains mRNA for bone morphogenetic protein-2a, a potent factor for osteoblastic differentiation, and cells that are capable of osteoblastic differentiation. These cells may be the ones from which vascular calcifying cells are derived. These and other recent findings indicate that calcification is an active process and not simply a passive precipitation of calcium phosphate crystals, as once thought.

Although calcification is found more frequently in advanced lesions, it may also occur in small amounts in earlier lesions that appear in the second and third decades of life. Histopathological investigation has shown that plaques with microscopic evidence of mineralization are larger and associated with larger coronary arteries than are plaques or arteries without calcification. The relation of arterial calcification to the probability of plaque rupture is unknown. Although the amount of coronary calcium correlates with the amount of atherosclerosis in different individuals and to a lesser extent in segments of the coronary tree in the same individuals, it is not known if the quantity of calcification tracks the quantity of atherosclerosis over time in the same individuals. Further research is needed to better elucidate the relation of calcification to the pathogenesis of both atherosclerosis and plaque rupture.

In vivo epidemiological evidence and postmortem studies show that the prevalence of coronary calcium deposits in a given decade of life is 10 to 100 times higher than the expected 10-year incidence of coronary heart disease events for individuals of the same age. This disparity is less evident in the elderly and symptomatic than in the young and asymptomatic. Realization of this fact has generated the need to determine clinically useful threshold levels of coronary calcium content (such as the calcium score determined by electron beam computed tomography [EBCT]) in order to make appropriate management decisions. The limited available evidence linking radiographically detectable coronary calcium to future coronary heart disease events of death and infarction suggests that this link is strongest in symptomatic and very high-risk subjects. The results of ongoing epidemiological studies will be needed to further elucidate this connection.

Fluoroscopy, electron beam, and helical computed tomography can identify calcific deposits; EBCT and, to a lesser extent, double-helical CT have the enhanced capability to localize coronary calcification and detect smaller and less dense calcific deposits. Only EBCT can quantitate the amount or volume of calcium. The absence of calcific deposits on an EBCT scan implies the absence of significant angiographic coronary narrowing; however, it does not imply the absence of atherosclerosis, including unstable plaque. Similarly, calcification may frequently be seen in the absence of significant angiographic narrowing and before there has been sufficient plaque build-up to narrow the vessel to the extent that ischemia would be apparent on stress electrocardiograms or stress-thallium determinations.

According to the available evidence, a negative EBCT coronary calcium study, when no calcium is detected, does not absolutely rule out the presence of atherosclerotic plaque, including unstable plaque, but does imply a very low likelihood of significant luminal obstruction. The majority of patients who have had angiographically normal coronary arteries have negative EBCT scans and a low risk of a cardiovascular event within the next 2 to 5 years. Women tend to have low scores or negative scans before menopause.

On the other hand, a positive scan, that is, one in which some calcium is detected in at least one vessel, confirms the presence of atherosclerotic plaque. The greater the amount of calcification, the greater the likelihood of obstructive disease, but there is not a one-to-one relation, and the findings are not site specific. A high calcium score may be consistent with a moderate to high risk of a cardiovascular event within the next 2 to 5 years.

Unless the calcific area is greater than 2 mm, the reproducibility of coronary calcium detection with EBCT appears to be insufficient for serial assessment of coronary calcium levels in individual patients. However, because EBCT has been shown to be sufficiently accurate for predicting the presence of angiographic stenoses somewhere in the coronary arteries and for predicting the likelihood of clinical end points in symptomatic patients, it can be used as part of a cardiological examination done under the supervision of a physician knowledgeable about the significance of scan results and the management of coronary heart disease. Presently the data are insufficient to recommend coronary artery calcium screening in lieu of stress testing for most patients with chest pain, except in those with atypical chest pain, for whom a negative study may be useful by itself or in addition to exercise testing. The role of EBCT as a screening tool in asymptomatic patients with conventional risk factors is not yet clearly defined. It can be anticipated, however, that identifying the presence of premorbid coronary artery disease would influence the aggressiveness with which risk factor modification is approached. There is no role at present for application of the test to screen populations of young (less than 40 years old), healthy individuals with no risk factors. The importance of calcification in such individuals will have to await event data that are currently being obtained.

Introduction

The ability to detect and quantify coronary artery calcium using currently available imaging methods has created significant interest in developing appropriate applications for various clinical settings. This statement describes the pathophysiology of coronary artery atherosclerosis and calcification, the available epidemiological information related to coronary calcification, various diagnostic methods for detecting coronary calcification and, once identified, its significance and prognostic value. The statement concludes with potential practical applications for detecting coronary calcium in specific patient subgroups and comments on screening for coronary calcium, based on available epidemiological data and current understanding of the pathophysiology of coronary artery disease.

Pathophysiology of Coronary Artery Disease

Viewpoints on the pathophysiology of coronary atherosclerosis have dramatically changed in the last few years. The mechanisms of progression of coronary atherosclerosis and plaque instability and rupture in acute coronary syndromes are now more completely understood.^1 2 3 4

Lesions of Atherosclerosis
A new classification has been proposed to characterize atherosclerotic plaque progression into five phases, from fatty streak to the advanced complicated lesion.^2 5 The American Heart Association Committee on Vascular Lesions defined each of these phases by lesion morphological characteristics.⁶ This classification system relates the clinical phases of plaque evolution to the types of lesions seen pathologically so that clinicians and investigators alike may share a common language and understanding of these processes.

Vulnerable Lipid-Rich Plaque and Acute Coronary Syndrome
Plaque Disruption
In the process of atherogenesis, lipid accumulation, cell proliferation, and extracellular matrix synthesis may be expected to be linear with time. However, angiographic studies show that the progression of coronary artery disease in humans is neither linear nor predictable.^2 4 Indeed, recently it has become apparent that arteriographically mild coronary lesions may undergo significant progression to severe stenosis or total occlusion over a period of a few months.^7 8These lesions, which are often found in concert with more advanced atherosclerotic lesions, may account for as many as two thirds of patients in whom unstable angina or other acute coronary syndromes develop.^2 9 This unpredictable and episodic progression is likely caused by plaque disruption with subsequent thrombus formation that changes plaque geometry, leading to plaque growth and acute occlusive coronary syndromes.^10 11 Angioscopic studies performed in vivo have supported this theory.¹²

Recent pathological studies have shown that atherosclerotic plaques prone to rupture are commonly composed of a crescentic mass of lipids separated from the vessel lumen by a fibrous cap.¹³ In addition to a rather passive phenomenon of plaque disruption, the concept of an active phenomenon related to macrophage activity is evolving.¹⁴ Extracts from human and rabbit atherosclerotic plaques revealed macrophages and expression of metalloproteinases that induced an increase in the breakdown of extracellular matrix, suggesting that macrophages could be responsible for an active phenomenon of plaque disruption.^3 15 16

Thrombosis
Disruption of a vulnerable or unstable plaque with a subsequent change in plaque geometry and thrombosis may result in acute occlusion with unstable angina or other acute coronary syndromes. Histopathologically, plaque fissuring occurs in various shapes and sizes. The tear may be small, allowing blood to enter, expand, and uproot the plaque but not necessarily resulting in thrombus formation in the arterial lumen. If the tear is large, a subsequent, usually platelet-rich thrombus may form within the lumen and occlude the vessel¹⁷ ; such a thrombus may be either partially lysed or become replaced in the process of organization by the vascular repair response.^2 5 Of interest, an acute occlusive thrombus may be permeated by several channels and appear partially open at angiography.

Calcium Deposition in Coronary Artery Disease
Lesions of Atherosclerosis and Calcium Deposition
Atherosclerotic calcification begins as early as the second decade of life, just after fatty streak formation.¹⁸ With refined microscopic methods,¹⁹ the lesions of younger adults have revealed small aggregates of crystalline calcium among the lipid particles of lipid cores.^6 18 Calcific deposits are found more frequently and in greater amounts in elderly individuals and more advanced lesions.¹⁹ In most advanced lesions, when mineralization dominates the picture, components such as lipid deposits and increased fibrous tissue may also be present.

Calcium phosphate (hydroxyapatite, Ca₃[-PO₄]_2-xCa[OH]₂), which contains 40% calcium by weight, precipitates in diseased coronary arteries by a mechanism similar to that found in active bone formation and remodeling.²⁰ Electron microscopic evidence supports the theory by which hydroxyapatite, the predominant crystalline form in calcium deposits,^19 21 is formed primarily in vesicles that pinch off from arterial wall cells, analogous to the way matrix vesicles pinch off from chondrocytes in developing bone.^22 23 24 It has been postulated that vesicles, derived from dead foam and smooth muscle cell debris and contained within extracellular lipid-rich accumulations, may also serve as the sites of small calcium deposits.^5 6 A very close spatial association between cholesterol deposits and hydroxyapatite has also been demonstrated.²⁵Accordingly there may be various mechanisms of calcium deposition in atherosclerosis.

Although the biochemical sequence of events leading to atherosclerotic calcification is not well understood, recent attention has focused on a unique class of proteins known as Gla-containing proteins, which have a very high affinity for hydroxyapatite. Gla (gamma carboxyglutamate) is an unusual amino acid residue whose only known function is to bind calcium.^26 27 Indeed, it has been suggested that Gla proteins may be actively related to atherosclerotic calcification. They do not interfere with normal calcium homeostasis because they are not calcium chelators, but if precipitationof calcium occurs, available Gla-containing proteins would be expected to bind to the precipitate.¹⁹Decarboxylation of Gla residues to glutamyl residues greatly diminishes the affinity of Gla-containing proteins for hydroxyapatite.^26 27

Although passive absorption of Gla-containing proteins from serum cannot be entirely excluded, this seems unlikely¹⁹ because coronary arterial calcification seems to occur exclusively in atherosclerotic arteries and is absent in normal vessel wall.²⁸ There may be a mechanistic link between pathological processes leading to calcification and those leading to atherosclerosis. It is conceivable, for example, that atherosclerotic processes inhibit the synthesis and/or activity of γ-glutamate carboxylase, thus perhaps explaining why atherosclerotic arteries contain only about 30% of the carboxylase activity found in normal arterial segments.²⁹ Alternatively, it is also conceivable that cells in atherosclerotic lesions synthesize less carboxylase.

Overall, present findings lend credence to the idea that atherosclerotic calcification is not merely passive adsorption but instead is an organized, regulated process similar in many respects to bone formation. Recently, clear differences in the occurrence of arterial wall calcification were observed among genetically distinct inbred mouse strains,³⁰ indicating for the first time that there is a genetic component in this clinically significant trait.

Molecular Basis for Calcification
Fitzpatrick et al³¹ used in situ hybridization to identify mRNA of matrix proteins associated with mineralization in coronary artery specimens. Using undecalcified sections of postmortem coronary arteries, they found mineralization to be diffuse, rather than solely confined to the intima, and present in all atherosclerotic plaques. Specifically they identified a cell attachment protein (osteopontin), a protein associated with calcium (osteonectin), and a γ-carboxylated protein that regulates mineralization (osteocalcin). The traditional sectioning methodology, which uses decalcification, may miss a significant amount of mineralization. Indeed, hydroxyapatite was not detected in any coronary sections deemed normal by traditional light microscopy.

Osteopontin is a phosphorylated glycoprotein, regulated by local cytokines, with known involvement in the formation and calcification of bone. Immunohistochemistry of the specimens examined by Fitzpatrick for osteopontin demonstrated intense, highly specific staining in the outer margins of all diseased segments at each calcification front, although staining was evident throughout the entire plaque.³¹ Other studies showed that osteopontin can be seen in tissue demonstrating atherosclerotic involvement²⁰ and appeared to be present only in sites of concomitant coronary atherosclerotic disease.^32 33 Hirota et al³⁴demonstrated with Northern blotting that osteopontin mRNA expression is related to severity of atherosclerosis. On the other hand, osteonectin expression of mRNA decreased with development of atherosclerosis, suggesting a counterregulatory role. Shanahan et al³⁵ and Ikeda et al³² independently demonstrated that the predominant cell type in the areas associated with this ectopic bone protein expression are macrophage-derived foam cells, although some smooth muscle cells could also be identified.

Giachelli and associates,³⁶ using immunochemistry and in situ hybridization, demonstrated that medial smooth muscle cells in uninjured arteries contain very low levels of osteopontin and mRNA. However, injury to either the adult rat aorta or carotid artery initiated a time-dependent increase in both osteopontin protein and mRNA within the arterial smooth muscle cells, suggesting a possible role for osteopontin in the proliferative and migratory phases of arterial injury. They also showed that basic fibroblast growth factor, transforming growth factor-ß, and angiotensin II, all proteins implicated in the arterial injury response, elevated osteopontin expression in confluent vascular smooth muscle cells in vitro.

Finally, Bostrom et al²⁰ recently identified bone morphogenetic protein-2a, a potent factor for osteoblastic differentiation, in calcified human atherosclerotic plaque. Cells cultured from the vascular wall formed calcified nodules similar to those found in bone cell cultures and responded to transforming growth factor ß.³⁷ The predominant cells in these nodules had immunocytochemical features characteristic of microvascular pericytes, which are capable of osteoblastic differentiation. These findings provide additional evidence that arterial calcium in atherosclerosis is a regulated process similar to bone formation rather than a passive precipitation of calcium phosphate crystals.

Role of Calcium in Remodeling
The role of mineralization in the pathogenesis and fate of the coronary plaque is unknown. Although coronary calcification has in the past been regarded as a passive process of adsorption or precipitation, evidence reviewed here suggests that this may not be the case.¹⁹ If it is true that coronary calcification is an organized, regulated process, then to what end is this organization and regulation directed? That is, does calcification serve some functional role?

Teleologically, it can be speculated that coronary arterial calcification may, in a manner analogous to collateral formation, represent an attempt to protect threatened myocardium by strengthening weakened atherosclerotic plaque prone to rupture. Calcified lesions and fibrotic hypocellular lesions are much stiffer than cellular lesions,³⁸ and biomechanical data suggest that calcified areas are unlikely to be associated with sites of plaque rupture.³⁹ Demer⁴⁰ has shown that the presence of calcification alters the mechanical properties of the plaque. In vivo evidence of the relative stability of calcified lesions has been obtained with intravascular ultrasound (IVUS).⁴¹ Thus, coronary calcification might represent an attempt by the arterial wall to stabilize itself, thereby minimizing the risk of plaque rupture. For example, if a plaque develops a heavily calcified cap, it is about five times stiffer than a cellular lesion or normal vessel wall and very resistant to rupture.^13 38 Over the short term this may lead to increased stress near the junction of the cap and the adjacent intima, and it is here, at the interface between a calcified and noncalcified atherosclerotic section, that plaque rupture often occurs. Although the dissection that follows angioplasty is not a model for natural plaque disruption, focal calcification is a major reason for dissection after balloon angioplasty, and it may influence the length and severity of balloon angioplasty–induced dissections.^42 43 One theory is that with more extensive calcification and fibrosis of the vessel, these weak points may be eliminated and the risk of rupture correspondingly decreased. It can be speculated that the vessel is rendered less vulnerable to rupture only when extensive calcification has occurred, whereas the early or intermediate stages of calcification may actually enhance plaque vulnerability. This may help explain why calcification alone is not an ideal prognostic indicator for plaque rupture in heterogeneous populations¹³and is compatible with the high frequency of calcification found in older populations,^44 45 46 who tend to have the largest plaque burden.

Coronary remodeling associated with the development and progression of atherosclerotic disease is a recently described phenomenon⁴⁷ whereby the luminal cross-sectional area and/or external vessel dimensions become enlarged to compensate for increasing areas of mural plaque.⁴⁸ Coronary artery calcium is an intimate component of some plaques. In a histopathological investigation, Clarkson et al⁴⁹ have shown that plaques with microscopic evidence of mineralization were much larger and were associated with much larger coronary arteries than sections without microscopic evidence of calcification; this was true in both humans and nonhuman primates. The compensatory enlargement of atherosclerotic coronary segments may explain why coronary angiography frequently underestimates the severity of coronary disease as compared with histopathological studies. Studies attempting to correlate the site and amount of coronary calcium with percent luminal narrowing at the same anatomic site have shown a positive but nonlinear relation with large confidence limits.^50 51 However, coronary plaque and its associated coronary calcification may have only a poor correlation with the extent of histopathological stenosis.^49 52 In situ coronary calcium, on the other hand, appears to be closely associated with plaque size.⁵²

A recent study by Rumberger et al⁵³ emphasized that the total area of coronary artery calcification, determined by EBCT, is correlated in a linear fashion with the total area of coronary artery plaque on a segmental, individual coronary artery and whole coronary artery system basis. However, the areas of coronary calcification were approximately one fifth that of the associated coronary plaque. In addition there were clear areas of plaque without associated coronary calcium. These data suggest that one size of coronary plaque is most commonly associated with coronary calcium, but in smaller plaques the calcium is either not present or is undetectable with currently available imaging modalities. Thus, mild coronary plaque may be present without detectable coronary calcium at that same anatomic site.

Coronary Mineralization and Acute Coronary Artery Syndromes
Although calcification is a consistent finding in areas of significant focal coronary artery narrowing,²⁸ its presence also has implications in coronary thrombotic syndromes and coronary dissection after angioplasty. Some believe that mildly or moderately stenotic plaque is more likely to rupture and lead to coronary syndromes.² Hangartner et al⁵⁴ have shown in pathological examination of the hearts of 54 men who had stable angina that, for stenoses of more than 50% diameter narrowing, 48% were caused by concentric fibrous (hard) plaques, 28% by lipid-rich (soft) plaques, 12% by eccentric fibrous plaques, and 12% by eccentric lipid rich plaques. Forty-four percent of plaques causing stenoses of 30% to 50% were eccentric and were often in series with segments of higher-grade stenoses. Most patients had mixtures of all plaque types in varying proportions, but in the mean, two thirds were fibrotic and one third were lipid rich. These observations underlie the apparent paradox that the more extensive the coronary calcification, the more likely that a coronary event may occur. Calcification can be detected in some mildly or moderately stenotic plaques, which some believe to be the type more likely to rupture and lead to coronary syndromes,² while Cheng et al³⁹suggest that a calcified plaque itself is less, not more, likely to rupture. Rather, the presence of calcified plaque implies the likely association of lipid-rich and possibly unstable plaque. There are additional implications for predisposition to plaque rupture and myocardial infarction.^42 55 56 Van der Wal et al⁵⁷demonstrated that the site of intimal rupture or erosion of thrombosed coronary plaques is characterized by an inflammatory process, regardless of the dominant plaque morphology. Demer et al⁵⁸ have argued that the presence of a soft plaque, with a point of weakness induced by inflammation adjacent to an area of calcification, predisposes the plaque to rupture because of the presence of a tissue interface of differing physical properties that is subjected to the pulsatile changes of arterial pressure.

In Vivo Imaging Methods

Coronary artery calcification is potentially detectable in vivo by the following methods: plain film roentgenography; coronary arteriography; fluoroscopy, including digital subtraction fluoroscopy; cinefluorography; conventional, helical, and electron beam computed tomography; intravascular ultrasound; magnetic resonance imaging; and transthoracic and transesophageal echocardiography. In current practice, fluoroscopy and EBCT are most commonly used to detect coronary calcification noninvasively, while cinefluorography and IVUS are used by coronary interventionalists to evaluate calcification in specific lesions before angioplasty.

Plain Chest Radiographs
Coronary calcification is not easily detected on chest radiographs. Accuracy is only 42%, compared with fluoroscopy, which in itself is not extremely sensitive.⁵⁹The chest film, while readily available and inexpensive, has low sensitivity for detecting coronary calcification.⁶⁰

Fluoroscopy
Fluoroscopy has frequently been used to detect calcification in coronary arteries. Table 1, adapted from Detrano and Froelicher,⁶¹ summarizes seven studies examining fluoroscopic detection of coronary calcification in 2670 patients undergoing coronary arteriography.^{62 63 64 65 66 67 68} Sensitivity in detecting significant stenoses (greater than 50% diameter obstructions) ranged from 40% to 79%, with specificity ranging from 52% to 95%.

Table 1. Correlation of Fluoroscopic Detection of Coronary Calcification With Presence of Significant* Angiographic Narrowing  {not copied because of low relevance}

In a fluoroscopic study of 613 asymptomatic male aircrew members who underwent coronary arteriography because of one or more abnormal screening tests,⁶⁹ coronary artery calcification had a 66.3% sensitivity and a 77.6% specificity in determining angiographically significant coronary stenosis (greater than 50% diameter narrowing). The positive predictive value was 37.7% and negative predictive value was 91.9%; for disease with greater than 10% stenosis, sensitivity was 60.6% and specificity 85.9%. The authors concluded that a fluoroscopically negative calcification test indicated a low likelihood of significant coronary artery disease, whereas a positive calcification test substantially increased the likelihood of angiographically significant coronary artery disease.

Although fluoroscopy can detect moderate to large calcifications, its ability to identify small calcific deposits is low. In one study⁷⁰ only 52% of calcific deposits seen on high-resolution EBCT images could be detected fluoroscopically (P<.001). The mean calcium density in lesions detected by EBCT was +99 HU, whereas for lesions detected by fluoroscopy it was +546 HU, signifying that only larger, more highly calcified plaques are detectable with fluoroscopy as compared with CT. This may explain why calcification detected by EBCT is more sensitive but less specific than fluoroscopy, as discussed below.

Fluoroscopy is widely available in both inpatient and outpatient settings and is relatively inexpensive, but it has several disadvantages. In addition to only a low to moderate sensitivity, fluoroscopic detection of calcium is dependent on the skill and experience of the operator as well as the number of views studied. Other important factors include variability of fluoroscopic equipment, the patient's body habitus, overlying anatomic structures, and overlying calcifications in structures such as vertebrae and valve annuli. With fluoroscopy, quantification of calcium is not possible, and film documentation is not commonly obtained.

Conventional Computed Tomography
Because calcium attenuates the x-ray beam, computed tomography (CT) is extremely sensitive in detecting vascular calcification. In a study evaluating CT detection of calcium as a marker of significant angiographic stenosis,⁷¹sensitivities of 16% to 78% were found, depending on which vessel included the calcified plaque. Specificities were 78% to 100% and positive predictive values 83% to 100%, suggesting that significant coronary artery disease was likely to be present when coronary calcification was seen on CT.

Computed tomography, fluoroscopy, and angiography were compared in a study of 47 patients with a mean age of 57 years.⁷² The CT scans showed calcification in 62% of vessels with significant lesions on angiography, whereas fluoroscopy showed calcium in only 35%. In a group without angina, coronary calcification was found by CT in only 4%, and no patient had significant stenosis on coronary arteriography. In this study CT detected calcification in all patients in whom fluoroscopy showed calcification and in all patients in whom angiography showed stenosis. Overall, CT showed calcification in 50% more vessels than did fluoroscopy.

Another study of suspected coronary artery disease patients⁷³ found that 90% of a group of 108 patients with calcification detected by conventional CT had significant stenosis (greater than 75% narrowing), whereas, of 121 patients with significant stenosis on angiography, 80% had calcification on CT. Sensitivity was 65%, specificity, 87%.

Therefore, while conventional CT appears to have better capability than fluoroscopy to detect coronary artery calcification, its limitations are slow scan times resulting in motion artifacts, volume averaging, breathing misregistration, and inability to quantify amount of plaque.

Helical or Spiral Computed Tomography
Helical CT has considerably faster scan times than conventional CT—on the order of 1 second—but imaging as fast as 0.6 second is possible. Overlapping sections also improve calcium detection. Shemesh et al⁷⁴ reported coronary calcium imaging by helical CT as having a sensitivity of 91% and a specificity of 52% when compared with angiographically significant coronary obstructive disease. Double helical CT was useful in predicting the absence of coronary artery disease in elderly women in the absence of calcification.⁷⁵ However, other preliminary data have shown that even at these accelerated scan times, and especially with single helical CT, calcific deposits are blurred due to cardiac motion, and small calcifications may not be seen.⁷⁶ Still, helical CT remains superior to fluoroscopy and conventional CT in detecting calcification. Double-helix CT scanners appear to be more sensitive than single-helix scanners in detection of coronary calcification because of their higher resolution and thinner slice capabilities.

Electron Beam Computed Tomography
General Description
Electron beam computed tomography uses an electron gun and a stationary tungsten "target" rather than a standard x-ray tube to generate x-rays, permitting very rapid scanning times. Originally referred to as cine or ultrafast CT, the termEBCT is now used to distinguish it from standard CT scans because modern spiral scanners are also achieving subsecond scanning times. For purposes of detecting coronary calcium, EBCT images are obtained in 100 ms with a scan slice thickness of 3 mm. Thirty to 40 adjacent axial scans are obtained by table incrementation. The scans, which are usually acquired during one or two separate breath-holding sequences, are triggered by the electrocardiographic signal at 80% of the RR interval, near the end of diastole and before atrial contraction, to minimize the effect of cardiac motion. The rapid image acquisition time virtually eliminates motion artifact related to cardiac contraction. The unopacified coronary arteries are easily identified by EBCT because the lower CT density of periarterial fat produces marked contrast to blood in the coronary arteries, while the mural calcium is evident because of its high CT density relative to blood. Additionally, the scanner software allows quantification of calcium area and density. An arbitrary scoring system has been devised based on the x-ray attenuation coefficient, or CT number measured in Hounsfield units, and the area of calcified deposits.⁷⁷ A screening study for coronary calcium can be completed within 10 or 15 minutes, requiring only a few seconds of scanning time. Electron beam CT scanners are more expensive than conventional or spiral CT scanners and are available in relatively fewer sites.

Efficacy
Tanenbaum et al⁷⁸ were the first to report use of EBCT for detecting calcific deposits in the coronary arteries. The amount of calcification detected on 50 ms scans with 1.5 mm² pixel size was compared with coronary arteriography in 54 patients. Stenosis of the coronary arteries was considered significant when luminal narrowing was 70%; in the left main coronary artery the accepted significant stenosis was 50%. In this series angiograms showed significant coronary artery disease in 43 patients; 88% of these had detectable calcium in at least one coronary artery. Specificity for significant stenosis in this study was 100%. The scans were confined to examination of the proximal 76 mm of the coronary arteries and not distal vascular sections.

In 1992 Agatston et al⁷⁰ reported the first large clinical series in which EBCT was used to detect calcification of the coronary arteries. Five hundred eighty-four consecutive patients with a mean age of 48 years underwent 100 ms EBCT scans with 3 mm thick slices (0.46 mm² pixel size); 50 also underwent fluoroscopic examination. One hundred nine patients had coronary artery disease established by a history of myocardial infarction (22) or angiographic evidence of greater than 50% diameter narrowing on coronary angiography (87). The remaining 475 patients had no history of coronary disease. Patients with a history of coronary artery disease consistently had more calcium than patients of comparable age with no history of coronary artery disease (P<.0001). A total calcium score of 50 (a weighted sum of x-ray density and total calcium area) resulted in a sensitivity of 71% and specificity of 91% for patients in the 40- to 49-year-old age group with at least 50% stenosis. A total calcium score of 300 in the 60- to 69-year-old age group with similar severity of stenosis had a sensitivity of 74% and a specificity of 81%. The negative predictive value of a zero calcification score was 98% (age 40 to 49), 94% (age 50 to 59), and 100% (age 60 to 69). Electron beam CT showed calcium in 90%, and fluoroscopy showed it in 52% of patients with established coronary artery disease, although only 87 patients had angiographic documentation. The authors concluded that EBCT appeared to be an excellent technique for detecting and quantifying calcification of coronary arteries. The study additionally showed that the mean total calcium score increased with age.

Breen et al⁷⁹ studied 100 patients aged 23 to 59 years who underwent EBCT and angiography. Significant obstruction was defined as greater than 50% narrowing of the vessel diameter on the angiogram. Sensitivity of detecting any calcium (ie, calcium score greater than 0) in individuals with significant angiographic stenosis was 100%; specificity was 47%. In patients whose angiograms showed stenosis greater than 10%, sensitivity for detecting any calcium by EBCT was 94%, and specificity was 72%. In this series eight patients with calcification had no angiographic evidence of coronary artery disease, and 28 patients with calcification had mild or moderate coronary artery disease.

Fallavollita et al⁸⁰ compared EBCT detection of calcium with coronary angiography in 106 patients under the age of 50 and found an 85% sensitivity and 45% specificity in patients with significant stenosis, defined as greater than 50% diameter narrowing on angiography. For multivessel disease, sensitivity was 94%, while in single-vessel disease it was 75%. Positive predictive value was 66%. Because negative predictive value was only 70%, the authors emphasized that the absence of EBCT calcium may not exclude significant coronary disease in this younger patient group. However, it should be emphasized that only 20 3-mm EBCT sections were examined, as opposed to 30 to 40 sections in most other angiographic comparison studies, and distal calcification, commonly seen in the right coronary artery,⁵⁰ may have been missed in the analysis. Thus, the negative predictive value in single-vessel disease may have been underestimated.

A larger multicenter study⁸¹ that looked at coronary calcification as an indicator of significant stenosis involved 431 patients with symptoms of coronary artery disease (CAD) (251 men and 180 women; mean age 56 years). In this group, sensitivity of any detectable calcification by EBCT as an indicator of significant stenosis (greater than 50% narrowing) was 92% and specificity 43%. When these CT images were reinterpreted in a blinded and standardized manner, however, specificity was only 31%.⁸²

In a more recent multicenter study⁵¹ of 710 enrolled patients, 427 had significant angiographic disease, and coronary calcification was detected in 404, yielding a sensitivity of 95%. Of the 23 patients without calcification, 83% had single-vessel disease on angiography. Of the 283 patients without angiographically significant disease, 124 had negative EBCT studies.

Although EBCT is very sensitive in defining coronary vascular calcification, the extent and site of calcification does not equate with site-specific stenosis. Bormann et al⁸³ found that calcium scores were not predictive of a significant stenosis at the calcification site and that no receiver-operator characteristic curve could be found that would suggest a clinically useful calcium score as an indicator of more than 70% stenosis at the same anatomic site. In this study, however, only one patient had significant stenosis in the absence of calcification. In a series of 150 patients from two institutions undergoing EBCT scans and coronary arteriography, Stanford et al⁸⁴ found only one patient who had greater than 50% narrowing in the absence of calcification.

The presence and amount of calcium detected in a coronary artery by EBCT indicates the presence and amount of associated atherosclerotic plaque.⁵³Additionally, in a recent review, Rumberger et al⁸⁵ suggested that the magnitude of the calcium score can be used to a high specificity in predicting associated stenosis somewhere within the epicardial coronary system, but the extent of calcification at a given anatomic site may be less useful in predicting luminal narrowing identified at angiography. Table 2 provides a summary of seven studies relating EBCT calcification to significant coronary stenosis.^{50 51 79 80 8687 88} In these studies the absence of calcification indicated a low likelihood of significant luminal obstruction. Except for the Fallavollita study,⁸⁰ which did not use the same imaging protocol as the others, the negative predictive values ranged from 84% to 100%. Predictive values are based on prior probability; because most patients in these studies had indications for angiography, these predictive values would not apply to asymptomatic patient groups or populations.

Table 2. Calcium Detected by Electron Beam Computed Tomography (EBCT) as an Indicator of Significant Angiographic Stenosis*   {Not included because of low relevance}

Reproducibility of Results
Reproducibility of the calcium measurement or calcium score is essential to assess progression, stabilization, and/or regression of disease in individual patients and to conduct longitudinal studies based on a change in coronary calcium determination. Janowitz et al⁸⁹ evaluated calcific plaque in 25 symptomatic and asymptomatic patients 406 days apart. Subjects with proven obstructive coronary artery disease on angiography had a 48% increase in calcium score compared with 22% in asymptomatic subjects. Patients with obstructive coronary artery disease had 55 new calcific deposits on the follow-up study versus 18 in the asymptomatic group. Although Janowitz et al concluded that EBCT may be useful for studying both the natural history of coronary disease and the effects of intervention, they did not consider that interscan differences might result from interscan measurement error.

Several studies have shown a variability in repeated measures of coronary calcium by EBCT; therefore, use of serial EBCT scans in individual patients to track the progression or regression of calcium is problematic. Bielak et al⁸⁶ studied 256 patients who had EBCT evaluation of calcium and coronary angiography. A repeat EBCT scan was done immediately after the initial scan but after the patient had gotten off the scanner table and walked briefly around the room. These investigators found that segmental areas with small amounts of calcification (less than 2 mm²) were seen at a second examination only 50% of the time (P<.0001). Other investigators^90 91 92 have suggested that a large increase in calcium score is needed before a change in calcium score can be attributed to progression of pathology rather than measurement error. Thus, although calculation of the total calcium score using EBCT is quantitative^{50 51 53 77 79 82 88 93 94} and operator independent,⁹⁵ reproducibility varies from excellent⁹⁶ to moderate,⁹¹ depending on the laboratory and, most likely, the actual magnitude of the calcium score. Therefore, the conclusions of Janowitz et al can most probably be applied to investigations in which large groups of patients are studied but not necessarily to individual patients undergoing routine clinical follow-up. The variability in repetitive measurements of calcium score is largely related to scan misregistration secondary to patient motion. Preliminary application of 6 mm thick slice scanning⁹² rather than the traditional 3 mm EBCT scanning has been suggested to halve interscan variability. Further improvements in electrocardiographic triggering algorithms and shorter total scan time have recently been implemented and should further reduce interscan variability. Also, careful patient instruction about breath holding can reduce respiratory misregistration.

Costs and Risks of Scanning
Assessment of coronary calcification by EBCT can be done in virtually any subject and provides anatomic rather than physiological information. Thus, no preparation or discontinuation of medications is required before testing, which is totally noninvasive, involves minimal patient cooperation, and produces results available for qualitative evaluation on an immediate basis. Quantitative review of calcium scoring using EBCT requires additional analysis but is available generally within 10 to 20 minutes. The current total charge for an EBCT examination (limited CT of the chest) and interpretation is approximately the same as the charge for a routine nurse-monitored treadmill test. Charges vary in different parts of the United States but average between $300 and $400. This is roughly one half the charge for a stress echocardiogram and one third the charge for a stress radionuclide examination. Radiation dosimetry for a single screening EBCT scan for coronary calcium has an effective (integrated over thorax) radiation dose of 82 mrem for males and approximately 150 mrem for females (accounting for breast irradiation).^88 97 Although it is difficult to make direct comparisons due to differences in dose delivery and localization, a posteroanterior and lateral chest x-ray combination involves approximately 10 mrem and a screening two-view mammogram about 35 mrem. A thallium scan delivers a highly localized dose of approximately 1 rem to the thorax and abdomen⁹⁸ ; conventional coronary arteriography results in radiation doses two to three orders of magnitude or greater than that from an EBCT coronary calcium scan.⁹⁹ However, even though the EBCT radiation dose is minimal, indiscriminate use or mass screening is not condoned. At present the AHA recommends that it be done only at the request of a physician and for specific clinical indications as outlined above.

Intravascular Ultrasound
Intravascular ultrasound is a newer method for detecting coronary atherosclerosis.¹⁰⁰ By using transducers with rotating reflectors mounted on the tips of catheters, it is possible to obtain cross-sectional images of the coronary arteries during cardiac catheterization. The sonograms provide information not only about the lumen of the artery but also about the thickness and tissue characteristics of the arterial wall. Calcification is seen as a hyperechoic area with shadowing: fibrotic noncalcified plaques are seen as hyperechoic areas without shadowing.¹⁰¹ Friedrich and colleagues¹⁰² reported on the ability of IVUS to detect the histological extent of in situ coronary calcium. Examining 50 fresh human coronary artery vessel segments and using histological confirmation, the sensitivity of IVUS for dense, coherent calcification was 90%, with a specificity of 100%. However, for small accumulations of microcalcification and/or scattered calcification (areas less than 0.05 mm²), sensitivity was only 64%, although high specificity was preserved.

Rickenbacher et al,¹⁰³ using IVUS, found that although intimal hyperplasia was seen early after cardiac transplantation, coronary calcification developed more slowly and was detected in 2% to 12% of patients within 5 years, increasing to 24% 6 to 10 years after transplantation.

Mintz et al⁴¹ compared IVUS to angiography and found that angiography was significantly less sensitive than IVUS in detecting calcification at the site of a target lesion. This finding was confirmed by Tuzcu et al,¹⁰⁴ who also found calcium by angiography at another site in the coronary tree in two thirds of patients without calcium at the target site.

The disadvantages in use of IVUS, as opposed to other imaging modalities, are that it is invasive and currently performed only in conjunction with selective coronary angiography, and it visualizes only a limited portion of the coronary tree. Thus, it has no role in screening for coronary artery disease. Although invasive, the technique is clinically important because it can show atherosclerotic involvement in patients with normal findings on coronary arteriograms^{102 104 105106} and helps define the morphological characteristics of stenotic lesions before balloon angioplasty and selection of atherectomy devices.^42 104

Magnetic Resonance Imaging
The ability to detect coronary calcification with magnetic resonance imaging (MRI) is limited. Calcium is almost always characterized by low signal intensity on both T1- and T2-weighted spin-echo (static dark blood) images primarily as a result of the low density of mobile protons in calcified lesions.¹⁰⁷ In addition, because of the sensitivity of MRI to the heterogeneous magnetic susceptibility found in calcified tissue, gradient-echo magnitude (static or dynamic bright blood) images also typically depict calcified lesions as discreet areas of reduced signal intensity.¹⁰⁸ However, the theory is more complex for particulate calcium, as T1 relaxation may be enhanced by surface relaxation mechanisms, and this may result in a hyperintense signal.¹⁰⁹ Experimental studies have shown that calcium particles with greater surface area create greater T1 relaxivity, thus negating the decreases in signal intensity caused by reductions in both proton density and T2. For concentrations of calcium particulate of up to 30% by weight, the intensity on standard T1-weighted images increases but then subsequently decreases with increasing concentrations.¹⁰⁹ Because microcalcifications do not substantially alter the signal intensity of voxels that contain a large amount of soft tissue, the net contrast in such calcium collections is low. Therefore, MRI detection of small quantities of calcification is difficult, and there are no reports or expected roles for MRI in detection of coronary artery calcification.

Transthoracic and Transesophageal Echocardiography
Transthoracic (surface) echocardiography is exquisitely sensitive to detection of mitral and aortic valvular calcification; however, visualization of the coronary arteries has been documented only on rare occasions, because of the limited available external acoustic windows. Thus, there are no practical applications for transthoracic echocardiographic localization of coronary artery calcifications. Transesophageal echocardiography is a widely available methodology that often can visualize the proximal coronary arteries.^110 111 However, neither of these methods has sufficient density, resolution, or available acoustic windows to reliably define in situ coronary artery calcium.

Epidemiological Considerations

Epidemiology
In 1961 Blankenhorn¹¹² summarized the evidence that coronary artery calcification occurred only in sites involved with atherosclerosis. This observation has been confirmed by several investigators. Because it is now accepted that the initial response of the artery to atherosclerosis is adaptive (arterial enlargement associated with atherosclerosis), and that, for most people, extensive atherosclerosis is not associated with coronary symptoms,^47 48 49 it is additionally important to know the prevalence of calcification in atherosclerotic lesions. Few publications have addressed this question, partly because methods to determine the presence of atherosclerotic plaque and calcification in vivo are imprecise. Furthermore, the definition of coronary artery disease may be based on clinical symptomatology or angiography, whereas descriptions of atherosclerosis may come from autopsy studies, IVUS, or other imaging methods. Nevertheless, the literature reviewed suggests atherosclerotic plaque is present in 50% of individuals aged 20 to 29 years, rising to 80% in individuals aged 30 to 39.¹¹³Calcification is present in 50% of individuals aged 40 to 49 and 80% of individuals aged 60 to 69,^{55 113 114 115 116} whereas significant stenosis is present in only 30% of individuals aged 60 to 69.¹¹³ For individuals aged 30 to 39 with symptomatic coronary artery disease, calcification may be present in 72%^55 115116 and stenosis in 60%.¹¹⁷

In autopsy studies a modest correlation has been observed between percent coronary stenosis and extent of calcification.²⁸ Mautner et al¹¹⁸ observed that 54% of coronary segments with greater than 75% stenosis had coronary calcification on directed EBCT scanning, but calcification was present in only 41%, 23%, and 6% of those with stenosis of 51% to 75%, 26% to 50%, and 1% to 25%, respectively. Overall, more calcified sites were associated with nonstenotic disease (632) than stenotic disease (368). In another analysis of these data,¹¹⁸ 93% of coronary arteries with at least one stenosis greater than 75% had calcification, whereas only 20% of arteries with less than 50% stenosis and only 4% of those arteries with less than 25% stenosis contained calcific deposits. Data were not presented for the group with 50% to 75% stenosis. These studies both relied on histology to define percent stenosis as an area of plaque per area within the internal elastic lamina rather than using percent stenosis determined by reduction in luminal diameter as seen on two-dimensional projections of angiographic images. Because area varies by the square of the radius, histologically estimated coronary stenosis is considerably greater than that provided by coronary angiography¹¹⁹ ; thus, 50% and 75% area stenosis on histopathology may correlate with 15% and 30% to 50% diameter stenosis by angiography, respectively.

A summary of the literature relating coronary calcification to clinical disease is complicated by the evolution of technology for identifying calcification (vide supra). Thus, in an early study using fluoroscopy,¹²⁰ prevalence of calcium in patients with and without symptoms was, respectively, 28% and 2% in persons aged 30 to 40 years and 95% and 56% in persons aged 60 to 70. On the other hand, a more recent study using EBCT⁷⁷ showed prevalences of 100% and 25% in younger persons and 100% and 74% in older persons with and without symptoms.

This dilemma is further confounded by the fact that while only a minority of patients undergoing catheterization have nonobstructive disease, this population likely represents the vast majority of apparently healthy middle-aged adults. Furthermore, recent clinical studies suggest that patients with mild coronary artery disease (less than 50% stenosis) may be at relatively high risk of developing clinical events,¹²¹ that the angiographic degree of stenosis is a poor predictor of subsequent culprit lesions, and that angiography cannot differentiate stable from unstable lesions with a substantial degree of confidence.²

Risk Factors for Coronary Calcification
Age and gender are the most important risk factors for coronary calcification, ranging from 14% for men and women less than 40, to 93% to 100% for men older than 70, and 77% to 100% for women older than 70.^{122 123 124 125}

The Framingham study¹²⁶ puts the 8-year risk of coronary heart disease for the average middle-aged person at between 1% and 5%, depending on age and risk factors, with an expected 8-year incidence of events ranging from less than 1% to 15% for persons younger than 40 to older than 80 years. Comparing these figures with the prevalence of calcification described above, it is evident that prevalence of calcification is much higher than risk of events. Thus, only a small proportion of persons with atherosclerosis and detectable coronary calcium will eventually develop clinical coronary events, and effective risk stratification will require a threshold calcium volume (at present undefined), score, or distribution pattern of calcification to define a high risk versus low or intermediate coronary calcium screen.¹²⁷

Several investigators have studied risk factors for their association with coronary artery calcifications.^{46 123 128 129 130 131} Elevated plasma cholesterol has most consistently been shown to be associated with coronary calcification.^{45 122 125128 129 130 131} Diminished HDL,^129 130 cigarette smoking,^46 123 130 elevated blood pressure,^123 130 obesity,^129 130 number of risk factors,^123 131 diabetes,⁴⁶and elevated triglycerides¹²⁹ have all been shown to be associated with coronary calcification in one or more studies in one or more patient groups.

Coronary Calcification and Clinical Outcomes
Although the presence or absence of calcification is related to overall atherosclerotic plaque burden, it is event data (angina, myocardial infarction, necessity for percutaneous transluminal coronary angioplasty [PTCA], or coronary artery bypass surgery) that are important in determining the clinical significance of coronary artery calcification. Little et al¹²¹ have shown that acute occlusions resulting in myocardial infarction often occurred in vessels with less than 50% angiographically determined stenosis, but it is also important to note that these patients frequently had concomitant severe angiographic stenoses. This study, however, did not assess calcification. Brundage et al¹³² reviewed several series reporting the 5-year mortality and incidence of myocardial infarction. In this review there was a 7% 5-year mortality in 1275 patients with less than 50% angiographic stenosis versus a 3% mortality in 4250 persons with normal arteriograms. The 5-year incidence of myocardial infarction was 5% in 188 patients with mild stenosis versus 1% in the 573 persons with normal arteriograms. Brundage concluded that infarction and death were two to three times more common in persons with mild plaque than in those without plaque.

It is important to study not only event data but also the progression or possible regression of disease, because Waters et al¹³³ have shown coronary atherosclerotic progression is a strong independent predictor of future coronary events.

Margolis and colleagues⁶⁶ studied 800 patients referred for cardiac catheterization predominantly for angina pectoris (90%). They observed that symptomatic patients with calcification demonstrated on conventional fluoroscopy had a 5-year survival rate of 58% versus 87% in those without detectable calcium. Furthermore, the prognostic significance of coronary artery calcification appeared to be independent of age, gender, and angiographically diseased vessels. In addition, calcification was independent of exercise and left ventricular function tests. On the other hand, a smaller study by Hudson and Walker¹³⁴ of 78 patients without cardiac symptoms concluded that there was no difference in 5-year survival whether or not calcium was present. Although they suggested that coronary calcium may have a different significance in symptomatic and asymptomatic populations, both studies^66 134 suffered from methodological problems regarding selection and measurement bias.

Detrano et al¹³⁵ studied survival in asymptomatic, high-risk subjects with coronary artery calcification detected on fluoroscopy. These investigators followed 1461 subjects with a greater than 10% risk of having a coronary event within 8 years. (A coronary event was defined as angina, documented myocardial infarction, myocardial revascularization, or death from coronary heart disease.) Events at 1 year occurred in 5.4% of 691 subjects with coronary calcification versus 2.1% of the 768 subjects without fluoroscopic calcium (P=<.001). One-vessel calcification incurred an event risk of 5.4%; two-vessel, 5.6%; and three-vessel, 6.2%. Detrano et al found that radiographically detectable calcium was associated with a risk for having an event 2.7 times greater compared with the group with no calcification. They also found that the presence of calcification was an independent predictor of at least one coronary event when controlled for age, gender, and other risk factors. However, it should be emphasized that three deaths due to coronary heart disease and two nonfatal myocardial infarctions occurred in subjects without detectable coronary calcium. Their conclusions were that the presence of coronary calcium detected fluoroscopically identified an increased risk of a cardiac event in asymptomatic high-risk subjects at 1 year, and this increased risk was independent of that incurred by standard risk factors.

There are limited data available on the prognostic significance of coronary calcium detected by conventional x-ray computed tomography. Naito and colleagues¹³⁶followed a group of 241 older individuals (136 men, 105 women) for an average of 4 years. Among 82 patients with coronary calcium, 4.9% developed myocardial infarction, whereas none of the 159 patients without coronary calcium experienced infarction. However, mortality (all causes) was no different between the two groups. In women overall mortality was 26% (3.7% due to infarction) in the calcium group and 9% (0% due to infarction) in the noncalcium group. In men total mortality was 13% (5.5% due to infarction) in the calcium group and 12% (0% due to infarction) in the noncalcium group. However, the mean age of these persons was 61 years, and there were no data for younger individuals.

Investigators in a recently published multicenter EBCT calcium study⁸² looked at event data in 501 symptomatic patients who were studied with both EBCT for calcium and coronary angiography. The majority of these patients had symptoms of coronary artery disease. In this group 1.8% died and 1.2% had nonfatal myocardial infarctions during a mean follow-up period of 31 months. A threshold of 100 or greater in the calcium score was shown to be highly predictive in separating patients with cardiac events at follow-up from those without events and calcium scores of less than 100. In this study, logistic regression, which included, in addition to calcium score, age, gender, and coronary angiographic findings as independent variables, showed that only log calcium score predicted events.

Mautner et al⁵⁰ looked at the amount of calcification detected on EBCT examinations and the percent blockage determined histomorphometrically. In the 1426 segments from coronary arteries of patients with histories of symptomatic coronary artery disease, calcium was present on EBCT in 41%. In the 1535 segments from asymptomatic coronary artery disease patients, EBCT detected calcium in 24%, whereas in normal control subjects, only 4% had calcium. EBCT had a sensitivity of 94% for detecting calcium in a coronary artery versus a specificity of 76%. The positive predictive value was 84%; the negative predictive value was 90%. Mautner et al concluded that the EBCT calcium score appeared to be an effective predictor of coronary artery disease. In this study the symptomatic CAD group was defined as having a history of angina or myocardial infarction and a narrowing greater than 75% in at least one section of a coronary artery. The asymptomatic CAD group had at least one segmental narrowing greater than 75% but no symptoms. The control group had no symptoms and no narrowing greater than 75%.

Arad et al¹³⁷ followed 1173 initially asymptomatic patients for an average of 19 months. Nineteen patients had 27 cardiovascular events, including one death, seven myocardial infarctions, and one nonhemorrhagic stroke. In addition 18 patients developed symptoms requiring coronary bypass surgery (8) or PTCA (10). Electron beam CT coronary calcium scores were correlated with subsequent events, depending on the threshold for the lower limit of calcium score. For coronary artery calcium score thresholds of 100, 160, and 680, EBCT had sensitivities of 89%, 89%, and 53%, and specificities of 77%, 82%, and 95%, respectively. Negative predictive values were greater than 99%, and odds ratios ranged from 22.2 to 35.6:1 (P<.00001) for these thresholds. Other risk factors, such as presence of hypercholesterolemia, low HDL cholesterol, hypertension, diabetes, and family history failed to predict subsequent events. Extrapolation of the results of this study to other asymptomatic populations must be done with caution, because there were only eight major coronary events (death or myocardial infarction), and patients were self-selected for entry into the study.

Although these correlative studies indicate that patients with greater amounts of coronary calcification are more likely to suffer a clinical event compared with patients without calcification or lesser amounts, it is important to note that they do not address the relation of calcification to the process or likelihood of plaque rupture.¹³⁸ These studies evaluated calcium in the entire coronary vasculature, and it is not known whether events were a consequence of ruptured plaques that were or were not calcified.

In addition to the rationale that detection of coronary artery calcium is useful in identifying those at risk for acute coronary events, early detection of mild coronary atherosclerosis is of potential value also, particularly if the process can be slowed, arrested, or reversed. There are substantial data to indicate that lowering serum cholesterol in patients with known coronary artery disease (secondary prevention) reduces the incidence of nonfatal infarction, fatal infarction, cardiovascular mortality, and all-cause mortality.¹³⁹ Although some have questioned the wisdom of routine cholesterol screening in asymptomatic populations,¹⁴⁰ there is mounting evidence that risk reduction and lipid lowering in patients with elevated cholesterol without clinical disease (ie, high-risk, asymptomatic individuals) is efficacious.^{141 142 143}

Practical Applications of Coronary Calcium Detection

Based on the prior discussion, clinical application of information about coronary artery calcification is predicated on its noninvasive detection and quantification as a surrogate measure of atherosclerotic plaque. This limits imaging to conventional and digital fluoroscopy and x-ray computed tomography. Of these methods, only EBCT presently has the capability to quantify coronary artery calcification.

Potential uses for coronary calcium assessments, based on this information, fall into three broad categories that reflect common clinical practice dilemmas or decision paradigms:

1. Evaluation of patients with chest pain: results used in the decision to perform adjunctive or additional noninvasive stress testing, a coronary angiogram, or proceed with medical therapy for angina pectoris, etc.

2. Screening of asymptomatic subjects: goals include identifying subjects for aggressive risk factor management, further diagnostic workup with exercise testing and angiography, and exclusion from high-risk occupations (airline pilot, etc).

3. Following progression of coronary atherosclerosis: scan results at follow-up intervals help determine efficacy of pharmacological or nutritional intervention (eg, lipid-lowering agents, antioxidants, or other therapies) for retarding progression of atherosclerosis.

The following section reviews the data supporting or contesting the validity of each of these potential applications.

Evaluation of Patients With Chest Pain
Patients with chest pain of an indeterminate nature, particularly if coronary status is unknown (no prior infarction, angiogram, or other definitive diagnostic result), frequently require further investigations before a decision can be made regarding the proper course of treatment. In ambulatory subjects, electrocardiographic exercise testing, with or without cardiac imaging, is commonly used. If the results are markedly abnormal, the patient is frequently referred for a coronary angiogram to obtain information about the need for revascularization. If the results are completely normal, the patient may be reassured or referred for further evaluations directed at noncardiac causes of chest pain. In the case of equivocal results, the decision to administer medical therapy, perform an alternative form of stress testing, implement aggressive risk factor management, or perform coronary angiography will depend on other aspects of the patient's symptoms and wishes as well as the physician's judgment. The physician's decision on the need for further testing and/or responses to the results of initial testing should consider the specificity and sensitivity of exercise test results, along with the pretest probability based on symptoms, risk factors, etc, in estimating the postexercise test probability of not only angiographically severe disease but also morbid events.¹⁴⁴ Specific results are more important when they are present and sensitive results when they are absent.

Numerous studies^{62 63 64 67 77 87 116 145 146 147 148 149 150} have shown that coronary calcium assessment using fluoroscopy or EBCT has a sensitivity for significant angiographic stenoses comparable to that of exercise tests when used with symptomatic patients, although specificity is lower. Moreover, three of these studies have shown that symptomatic patients with coronary calcium have at least a fourfold increased risk of death or infarction when compared with those with less or no calcification.^{68 82 86 151} The fluoroscopic finding of at least one definitely calcified coronary vessel⁶⁶ or the EBCT finding of a coronary calcium score exceeding 100 (or calcium phosphate mass exceeding 20 mg⁸² ) has been shown to be highly predictive of the presence of advanced coronary plaque and stenosis. This can be helpful in decisions to proceed with additional noninvasive stress testing or even to proceed to angiography in this patient subset with chest pain of uncertain origin. In general, greater degrees of calcification are consistent with greater amounts of atherosclerotic plaque and more advanced associated coronary luminal narrowing.^85 93

There are four published studies of sensitivities and specificities with regard to angiography that involved comparisons of radiographic coronary calcifications and exercise testing results in the same symptomatic subjects.^{67 135 152 153} They are listed in Table 3. In addition to these studies, Spadaro and colleagues¹⁵⁴ have published an abstract reporting the results of a comparison of EBCT with exercise thallium scintigraphy in 150 patients. These authors did not report sensitivity and specificity but did report an overall accuracy of 79% for EBCT compared with 63% for thallium scintigraphy. Detrano et al,⁶⁸ using multivariate techniques, have shown that fluoroscopic coronary calcium adds independent information for predicting angiographic stenosis when both fluoroscopy and exercise thallium scintigraphy are done before angiography. The advantage of using an assessment of coronary artery calcification under these circumstances is that it can be done regardless of the patient's ability to exercise to a maximum workload and regardless of the presence of resting electrocardiographic abnormalities. Additionally, noninvasive assessment of coronary artery calcium, an anatomic and nonphysiological evaluation of the coronary arteries, is not influenced by concomitant use of various cardiotonic and vasoactive drugs that may confound performance and/or interpretation of an exercise test. However, conventional stress testing provides a physiological basis for the chest pain syndromes and valuable guidance on subsequent pharmacological therapy of angina if the test is deemed positive. Also, there is general consensus as to the interpretation of stress testing, with and without imaging of the myocardium. On the other hand, there is no consensus on the proper threshold of calcium score for a positive versus a negative EBCT scan.

Table 3. Comparisons of Radiographic Coronary Artery Calcifications and Exercise Testing Results in Symptomatic Subjects*  {table not reproduced for lack of relevant information}

Perhaps the most valuable finding in the symptomatic patient is a negative EBCT scan for coronary calcium. As discussed earlier, the negative predictive value of an EBCT calcium scan for significant (ie, 50% or greater diameter stenosis in any major coronary vessel) is greater than 90% and perhaps closer to 95% in some circumstances. Electron beam CT scanning may then be an appropriate first test in individuals with atypical cardiac symptoms, in whom the pretest likelihood for ischemic disease is considered low by the clinician. Persons in this group are likely to have a negative or minimal calcium score. Those with a zero or very low calcium score (generally less than 10) could be reassured and further testing directed at noncardiac sources of chest pain. Conversely, if the calcium score is consistent with moderate or severe atherosclerotic plaque development, then additional cardiac evaluation, including formal stress testing, may be an appropriate next step.

Thus, there are insufficient data to recommend coronary artery calcium scanning in lieu of stress testing as a broad generalization in patients with chest pain. However, there are sufficient data to suggest that coronary calcium evaluation, especially with EBCT, is ready for clinical application in the de novo patient with chest pain, particularly with an "atypical" chest pain presentation, and that this anatomic evaluation may be useful by itself or in addition to exercise testing.⁸⁵

Screening of Asymptomatic Subjects
Asymptomatic subjects differ from symptomatic subjects (patients) in two regards. First, according to the data obtained from the Framingham study, asymptomatic subjects have a relatively low risk of morbid coronary events.¹⁵⁵Risk is plotted against age in the Figure, using data from Anderson et al,¹⁵⁵which also depicts the prevalence of coronary calcification in another group of asymptomatic men of the same ages. The risk of events is seen to be lower than the prevalence of calcification at all ages and is particularly lower in younger patients. Second, because a major objective of both medical and invasive therapy is to relieve symptoms, which asymptomatic subjects lack, there is less incentive to screen such subjects.

Figure 1. Coronary artery calcium prevalence, 10-year event risk, and prevalence/risk ratio in asymptomatic men (see text for risk profile).¹⁵⁵ Event risk and calcium prevalence are plotted against right axis, and prevalence/risk ratio is plotted against left axis. Prevalence/risk curve decreases with age, suggesting that although serious over-prediction will occur in the young, over-prediction will be only moderate in the elderly.

However, if a threshold for fluoroscopic or EBCT scan results could be determined that effectively stratifies asymptomatic subjects into very high- versus very low-risk groups, aggressive risk factor modification could be selectively applied to the high-risk group. This presumes that such thresholds exist and that they can be determined; that is, that one knows how much calcium separates very low from very high risk. However, if risk increases with the amount of calcium in a continuous graded manner without a sudden, discrete step at a certain score or mass level, and the slope of this risk versus mass curve is sufficiently steep, evaluation of coronary calcium mass may still be an effective method of selecting those who could benefit most from aggressive risk factor management. Such an approach to the use of the exercise electrocardiogram has been proposed by some to screen asymptomatic individuals¹⁵⁶ but, by and large, has not been accepted by the medical community as cost-effective.^157 158

Clinical follow-up studies of asymptomatic subjects undergoing coronary calcium screening are under way, and some results are available.^{134 135 137 159 160 161}Three of these studies^{137 159 161} involved self-referred subjects who underwent scanning as part of commercial, media-advertised heart evaluation programs. Two are studies of the South Bay Heart Watch cohort consisting of 1461 asymptomatic adult Los Angeles residents with very high coronary risk who underwent fluoroscopy between 1990 and 1992¹³⁵ and then EBCT between 1993 and 1994.¹⁶⁰ One report concerns a cohort of 153 subjects undergoing upper GI fluoroscopy, during which assessment of coronary calcification was performed.¹³⁴Table 4 summarizes the results of these six studies. From these limited data obtained from asymptomatic patients, it appears that the presence of coronary calcium is predictive of future revascularization, based on the results of the one published report by Arad et al.¹³⁷ However, the use of revascularization and angina as end points has been criticized⁶¹ because the reporting of chest pain and subsequent evaluation and invasive intervention can occur partly because of psychological concerns or economic motivation generated when an anxious patient reports a high coronary calcium score to his or her physician. The high ratio (3.2:1) of the number of these revascularizations to the number of coronary deaths and infarctions from the commercially recruited cohorts of Balogh¹⁵⁹ and Arad¹³⁷ suggest that this referral bias may have been present.

Table 4. Coronary Heart Disease Events in Asymptomatic Subjects Followed After Coronary Calcium Studies*

{not reproduced for lack of relevance}

Second, the presence of coronary calcium in asymptomatic subjects is probably predictive of future nonfatal myocardial infarctions, because 86% (32 of 37 total) of those who suffered nonfatal infarctions had coronary calcium in four of the studies. And third, in the two studies reporting coronary heart disease death, only 12 of 16 subjects suffering this event had radiographically detectable coronary calcium. One of these studies involved 15 coronary deaths¹³⁵ and the other only one coronary death.¹³⁷ Eleven of the 15 deaths in the first cohort occurred in subjects with coronary calcium (P=.07). Although these results are suggestive, the paucity of data in asymptomatic patients supporting an association between coronary calcification and hard events, particularly coronary heart disease death, is evident, and more data are needed before a conclusion regarding the predictive value for coronary death can be made.

In summary, there are insufficient data to determine whether the relation between coronary calcium and coronary heart disease risk warrants the use of calcium screening in low-risk, asymptomatic subjects. The widespread proliferation of screening programs for coronary calcium as a single, isolated diagnostic modality in such persons should be discouraged. The manifest relation between calcification and atherosclerosis suggests that EBCT may have a role in establishing susceptibility (as opposed to merely quantitating risk) for coronary disease. The role of EBCT as a screening tool in asymptomatic patients with conventional risk factors is not yet clearly defined. It can be anticipated, however, that identifying the presence of premorbid coronary artery disease would influence the aggressiveness with which risk factor modification is approached.

Following Progression of Coronary Atherosclerosis
The mechanism of coronary atherosclerosis progression involves changes in plaque volume and composition.⁶ Because the effects of these changes on coronary lumen morphology is variable, the use of coronary angiography to follow disease progression has been criticized.¹⁶²

Intracoronary ultrasound¹⁶³ shares with coronary angiography the problems of invasiveness and high cost but provides a much truer assessment of atherosclerotic volume and composition. However, because of its high cost, invasiveness, and difficulty in accessing the entire coronary tree, intracoronary ultrasound cannot be used for routine serial assessments in clinical situations.

Because EBCT coronary calcium scanning results in quantitative evaluations of coronary calcium, the concept of using this technique for serial assessments of atherosclerosis is attractive.¹⁶⁴ To be applicable, however, coronary calcium measurements should accurately track atherosclerotic volume and have sufficient interstudy reliability so that differences over time cannot be attributed to measurement error. Unfortunately, neither of these conditions has been sufficiently satisfied. There have been no studies in animals or humans showing that coronary calcium mass correlates over time with the volume of atherosclerosis in individual subjects. The reproducibility studies done to date^{8691 165 166} show that changes in calcium score of as much as 50% may be necessary to be certain that a real change has taken place. This wide margin is largely caused by misregistration due to respiratory and cardiac motion during the 2- to 3-second intervals between the acquisition of image slices. (Recent modifications in EBCT scanners have reduced misregistration artifacts by shortening the intervals between scans to less than 1 second.)

Because neither the theoretical tenet on which this would be based nor the necessary reliability of the EBCT image acquisition process has been proved, there are insufficient data at this time to recommend the use of coronary calcium studies in assessing progression of coronary atherosclerosis.

Other Potential Indications
Other potential indications for coronary calcium assessments that have received little discussion and even less scientific validation include the screening of heart transplant donors and evaluation of children with familial hyperlipidemias¹⁶⁶ and patients with Kawasaki disease.¹⁶⁷ Recommendations regarding these indications cannot be made at this time due to the lack of supporting data.

Conclusions

Atherosclerotic calcification is an organized, regulated process similar to bone formation that occurs only when other aspects of atherosclerosis are also present. Nonhepatic Gla-containing proteins like osteocalcin, which are actively involved in the transport of calcium out of vessel walls, are suspected to have key roles in the pathogenesis of coronary calcification. Osteopontin and its mRNA, known to be involved in bone mineralization, have been identified in calcified atherosclerotic lesions. Calcified human atherosclerotic plaque also contains mRNA for bone morphogenetic protein-2a, a potent factor for osteoblastic differentiation, and cells that are capable of osteoblastic differentiation. These cells may be those from which vascular calcifying cells are derived. These and other recent findings indicate that calcification is an active process and not simply a passive precipitation of calcium phosphate crystals, as once thought.

Although calcification is found more frequently in advanced lesions, it may also occur in small amounts in earlier lesions, which appear in the second and third decades of life. Histopathological investigation has shown that plaques with microscopic evidence of mineralization are larger and associated with larger coronary arteries than plaques or arteries without calcification. The relation of arterial calcification to the probability of plaque rupture is unknown. Although the amount of coronary calcium correlates with the amount of atherosclerosis in different individuals and to a lesser extent in segments of the coronary tree in the same individuals, it is not known if the quantity of calcification tracks the quantity of atherosclerosis over time in the same individuals. Further research is needed to better elucidate the relation of calcification to the pathogenesis of both atherosclerosis and plaque rupture.

Epidemiological evidence and postmortem studies show that the prevalence of coronary calcium deposits in a given decade of life is 10 to 100 times higher than the expected 10-year incidence of coronary heart disease events for individuals of the same age. This disparity is less evident in the elderly and symptomatic than in the young and asymptomatic. Realization of this fact has generated the need for determination of clinically useful threshold levels of coronary calcium content (such as the calcium score determined by EBCT) to make appropriate management decisions. The limited available evidence linking radiographically detectable coronary calcium to future coronary heart disease events of death and infarction suggests that this link is strongest in symptomatic and very high-risk subjects. The results of ongoing epidemiological studies will be needed to further elucidate this connection.

Fluoroscopy, electron beam, and helical computed tomography can identify calcific deposits. Electron beam CT and, to a lesser extent, double-helical CT have the enhanced capability to localize coronary calcification and detect smaller and less dense calcific deposits. Only EBCT can quantitate the amount or volume of calcium. The absence of calcific deposits on an EBCT scan implies the absence of significant angiographic coronary narrowing; however, it does not imply the absence of atherosclerosis, including unstable plaque. Similarly, calcification may frequently be seen in the absence of significant angiographic narrowing and before there has been sufficient plaque build-up to narrow the vessel to the extent that ischemia would be apparent on stress electrocardiograms or stress thallium determinations.

Table 5 summarizes the conclusions that can be drawn from a negative EBCT coronary calcium study, based on the available evidence, when no calcium is detected. Similarly, Table 6 provides guidelines for interpreting the results of a positive scan, with some calcium detected in at least one vessel.

Table 5. Absence of Detectable Coronary Artery Calcification Using Electron Beam Computed Tomography (Negative Test)

Table 6. Presence of Detectable Coronary Artery Calcification Using Electron Beam Computed Tomography (Positive Test)

Unless the calcific area is greater than 2 mm, reproducibility of coronary calcium detection with EBCT appears to be insufficient for serial assessment of coronary calcium levels in individual patients. However, EBCT has been shown to be sufficiently accurate for predicting the presence of angiographic stenoses somewhere in the coronary arteries and for predicting the likelihood of clinical end points in symptomatic patients. This evaluation should be done under the supervision of a physician knowledgeable about the significance of scan results and in management of coronary heart disease. There is no role at present for use of the test to screen populations of young (less than 40 years old), healthy individuals with no risk factors. The importance of calcification in such individuals will have to await event data that are currently being obtained.

Acknowledgments

The authors thank Richard Cohen, MD, and James Willerson, MD, for their helpful comments during the preparation of this manuscript. We also thank Jeanette Allison for her superb secretarial assistance.