Histopathologic studies have demonstrated the correlation of coronary artery calcification with the presence of atherosclerosis (1- 2), and there appears to be a modest relationship between the extent of coronary artery calcification and severity of luminal stenosis (3- 5). Angiography (6- 7) and intravascular ultrasound (8- 9) have demonstrated this relationship, and in spite of the lack of a site-by-site correspondence of calcification and luminal stenosis (10), a high calcium score (CS) predicts the presence of obstructive luminal disease, as demonstrated by single photon emission computed tomography (SPECT) imaging (11), with good reliability. Nonetheless, neither the severity nor the site of a vessel obstruction detected at angiography can be used to predict the occurrence of myocardial infarction (MI) or cardiac death (hard coronary events [HCEs]) (12- 13). On the contrary, plaque burden appears to be a better determinant of future coronary events (14- 16). Nuclear myocardial perfusion imaging and stress echocardiography are invaluable in detecting the presence of obstructive coronary stenosis. However, though very helpful to assess abnormal myocardial perfusion and its attendant prognosis, these imaging modalities cannot be used to estimate the total plaque burden. On the contrary, electron beam tomography (EBT) is an anatomic test employed to visualize the presence of plaque noninvasively.

Much controversy still surrounds the role of coronary calcification for the prediction of coronary events. Several published reports support the view that selected asymptomatic and symptomatic individuals with coronary calcium on a screening EBT scan are at high risk for development of clinical events (17- 21). Prior American College of Cardiology/American Heart Association guidelines affirm the strong negative predictive value of a normal EBT, although they are less supportive of the positive predictive value of coronary calcification (22).

In this observational prospective study, we sought to determine the incidence of HCEs, defined as MI or death, in a cohort of asymptomatic individuals with high CSs (≥1,000) who did not undergo initial stress testing or coronary angiography driven by the results of the EBT screening procedure. To gauge the prognostic significance of these high CSs, indicative of an extensive plaque burden, we compared the event rate recorded in our cohort with that reported in the literature for symptomatic patients with severe abnormalities on myocardial functional studies.

Data gathering started after study approval by the Medical Investigation Committee in our institution. All records for asymptomatic subjects submitted to EBT scanning during the period of June 1, 1996, to March 30, 1999, were reviewed. To be entered in the database, individuals had to be asymptomatic for cardiovascular disease and have no prior history of coronary artery disease (CAD). Furthermore, all patients had to have an initial CS ≥1,000 (Agatston method) (23). Individuals whose EBT results triggered further cardiovascular testing, including functional testing or angiography, were excluded from the final analysis.

A total of 104 asymptomatic subjects with a CS ≥1,000 were identified. We could not obtain follow-up information on two subjects, and four patients underwent interventions triggered by their EBT results and were excluded from the analysis. Therefore, our final sample included 98 patients.

Information regarding the patients’ demographics, past medical history and clinical data were obtained through questionnaires distributed at the time of the initial EBT screening and—where available—medical record review. The accuracy of the medical records was confirmed at the time of patients’ first interview. The information we collected regarded only the presence of categorical risk factors, and most risk factors were self-reported by the individuals undergoing EBT testing. No continuous variables were obtained. Systemic arterial hypertension was defined as current use of anti-hypertensive medications or known but untreated hypertension. Current smoking or cessation of smoking within three months of testing was defined as positive smoking status. Hypercholesterolemia was defined as currently receiving cholesterol- lowering medications or the presence of known but untreated hypercholesterolemia. Patients receiving insulin or oral hypoglycemic agents were classified as diabetic.

Telephone surveys were conducted at 3, 6, 9, 12, 18, 24 and 36 months from enrollment to collect information regarding the occurrence of HCEs (MI and death). All events were verified by means of medical chart and death certificate review. If a patient suffered more than one HCE during the follow-up period (two consecutive MIs or MI followed by death), only the first event was counted. The time from EBT screening to first event was recorded in months, and patient-months were used as an estimate of event-free survival (patients-months were calculated as the sum of all event-free months/number of events). Enrollment ceased six months before the last telephone contact.

Electron beam tomography imaging was performed using an Imatron C-100 scanner (Imatron, South San Francisco, California), with a 3-mm slice thickness and 100-ms imaging time. Tomographic imaging was triggered at 80% of the R-to-R interval on the 12-lead electrocardiogram during two end-inspiratory breath-holding periods. This was done to ensure minimal motion artifact and image distortion.

Scoring of all calcified areas was performed using the standard methodology described by Agatston et al. (23). A single investigator (A.Z.) blinded to the clinical data reviewed all EBT studies and confirmed the CSs. We did not consider it necessary to measure intra-reader scoring variability, because the latter is negligible with CSs in the range we selected (24).

Categorical variables were compared using the Fischer exact test, and continuous variables were compared using the Student t test. Demographic data were compared using the likelihood ratio derived from chi-square. Curves describing the probability of event-free survival were obtained by plotting the number of patients free of events at each follow-up point. A patient who suffered an event at any point continued to be counted in the event group at the next follow-up. All calculations of significance were two-tailed, and a p < 0.05 was deemed necessary to achieve significance.

A MEDLINE search was conducted to identify recent publications that reported on hard outcome of patients found to have severe abnormalities on myocardial perfusion imaging (MPI) or stress echocardiography. Because such abnormalities probably point to the presence of multifocal obstructive disease, we considered these historical controls as plausible matches for our patients with extensive coronary artery calcification.

The clinical characteristics of the subjects submitted to EBT screening are shown in (Table le1). The majority of our patients were self-referred, and only a small number were referred by primary care physicians. Of the original 104 patients screened, two patients could not be reached for follow-up and were excluded. Four patients underwent cardiac catheterization after the EBT study. Two of these patients underwent percutaneous coronary angioplasty with stent placement, and one underwent coronary bypass surgery, while the fourth patient was treated medically. These four patients were also excluded from analysis. The remaining 98 patients were followed for an average of 17 ± 11 months (range: 4 to 36 months). The average CS for the entire cohort was 1,328 ± 287 (range: 1,000 to 2,010).

In total, we recorded 35 HCEs during the follow-up period. The mean age of patients with and without events during follow-up was not statistically different (61 ± 9 years and 63 ± 8 years, respectively, p = NS). However, the CSs of patients who sustained an HCE during the follow-up period was significantly greater than that of patients who did not (1,561 ± 270 vs. 1,199 ± 200, p < 0.001). Treatment with HmG-CoA reductase inhibitors was employed at any time before or after the screening EBT scan in 27% of the patients without an HCE and 26% of the patients with an HCE (p = NS). The same was true of therapy with beta-adrenergic blocking agents (24% vs. 20%, p = NS).

Cardiac mortality comprised 12 events: seven deaths were not preceded by a documented MI and happened an average of 19 ± 7 months from the screening EBT. The remaining five deaths were preceded by an MI suffered after an average time of 15 ± 6 months from the screening EBT. Death in these five cases occurred within zero to nine months from MI. Of the 12 patients who died, two patients had developed angina after a follow-up of six and eight months from EBT screening and underwent percutaneous coronary artery angioplasty at that time. Both of these patients died within five months of the procedure.

Morbid events included 23 nonfatal MIs recorded at an average of 18 ± 7 months from the screening EBT. There was no statistical difference in age and CS between the patients who suffered a fatal cardiac event and those who suffered a nonfatal MI (mean age: 61 ± 10 years for both groups and CS = 1,555 ± 214 vs. 1,563 ± 270, respectively; p = NS).

The annual absolute event rate for the entire cohort was 25%. In the study group as a whole, there was one event per 30.7 patient-months. For patients with a CS ≥1,500, the event rate was 1 per 24.5 patient-months. The difference in time-course between these two patient groups was statistically significant (p < 0.05). If only the patients who died were considered, the time to event was further reduced to 18.4 patient-months, and this too was statistically significant (p < 0.02) compared with the entire cohort.

Although the longest follow-up was 36 months, all HCEs occurred within 28 months from the time of EBT screening. Eleven patients were followed for longer than 28 months and up to 36 months but suffered no events. A curve describing the probability of event-free survival at each individual follow-up time is presented in (Figure 1).

Our cohort showed a greater mortality and morbidity than historical controls submitted to functional stress imaging. Hachamovitch et al. (25) published data on the outcome of 2,200 consecutive patients with no prior history of CAD submitted to dual isotope SPECT imaging. All patients had symptoms, either atypical (70%) or typical (30%). During a mean follow-up of 566 ± 142 days, the HCE rate for patients with severely abnormal scan results was 10% (approximately 7% per year). The same group reported on the outcome of 5,183 patients followed for an average of 1.8 years after having been submitted to MPI (26). Of 5,183 patients, 64% had experienced either typical or atypical angina before testing, and 21% had suffered an MI. The yearly HCE rate was 7.1% in the patients with severely abnormal scan results and 12.3% for all patients with moderately and severely abnormal test results considered together. An extensive review of outcome studies performed in a variety of patients submitted to MPI showed an average annual HCE rate of 7.4% in individuals with abnormal scans and a rate of 0.6% in subjects with normal scans (27).

Elhendy et al. (28) studied a group of 563 high-risk diabetic patients by means of exercise echocardiography. Again, over two thirds of the patients studied had anginal symptoms or had suffered a prior MI. Patients with inducible wall motion abnormalities in multiple territories showed an HCE rate of 32.8% at five years (approximately 7% annualized event rate).

In every comparison, our asymptomatic subjects with CSs ≥1,000 demonstrated a statistically greater event rate (p < 0.0001) than patients with severely abnormal stress test results.

This study shows that the presence of a high coronary artery CS on EBT imaging in asymptomatic patients portends a very high risk for HCE. Previous EBT outcome studies included very few cases with a CS ≥1,000, and this is the first large series of asymptomatic patients with such score values. Although medical treatment of these individuals may have been intensified by the detection of extensive coronary calcification, the screening test was not immediately followed by any coronary intervention or further risk stratification by means of functional studies. Stress testing or coronary interventions were performed during the follow-up procedure only as warranted by clinical circumstances such as the appearance of angina. Therefore, this analysis represents a true natural history of extensive asymptomatic CAD.

Margolis et al. (29) reported that symptomatic patients with coronary calcification seen during angiography had a five-year survival rate of 58% compared with 87% for patients without calcification. They further noted that the prognostic significance of coronary artery calcification appeared to be independent of information obtained during angiography. The presence of calcification of the coronary arteries on fluoroscopy was an independent predictor of death and MI in a study by Detrano et al. (30). These authors reported on the outcome of 1,461 asymptomatic individuals followed for 55 months after having undergone fluoroscopy screening to identify coronary artery calcification. During the follow-up period, there were 35 coronary deaths and 43 nonfatal MIs. Subjects with calcification in more than one vessel were 2.2 times more likely to suffer an event than subjects with one or no calcified vessels. Furthermore, the number of calcified vessels was an independent predictor of events. Given the much lower sensitivity of fluoroscopy compared with EBT, it is plausible that the calcifications seen on fluoroscopy by Margolis et al. (29) and Detrano et al. (30) were rather extensive.

The value of EBT screening for coronary calcium and as a predictor of cardiovascular events is still being disputed (22) and is undergoing intense investigation. In 1,172 self-referred asymptomatic individuals, the presence of a CS >160 was associated with an odds ratio of 22 for either developing an HCE or requiring a coronary revascularization procedure after an average of 3.6 years from EBT screening (19). A recent trial showed that a CS >75th percentile in asymptomatic physician-referred individuals is highly predictive of the occurrence of HCEs in the short term (17). Because a CS ≥1,000 places an individual in at least the 75th percentile, and perhaps even above the 90th percentile for most adults of both genders (31), our analysis falls in line with other studies that proved these relative score thresholds to be indicative of substantially elevated cardiovascular risk (17).

Several explanations could be offered for the occurrence of the extremely elevated event rate we observed. The most plausible is that extensive calcification of the coronary arteries indicates disseminated atherosclerosis. The calcified areas should not necessarily be considered the areas at immediate risk of sudden rupture or erosion but rather markers for the presence of unstable atherosclerotic plaques admixed between the more heavily calcified areas (1). A recent report by Huang et al. (32) supports the hypothesis that calcified plaques are less prone to rupture. These investigators conducted an in vitro study of the elastic properties of 10 stable plaques and 10 plaques that had undergone rupture before the patient’s death. The authors concluded that calcified atherosclerotic plaques were more stable and less prone to rupture than plaques with a large lipid pool. These results, however, are in contrast with other experimental data produced with in vitro simulations by Veress et al. (33), who suggested that atherosclerotic plaques show an increased tendency to rupture when a calcium layer is added to a stable plaque. Furthermore, Demer (34) demonstrated an extreme fragility of calcified rabbit coronary arteries exposed in vitro to stretching with angioplasty balloons, while Fitzgerald et al. (35) observed a greater propensity to dissection during in vivo angioplasty of calcified human coronary arteries than noncalcified arteries (35). Finally, Mascola et al. (36) reported that the culprit coronary artery responsible for an acute event is most often calcified. In the latter analysis, the investigators concluded that the weight of the evidence favors the concept that coronary calcification is an important risk factor for the development of HCEs. Nonetheless, although the culprit artery is more often calcified, the specific region of plaque rupture or erosion might not be.

Although the issue of the relationship between arterial fragility and focal calcification remains controversial, it is clear that the total atherosclerotic plaque burden is a more determinative factor in the occurrence of HCEs than the presence of focal luminal stenoses (14,37). Goldstein et al. (14) showed that the presence of multiple complex plaques predicted the recurrence of acute coronary events that did not necessarily occur at the site of the most severe obstruction. This happened in spite of a more frequent use of coronary revascularization procedures in patients with multiple complex plaques.

The presence of multiple perfusion abnormalities on a myocardial perfusion study in patients with anginal symptoms indicates the development of obstructive luminal disease, which may well be caused by stable plaques not prone to rupture. Thus, functional abnormalities, although very prognostically important (38), may not necessarily carry as severe a prognosis as extensive atherosclerotic disease. On the other hand, the actual risk of an HCE in patients with severely abnormal functional tests could be greater than reported in the literature. In fact, it is likely that the majority of individuals with very abnormal test results would undergo coronary angiography and revascularization early after test completion. This might limit the occurrence of events, although large-scale randomized trials have shown that improved event-free survival can be achieved only in specific subsets of symptomatic patients submitted to elective revascularization procedures (39- 41).

Our study was subject to a few limitations. The size of our patient cohort was small, although this is the largest published series of asymptomatic individuals with a CS ≥1,000 to date. Today it would be extremely rare to find individuals with a CS ≥1,000 not submitted to further testing, because the interest and attention of clinicians to the prognostic significance of coronary calcification has exponentially increased since the time of our study inception. The majority of patients we followed were self-referred. This might represent a more health conscious group of individuals or a group of subjects that, despite the declared absence of chest discomfort, actually suffered atypical symptoms and wondered about their nature. This preselection bias could be overcome only by conducting a population-based study where no intervention after the EBT screening is allowed. This approach is becoming less practical and ethical as more outcome studies supporting the negative impact of coronary calcification are published. The design of our study did not include the measurement of markers of inflammation such as high sensitivity C-reactive protein, an important marker of risk of future coronary events (42- 44). It is conceivable that the individuals who suffered an event in our study might have had the highest C-reactive protein levels, therefore representing a more unstable group. Interestingly, all events happened in the first 28 months of follow-up, with no event registered after that period. In theory, it is plausible that the unstable patients suffered an event earlier, while in the remaining patients calcification of the coronary arteries indicated the presence of a more stable form of disease.

The prognostic significance of coronary calcification is undergoing intense investigation, and it is slowly becoming accepted as an additional significant risk factor for coronary heart disease. Our study shows that the presence of high CSs (≥1,000) on a screening EBT in asymptomatic persons portends a very high risk of HCEs. The risk seems to be greater than severe abnormalities seen on functional myocardial studies conducted in symptomatic patients. These data may need to be taken into consideration when designing randomized trials of asymptomatic individuals expected to undergo EBT screening. Patients with such high scores deserve aggressive medical management and further risk stratification with serologic testing for markers of inflammation, and they are likely to require functional stress testing to demonstrate the presence of inducible myocardial ischemia.