| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Cardiac Imaging |
1 From the American Cardiovascular Research Institute, Atlanta, Ga (L.J.S.); Division of Cardiology, Tulane University School of Medicine, 1430 Tulane Ave, SL48, New Orleans, LA 70112 (P.R.); Cedars-Sinai Medical Center, Los Angeles, Calif (E.S., D.S.B.); and EBT Research Foundation, Nashville, Tenn (T.Q.C.). Received August 14, 2002; revision requested October 15; final revision received March 11, 2003; accepted March 20. Address correspondence to P.R. (e-mail: praggi@tulane.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: We followed up a cohort of 10,377 asymptomatic individuals undergoing cardiac risk factor evaluation and coronary calcium screening with electron-beam CT. Multivariable Cox proportional hazards models were developed to predict all-cause mortality. Risk-adjusted models incorporated traditional risk factors for coronary disease and coronary calcium scores.
RESULTS: Cardiac risk factors such as family history of coronary disease (69%), hypercholesterolemia (62%), hypertension (44%), smoking (40%), and diabetes (9%) were prevalent. The frequency of coronary calcium scores was 57%, 20%, 14%, 6%, and 3% for scores of 10 or less, 11–100, 101–400, 401–1,000, and greater than 1,000, respectively. During a mean follow-up of 5.0 years ± 0.0086 (standard error of the mean), the death rate was 2.4%. In a risk-adjusted model (model 
2 = 388.2, P < .001), coronary calcium was an independent predictor of mortality (P < .001). Risk-adjusted relative risk values for coronary calcium were 1.64, 1.74, 2.54, and 4.03 for scores of 11–100, 101–400, 401–1,000, and greater than 1,000, respectively (P < .001 for all values), as compared with that for a score of 10 or less. Five-year risk-adjusted survival was 99.0% for a calcium score of 10 or less and 95.0% for a score of greater than 1,000 (P < .001). With a receiver operating characteristic curve, the concordance index increased from 0.72 for cardiac risk factors alone to 0.78 (P < .001) when the calcium score was added to a multivariable model for prediction of death.
CONCLUSION: This large observational data series shows that coronary calcium provides independent incremental information in addition to traditional risk factors in the prediction of all-cause mortality.
© RSNA, 2003
Index terms: Computed tomography (CT), electron beam, 54.1211 • Coronary vessels, calcification, 54.812
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
All screened individuals provided informed consent to undergo electron-beam CT screening, and our study received Human Investigations Committee approval. Furthermore, separate approval from the Human Investigations Committee was obtained, along with informed consent, for the patient interviews, collection of data and follow-up, and corroboration of the occurrence of death.
Data Collection
We (T.Q.C., P.R.) collected information in regard to the presence of categoric cardiac risk factors in every patient. Risk factor data were derived through patient interview, referring physician contact, and existing medical record data. Systemic arterial hypertension was defined as a documented history of high blood pressure or treatment with medication, diet, and/or exercise. A history of current smoking or cessation of smoking within 3 months before testing was defined as positive smoking status. Hypercholesterolemia was determined on the basis of the answers to the following questions: "Has your physician ever told you that you need medications for high cholesterol?" "Are you currently taking cholesterol medications?" Answers to these questions often identified patients who were currently receiving cholesterol-lowering medications. Individuals were classified as having diabetes mellitus if they had received a previous diagnosis of diabetes mellitus that was determined with blood glucose levels or if they had received treatment with insulin or oral hypoglycemic agents.
Estimation of Framingham Risk
We calculated an expected 10-year risk of cardiac death or nonfatal myocardial infarction with range estimates published within the National Cholesterol Education Panel III risk calculator (11). A global risk score was devised for each patient on the basis of sex, age, and a history of hyperlipidemia, hypertension, and cigarette smoking. For each of the traditional risk factors for cardiovascular disease, individual points were summed for age and sex subsets of this cohort. Furthermore, the corresponding points for hyperlipidemia were based on a range of a total cholesterol level from 130 to 320 mg/dL (3.38–8.32 mmol/L) and on a range of a high-density lipoprotein cholesterol level from 25 to 55 mg/dL (0.65–1.43 mmol/L). Treated hypertension adds an average of 1%–2% risk for women and men, though the range was as high as 5%. Current history of cigarette smoking added a range in risk estimates from 4% to 12%. For the risk with diabetes mellitus, which is consistent with the coronary heart disease risk equivalent, a 20% risk of death or myocardial infarction was calculated. We then summed all point values for each of the previously mentioned risk factors to obtain a global risk score or the expected Framingham risk of death or myocardial infarction at 10 years. The median risk estimate was 15% (range, 1%–65%). Estimates were performed by one of us (L.J.S.).
Electron-Beam CT
Each patient signed an informed consent prior to undergoing screening. Electron-beam CT was performed with a scanner (C-100 or C-150; Imatron, South San Francisco, Calif), and images were obtained with 100-msec scanning time. The CT section thickness was 3 mm, and, in total, 40 sections were obtained starting at the level of the carina and proceeding to the level of the diaphragm. CT was electrocardiographically triggered at 60%–80% of the R-R interval. Coronary calcification was defined as a plaque of at least 3 consecutive pixels (area, 1.03 mm2) with attenuation of 130 HU or greater. Quantitative calcium scores were calculated according to the method described by Agatston et al (12). One of two experienced investigators (T.Q.C., P.R.) with 6 and 8 years of experience, respectively, reviewed all electron-beam CT scans in random order. Since calcium scoring was performed only once in each patient, intra- and interinvestigator score variability was not calculated.
Follow-up Procedures
Epidemiologic methods for follow-up included ascertainment of events by individuals who were blinded to historical and calcium score results (13–15). The occurrence of all-cause death was verified with the National Death Index (16). Individuals who underwent cardiovascular screening were followed up for a mean of 5 years ± 0.0086 (standard error of the mean), with a range of 2–5 years, and we were able to perform follow-up in 100% of the patients.
Statistical Analysis
Statistical analysis was performed by two experienced statistical analysts (L.J.S., E.S.) who were blinded to patient identifiers. Categoric variables were compared with the 2 statistic. Comparisons of categoric variables with continuous measures were calculated with analysis-of-variance techniques. A general linear model was used to compare continuous measures of coronary calcium in subsets of the population according to age and sex groupings. A first-order test for interaction of coronary calcium according to sex was performed.
The primary end point for this analysis was all-cause mortality. For the outcome analysis, we used univariable and multivariable Cox proportional hazards models to estimate the time to all-cause death (13,15). For the univariable Cox proportional hazards regression models, we evaluated the statistical significance of cardiac risk factors and coronary calcium scores in separate models. Univariable models with a probability value of less than .20 were considered for the multivariable models. A final multivariable model was constructed that included variables with a P value of less than .10. Relative risk ratios and 95% CIs were calculated for the univariable and multivariable models. Stepwise Cox proportional hazards models were used to identify the strongest estimators of outcome by using a forward likelihood ratio method.
A concordance index, event classification ability, was determined for a model containing cardiac risk factors and then again with coronary calcium added to the Cox proportional hazards regression analysis. A split-sample test and validation set were used to verify the multivariable model results. Unadjusted and risk-adjusted (ie, controlling for cardiac risk factors) Cox proportional hazards survival curves were generated to compare coronary calcium scores of 10 or less, 11–100, 101–400, 401–1,000, and greater than 1,000.
In a final analysis, we evaluated the use of a predicted Framingham risk score in relation to the coronary calcium score. Initially, Cox proportional hazards models included the added value of coronary calcium scores above and beyond the estimated Framingham risk score. A receiver operating characteristic (ROC) curve (95% CI) was calculated and compared the ability to classify events on the basis of the Framingham risk score and a combination of the Framingham risk score and the coronary calcium score. Additionally, the predicted 5-year mortality from the multivariable model was compared for low, intermediate, and high Framingham risk groups according to their coronary calcium score.
Continuous variables are presented as the mean and standard error of mean. The level of significance was chosen at a P value of less than .05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
|
|
|
|
Additionally, when compared with a model that was based on risk factors alone, the concordance index increased from 0.72 for cardiac risk factors to 0.78 when calcium was added to the multivariable model (P < .001). From this risk-adjusted model, the relative risk of mortality was 1.64, 1.74, 2.54, and 4.03 times greater for calcium scores of 11–100, 101–400, 401–1,000, and greater than 1,000, respectively, compared with the relative risk for a calcium score of 10 or less.
Split Sample: Test and Validation Sets
By using a split-sample approach, training and validation samples of the predictive accuracy of coronary calcium revealed similar results. The 2 test result for the training set was 18, and the 2 test for the validation set was 21 (Table 6). In both samples, the relative risk of death adjusted for all risk factors increased proportionally with the extent of coronary calcification.
|
|
|
2 = 111, P < .001). In a model that included both the estimated Framingham risk and coronary calcification, the calcium score was highly predictive of time to all-cause mortality (model 2 = 238, P < .001; coronary calcium score 2 = 69, P < .001). An ROC curve analysis revealed that both the estimated Framingham risk (area under the ROC curve [Az] = 0.67; 95% CI: 0.63, 0.71; P < .001) and coronary calcium scores (Az = 0.73; 95% CI: 0.70, 0.76; P < .001) were significantly predictive of death. In this analysis, coronary calcium showed superior outcome classification ability when compared with the estimated Framingham risk (Fig 3; Az = 0.73 vs 0.67, P < .001).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Calcification of the arterial wall is associated with the majority of atherosclerotic lesions, though only those in the more advanced stage of development may be visible at electron-beam CT screening (25). Electron-beam CT is widely used as a noninvasive tool to assess the presence and extent of coronary calcification, and more than 10 years of published evidence is available about its diagnostic and prognostic accuracy. In a recent expert consensus document prepared by the American College of Cardiology and the American Heart Association, key limitations to the existing body of evidence were discussed (1). Consistent with findings in prior publications about other noninvasive tests for coronary artery disease, early evidence was based on small highly selected patient samples and, for purposes of prognosis, revascularization often was used as an end point. In this era of evidence-based medicine, ever-increasing rigor is being applied to the evaluation of new technologies, and current standards require the use of risk-adjusted analysis for the estimation of clinical outcomes. The present analysis provides supportive evidence that there is a linear relationship between the extent of coronary artery calcification and all-cause mortality. All-cause mortality is an appropriate end point to follow, since in the United States, when one accounts for both cardiac and systemic forms of the disease, nearly three-fourths of all deaths are related to atherosclerosis (26,27). Furthermore, this end point is not affected by the reporting and misclassification bias potentially introduced by a physician’s filing of a death report (28). Finally, as noted according to Bayesian theory, the a priori limitation for low-risk populations imposed by the use of infrequent cardiac-specific end points may be overcome with the use of all-cause mortality as the primary outcome (7). Thus, our analysis that was conducted by using established epidemiologic methods in a sufficiently large population showed that the extent of coronary calcium is highly correlated with mortality risk.
Disruption of atherosclerotic plaques is the most frequent cause of acute thromboembolic events, and these events include sudden death, acute coronary syndromes, and stroke (19). Further, calcification of the arterial wall often is found in ruptured plaques, as well as in plaques that show superficial erosion (29–31). It is not clear whether coronary calcium renders the individual plaque unstable, but as a marker and a measure of the atherosclerotic burden, it indicates an individual’s predisposition to develop thromboembolic and ischemic events (32). Our prognostic models that were used to estimate all-cause mortality revealed that age and coronary calcium were the strongest risk markers in this 10,377-subject registry.
Risk Estimation
An evolution in thought has increasingly linked the intensity of therapeutic intervention to the estimation of risk (23,33). It is now recommended that a compilation of risk that is based on established cardiac risk factors be used to determine various outcome strata. This recommendation recently was implemented in the development of the National Cholesterol Education Program III guidelines in which the approach offered by the Framingham risk score to formulate a 10-year risk of cardiac death or myocardial infarction was embraced (34). In our article, many of the established risk factors were predictive of mortality, but when the coronary calcium score was added to the model, it was a highly significant predictor of outcome and added independently to the risk prediction (P < .001 for concordance index increase).
Although many noninvasive markers have been evaluated for inclusion in the array of cardiac risk estimators, to date, the evidence for their utility has not been overwhelming. For any new marker or test to be judged as a valuable additional tool for the prediction of events, it should increase the precision of outcome estimation beyond information derived from integrated risk scores, such as the Framingham risk model (22). Nonetheless, current evidence also shows that established cardiac risk factors possess a limited ability in the estimation of risk. On average, only 50%–60% of the variability in outcome is explained by risk factors (22,33–35). Hence, there is a need to improve risk prediction. Markers of vascular wall abnormalities have been used in large epidemiologic studies, such as the Atherosclerotic Risk Factors in the Community study, to assess the future risk of major adverse cardiac events (36). In the Atherosclerotic Risk Factors in the Community study, the hazard ratio for events with an intima-media thickness of the carotid artery of 1 mm or thicker was increased 5.1-fold in women and 1.9-fold in men (36). It appears that coronary calcium may be an additional marker that provides unique information for risk assessment. A frequent criticism of the available literature about electron-beam CT is that findings therein have not firmly established that coronary calcification adds incremental value in addition to established cardiac risk factors (7). In an older population with high pre-test probability of disease, Detrano et al (37) reported that the concordance index, which reflects a measure of event classification, did not change when the calcium score was added to a model containing the Framingham risk index. Conversely, in an extended follow-up of their prior series, Arad et al (10) recently reported that moderate calcium scores (ie, 
160) were associated with a relative risk of cardiac death or myocardial infarction that was 10-fold higher than that which was estimated with the Framingham risk index alone. Furthermore, Raggi et al (5) showed that coronary artery calcium provided incremental prognostic information to predict hard cardiac events in a cohort of 676 individuals who were followed up for approximately 3 years. Finally, in the current study, 21.5% of mortality information—or incremental value—was attributable to coronary calcium, and this finding indicates that calcification of the coronary arteries may provide additional documentation of risk independent of the presence of conventional cardiac risk factors.
Study Limitations
Although the current article includes a rigorous analysis of the prognostic value of coronary calcium, the majority of patients referred for electron-beam CT screening had cardiac risk factors and, as such, may not be representative of the general population. Furthermore, it is expected that the inclusion of measured risk factors, such as systolic blood pressure, blood glucose level, and cholesterol values would provide a better estimation of risk than historical data alone. However, in many imaging laboratories, patients are referred on the basis of a brief history and measured values often are unavailable. Thus, we believe that our analysis is probably a close representation of the reality of daily laboratory practices. Furthermore, the use of categoric risk factors instead of continuous variables has been shown to constitute a valid approach to risk assessment (38). Hence, we believe that a risk assessment approach that is based on historical risk factors rather than on continuous variables does not significantly weaken the assumptions made in this study. Additionally, the National Death Index data do not include the cause of death and, as such, our models include mortality possibly unrelated to atherosclerotic disease. Furthermore, we do not have data on cardiac-specific procedures.
In conclusion, we collected mortality data in regard to 10,377 asymptomatic individuals with cardiac risk factors who were referred by their primary care physicians for coronary calcium screening with electron-beam CT. This large observational data series strongly indicates that coronary artery calcium is an independent estimator of all-cause mortality. Our results show that survival at 5 years worsens substantially as the screening calcium scores increase from levels of 10 or less to those of greater than 1,000. Therefore, it appears justified to use coronary calcium screening to identify intermediate-risk patients with traditional risk factors for whom aggressive risk-reducing strategies for the treatment of atherosclerotic disease should be indicated.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, L.J.S., P.R.; study concepts, L.J.S., P.R., D.S.B., T.Q.C.; study design, L.J.S.; literature research, L.J.S., D.S.B., P.R., T.Q.C.; clinical studies, T.Q.C., P.R.; data acquisition, L.J.S., P.R., T.Q.C.; data analysis/interpretation, D.S.B., P.R., L.J.S., E.S.; statistical analysis, L.J.S.. E.S.; manuscript preparation and definition of intellectual content, all authors; manuscript editing, revision/review, and final version approval, L.J.S., P.R.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. He, K. Liu, M. L Daviglus, E. Mayer-Davis, N. S. Jenny, R. Jiang, P. Ouyang, L. M Steffen, D. Siscovick, C. Wu, et al. Intakes of long-chain n-3 polyunsaturated fatty acids and fish in relation to measurements of subclinical atherosclerosis Am. J. Clinical Nutrition, October 1, 2008; 88(4): 1111 - 1118. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Henneman, J. D. Schuijf, G. Pundziute, J. M. van Werkhoven, E. E. van der Wall, J. W. Jukema, and J. J. Bax Noninvasive evaluation with multislice computed tomography in suspected acute coronary syndrome plaque morphology on multislice computed tomography versus coronary calcium score. J. Am. Coll. Cardiol., July 15, 2008; 52(3): 216 - 222. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. C. Hoang, H. Ghandehari, V. A. Lopez, M. G. Barboza, and N. D. Wong Global Coronary Heart Disease Risk Assessment of Individuals With the Metabolic Syndrome in the U.S. Diabetes Care, July 1, 2008; 31(7): 1405 - 1409. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E J HALL and D J BRENNER Cancer risks from diagnostic radiology Br. J. Radiol., May 1, 2008; 81(965): 362 - 378. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Raggi and M. Kleerekoper Contribution of Bone and Mineral Abnormalities to Cardiovascular Disease in Patients with Chronic Kidney Disease Clin. J. Am. Soc. Nephrol., May 1, 2008; 3(3): 836 - 843. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P. Schenker, S. Dorbala, E. C. T. Hong, F. J. Rybicki, R. Hachamovitch, R. Y. Kwong, and M. F. Di Carli Interrelation of Coronary Calcification, Myocardial Ischemia, and Outcomes in Patients With Intermediate Likelihood of Coronary Artery Disease: A Combined Positron Emission Tomography/Computed Tomography Study Circulation, April 1, 2008; 117(13): 1693 - 1700. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Hemingway, C. Langenberg, J. Damant, C. Frost, K. Pyorala, and E. Barrett-Connor Prevalence of Angina in Women Versus Men: A Systematic Review and Meta-Analysis of International Variations Across 31 Countries Circulation, March 25, 2008; 117(12): 1526 - 1536. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Gottlieb and J. A.C. Lima Screening High-Risk Patients With Computed Tomography Angiography Circulation, March 11, 2008; 117(10): 1318 - 1332. [Full Text] [PDF]
|
|
|
|
 |
|
 T. Winzenberg and G. Jones Recommended Calcium Intakes in Children: Have We Set the Bar Too High? IBMS BoneKEy, February 1, 2008; 5(2): 59 - 68. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Rutten, I. Isgum, and M. Prokop Coronary Calcification: Effect of Small Variation of Scan Starting Position on Agatston, Volume, and Mass Scores Radiology, December 1, 2007; 246(1): 90 - 98. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Leontiev and T. J. Dubinsky CT-Based Calcium Scoring to Screen for Coronary Artery Disease: Why Aren't We There Yet? Am. J. Roentgenol., November 1, 2007; 189(5): 1061 - 1063. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Burrill, Z. Dabbagh, F. Gollub, and M. Hamady Multidetector computed tomographic angiography of the cardiovascular system Postgrad. Med. J., November 1, 2007; 83(985): 698 - 704. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. T. Wilson, P. Gopalakrishnan, and T. Tak Noninvasive Cardiac Imaging with Computed Tomography Clin. Med. Res., October 1, 2007; 5(3): 165 - 171. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Bax, L. H. Young, R. L. Frye, R. O. Bonow, H. O. Steinberg, and E. J. Barrett Screening for Coronary Artery Disease in Patients With Diabetes Diabetes Care, October 1, 2007; 30(10): 2729 - 2736. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Nasir, L. J. Shaw, S. T. Liu, S. R. Weinstein, T. R. Mosler, P. R. Flores, F. R. Flores, P. Raggi, D. S. Berman, R. S. Blumenthal, et al. Ethnic Differences in the Prognostic Value of Coronary Artery Calcification for All-Cause Mortality J. Am. Coll. Cardiol., September 4, 2007; 50(10): 953 - 960. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A. Rumberger The Ethnic Rosetta Stone: Translating Risk Factors, Plaque Scores, and Mortality J. Am. Coll. Cardiol., September 4, 2007; 50(10): 961 - 963. [Full Text] [PDF]
|
|
|
|
 |
|
 T. Schepis, O. Gaemperli, P. Koepfli, M. Namdar, I. Valenta, H. Scheffel, S. Leschka, L. Husmann, F. R. Eberli, T. F. Luscher, et al. Added Value of Coronary Artery Calcium Score as an Adjunct to Gated SPECT for the Evaluation of Coronary Artery Disease in an Intermediate-Risk Population J. Nucl. Med., September 1, 2007; 48(9): 1424 - 1430. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Nasir, M. J. Budoff, N. D. Wong, M. Scheuner, D. Herrington, D. K. Arnett, M. Szklo, P. Greenland, and R. S. Blumenthal Family History of Premature Coronary Heart Disease and Coronary Artery Calcification: Multi-Ethnic Study of Atherosclerosis (MESA) Circulation, August 7, 2007; 116(6): 619 - 626. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Raggi and A. Bellasi Review: Imaging to assess effect of medical therapy in patients with diabetes mellitus The British Journal of Diabetes & Vascular Disease, July 1, 2007; 7(4): 157 - 164. [Abstract] [PDF]
|
|
|
|
 |
|
 S. K. Mehta, J. E. Rame, A. Khera, S. A. Murphy, R. M. Canham, R. M. Peshock, J. A. de Lemos, and M. H. Drazner Left Ventricular Hypertrophy, Subclinical Atherosclerosis, and Inflammation Hypertension, June 1, 2007; 49(6): 1385 - 1391. [Abstract] [Full Text] [PDF] |
