Author + information
- Received May 16, 2010
- Revision received July 21, 2010
- Accepted August 2, 2010
- Published online December 1, 2010.
- Matthew J. Budoff, MD⁎,⁎ (, )
- John E. Hokanson, MPH, PhD‡,
- Khurram Nasir, MD, MPH§,
- Leslee J. Shaw, PhD∥,
- Gregory L. Kinney, PhD‡,
- David Chow, BS⁎,
- Daniel DeMoss, BS⁎,
- Vivek Nuguri, BS⁎,
- Vahid Nabavi, BS⁎,
- Raghu Ratakonda, BS⁎,
- Daniel S. Berman, MD† and
- Paolo Raggi, MD∥
- ↵⁎Reprints requests and correspondence:
Dr. Matthew Budoff, Los Angeles Biomedical Research Institute, 1124 West Carson Street, Torrance, California 90502
Objectives This study examined a large cohort to assess whether progression of coronary artery calcium (CAC) was associated with all-cause mortality, and which among 3 different methods to assess CAC progression provided the best estimate of risk.
Background Serial assessment of CAC scores has been proposed as a method to follow progression of coronary artery disease, and it has been suggested that excessive CAC progression may be a useful noninvasive predictor of the patient's risk of future events. However, the optimal method to measure calcium progression has not been well established.
Methods The study sample consisted of 4,609 consecutive asymptomatic individuals referred by primary physicians for CAC measurement with electron beam tomography, who underwent repeat screening. Three general statistical approaches were taken: 1) the absolute difference between follow-up and baseline CAC score; 2) percent annualized differences between follow-up and baseline CAC score; and 3) difference between square root of baseline and square root of follow-up CAC score >2.5 (the “SQRT method”).
Results The average interscan time was 3.1 years, and there were 288 deaths. Progression of CAC was significantly associated with mortality regardless of the method used to assess progression (p < 0.0001). After adjusting for baseline score, age, sex, and time between scans, the best CAC progression model to predict mortality was the SQRT method (hazard ratio [HR]: 3.34; 95% confidence interval [CI]: 2.65 to 4.21; p < 0.0001), followed by a >15% yearly increase (HR: 2.98; 95% CI: 2.20 to 4.95; p < 0.0001). Progression was very limited and did not predict mortality in patients with baseline CAC = 0.
Conclusions The CAC progression added incremental value in predicting all-cause mortality over baseline score, time between scans, demographics, and cardiovascular risk factors. Serial assessment may have clinical value in assessing plaque progression and future cardiovascular risk.
As a sensitive marker of atherosclerosis, coronary artery calcium (CAC) has been proposed as a method to follow progression of coronary artery disease. However, randomized clinical trials that used sequential computed tomography (CT) scanning failed to uniformly demonstrate that CAC can to be used to monitor response to medical therapy for atherosclerosis. In spite of this, 1 trial (1) and a few observations (2–4) have shown that the progression of CAC is a marker of increased risk of future cardiovascular events. Similarly, early trials using quantitative invasive coronary angiography conclusively demonstrated that atherosclerosis progression is a harbinger of adverse outcomes. Hence, excessive CAC progression may be a useful noninvasive predictor of the patient's risk of future events. Nonetheless, the evidence so far accumulated is small (1–4), and little is known of the development of CAC and its significance in subjects without CAC at baseline. Therefore, in this study, we verified the all-cause mortality of patients who underwent at least 2 CT scans for quantification of CAC. We further assessed which of several methods to estimate CAC progression was more accurate to predict mortality, and whether absence of CAC at baseline and its development are independently associated with all-cause mortality.
The study sample consisted of 4,609 consecutive asymptomatic individuals referred by primary care physicians for CAC measurement with electron beam tomography (EBT), who underwent sequential scans at least 10 months apart. Patients underwent repeat scans, as per their primary physician, to assess change in atherosclerosis risk over time. This study received the approval of the institutional review board.
Patients with a history of coronary artery disease (i.e., admission to the hospital for chest pain, acute coronary syndrome, or myocardial infarction [MI], as well as prior coronary revascularization) were excluded. Also, patients suffering an interim event between scans were excluded. Subjects were given a risk-factor questionnaire to assess ethnicity and cardiovascular risk factors at baseline and follow-up. The presence and number of risk factors for each subject was calculated based on the National Cholesterol Education Program guidelines (5). Risk factors included: age (men age >45 years of age, women >55 years of age), current cigarette smoking, diabetes mellitus, history of premature coronary artery disease in first-degree relatives (men <55 years of age, women <65 years of age), hypertension, and hypercholesterolemia. Current cigarette smoking was defined as any cigarette smoking in the past month. Hypertension was defined by current use of antihypertensive medications or known and untreated hypertension. Hypercholesterolemia was defined as use of cholesterol-lowering medications or, in the absence of cholesterol-lowering medication use, as having a total serum cholesterol >200 mg/dl. Statin use was recorded at the time of scanning at both baseline and follow-up.
All study subjects underwent 2 EBT scans using an Imatron C-300 computed tomography scanner (GE Imatron, South San Francisco, California). Thirty to forty contiguous tomographic slices were obtained at 3-mm intervals beginning 1 cm below the carina and progressing caudally to include the entire coronary tree. Exposure time was 100 ms per tomographic slice, and total radiation dose was 0.6 mSv per scan.
All scans were analyzed with a commercially available software package (Neo Imagery Technologies, City of Industry, California). An attenuation threshold of 130 Hounsfield units (HU) and a minimum of 3 contiguous pixels were utilized for identification of a calcific lesion. Each focus exceeding the minimum criteria was scored using the algorithm developed by Agatston et al. (6), calculated by multiplying the lesion area by a density factor derived from the maximal HU within this area. The density factor was assigned in the following manner: 1 for lesions with peak attenuation of 130 to 199 HU, 2 for lesions with peak attenuation of 200 to 299 HU, 3 for lesions with peak attenuation of 300 to 399 HU, and 4 for lesions with peak attenuation >400 HU. The total CAC score was determined by summing individual lesion scores from each of 4 anatomical sites (left main, left anterior descending, left circumflex, and right coronary artery) (7).
Follow-up data collection
Epidemiologic methods for follow-up included ascertainment of death by individuals who were blinded to historical and CAC score results (8,9). The occurrence of all-cause death was verified with the National Death Index (10). Individuals who underwent cardiovascular screening were followed for a mean of 5.4 ± 3.4 years after the second scan (range 1.0 to 16.0 years). Follow-up was completed in 100% of patients; there were 4,609 asymptomatic subjects in this sample.
Categorical variables comparing CAC patient subsets with historical variables were compared using a chi-square likelihood ratio test. For comparing CAC subsets by age and other continuous measures, we employed analysis of variance techniques. A p value <0.05 was considered statistically significant. Time to death from all causes was estimated using a Cox proportional hazards model. The exposure variable is the progression of CAC score from baseline to follow-up scan. A priori methods for assessing progression were established. Three general approaches were taken: 1) the difference between follow-up and baseline absolute CAC score; 2) annualized percent differences between follow-up and baseline CAC score; and 3) previously established cut points for progression of CAC. Specifically:
1. the absolute difference between the second and first measure of coronary calcium [CAC(follow-up) – CAC(baseline)],
2. the square root transformed difference [√CAC(follow-up) – √CAC(baseline)] (the “SQRT method”) (11),
3. the natural logarithm plus 25 difference [(ln CAC(follow-up)+ 25) – (ln CAC(baseline) + 25)]. (the “MESA method”) (12),
4. the percent change [(CAC(follow-up) – CAC(baseline))/ CAC(baseline)].
The preceding 4 quantitative exposure variables divided by the time (in years) between the first and the second scan.
Pre-defined cut points:
1. a square root transformed difference > 2.5 (the difference beyond measurement error),
2. an annualized percent change >15%.
The primary outcome was mortality verified from the National Death Index. Cox proportional hazard modeling was performed as implemented in SAS software (version 9.2, SAS Institute, Cary, North Carolina). Covariates included in all models were baseline CAC score, age, and sex. The length of time between the 2 scans was included as a covariate in models other than those using annualized difference in progression of CAC. Hazard ratios (HR) and 95% confidence intervals (CI) are presented for continuous variables, for 1 SD of continuous variables (standardized HR), and for the pre-defined cut points. Standardized hazard ratios allow for direct comparison of the effect size of the different continuous exposure methods. Model comparisons for different approaches for assessing progression of CAC scores within each study group used the Akaike information criterion for non-nested models. A lower Akaike information criterion indicates a better fit to the data. We then tested the best fitting models by adding cardiovascular disease risk factors (hypertension, diabetes, smoking, pre-mature family history, and hypercholesterolemia) to assess for further adjustment of the hazard ratios.
To facilitate data interpretation and verification of proportional hazards assumptions, CAC scores were classified into the following categories: 0, 1 to 10, 11 to 100, 101 to 400, and ≥400 (no identifiable plaque, minimal plaque, mild plaque, moderate plaque, and extensive atherosclerotic plaque burden, respectively) (13).
The majority of patients (72.9%) were men, and the mean age was 60.1 ± 10.8 years. The mean interscan period (scan 1 to scan 2) was 3.1 ± 2.0 years (range 1 to 16 years). The prevalence of cardiovascular risk factors was high: current tobacco use: 6.2%; diabetes: 7.1%; high blood pressure: 23.6%, hypercholesterolemia: 41.2%; family history of coronary artery disease: 40.4%. The CAC score was a strong independent predictor of mortality (chi-square = 51.02, p < 0.0001). Mortality significantly increased with increasing CAC score.
There were 288 deaths among the 4,609 individuals in the full cohort (Table 1). A priori analyses were planned for all patients (Table 1), patients with baseline CAC score >0 (n = 2,866, 236 deaths) (Table 2), and those with baseline CAC score >30 (n = 2,183, 204 deaths) (Fig. 1).
For comparison purposes, subgroups were also examined: patients with baseline CAC = 0, the corollary of Table 2 (n = 1,743, 52 deaths) (Table 3), and those with baseline CAC ≤30, the corollary of Table 3 (n = 2,426, 84 deaths).
Progression of CAC was significantly associated with mortality regardless of the method used to assess progression (p < 0.0001). Standardized HRs ranged from 1.12 (95% CI: 1.09 to 1.15) for annualized absolute CAC score increase, to 1.49 (95% CI: 1.39 to 1.59) for the MESA method. The best model to predict mortality, however, was the SQRT method (HR: 3.34, 95% CI: 2.65 to 4.21, p < 0.0001), followed by the method that uses an annualized increase in CAC score >15% in those with baseline CAC >30 (HR: 2.98, 95% CI: 2.20 to 4.95, p < 0.0001), after controlling for baseline CAC, age, sex, and follow-up time between scans.
The effect of progression of CAC in individuals with CAC >0 at baseline is presented in Table 2. There were 236 deaths among 2,866 individuals. As in the full cohort, all methods for assessing progression of CAC were significantly related to mortality. The best model was again based on the SQRT method (HR: 3.66, 95% CI: 2.82 to 4.74, p < 0.001).
The same concept held true in individuals with a baseline CAC >30 (n = 2,183, 204 deaths), where, though all methods were predictive, the SQRT method provided the best fit of the data with a HR = 3.28 (95% CI: 2.48 to 4.32, p < 0.001). In secondary analyses, we examined the progression of CAC in individuals who had no detectable CAC at the baseline visit (n = 1,743, 52 deaths during follow-up) (Table 3). There was no significant relationship between development of CAC and mortality in these individuals.
Among individuals with baseline CAC score ≥30, progression of CAC was significantly associated with mortality. In this subgroup, the MESA method provided the best fit of the data (HR: 2.85; 95% CI: 2.20 to 3.70; p < 0.001).
Adjusting for cardiovascular disease risk factors did not appreciably alter the relationship between CAC progression and mortality. Progression adjusted for baseline CAC, age, sex, and time between scans yielded a HR = 3.34 (95% CI: 2.65 to 4.21). Further adjustment adding hypertension, hypercholesterolemia, diabetes, family history, and smoking resulted in HR = 3.32 (95% CI: 2.62 to 4.20, p < 0.0001). Similarly, HR did not change significantly for models of prediction including baseline CAC >0 (HR: = 3.62, 95% CI: 2.78 to 4.70, p < 0.0001) or CAC >30 (HR: 3.22, 95% CI: 2.43 to 4.27, p < 0.0001).
We further examined the combined effect of the presence of coronary calcium at baseline and progression of coronary calcium. Having CAC at baseline and significant progression of CAC was a significant predictor of future mortality (HR: 5.15, 95% CI: 3.67 to 7.22, p < 0.0001) (Table 4). Adjusting for cardiovascular disease risk factors did not appreciably alter the relationship between CAC progression and mortality (HR: 5.33, 95% CI: 3.74 to 7.60, p < 0.0001). Having baseline CAC without significant progression was marginally associated with mortality (HR: 1.42, 95% CI: 0.99 to 2.02, p = 0.054). Those without CAC at baseline but with significant progression of CAC did not have an increase in the rate of mortality (p = 0.97).
Progression in the setting of increasing baseline CAC using set cut points (1 to 10, 11 to 100, 101 to 400, and >400) was associated with increasing risk of all-cause mortality (Fig. 2).
This is the first large study following patients for 1 to 16 years and submitted to repeat CT scans, demonstrating incremental increase in mortality with progression of CAC over baseline score, time between scans, demographics, and cardiovascular risk factors. The CAC score in itself, as shown in multiple studies before, is a strong independent predictor of mortality. Because CAC can be located and quantified, changes over time can be analyzed so that progression or regression can be determined. At the moment, this is one of the few techniques that enables research to study this important question, as outcome may be related more to the continuous observation of changes over time than a single determination once in a lifetime. The observation that progression of CAC is associated with an adverse outcome extends prior reports linking CAC progression to coronary syndromes (1–3).
Prior smaller studies have suggested this effect. Raggi et al. (14) followed 269 asymptomatic subjects for 2.5 years after being submitted to sequential CT scans; of the 22 cardiovascular disease events, 20 occurred in patients with progression of CAC and 2 among patients without progression (p < 0.01). Another study of 225 subjects followed for an average of 3 years after the repeat CT scan, those with new cardiac events had a significantly greater annual change in CAC score than those who did not experience events (35% vs. 22%, p = 0.04) (15). Moreover, 78% of patients with events had >20% annual progression versus 37% of those not experiencing events (p < 0.001).
Among 817 asymptomatic subjects submitted to sequential CT scans approximately 2 years apart, mean absolute and percent changes in CAC score were 147% and 47%, respectively, in those who developed an MI after the second scan, compared with 63% and 26% in those without events (p < 0.001 and p = 0.01, respectively) (3). In another EBT study of 495 asymptomatic subjects receiving statins and submitted to more than 1 CT scan, ≥15% CAC progression was associated with a 17.2-fold (95% CI: 4.1 to 71.2) increased risk of MI when compared with patients without CAC progression (p < 0.0001) (4). These relatively small studies suggested that continued accumulation of CAC in asymptomatic individuals is associated with increased risk of future MI.
One large prospective study using CT to measure progression of CAC has also been reported. This prospective observational study evaluated 4,613 asymptomatic persons aged 50 to 70 years with EBT screening for CAC at baseline and again at 2 years and the follow-up lasted 4.3 years (1). This study demonstrated that the median (interquartile range) CAC scores increased by 4 (0, 38) units from baseline to follow-up in subjects who did not sustain a coronary event at any time during the study. In contrast, median (interquartile range) CAC scores increased by 247 (40, 471) units in subjects who experienced a first coronary disease event after the follow-up scan (p < 0.0001). Multivariable logistic regression analyses demonstrated that age (p = 0.03), male sex (p = 0.04), low-density lipoprotein cholesterol (p = 0.01), high-density lipoprotein cholesterol (p = 0.04), and 2-year change in CAC score (p < 0.0001) were significantly associated with subsequent coronary artery disease events. The MESA (Multi-Ethnic Study of Atherosclerosis) (12) is following patients after a second CAC screen, and will assess subsequent cardiovascular events; however, results are still 1 to 2 years away.
Baseline 0 scores were not predictive of progression or all-cause mortality in this study. This further validates the concept that a baseline 0 score has a significant warranty period for both future cardiovascular events and progression of atherosclerosis (16,17). Min et al. (16) have suggested a 0 calcium score affords at least a 5-year warranty period, and our study strongly supports that evidence with even longer follow-up and interscan periods.
Our study provides strong confirmatory evidence that CAC progression is associated with future cardiovascular events and, as radiation doses are being reduced to a minimum, that may be a useful tool in the prevention armamentarium to assess atherosclerosis progression noninvasively. Measuring CAC progression requires sequential CT scans, with a cumulative radiation exposure. Prior reports have raised concern about the excess risk of cancer with such an approach (18). However, such predictions are outdated as current gating technology reduces the radiation dose substantially, with an expected dose of <1 mSv per scan (7). Further advances have reduced the CAC dose to as low as 0.6 mSv, lower than screening mammography (19).
Strengths of this study include its large sample size and long follow-up. In addition, the scans were read in the same CT reading center that interprets the MESA and multiple other National Institutes of Health epidemiologic studies, with standardized protocols for acquisition and interpretation of CAC scans (20).
Prior event studies of CAC progression have reported on different techniques to assess progression. In this analysis, we compared previously reported techniques to determine the best predictor of outcomes with CAC progression. All methods for assessing progression of CAC were significantly related to mortality except in individuals with no detectable CAC at baseline (score = 0). Multiple studies have demonstrated the very low event rates in persons without CAC (score = 0) (17,21). In general, a square root transformed difference of >2.5 provided the best fit of the data. This (20) and other methods (21) account for interscan variability in CAC. This was true in the full cohort, among those with baseline CAC >0, and in those with a baseline CAC >30, or 3 pre-defined subgroups. The HRs for this cut point were 3.34 (95% CI: 2.65 to 4.21) in the full cohort, 3.66 (95% CI: 2.82 to 4.74) in those with baseline coronary calcium >0, and 3.28 (95% CI: 2.48 to 4.32) in those with CAC >30. These analyses indicate that regardless of cut points, progression of CAC is related to a >3-fold increase in mortality. The greatest risk was a combination of moderate (>100) or severe (>400) CAC score and progression (Fig. 2). This supports the hypothesis that whereas CAC represents lifetime accumulation of atherosclerosis and scarring after plaque rupture (22,23), progression demonstrates the quiescence or activity of the atherosclerotic pathway and is an independent predictor of survival.
The cardiovascular risk factors were taken by survey, rather than measured. Nonetheless, the prevalence of hypercholesterolemia, hypertension, and diabetes mellitus in our population was similar to that observed in other large, population-based studies of coronary heart disease (24). Additionally, the National Death Index obtained for this study did not include cause of death and, as such, our models include mortality possibly unrelated to atherosclerotic disease. However, the bias resulting from death misclassification does not occur in all-cause mortality models and in this age group, the prevalence of coronary heart disease deaths has been reported to be approximately one-third of death from all causes (25).
This is a single-center study, using the same CT reading center as used in multiple National Institutes of Health–sponsored studies. Reproducibility of CAC scanning and scoring from this laboratory has been previously reported (26,27). We used a minimum interscan time of 10 months. Only 311 patients had follow-up <1 year. There is no difference if different time cut points are used. We used 10, 12, and 18 months as minimum without affecting overall results, just lowering sample size (data not shown).
The progression of CAC adds significant incremental prediction ability of all-cause mortality, after adjustment for time between scans, demographics, risk factors, and baseline CAC scores. It appears that persons with scores >30 can be assessed for progression of CAC, and this adds incremental information regarding future prognostic risk. Though use of repeat CT testing to estimate an individual's risk associated with CAC “change” appears to be of value, a better understanding of what therapies may be of benefit and how clinicians should use these data in clinical practice remains to be determined.
The authors have reported that they have no relationships to disclose. Pim J. de Feyter, MD, PhD, served as Guest Editor for this article.
- Abbreviations and Acronyms
- coronary artery calcium
- computed tomography
- electron beam tomography
- Hounsfield unit
- myocardial infarction
- Received May 16, 2010.
- Revision received July 21, 2010.
- Accepted August 2, 2010.
- American College of Cardiology Foundation
- Arad Y.,
- Goodman K.J.,
- Roth M.,
- Newstein D.,
- Guerci A.D.
- Arad Y.,
- Spadaro L.A.,
- Roth M.,
- Newstein D.,
- Guerci A.D.
- Raggi P.,
- Callister T.Q.,
- Shaw L.J.
- Expert Panel on Detection, Evaluation, and Treatment of High Blood Pressure Cholesterol in Adults
- Agatston A.S.,
- Janowitz W.R.,
- Hildner F.J.,
- Zusmer N.R.,
- Viamonte M. Jr..,
- Detrano R.
- Budoff M.J.,
- Achenbach S.,
- Blumenthal R.S.,
- et al.
- Shaw L.J.,
- Peterson E.D.,
- Kesler K.L.,
- et al.
- Centers for Disease Control and Prevention
- Kronmal R.A.,
- McClelland R.L.,
- Detrano R.,
- et al.
- Min J.K.,
- Lin F.Y.,
- Gidseg D.S.,
- et al.
- Blaha M.,
- Budoff M.J.,
- Shaw L.J.,
- et al.
- Budoff M.J.,
- Nasir K.,
- McClelland R.L.,
- et al.
- Arnett D.K.,
- McGovern P.G.,
- Jacobs D.R. Jr..,
- et al.
- National Center for Chronic Disease Prevention and Health Promotion