This Article Free upon publication Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Clouse, M. E. Articles by Krumholz, H. M. Search for Related Content PubMed Articles by Clouse, M. E. Articles by Krumholz, H. M. Related Collections Health policy and outcome research Primary prevention CT and MRI Chronic ischemic heart disease

(Circulation. 2006;113:125-146.)
© 2006 American Heart Association, Inc.

Controversies in Cardiovascular Medicine

How useful is computed tomography for screening for coronary artery disease?

Noninvasive Screening for Coronary Artery Disease With Computed Tomography Is Useful

Melvin E. Clouse, MD

From the Department of Radiology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Mass (M.E.C.); Section of Cardiology, Department of Medicine, Beth Israel-Deaconess Medical Center, Boston, Mass (J.C.); and the Sections of Cardiovascular Medicine, Department of Medicine; Health Policy and Administration, Department of Epidemiology and Public Health; and Robert Wood Johnson Clinical Scholars Program, Department of Medicine, Yale University School of Medicine, and Center for Outcomes Research and Evaluation, Yale-New Haven Health, New Haven, Conn (H.M.K.).

Correspondence to Melvin E. Clouse, MD, Vice Chairman, Director of Research, Beth Israel Deaconess Medical Center, Deaconess Professor of Radiology, Harvard Medical School, 1 Deaconess Rd, Room 302, Boston, MA 02215 (e-mail mclouse{at}bidmc.harvard.edu); or Dr Harlan M. Krumholz, Yale University School of Medicine, 333 Cedar St, PO Box 208088, New Haven, CT 06520-8088 (e-mail harlan.krumholz@yale.edu).

	Introduction

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
The introduction of new ideas and concepts that lead to change in practice has always caused some degree of controversy, especially in medicine. At first glance, the concept of noninvasive imaging for calcium as a screen to identify patients at high risk for future coronary events would seem the most intense; however, one must only reflect on past controversies to gain an appropriate perspective. The controversy over radical mastectomy versus segmental resection or lumpectomy with radiation therapy has raged for the past 50 years, and only recently have data from the 20-year follow-up of a randomized trial comparing these forms of treatment been put forward.^1,2 The process of establishing the chest roentgenogram as a standard diagnostic method in the diagnosis of respiratory disease spanned 30 years and was opposed by many of the leading physicians of the day, including Osler,³ who believed a good clinical examination was superior. In 1915, Crane⁴ stated that the chest roentgenogram that "claims a delicacy, rapidity and precision outranking the stethoscope and the percussion finger must expect to run a gauntlet of merciless criticism." The chest roentgenogram largely came into general use in the 1930s, when it was recognized that &15% of the deaths in the United States were due to tuberculosis, and a massive screening process was instituted after World War II.⁵ Establishment of the chest roentgenogram as a diagnostic tool was based largely on the belief in technology and innovation; to date, however, no prospective randomized studies have been conducted to determine whether the chest roentgenogram has indeed affected the outcome of patients with cardiopulmonary diseases. Thankfully, the coronary artery calcium (CAC) examination has been placed under intense scrutiny, and although the construct and ethics of a prospective randomized study have yet to be decided, it is appropriate to review and discuss how it may help in treating patients with subclinical atherosclerosis and to determine its absolute predictive value and its relationship to the Framingham Risks score and National Cholesterol Education Panel Adult Treatment Panel III (NCEP ATP III) guidelines because it is the only noninvasive test available to evaluate insults to the arterial wall from all risk factors causing atherosclerosis.

The concept of imaging coronary arteries for calcification in vivo arose shortly after the discovery of x-rays by scientists who demonstrated calcification within the coronary arteries but were limited by current technology.^6–8 After publications by Habbe and Wright⁹ and Van der Straeten,¹⁰ Blankenhorn and Stern, in a landmark article, scientifically established the fact that calcification in the coronary arteries is directly related to atherosclerosis.^9–13 Recent studies have confirmed that the development of arterial calcification is intimately associated with vascular remodeling and atherosclerotic plaque and is controlled largely by cellular and subcellular mechanisms.^14–17 Histopathological studies have also shown that calcification is found more frequently in advanced atherosclerotic plaque and is associated with plaque in larger arteries than in peripheral coronary arteries.^18–20 In 1903, Monckeberg described calcification that occurs in the media, usually in the peripheral and visceral arteries and only occasionally in the coronary arteries, and is not associated with atherosclerosis.²¹

Historically, cardiac fluoroscopy was frequently used to detect calcium in the coronary arteries because it was much more sensitive for detecting calcium than the standard chest roentgenogram. In 1987, Detrano and Froelicher²² summarized studies involving 2670 patients undergoing coronary arteriography and equated the findings of calcification for detecting significant stenosis (>50% diameter) with a sensitivity of 40% to 79% and specificity of 52% to 95%. That same year, Reinmueller and Lipton²³ studied a small group of patients and demonstrated that CT was much more sensitive for detecting calcification than fluoroscopy (62% versus 35%). However, image quality was degraded by cardiac motion.

The new era for imaging CAC began with the introduction of the high-speed cine-CT scanner. The cine-CT/ultrafast CT scanner, later designated the electron-beam CT (EBCT), performed 3-mm-thick cross-sectional slices in 50 to 100 ms with exposure gated to 80% of the R-R interval. Thus, the heart could be examined in a single breathhold with the x-ray beam passing from the source through the body to a detector array; the recorded data were transformed through a filtered back-projection reconstruction technique into 2D images.

Agatston et al,²⁴ guided by a method originally conceived by David King (Imatron), used the EBCT to quantify CAC, working on the premise of using the calcium score as an independent predictor for future myocardial events, as indicated by Margolis et al²⁵ in 1980. They established the scientific basis for the scoring system based on an x-ray attenuation coefficient or CT numbers measured in Hounsfield units by selecting the maximum calcium density within the area. The area of calcium was calculated from the field of view and the image matrix that, on the standardized protocol, relate to 3 pixels or 1 mm² with a density of 130 Hounsfield units. Statistical analysis was performed on the log-transformed total score and on the square root of the number of lesions to normalize the data. After completing a scan with the same parameters using a high-resolution volume mode with 3-mm-thick slices, they repeated the same scan in a single-slice mode with 20 and then 40 contiguous slices throughout the heart with no interslice gaps. Callister et al²⁶ improved the reproducibility of the calcium score, especially in the lower ranges, by introducing the volume score (isotropic interpolation) method.

The ability to identify individuals at high risk and thus to direct appropriate therapies to prevent further myocardial events would be a great benefit to society because cardiovascular disease is the most important health problem in America and the Western world, accounting for 38.5% of all deaths. The death rate from cardiovascular disease is greater than the second through the seventh leading causes of adult death, including cancer, AIDS, accidents, homicides, infections, and diabetes mellitus. The total cost for treating far-advanced, ie, end-stage, cardiovascular disease is enormous. The estimation has increased from $286.5 billion in 1999 to $368.4 billion in 2004, accounting for a third of the cost of our $1 trillion healthcare economy. The cost for physician care and testing is only 10% ($31 billion), with the remainder being for patient care.^27,28 The current methods of diagnosis and treatment have had little effect on the outcome of a disease that is largely preventable with institution of strict risk factor modification and statin therapy if discovered early.^29–34 Up to 50% of patients with atherosclerotic disease present with either ischemic heart disease or sudden death, and for 150 thousand individuals, a fatal heart attack is the first symptom of heart disease.^27,28 Fifty percent of these myocardial infarctions (MIs) occur in patients with no prior history of disease, and 68% of these are due to lesions representing a stenosis diameter <50%. Cholesterol is perceived as one of the most important risk factors for coronary artery disease, but &35% with established heart disease have total cholesterol levels <250 mg/dL; thus, cholesterol has failed to predict up to one third of future deaths resulting from coronary artery disease. In a recent study, 204 men <55 years of age and women <65 years of age presenting with acute MI had cholesterol tests performed within 12 hours of admission. Sixty-eight percent had LDL cholesterol levels <131 mg/dL, 41% had LDL cholesterol levels <100 mg/dL, and 38% had LDL cholesterol levels >130 mg/dL. Only 25% of these patients, all of whom subsequently suffered MI, would have qualified for lipid-lowering therapy under the current NCEP ATP III guidelines.³⁵

	Comparison of EBCT CAC Score With Other Noninvasive Tests

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
Patients with typical angina/symptoms of coronary heart disease normally undergo routine noninvasive tests such as exercise ECG, echocardiography, myocardial scintigraphy, or pharmacological stress tests. These tests are used when patients are symptomatic with far-advanced disease, are based on indirect signs of atherosclerosis that result from inadequate myocardial perfusion, and have a high pretest probability of being positive.

The consensus statement reports on a large meta-analysis with high sensitivities, specificities, and accuracy for the exercise treadmill test (ETT) in the range of 68%, 77%, and 73%, respectively, for ECG; 89%, 80%, and 89% for myocardial perfusion; and 85%, 84%, and 87% for pharmacological scintigraphy/echocardiography compared with 91%, 49%, and 70% for EBCT.³⁶ Others give lower and more variable sensitivity and specificities of 85% to 77% for echocardiography, 87% to 63% for myocardial scintigraphy, and 84% to 44% for pharmacological treadmill testing, depending on the number of vessels involved.^37–39 Such results are influenced by gender, age, cardiac rhythm, and inability to exercise.

Haberl et al,⁴⁰ like most investigators, reported a higher sensitivity and specificity and less variability for EBCT. With cut points for calcium scores of >20th, >100th, and >75th percentile of age groups, the sensitivity for detecting stenoses decreased to 97%, 93%, and 81%, respectively, for men and 98%, 82%, and 76% for women. Specificity increased up to 77% for both. Sensitivity and specificity are related to the cut points for the calcium score for which there currently is no agreement. The negative predictive value for a zero calcium score was 99%.⁴⁰ Kajinami et al⁴¹ also reported an overall accuracy of 85% for EBCT compared with 71% for myocardial scintigraphy.

Regardless of the variability of the reported data, the ETT/myocardial perfusion tests provide a high accuracy for predicting future myocardial events.⁴² Therefore, they are an essential part of the diagnostic armamentarium. They are performed to detect the possibility of flow-limiting lesions (far-advanced disease) but when negative give no information as to the presence of significant plaque burden and do not identify patients with subclinical atherosclerosis who may be at risk for future myocardial events, thereby alerting the patient/and physician to vigorously pursue preventive measures. Therefore, the calcium examination should be used in low-yield situations such as atypical chest pain to screen and possibly reduce the number of patients subjected to invasive procedures when the above noninvasive tests are not conclusive. Intravascular ultrasound is a more accurate method for plaque evaluation, but its usefulness in routine clinical decision making is limited because of its invasive nature.

	CAC Scoring

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
To be used effectively, EBCT CAC must be validated. How accurate is it for identifying calcium? How reproducible is the score? What variation is there between 2 scans taken several minutes apart in the same patient? The reproducibility and variability of the EBCT calcium score have been studied extensively. Earlier reports have shown significant variability, between 14% and 38%; however, these imaging algorithms are no longer up to date. Previously, the limitations on slice number, suboptimal gating, and table motion led to higher interscan variability. Hardware for EBCT has improved significantly, and there has been marked improvement in the reproducibility of the calcium score. A recent study of 1311 asymptomatic individuals undergoing 2 scans 3 minutes apart resulted in an average interscan variability of 15% to 17%.⁴³ Another study using a newer protocol demonstrated a mean interscan variability of 16% to 19% and a median variability of 4% to 8.9% for the Agatston and volumetric scores.⁴⁴ There was also significant improvement in the quantification of calcium score with the introduction of the volumetric method. The inherent issue of cardiac motion will continue to be a problem, especially for the right coronary and left circumflex arteries.^45,46 Several investigators have suggested triggering exposure to 40% of the R-R interval and have reported an interscan variability of 11.5%. However, others have not found this necessary.⁴⁴

Multirow detector computed tomography (MDCT) has recently been introduced for CAC scoring. Investigators have found significant interscan variability and reproducibility with single-slice scanners at rotational speeds of 800 ms. The variability has been most marked using densities of 90 rather than 130 Hounsfield units.⁴⁷ MDCT technology for CAC scoring is improving rapidly. Initial reports were from dual and 4-slice scanners with variabilities of 25.2% for overlapping images with volume scoring and 45.5% for Agatston scoring.⁴⁸ MDCT scanners can image a section of the heart simultaneously with ECG gating in either the prospective (ECG triggering) or retrospective mode for segmented reconstruction. This allows a gapless helical scan of the entire heart. Prospective gating usually produces 3-mm-thick slices with a temporal resolution of 200 or 250 ms. Temporal resolution of 100 to 125 ms can be achieved with the retrospective mode with overlapping slices but with a marked increase in radiation dose. Now, 4-MDCT and 8-MDCT scanners are being replaced with 16-MDCT scanners. Reconstruction algorithms have improved with retrospective gating. Furthermore, we can expect 32- and 64-MDCT scanners to have rotational speeds of 330 ms, which will allow temporal resolutions of 175 or 87 ms to improve resolution and to reduce cardiac motion.

A recent study of 32 patients demonstrated a variability of 20.4% for Agatston scoring and 13.9% for volumetric scoring for MDCT.⁴⁹ Another recent publication comparing MDCT with EBCT shows high correlation of scores at every calcium level and similar areas under receiver-operating characteristic (ROC) curve.⁵⁰ A more recent report of 100 patients undergoing both MDCT and EBCT shows similar sensitivity and specificity of 98.7% and 100%, respectively. The variability of the volume score was 20%; the mass score was 20.3%.⁵¹

There is significant discussion as to the most appropriate scoring method, ie, Agatston, volume, and mass scores. However, regardless of imaging technology and methods of obtaining and measuring calcium score, the Agatston method is the standard now and for the foreseeable future. This is predicated on the significant available database for these scores and outcomes data currently in use because clinicians know the significance of a certain score using the Agatston method. Volume scores are similar, although slightly lower, and mass scores are significantly lower. Almost all scoring software now gives all 3 scores simultaneously for each subject; therefore, all are readily available.

	Radiation Dose

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
Radiation dose for CT scanning is significant, and every effort is being made to reduce the dose. MDCT scanning is usually performed with prospective gating with 3-mm slice thickness. The effective radiation dose for MDCT scoring was 1 to 1.5 mSv for men and 1 to 1.8 mSv for women using 100 to 140 mA and 140 kV. The equivalent dose for EBCT is 0.7 to 1.0 mSv for men and 1.3 mSv for women. These dose rates are based on prospective triggering rather than retrospective triggering using thinner overlapping slice segments that improve spatial resolution.⁵² To reduce radiation exposure using these retrospective gating algorithms with 4-MDCT, Mahnken et al⁵³ report an effective radiation dose of 3.01 mSv (range, 2.5 to 4.18 mSv) for men and 4.44 mSv (range, 3.28 to 5.88 mSv) for women. Trabold et al⁵⁴ and Flohr et al⁵⁵ report dose rates for 16-MDCT that are similar to the previous MDCT scanners. Hirota et al,⁵⁶ however, report effective dose rates with gated studies of 2.6 and 4.1 mSv using 100 and 150 mA and 120 kV, respectively. The dose rates for CT coronary arteriography are much higher, 9.3 and 11.3 mSv using 300 mA. Most investigators did not use ß-blockers to reduce heart rate and cardiac motion. Although both EBCT and MDCT have inherent limitations using 100- and 200-ms exposures, EBCT (e-speed) does image at 50 ms, if needed, for high resolution. Reducing the heart rate from 75 to 65 bpm increases the diastolic phase from 530 to 620 ms, whereas the systolic phase is increased only from 270 to 300 ms. Scanning at a lower heart rate would significantly reduce the in-plane motion of the coronary arteries; however, a significant number of patients will have heart rates >75 bpm, which may predicate the use of retrospective gating.

	Significance of CAC

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
Recent reports have confirmed that atherosclerosis is the only disease associated with coronary calcification and that calcification is intimately associated with plaque.^{14,18,36,56,57} CAC is an active process seen in all stages of plaque development. It is strongly correlated with age and increases significantly after 50 years of age. It parallels the prevalence of atherosclerotic plaque development as demonstrated by intracoronary ultrasound, which shows significant noncalcified plaque of 17% in 20-year-old individuals, increasing to 85% in individuals >50 years of age.⁵⁸ The EBCT calcium score follows the same pattern of calcification in all age groups and progresses rapidly after 50 years of age.⁵⁹ There is a slight gender variation in women, with lower scores in the early decade, but this is eliminated in the 65 to 70 years of age group.⁶⁰

The correlation of plaque calcification within noncalcified plaque as demonstrated by EBCT was established by Simons et al⁵⁷ and Rumberger et al^61–63 with excellent histological studies on randomly selected hearts quantifying CAC and total plaque by measuring direct histological plaque area and percent luminal stenosis. These studies demonstrated that the calcium score correlated linearly with total plaque area and that calcified plaque accounted for only 20% of the total plaque burden. In addition, a calcium area 1 mm in diameter predicted mild stenosis, whereas a calcified area of 3 mm was more likely to be associated with significant luminal narrowing. These studies also noted that calcium is a reflection of total plaque burden but that the calcium score does not translate in a one-to-one fashion to direct luminal narrowing. A study by Sangiorgi et al⁶⁴ suggests that this is related to the remodeling phenomenon reported by Glagov et al.⁶⁵ Baumgart et al⁶³ confirmed the direct association of CAC score with hard and soft plaque using intracoronary ultrasound and arteriography. For plaques with and without calcification, the sensitivity was 97% and 47% and specificity was 80% and 75%, with an overall accuracy of 82% and 69% respectively, thus confirming the high sensitivity for detecting calcium and the high negative predictive value of a negative EBCT score.⁶³

In addition, EBCT has demonstrated its ability to quantify atherosclerotic plaque and, by virtue of the score, measure the severity, ie, stage of disease, in the coronary artery in direct comparison to pathological studies, regardless of age and gender.^61,62 The scores are reproducible and interscan variability is sufficient for use in research and clinical studies. The 4-, 8-, and 16-MDCT scanners have been shown to be comparable with respect to quantifying the calcium score.

The most important application of the EBCT CAC examination is the high negative predictive value of a zero CAC score. It indicates that no calcium is present. It also indicates that there is little likelihood of significant arterial stenosis (negative predictive value, &95% to 99%). A negative score is consistent with a low risk for hard coronary event (0.1% per year) or any event in the next 2 to 5 years.^36,57

Although there may be controversy over the use of the calcium score to diagnose obstructive disease, there is little controversy in its ability to detect calcified plaque. The ability of the CAC to estimate total plaque burden, ie, stage of disease, is the most significant predictor for future myocardial events.⁶⁶ Therefore, the importance of the CAC score lies in its ability to identify individuals at risk and to integrate this information with other risk factors for risk stratification and goal-directed prevention.

	CAC as a Predictor for Future Myocardial Events

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
It has long been known that CAC is related to atherosclerosis, and individuals dying of coronary artery disease have significantly more calcification than that seen in age-matched control subjects.⁶⁷ In addition, calcification is the best indicator for severity, ie, stage of the disease. It would seem intuitive that calcification represents the sum total of insults to the arterial wall from all risk factors. It therefore should be an important predictor for future myocardial events and should be compared with the standard risk factors and NCEP guidelines. It is also known that a major portion of acute ischemic cardiac events occur from rupture of vulnerable plaques that are hemodynamically insignificant in asymptomatic individuals. Thus, it is important to evaluate the significance of CAC as a predictor for future myocardial events.

Five recent studies have evaluated the significance of CAC as a predictor for future myocardial events since the initial article by Arad et al⁶⁸ in 1996. These articles have been selected for this review and include data from a total of 17 976 subjects who were self-/physician referred and 6897 prospectively enrolled for EBCT CAC studies. The mean age varied from 52 to 59 years; 51% to 79.45% were men, and 20.6% to 49% were women. The participants were asymptomatic with no prior history of coronary artery disease. In the self-/physician-referred group, most were 40 to 70 years of age, with equal numbers <40 and >70 years of age. In the St Francis Heart Study (SFHS), the mean age was 53±11 years; in the South Bay Heart Watch Study (SBHW), the mean age was >45 years, and most had at least 1 abnormal risk factor that would place them in the intermediate- to high-risk category for Framingham Risk Score (FRS)/NCEP ATP II guidelines (>10% estimated 8- to 10-year risk for developing coronary heart disease [CHD]). On evaluation, the conventional risk factors were reported to be in the range of 45% for hypertension, 10% for diabetes, 60% for hypercholesterolemia, and 40% for smoking. The mean follow-up was 32±7 to 51±9 months in the self-/physician-referred group and 4.3 and 8.5 years for the SFHS and the SBHW groups, respectively.

The first publication to assess the potential predictive value of CAC for future myocardial events was an analysis of 1173 in a 19-month follow-up that reported sensitivities of 89%, 89%, and 50% (inadequate number of subjects) and specificities of 77%, 82%, and 95% for calcium scores of 100, 160, and 680, respectively. Odds ratios (ORs) ranged from 20.0 to 35.4 (P<0.00001 for hard and soft events). The ROC curve analyses comparing the NCEP II guidelines to EBCT scores were 0.74 and 0.91, respectively, indicating the possible significance of the EBCT CAC score as a significant predictor.⁶⁸ A recent article from the same investigators reported on a 3.6-year follow-up of 1172 patients with a 99% response rate. CAC scores remained independently associated with outcomes of hard and soft cardiac events after adjustment for self-reported standard risk factors. The areas under the ROC curve were 0.84 and 0.86 for the prediction of all coronary events and nonfatal MIs and death, respectively, and CAC scores >160 and <160 were associated with an OR of 15.8 and 22.2, respectively. Hard coronary events progressed with increasing CAC scores (P<0.0001).⁶⁹ Raggi et al⁷⁰ compared a group of 172 patients who had EBCT imaging within 60 days of an unheralded MI with 632 self-/physician-referred asymptomatic patients with a 32±7-month follow-up. The groups’ demographics, including age and calcium scores, were similar. The annualized event ranged from 0.09 to 1.05 (12-fold difference) between the lowest and highest quartiles in patients identified by conventional risk factors and 0.045 to 2.7 (59-fold difference) when grouping was done according to CAC quartiles, indicating that although standard risk factors are important, CAC percentiles are substantially more important for identifying patients at risk.⁷⁰ In a previous study, these same authors, analyzing 676 patients and using 10 122 asymptomatic patients as control subjects, demonstrated that CAC score percentiles were a significant predictor for coronary events and incrementally added to the prognostic value of traditional risk factors for CAD (P<0.001). Area under the ROC curves for hard events, when added to conventional risk factors, was significantly larger than conventional risk factors alone as predictors (0.84 versus 0.71; P<0.001). The area under the curve using CAC score percentiles alone was significantly greater than conventional risk factors (0.82 versus 0.71; P=0.028). The authors conclude that age- and sex-specific CAC score percentiles provide the best predictive model and add incremental predictive information to conventional risk factors.⁷¹

Kondos et al⁷² reported on a group of 5635 asymptomatic patients (64% response). The mean age was 59±9 years, with a follow-up of 37±13 months. The prevalence of CAD risk factors was less than reported in the National Health and Nutrition Survey (NHANES) and Atherosclerosis Risk in Communities (ARIC) except for hypercholesterolemia, which was higher. Using univariate and multivariate analysis comparing those with and without events demonstrated that increasing age, smoking, diabetes, and hypertension were all significant (P<0.001). The probability value was not significant for individuals with or without hypercholesterolemia. Patients with CAC scores in the first quartile (1.0–3.8) had a relative risk (RR) of 1.76 (95% CI, 0.39 to 7.88) compared with those in the top quartile with scores >170 (RR, 7.24; 95% CI, 2.01 to 26.15) of developing a hard coronary event compared with those without CAC. In another large study of 10 377 self-/physician-referred patients, the authors demonstrated that the 5-year risk-adjusted survival was 99.0% for a CAC score <10 and 95.0% for those with CAC scores >1000 (P<0.001).⁷³ The area under the ROC curve of 0.72 for conventional risk factors increased to 0.78 when CAC scores were added to the model (P<0.001). Wong et al⁷⁴ has also reported on 926 asymptomatic individuals with mean age of 54 years who were followed up for 3.3 years. After adjustment for age, gender, and other risk factors the RR (CAC score of 81 to 270 and >270 compared with 0) for hard coronary events was 4.5 (P<0.05) and 8.8 for soft events (P<0.001).

To date, there have been 3 prospective study reports: 2 from the SBHW that focused on different analyses and 1 from the SFHS. Guerci et al⁷⁵ reports a prospective study of 5585 subjects of approximately the same age (59±5 years) that followed baseline CAC scores and FRS with a 4.3-year follow-up at SFHS. A score of >100 predicted all cardiovascular events, all coronary events, nonfatal MI, and coronary deaths with an RR of 9.5 to 10.7 at 4.3 years compared with a score of <100. The area under the ROC curve for was 0.71 and 0.81 for CAC scores.⁷⁵

The SBHW study began in 1990 as a prospective study to determine the prognostic accuracy of cardiac fluoroscopy in 1461 asymptomatic patients >45 years of age with at least 1 abnormal risk factor (>10% estimated risk for developing CHD by early Framingham risk equation) selected from a community mailing campaign. Beginning in 1992, the investigators began using EBCT. An early report by Secci et al,⁷⁶ who selected 326 of 462 original study participants, noted after a follow-up of only 2.7 years that the prediction of nonfatal MI and death based on the calcium score did not reach statistical significance (OR, 3.1; P=0.07).⁷⁶ Detrano et al⁷⁷ later reported on the same SBHW group of 1196 asymptomatic high-coronary-risk subjects with a mean age of 66 years. The ROC curves from the Framingham model, their own data-derived risk model, and the CAC score were 0.69±0.05, 0.68±0.05, and 0.64±0.05, respectively (P=NS), demonstrating that the EBCT, although no better a predictor than FRS, nevertheless was equal to the sum of all risk factors in predicting cardiac events. This report was a major factor for the final report from the ACC/AHA consensus document largely because of the incomplete representation of the data.³⁶ The ACC/AHA panel neglected to mention that the Detrano group also did not find the Framingham risk model to be a significant predictor. A more recent study by Park et al,⁷⁸ also from SBHW, selected 967 subjects from the 1461 participants and conducted a Cox regression analysis with C-reactive protein 10 mg/L to estimate the risk-factor–adjusted RRs of CAC and C-reactive protein for occurrence of hard and soft coronary events. CAC was a predictor for MI/coronary death (P<0.005) and any cardiovascular event (P<0.0001); C-reactive protein was a predictor of any cardiovascular event (P<0.003).⁷⁸ Risk group analyses showed that the risk increased with increasing CAC and C-reactive protein combined (P<0.003 for MI/coronary death and P<0.001 for any cardiovascular event). Greenland et al,⁷⁹ also from the SBHW group, reported on 1312 subjects in a long-term follow-up with a median of 7 years. Excluded from the original 1461 participants were 269 with diabetes and 14 with missing data or coronary events before the CAC was performed. Compared with a CAC score of 0, a CAC score of >300 was predictive (hazard ratio, 3.9; P<0.001). Across all FRS categories, CAC was predictive of risk among patients with an FRS >10% (P<0.001) but not <10%. The ROC curves for FRS alone were 0.63 and 0.68 for FRS plus CAC score, demonstrating the importance of incorporating the CAC with conventional risk factors. Except for the early studies of the SBHW, all prognostic studies using EBCT have demonstrated independent and incremental value compared with FRS analysis for predicting future cardiac events.

	CAC in Clinical Studies

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
The scientific basis for CAC examination has been validated, and the scanning technology has undergone intense evaluation. The reproducibility and interscan variability have improved sufficiently to be used in clinical studies for further evaluation of its usefulness. To this end, there are several specific clinical uses of importance with associated publications that should be reviewed.

First, the CAC score can be followed to document the change over time to compare the rate of progression/stabilization/regression to correlate the score with hard and soft coronary events as it relates to strict risk factor modification similar to the studies of Nissen⁸⁰ with intracoronary ultrasound. Janowitz et al⁸¹ first reported a pilot study of a small group of patients by angiography with obstructive disease who showed a 48% increase in CAC scores compared with a 22% increase in score for those without obstructive disease. Budoff and Raggi⁸² reported on 1178 patients from 9 investigations in a meta-analysis. The studies show rates of progression with variation between 18% and 44%.⁸²

Other investigators have extended this concept of tracking the calcium score after statin therapy. Callister et al⁸³ found the calcium score to increase in those individuals not treated with statin therapy and observed a significant reduction in the calcium score in those treated with statin therapy and whose final LDL cholesterol levels decreased to <120 mg/dL.⁸³ Even individuals treated less aggressively demonstrated an increase in volume score significantly lower than those who were untreated. Budoff et al,⁸⁴ in a similar observational study, showed that hypercholesterolemic patients on statin therapy had an annual rate of progression in their calcium score of 15% compared with a 39% increase in the nontreated group. This represents a 61% reduction in progression with statin therapy (P<0.001).

So does the progression of the calcium score, ie, atherosclerosis, translate into hard coronary events? Studies of patient outcomes observed over time for evidence of coronary calcium progression have been reported. In a retrospective study of 817 asymptomatic patients who were followed up for 2.2±1.3 years, the mean absolute and percentage CAC volume scores from those with MI were 147±152% and 47±50%, respectively, compared with 63±128% and 26±32% (P<0.001, P<0.01) for those without coronary events.⁸⁵ In another study, 225 moderate- to high-risk asymptomatic subjects with calcium scores >20 were followed up for 1 to 7 years. The annual event rate for patients who demonstrated coronary calcium score progression >35% per year showed a relative risk of a coronary event of >17.7 compared with those whose calcium scores progressed <20% per year. The only other independent predictor was age. Hypertension, diabetes, cholesterol, tobacco use, family history, coronary artery disease, and gender failed to predict events.⁸⁶ Thus, progression of coronary artery atherosclerosis can be observed noninvasively by monitoring the progression of the calcium score. It also gives the clinician a method to measure the effectiveness of therapy and to allow better assessment of the process associated with progressive disease.

Second, the NCEP II guidelines have been used to identify patients with subclinical atherosclerosis at high or low risk for future myocardial events. One of the most important benefits of the CAC score is identifying early asymptomatic disease with or without calcium because the negative predictive value is &99%. A recent study to determine the relationship between NCEP ATP II guidelines and EBCT for treatment of asymptomatic atherosclerosis involved 304 asymptomatic women who had EBCT evaluation and were classified as NCEP high and low risk according to LDL levels and EBCT positive or negative according to the presence or absence of calcified plaque. Forty-two percent (n=227) were EBCT positive and 58% were EBCT negative (0 score). Women who were EBCT positive had higher cholesterol and triglyceride levels than EBCT-negative women. However, NCEP-higher-risk women made up 53.5% of the EBCT-positive group and 37.7% of the EBCT-negative group; 46.5% of the NCEP-lower-risk group were EBCT positive and 62.3% were EBCT negative. Using NCEP guidelines, 46.5% (n=59) of the EBCT-positive patients would not have received therapy, and 37.7% (n =66) in the EBCT-negative group would have had unnecessary statin therapy. Thus, only 58.9% of the study population would have been appropriately identified by NCEP guidelines. There was no difference between groups when age was eliminated from the Framingham risk calculation. Further studies of this type are needed to determine whether CAC scoring can truly help in triaging those patients at risk for future events and to decide whether individuals should have calcium scoring before being placed on statin therapy.⁸⁷ This study also indicates the possible benefit of incorporating the CAC score with the FRS, as demonstrated by Greenland et al.⁷⁹

There is a possible cost-saving benefit for using EBCT in the workup of patients with new-onset chest pain in the category of low or intermediate pretest probability. Physicians are frequently faced with the task of evaluating patients with new-onset chest pain of questionable significance with low or intermediate pretest probability. The standard workup is the ETT. The ETT is known for its low sensitivity, specificity, and accuracy.^37,38,88 Patients may also undergo myocardial perfusion imaging, which adds significantly to the diagnostic and prognostic accuracy of ETT but adds significantly to the cost of the diagnostic workup because myocardial perfusion imaging costs more than ETT and CAC imaging combined. To approach this problem, Rumberger et al⁸⁹ studied an assimilated prototype ambulatory patient population model with a cut point for the EBCT calcium score of 168 to achieve a sensitivity and specificity of 71% and 90%, respectively, for >50% obstructive coronary artery disease. Their total direct testing costs increased in proportion to disease prevalence. The cost based directly on patients correctly diagnosed decreased as a function of disease prevalence. The cut point for a calcium score of 80 was also cost effective with a disease prevalence of 70%.⁸⁹ A more recent study published in 2000 used a cut point for the EBCT CAC score of 150 for a sensitivity of 79% and specificity of 89% for the prevalence of obstructive disease using a Bayesian cost model in a prospective group of low- to moderate-risk patients. The authors, using actual patient data and cost reimbursement rates for EBCT and ETT as initial testing, demonstrated a cost savings of 44% and 15% when disease prevalence was 0% and 100%, respectively. The authors agree that although the EBCT CAC does not give the prognostic information for a positive MPI test, this is balanced by the number of EBCT-positive patients below a score of 150 who could then implement strict risk factor modification to prevent progression to myocardial events.⁹⁰

	Conclusions

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 
Atherosclerosis is an indolent long-term, mostly preventable disease with significant plaque formation in the early years. The prevalence of the disease with significant plaque formation is 17%, 37%, 60%, 71%, and 85% in the second through sixth decades of life. These data are in parallel with the development of calcification in the coronary arteries from large databases and the incidence of hard (nonfatal MI, death) and soft cardiovascular events. Atherosclerotic cardiovascular disease accounts for 38.5% of all deaths in the United States. This is in contrast to 15% of all deaths from tuberculosis in 1930, when a large screening effort was instituted whereby individuals could stop at a screening van on the street for a chest x-ray.⁵ For those who die from CHD, 84.7% are >65 years of age. Eighty percent of CHD mortality in individuals <65 years of age occurs during the first attack, and 57% of men and 64% of women who die suddenly of CHD have had no previous symptoms.^27,28 In addition, the cost is $368.4 billion (more than one third of the total US healthcare costs) for treating this end-stage disease when premature death can be prevented with risk factor modification. The current method of predicting these events comes from the excellent studies that have evolved from the Framingham Heart Study’s identification of risk factors and include age, hypertension, hypercholesterolemia, smoking, diabetes, obesity, and family history. The FRS is the reference standard for comparing any test that may be helpful for increasing risk prediction that may lead to treatments to prevent progression, ie, to reduce cardiovascular events and premature death.

The current controversy questions whether CAC should be used as an additional predictor for future coronary events. The answer is yes for the following reasons. The scientific basis for the examination has been validated. The technology has reached the point where the CAC score is accurate and reproducible, the interscan variability is acceptable by either EBCT or MDCT, and the examination can be performed throughout the country. The examination is not operator dependent and has a very high negative predictive value. It is the only noninvasive method to estimate total plaque burden, which is the most important predictor for future cardiac events. The real nexus of the controversy is how important are the data derived from patient/physician referral and prospective enrollment such as the SBHW/SFHS in contrast to the methods used to derive the FRS, ie, prospective randomized studies. After analyzing the data from 5 reports involving self-/physician-referred, SBHW, and SFHS subjects, one must conclude that there is very little difference in the data. The area under the ROC curves in all publications varies from 0.6 to 0.74 for FRS (NCEP II) and from 0.84 to 0.91 for CAC score cut points in the highest tertile or quartile scores. The incidence of hard and soft coronary events increases with increasing CAC score. In the most recent publication from SBHW, the areas under the ROC curve were 0.63 for FRS and 0.68 for FRS plus CAC. Both FRS and CAC in a graded fashion were independent predictors in all FRS categories >10% but not <10%. In addition, these findings demonstrate that the CAC score adds to coronary event prediction over and above that predicted by FRS. This report is also the longest reported (8.5 years) follow-up; however, the significance of the 5 (self-/physician-referred) reports cannot be discounted. The SBHW, although a prospective study, selected intermediate- to high-risk subjects.⁷⁷ Therefore, the selection bias is such that there would be very little difference in risk from the beginning and the difference in any risk factor, ie, CAC scores and FRS, may be small. In addition, any difference would become obvious only with long-term follow-up as demonstrated. This is probably the reason why the first report showed no difference in predictability of either the FRS and the CAC score and longer follow-up uncovered this significance. In addition, the authors used 6-mm slices, which resulted in undersampling the volume, lowering all scores, especially in the lower percentile, and possibly reducing the low scores to zero.

The patient selection for the self-/physician referrals may include individuals with a variable number of risks, thereby accentuating the differences at an earlier stage. However, one cannot discount the incidence of disease in the population regardless of selection. Any population study selecting patients >45 years of age would have an incidence of significant atherosclerosis of 71% to 85% and a majority in the high-risk group from the outset, as evidenced by the demographics of disease prevalence and cardiovascular events reported in these age groups.

With respect to selection in a prospective random versus prospective selection, it would be interesting to analyze the Framingham study data from the original group. That selection was from community volunteers who were not randomized versus those who were later randomly recruited. One would expect that a larger number of low-risk individuals would be included in the nonrandom group, that initially the incidence of cardiovascular events would be less in this group, but that later both the incidence of disease and the event rate would be the same. It is not the purpose of this communication to discuss how a prospective random study should be done and if it is ethical to perform one. However, it is mentioned to highlight the problem of constructing a prospective randomized study when the disease prevalence is the same relative to the age group and to ask whether the expected outcome would be different from the data collected from the self/physician/prospective studies reviewed.

Regardless, the introduction of the CAC test is no different from the lifecycle of any new test such as the chest x-ray that had been viewed as a disruptive technology from any industry.⁹¹ Almost all leading physicians and established professionals other than those performing the new test will initially oppose it, viewing it as useless; however, the emergent disruptive technology eventually emerges above the performance trajectory line when significant benefit is demonstrated. The disruptive technology nearly always wins, as in the case of the chest x-ray, when superior results are validated. The reason the self-/physician-referred centers flourish is that the highly educated and motivated patients see the benefit of knowing whether they have significant disease and what they (not their doctor) should do about it. They do not want to wait until they have angina (far-advanced heart disease) to know. It is imperative that we take a proactive integrated approach to earlier coronary risk assessment.

The conclusion from this review is that the CAC score should be (1) added to/integrated with the FRS in the intermediate- and high-risk groups as suggested by Greenland et al⁷⁹ and should be used as a guide for therapy in addition to correction of other risk factors such as changes in lifestyle, diet, weight reduction, and exercise; (2) incorporated into treatment evaluation (statin therapy) by following the CAC score over time to evaluate its significance as a predictor for future cardiac events as it relates to progression, regression, or stabilization; (3) included in the decision tree for ETT to screen those patients with a low test probability and to reduce invasive testing and thereby reduce costs; and (4) used to evaluate the significance of the CAC versus the FRS. Are all subjects with high-risk FRS truly at risk, and how does this relate to the CAC score? It is well known that the FRS does not predict all future myocardial events and the NCEP ATP II guidelines were, by today’s standard, deficient by evidence of the change in cut points for total and LDL cholesterol levels in the ATP III guidelines. What cut points should be used for total cholesterol, LDL, and triglycerides? Likewise, are patients with any calcium score at high risk, and if not, what score level (cut points for CAC score) should be considered high risk? Rumberger et al⁹² demonstrated that calcium score cut points can be associated with severity of disease and reported ranges in sensitivity and specificity for luminal stenosis varying from >20% to 100% associated with specific calcium scores. The next question is also 2-fold: Are the current guidelines from the databases sufficient to recommend statin therapy when the CAC score is above the 75th percentile, or should they be recommended above the 50th percentile?

I recognize that these are thorny issues but am hopeful that the medical establishment can unite to reconcile differences and promote an inclusive approach to heart disease that recognizes the preventable nature of the most important health problem facing the nation.

	References

Top Introduction Comparison of EBCT CAC... CAC Scoring Radiation Dose Significance of CAC CAC as a Predictor... CAC in Clinical Studies Conclusions References EBCT for the Detection... Do Calcium Scores Correlate... Does EBCT Improve Prediction... Will EBCT Change Clinical... Does EBCT Improve Outcomes? Does EBCT Screening Have... Is EBCT a Risk-Free... The Future of EBCT... Conclusions References  References

 

1. Cotlar AM, Dubose JJ, Rose DM. History of surgery for breast cancer: radical to the sublime. Current Surg. 2003; 60: 329–337.[CrossRef]
   
2. Fisher B, Anderson S, Bryant J, Margolese G, Deutsch M, Fisher ER, Jeong H, Wolmark N. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med. 2002; 347: 1233–1241.[Abstract/Free Full Text]
   
3. Osler W. The Principles and Practice of Medicine. Edinburgh, Scotland: Young & Pentland; 1898.
   
4. Crane AW. Pulmonary roentgenology. AJR Am J Roentgenol. 1915; 2: 765–770.
   
5. Ochsner SF. Pulmonary tuberculosis: contributions of radiology in diagnosis and treatment. South Med J. 1986; 79: 1416–1424.[CrossRef][Medline] [Order article via Infotrieve]
   
6. Lenk R. Röntgendiagnose der Koronarsklerose in vivo. Fortschr Geb Rontgenstrahlen. 1927; 35: 1265–1268.
   
7. Snellen HA, and Nauta JH. Zur Röntgendiagnostik der Koronarverkalkungen. Fortschr Geb Rontgenstrahlen. 1937; 56: 277–286.
   
8. Wosika PH, Sosman MC. Roentgen demonstration of calcified coronary arteries in living subjects. JAMA. 1934; 102: 591–593.
   
9. Habbe EJ, Wright HH. Roentgenographic detection of coronary arteriosclerosis. Am J Roentgenol Rad Ther. 1950; 63: 50–62.[Medline] [Order article via Infotrieve]
   
10. van der Straeten PP. La coronarographie post mortem de l’homme âgé. Acta Cardiol. 1955; 10: 15–43.[Medline] [Order article via Infotrieve]
    
11. Blankenhorn DH, Stern D. Calcification of the coronary arteries. Am J Roentgenol Radium Ther Nucl Med. 1959; 81: 772–777.[Medline] [Order article via Infotrieve]
    
12. Kim MK. Calcification of matric vesicles in human aortic valve and aortic media. Fed Proc. 1976; 35: 156–162.[Medline] [Order article via Infotrieve]
    
13. Tanimura A, McGregor DH, Anderson HC. Calcification in atherosclerosis, I: human studies. J Exp Pathol. 1986; 2: 261–273.[Medline] [Order article via Infotrieve]
    
14. Bostrom K, Watson KE, Horn S, Wortham HC, Herman IM, Demer LL. Bone morphogenetic protein expression in human atherosclerotic lesions. J Clin Invest. 1993; 91: 1800–1809.[Medline] [Order article via Infotrieve]
    
15. Ideda T, Shirasawa T, Esaki Y, Yoshiki S, Hirokawa K. Osteopontin mRNA is expressed by smooth muscle-derived foam cells in human atherosclerotic lesions of the aorta. J Clin Invest. 1993; 92: 2814–2820.[Medline] [Order article via Infotrieve]
    
16. Hirota S, Imakita M, Kohri K, Ito A, Morii E, Adachi S, Kim HM, Kitamura Y, Yutani C, Nomura S. Expression of osteopontin messenger RNA by macrophages in atherosclerotic plaques: a possible association with calcification. Am J Pathol. 1993; 143: 1003–1008.[Abstract]
    
17. Shanahan CM, Cary NR, Metcalfe JC, Weissberg PL. High expression of genes for calcification-regulating proteins in human atherosclerotic plaque. J Clin Invest. 1994; 93: 2393–2402.[Medline] [Order article via Infotrieve]
    
18. Wexler L, Brundage B, Crouse J, Detrano R, Fuster V, Maddahi J, Rumberger J, Stanford W, White R, Taubert K. Coronary artery calcification: pathophysiology, epidemiology, imaging methods, and clinical implications: a statement for health professionals from the American Heart Association Writing Group. Circulation. 1996; 94: 1175–1192.[Free Full Text]
    
19. Rumberger JA, Simons DB, Fitzpatrick LA, Sheedy P, Schwartz RS. Coronary artery calcium area by electron-beam computed tomography and coronary atherosclerotic plaque area: a histopathologic correlative study. Circulation. 1995; 92: 2157–2162.[Abstract/Free Full Text]
    
20. Stary HC. The sequence of cell and matrix changes in atherosclerotic lesions of coronary arteries in the first forty years of life. Eur Heart J. 1990; 11 (suppl E): 3–19.[Free Full Text]
    
21. Monckeberg JG. Uber die Heine Mediaverkalkung der Extremitatenarterien und ihr Verhalten sur Arterioskelerose. Virchos Arch (Pathol Anat). 1903; 171: 141.[CrossRef]
    
22. Detrano R, Froelicher V. A logical approach to screening for coronary artery disease. Ann Intern Med. 1987; 106: 846–852.[Medline] [Order article via Infotrieve]
    
23. Reinmuller R, Lipton MJ. Detection of coronary artery calcification by computed tomography. Dynamic Cardiovasc Imaging. 1987; 1: 139–145.
    
24. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol. 1990; 15: 827–832.[Abstract]
    
25. Margolis J, Chen J, Kong Y, Peter RH, Behar VS, Kisslo JA. The diagnostic and prognostic significance of coronary artery calcification: a report of 800 cases. Radiology. 1980; 137: 609–616.[Abstract/Free Full Text]
    
26. Callister TQ, Cooil B, Raya SP, Nicholas JL, Russo DJ, Raggi P. Coronary artery disease: improved reproducibility of calcium scoring with an electron-beam CT volumetric method. Radiology. 1998; 208: 807–814.[Abstract/Free Full Text]
    
27. Heart and Stroke. Dallas, Tex: American Heart Association; 1999.
    
28. Heart Disease Statistical Update and Stroke Statistics. Dallas, Tex: American Heart Association; 2004.
    
29. Wald NJ, Law M, Watt HC, Wu T, Bailey A, Johnson AM. Apolipoproteins and ischaemic heart disease: implications for screening. Lancet. 1994; 344: 75–79.[CrossRef][Medline] [Order article via Infotrieve]
    
30. Waters D, Craven TE, Lesperance J. Prognostic significance of progression of coronary atherosclerosis. Circulation. 1993; 87: 4: 1067–1075.[Abstract/Free Full Text]
    
31. Buchwald H, Matts JP, Fitch LL, Campos CT, Sanmarco ME, Amplatz K, Castaneda-Zuniga WR, Hunter DW, Pearce MB, Bissett JK, et al. Changes in sequential coronary arteriograms and subsequent coronary events: Surgical Control of the Hyperlipidemias (POSCH) Group. JAMA. 1992; 268: 1429–1433.[Abstract]
    
32. Brown BG, Zhao X-Q, Sacco DE, Albers JJ. Lipid lowering and plaque regression: new insights into prevention of plaque disruption and clinical events in coronary disease. Circulation. 1993; 87: 1781–1791.[Abstract/Free Full Text]
    
33. Haskell WL, Alderman EI, Fair JM, Maron DJ, Mackey SF, Superko R, Williams PT, Johnstone IM, Champagne MA, Krauss RM, et al. Effects of intensive multiple risk factor reduction on coronary atherosclerosis and clinical cardiac events in men and women with coronary artery disease: the Stanford Coronary Risk Intervention Project (SCRIP). Circulation. 1994; 89: 975–990.[Abstract/Free Full Text]
    
34. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet. 1994; 344: 1383–1389.[CrossRef][Medline] [Order article via Infotrieve]
    
35. Akosah KO, Schaper A, Cogbill C, Schoenfeld P. Preventing myocardial infarction in the young adult in the first place: how do the National Cholesterol Education Panel III Guidelines Perform? J Am Coll Cardiol. 2003; 41: 1475–1479.[Abstract/Free Full Text]
    
36. O’Rourke RA, Brundage BH, Froelicher VF, Greenland P, Grundy SM, Hachamovitch R, Pohost GM, Shaw LJ, Weintr