Key Points
- •
Global risk assessment approaches for coronary heart disease, such as Framingham risk scores, underestimate long-term risks, especially in young men and postmenopausal women with multiple risk factors.
- •
Screening modalities, such as non–contrast-enhanced CT detection of coronary artery calcium, improve the ability to “accurately” predict risk in vulnerable groups and add information above and beyond global risk assessment.
- •
Absence of coronary artery calcification is associated with a very low risk of future CAD, significant stenosis, myocardial ischemia, and acute coronary syndrome during the next 5 years in nonsmokers.
- •
Guidelines reiterate that intermediate-risk individuals are the best patients for coronary artery calcium testing.
- •
Serial coronary artery calcium testing is currently not recommended because data for its prognostic significance are limited.
- •
Coronary artery calcium testing also appears to improve lifestyle changes and medication adherence.
- •
Coronary CTA has the potential to provide comprehensive information about the location, severity, and characteristics of atherosclerotic plaque, especially noncalcified plaque; however, there are currently no recommendations for its use as a screening tool in asymptomatic persons.
- •
The prognostic value of CTA-based plaque subtypes and burden above and beyond coronary artery calcium scores is not yet established.
Cardiovascular disease (CVD) is the leading cause of mortality worldwide, with coronary artery disease (CAD) accounting for nearly half of all CVD deaths. Although it is highly prevalent in the Western world, in the next 15 years, an estimated 25 million people will die of stroke or heart disease, with 80% of this burden in developing countries. Emerging data suggest that although the mortality rate after the occurrence of clinical heart disease (e.g., myocardial infarction [MI]) has significantly decreased during the last two decades, the incidence of new onset of CAD has remained relatively stable during this period. This finding points to the fact that although we have made great strides in secondary prevention, we have failed in our primary efforts of decreasing the rate of new-onset CAD.
In approximately half of the individuals, the initial presentation of CAD is either MI or sudden death. Unfortunately, conventional risk factor assessment predicts only 65% to 80% of future cardiovascular events, leaving many middle-aged and older individuals to manifest a major cardiovascular event despite being classified as low risk by the Framingham risk estimate. Because half of first major coronary events occur in asymptomatic individuals, clinicians who want to implement appropriate primary prevention therapy must be able to accurately identify at-risk individuals.
Clinical decision making for primary prevention of CAD in asymptomatic individuals is traditionally guided by an initial estimate of the impact of single or clustered laboratory and physical factors as they relate to the risk of a coronary event. Preventive strategies are then modified and implemented after economic (personal, insurance provider, national impact) and individual (compliance, side effects) consequences of treatment versus no treatment are taken into account. Recommendations for diet, weight loss, and exercise offer little or no risk to the patient and yield significant long-term benefits. Most decisions for clinical (i.e., pharmacologic) intervention, specifically those related to lipid lowering, are driven by perception of risk and attainment of goals for a given individual that are derived from large studies applied to both heterogeneous and homogeneous populations.
Screening for subclinical atherosclerosis in an effort to better identify persons at risk for CAD has been of increasing interest during the past decade. Cardiac computed tomography (CT) has shown promise in this regard, with a significant base of research and recent guidelines that have been established to support its use in selected groups of patients. Its use to better identify patients who might benefit from initiation or intensification of risk factor modification efforts is paramount to prevention efforts. This chapter details the current and future role of cardiac CT. The following discussion examines the methods, value, and future of cardiac CT in overall CAD in primary risk stratification and preventive strategies.
Current Coronary Artery Disease Risk Stratification Guidelines
A primary recommendation of the major advisory bodies is that all adults should undergo an office-based assessment as the initial step to identify those at higher risk for CAD events. One approach endorsed by both the American Heart Association (AHA) and the American College of Cardiology (ACC) and adopted by the National Cholesterol Education Program (NCEP) Adult Treatment Panel III (ATP III) is to apply a modification of the risk prediction algorithm derived from the Framingham Heart Study that incorporates a patient’s age, total cholesterol concentration, high-density lipoprotein cholesterol (HDL-C) concentration, smoking status, and systolic blood pressure to estimate a person’s 10-year risk for development of a “hard” coronary heart disease (CHD) event (MI or CHD death).
Three levels of risk are defined on the basis of the probability of the occurrence of CHD in the next 10 years: <10% (low), 10% to 20% (intermediate), and >20% (high). Individuals with CAD risk <10% are at presumed low risk of coronary events and can be reassured about their risk status without further risk assessment testing. Those with low risk are to be offered general public health recommendations in the short term, and they can usually avoid further risk assessments for approximately 5 years. At the other end of the risk spectrum, high-risk patients are those with established CAD or other clinical forms of atherosclerotic disease, suffer from diabetes mellitus, or frequently are older patients with multiple other CAD risk factors in accordance with NCEP ATP III guidelines; high-risk asymptomatic people should have all CAD risk factors treated to reduce CAD-specific and total CVD risk. Finally, a sizable group of individuals fall into the intermediate-risk category. Patients in this group do not currently qualify for the most intensive risk factor interventions, yet they have one or more risk factors that exceed desirable levels or multiple borderline risk factors. Often, there is not a definitive need to begin or to intensify therapy; however, many such persons do have underlying subclinical atherosclerosis that would place them at higher future risk of CAD than global risk assessment would predict. Intermediate-risk patients are the most likely to benefit from further risk stratification testing, if it is feasible, practical, targeted, and effective at further defining risk or in motivating effective behavioral changes.
Determination of the intensity of treatment for primary prevention by use of global risk assessment algorithms such as the Framingham risk score (FRS) system employs data from population-based studies and does not take into account an individual’s actual burden of atherosclerotic disease, which is believed to be the main culprit in development of clinical CAD. Indeed, such assessments may fall short if they are solely relied on for management decisions for the individual patient as these risk estimates are significantly age dependent; as a result, there is a great tendency to underestimate risk in younger individuals, who may be more appropriate candidates for early initiation of aggressive preventive strategies to reduce risk for development of clinical CAD. In addition, the risk of a future CVD event in women is likely to be underestimated by these approaches, and nearly 90% of women younger than 70 years are considered low risk.
This point can be illustrated in a study by Akosah and colleagues, in which the authors assessed a simple question: How do NCEP guidelines classify young men and women presenting with an acute MI as their first manifestation of CAD? The study findings demonstrated that among adults aged <65 years with an acute MI, only 25% of patients would have met ATP III criteria, which are based on the FRS, for pharmacotherapy at the time of admission. The tendency of NCEP guidelines to underappreciate the risk for CAD was even more pronounced in women, with only 18% of women qualifying for pharmacotherapy for primary prevention; 58% of these patients had low-density lipoprotein cholesterol (LDL-C) concentration <130 mg/dL, and 40% had LDL-C <100 mg/dL.
Atherosclerosis and Coronary Artery Disease
Considering that atherosclerosis is the major underlying culprit for the development of clinical CAD, detection of individuals with subclinical atherosclerosis may aid in supplementing current global risk assessment approaches by more clearly identifying high-risk individuals who harbor advanced preclinical atherosclerosis. Screening studies to detect occult cancers, such as breast and colon cancer, are recommended in appropriate-risk adults to help prevent life-threatening conditions. Although atherosclerotic vascular disease accounts for more death and disability than all types of cancer, a screening tool to detect significant subclinical atherosclerosis and to target prevention of future cardiovascular events has not yet been adopted.
A minority of patients with CAD do not exhibit traditional risk factors, such as hypertension, elevated cholesterol, obesity, and smoking. In addition, many patients with such risk factors do not develop CVD. Furthermore, there is substantial variation in the severity of CAD at every level of risk factor exposure. This variation in disease is probably due to a number of factors, including genetic susceptibility, presence of intrinsic biochemical and extrinsic environmental risk factors that are yet to be identified, and duration of exposure to the specific level of risk factors.
Established noninvasive methods of evaluating CAD, such as stress testing, generally identify only patients with advanced atherosclerotic disease leading to a flow-limiting coronary stenosis and myocardial ischemia. The long-term risk of CAD, however, is more closely related to atherosclerosis plaque burden and stability than to the extent of a particular stenosis. There is growing interest in quantifying and characterizing atherosclerosis in its preclinical, pre–flow-limiting phase so that appropriate preventive strategies can be instituted before an adverse event occurs.
With understanding of the need for noninvasive tests to assess atherosclerotic plaque, cardiac CT has evolved rapidly, with increasing ability to visualize the amount of coronary artery calcium. Cardiac CT has been challenging, given rapid cardiac motion, small vessel diameters, tortuous anatomic patterns, and overlapping cardiac structures. Current 64-slice multirow detector computed tomography (MDCT) systems have faster gantry rotation speed, resulting in better temporal resolution and the better z -axis spatial resolution made possible by thin collimations with extensive volumetric acquisitions.
Non–Contrast-Enhanced Coronary Computed Tomography: Assessment of Coronary Artery Calcification
During the past decade, there has been marked increased interest in the clinical use of cardiac CT scanning to identify and to quantify the amount of coronary artery calcified plaque (CAC). Calcification of the atherosclerotic plaque occurs by an active process of mineralization with deposition of hydroxyapatite crystals and not simple mineral precipitation. It begins in the very early stages of atherosclerosis. Studies have demonstrated that electron beam tomography (EBT) is a highly reliable method for identification of arterial calcification with a high sensitivity for detection of significant atherosclerosis. Rumberger and colleagues have demonstrated a strong relationship ( r = 0.90) between CAC measured by EBT and direct histologic plaque areas in autopsy hearts. Although the total atherosclerotic plaque burden was associated strongly with the total calcium burden, not all plaques were found to be calcified. Moreover, within a given coronary artery, there is a poor correlation and wide variation between the degree of plaque calcification and extent of luminal stenosis on coronary angiography. This may be due, at least in part, to individual variations in coronary artery remodeling, whereby the luminal cross-sectional area or external vessel dimensions enlarge in compensation for increasing area of mural plaque.
Despite the lack of a site-by-site correlation between calcification and luminal stenosis, coronary calcium scores calculated by EBT give a close approximation of the total atherosclerotic burden. Because research has shown that burden of disease and cardiovascular risk are accounted for by more than focal luminal stenosis, non–contrast-enhanced cardiac CT is a potentially powerful tool for the identification of patients at risk.
Methods of Assessing Coronary Artery Calcium
Modalities for Coronary Artery Calcium Determination
In the near past, CAC had generally been assessed by EBT; however, with a rapid explosion of use of MDCT in recent years. This has been a widely used modality to assess the extent and severity of underlying coronary calcification. Neither modality requires intravenous administration of contrast material to determine CAC. In general, EBT uses a unique technology enabling ultrafast scan acquisition times in the high-resolution, single-slice mode with continuous, nonoverlapping slices of 3-mm thickness and an acquisition time of 100 msec/tomogram in a prospective manner. Electrocardiographic (ECG) triggering is done during end systole or early diastole at a time determined from the continuous ECG tracing recorded during scanning. Historically, the most common trigger time used is 80% of the R-R interval. However, this trigger occurs on or near the P wave during atrial systole, and the least cardiac motion among all heart rates occurs at 40% to 60% of the R-R interval.
The current generation of MDCT systems is capable of acquiring up to 320 sections of the heart simultaneously with ECG gating in either a prospective or retrospective mode. These MDCT systems have two principal modes of scanning, which depend on whether the patient on the CT couch is advanced in a stepwise fashion (axial, sequential, or conventional mode) or continuously moved at a fixed speed relative to the gantry rotation (helical or spiral mode). Coronary calcification is determined in axial mode with use of prospective ECG triggering at predetermined offset from the ECG-detected R wave. With prospective gating, the temporal resolution of the MDCT system is proportional to the gantry speed, which determines the time to complete one 360-degree rotation.
For reconstruction of each slice, data from a minimum of 180 degrees plus the angle of the fan beam are required (approximately 220 degrees of the total 360-degree rotation). The most commonly used 64-slice scanners have rotation gantry speeds up to 330 msec. MDCT imaging protocols vary among different camera systems and manufacturers. In general, 40 consecutive 2.5- to 3-mm-thick images are acquired per cardiac study. Calcified lesions are defined as two or three adjacent pixels with a tomographic density of >130 HU. Effective pixel size for a reconstruction matrix of 512 × 512 pixels with a common field of view of 26 cm is 0.26 mm 2 .
Measurement of Coronary Artery Calcium Burden
On non–contrast-enhanced cardiac CT, CAC is in general defined as a hyperattenuated lesion above a threshold of 130 HU with an area of three adjacent pixels (at least 1 mm 2 ). There are currently two CT calcium scoring systems widely used, the original Agatston method and the “volume” score method. The Agatston method involves multiplication of the calcium area by a number related to CT density and, in the presence of partial volume artifacts, can be variable. With this method, area for all pixels above a threshold of 130 HU is calculated at every 3-mm slice and multiplied by a density factor. Partial volume effects lead to higher peak values for small lesions (but not for large ones). On the other hand, the volume method developed by Callister and associates appears to somewhat resolve the issue of slice thickness and spacing by computing a volume above threshold. As a result, it appears to be less dependent on minor changes in slice thickness. However, our group has previously demonstrated in nearly 10,000 patients that there appeared to be an excellent correlation between the scoring methods, and they show similar characterization when applied properly. Both methods calculate lesion-specific scores within the left main, left circumflex, left anterior descending, and right coronary arteries and provide total scores for each artery and a sum total across all four arteries. An example of significant coronary calcium is shown in Figure 27-1 .
Multidetector Computed Tomography Compared With Electron Beam Tomography For Detection Of Coronary Artery Calcium
The comparability of MDCT- and EBT-derived CAC scores has been extensively explored. The MDCT protocols vary considerably in these studies, ranging from conventional CT to single-slice CT (with either retrospective or prospective gating) to MDCT. EBT imaging was performed with the standard protocol conventionally used in routine clinical practice. Coronary calcification was defined as >130 HU for EBT but varied from 90 to 130 HU for MDCT. Although high correlation coefficients were reported between EBT and MDCT CAC scores, there was significant variability in individual CAC scores (range, 17% to 84%). In general, the interscan agreement for the presence of CAC between EBT and 64-slice MDCT is excellent (99%). There was a significant linear relationship between the scores from the two scanners, and the interscanner variability between EBT and 64-slice MDCT was not significantly different. Bland-Altman analysis demonstrated a mean difference in scores of 8.3% by Agatston and 7.8% by volumetric calcium scoring. Compared with EBT, there were larger and more prevalent motion artifacts and larger mean Hounsfield units with 64-slice MDCT ( P < 0.001). At CAC scanning, 64-slice MDCT and EBT were comparable in Agatston and volumetric scoring. The interscan variability between scanners is similar to interscan variability of two calcium scores done on the same equipment. However, heart rate control was achieved for this study for calcium scores. Whether these results are repeatable without heart rate control needs to be further assessed.
Clinical Value of Coronary Artery Calcification In Asymptomatic Individuals
Efforts have been made to develop noninvasive diagnostic tools to help determine the extent of atherosclerosis in asymptomatic patients and to improve detection of those who would benefit from more intensive preventive therapies, such as lipid-lowering medication and aspirin. This potential of coronary CT in the risk assessment protocol and management strategy is in accordance with the current philosophy of the NCEP and other organizations that stress the importance of matching therapy to level of assessed risks. However, to establish the role of CAC testing in primary preventive strategies, important questions need to be answered. Is the information gained from coronary CT additive to assessments made by cheaper office-based estimations of risk? If so, which populations of patients are expected to benefit from testing?
Does Coronary Artery Calcium Independently Predict Coronary Artery Disease Events?
The likelihood of plaque rupture and the development of acute cardiovascular events is related to the total atherosclerotic plaque burden. Although controversy exists as to whether calcified or noncalcified plaques are more prone to rupture, extensive calcification indicates the presence of both plaque morphologies. There is a direct relationship between the CAC severity and the extent of atherosclerotic plaque; thus, the CAC score could be useful for risk assessment of asymptomatic individuals and potentially guide therapeutics.
Table 27-1 summarizes the findings of all major studies assessing the prognostic value of CAC burden among asymptomatic individuals. In general, there appears to be a consensus among all studies that CAC is an independent predictor of CAD adverse outcome as well as of all-cause mortality after traditional risk factors are taken into account. Among 1172 asymptomatic patients observed for 3.6 years after an initial EBT screening, no events occurred in patients with a normal study, and the negative predictive value was 99.8% in patients with a CAC score <100. These results showed a 5%, 7%, and 13% hard cardiac event rate in individuals with a CAC score ≥80, ≥160, and ≥600, respectively. The CAC score remained the best single predictor of risk after adjustment. Wong and colleagues also showed that the CAC score severity predicted subsequent cardiovascular events independent of age, gender, and patient risk factor profile. Raggi and coworkers studied more than 600 asymptomatic patients who were referred for screening EBT and then observed for 32 ± 7 months. Both the absolute CAC score and the relative CAC score percentiles adjusted for age and gender predicted subsequent death and nonfatal MI. Hard cardiac events occurred in only 0.3% of subjects with a normal EBT study, but this increased to 13% in those with a CAC score >400. A very high CAC score ≥1000 may portend a particularly high risk of death or MI (25% per year) in individuals who are not treated with standard secondary prevention measures.
| Author (Year) | Type of Study and Population | Follow-up (years) | Number of Events | Results |
|---|---|---|---|---|
| Arad et al (2000) | Observational study, referral based N = 1172; age, 53 ± 11 yr | 3.6 | 15 nonfatal MI, 21 revascularizations, 3 deaths | OR of 20 for CAC scores ≥160 compared with those with CAC scores <160 |
| Wong et al (2000) | Observational study, referral based N = 926; mean age, 54 yr | 3.3 | 6 nonfatal MI, 20 revascularizations, 2 CVA | Overall, patients with CAC score ≥271 had a risk ratio of 9 for a CHD event. |
| Raggi et al (2001) | Observational referral-based study N = 676; mean age, 52 yr | 2.7 | 21 nonfatal MI, 9 deaths | CAC score was predictive of hard CAD events, with an OR of 22 for the CAC score 90% percentile. |
| Kondos et al (2003) | Observational study, referral based N = 5635; age, 30-76 yr; 26% women | 3.1 | 37 nonfatal MI, 166 revascularizations, 21 deaths | RR of 124 in men with soft events in the highest quartile (CAC 170-7000) Higher CAC scores added incremental prognostic information to conventional CAD risk assessment in men for hard CHD events. |
| Shaw et al (2003) | Observation data series, referral based N = 10,377; age, 30-85 yr | 5 | 249 all-cause mortality | CAC score an independent predictor of mortality with RR 4.0 for score of 401-1000 |
| Greenland et al (2004) | Prospective population-based study N = 1312; age, >45 yr | 7 | 68 nonfatal MI, 16 deaths | HR of 3.9 for CAC score >301 CAC score able to modify predicted risk obtained from FRS alone (0.73 for FRS alone and 0.78 for FRS and CAC combined) |
| Arad et al (2005) | Prospective population-based study N = 4613; age, 50 to 70 yr | 4.3 | 40 nonfatal MI, 59 revascularizations, 7 CVA | RR for CAD events with CAC >100 was 11. Overall, it was superior to FRS in prediction of events (ROC curve of 0.79 versus 0.69; P = 0.006). |
| Vliegenthart et al (2005) | Prospective population-based study N = 1795; age, 62 to 85 yr | 3.3 | 40 nonfatal MI, 11 revascularizations, 38 CVA | Compared with those with CAC <100, the RR for events was 3.1, 4.6, and 8.3 for CAC 101-400, 401-1000, and >1000, respectively. There was a statistically significant high relative risk, >8, for those with CAC scores >1000 regardless of Framingham 10-year risk score ≤20% versus >20%. |
| Taylor et al (2005) | Prospective cohort study 1627 men, 356 women; age, 40 to 50 yr (Army based) | 3 | 9 ACS events | 2% of men with CAC had events versus 0.2% without CAC ( P <0.0001). Controlling for FRS, presence of CAC was associated with an independent 12-fold increase in relative risk. No events in women |
| LaMonte et al (2005) | Retrospective study 6835 men, 3911 women; age, 22-96 yr | 3.5 | 81 MI/CAD death, 206 revascularizations | Age-adjusted rates per 1000 person-years were computed according to 4 CAC categories: 0 CAC and incremental sex-specific thirds of detectable CAC. The rates were 0.4, 1.5, 4.8, and 8.7 for men and 0.7, 2.3, 3.1, and 6.3 for women. |
| Anand et al (2006) | Prospective study 510 asymptomatic type 2 diabetic subjects; age, 53 ± 8 yr | 2.2 | 22 (2 coronary deaths, 9 nonfatal MI, 3 ACS, 3 CVA, and 3 late revascularizations) | The overall rate of death or MI by CAC categories (<100, 101-400, 401-1000, and >1000) was 0 (n = 0), 2.6 (n = 2), 13.3 (n = 4), and 17.9% (n = 5), respectively ( P <0.0001). |
| Budoff et al (2007) | Observation data series, referral-based N = 25,253; mean age, 65 ± 11 yr | 6.8 | 510 all-cause deaths | Compared with those without CAC, the risk-adjusted relative risk ratios for CAC were 2.2-, 4.5-, 6.4-, 9.2-, 10.4-, and 12.5-fold for scores of 11 to 100, 101 to 299, 300 to 399, 400 to 699, 700 to 999, and >1000, respectively ( P <0.0001). |
| Detrano et al (2008) | Prospective population-based study | 3.4 | 162 CHD events (72 myocardial infarctions, 17 CHD deaths, 73 revascularizations) | Overall, the FRS-adjusted risk was 28% higher with doubling of CAC scores. CAC was equally predictive in all ethnic groups. |
| Becker et al (2008) | Prospective population-based study | 3.3 | 179 (65 cardiac death, 114 MI) | CAC score ≥75th percentile was associated with a significantly higher annualized event rate for MI (3.6% versus 1.6%; P <0.05). No cardiac events were observed in patients with CAC = 0. |
Larger studies have reported an approximately 10-fold increased risk with the presence of CAC. In one of the largest observational trials to date, Shaw and colleagues reported all-cause mortality among 10,377 asymptomatic patients (4191 women and 6186 men) who had a baseline EBT study and were then observed for 5.0 ± 3.5 years ( Fig. 27-2 ). Most subjects had cardiac risk factors including a family history of CAD (69%), hyperlipidemia (62%), hypertension (44%), and current cigarette smoking (40%). The CAC score was a strong independent predictor of mortality, with 43% additional predictive value contained within the CAC score beyond risk factors alone. Mortality significantly increased with increasing CAC score, within men and women separately as well as within each Framingham risk group (low-, intermediate-, and high-risk persons). In addition, in the South Bay Heart Watch study, 1196 asymptomatic patients were observed (median = 7.0 years), and it was demonstrated that the CAC score added predictive power beyond that of standard coronary risk factors and C-reactive protein (see Table 27-1 ). The results of the St. Francis Heart Study, which is a prospective registry of 5585 asymptomatic individuals, mirrored previous retrospective studies and confirmed the higher event rates associated with increasing CAC scores. CAC scores >100 were associated with relative risks of 12 to 32, thus achieving secondary prevention equivalent event rates >2%/year. The Rotterdam Heart Study investigated 1795 asymptomatic participants (mean age, 71 years) who had CAC and measured risk factors. During a mean follow-up of 3.3 years, the multivariate-adjusted relative risk of coronary events was 3.1 for calcium scores of 101 to 400, 4.6 for calcium scores of 401 to 1000, and 8.3 for calcium scores >1000. Similarly, in a younger cohort of asymptomatic persons, the 3-year mean follow-up in 2000 participants (mean age, 43 years) showed that coronary calcium was associated with an 11.8-fold increased risk for incident CHD (CAD) ( P <0.002) in a Cox model controlling for the FRS.
The Cooper Clinic Study included more than 10,000 adults who were 22 to 96 years of age and free of known CAD. During a mean follow-up of 3.5 years, 81 hard events (CAD death, nonfatal MI) occurred. Age-adjusted rates (per 1000 person-years) of hard events were computed according to four CAC categories: no detectable CAC and incremental sex-specific thirds of detectable CAC; these rates were, respectively, 0.4, 1.5, 4.8, and 8.7 (trend P <0.0001) for men and 0.7, 2.3, 3.1, and 6.3 (trend P <0.02) for women. The association between CAC and CAD events remained significant after adjustment for CAD risk factors. Of note, in the largest single cohort study, Budoff and colleagues showed risk-adjusted hazard ratios for total mortality ranging from 2.2 to 12.5 for CAC score categories of 11-100 to >1000 relative to 0, with CAC scores providing significant incremental information over risk factors. Finally, in a German study by Becker and coworkers, the extent of CAC was determined by MDCT in 924 patients (443 men, 481 women, aged 59.4 ± 18.7 years). During the 3-year follow-up period, the event rates for coronary revascularization, MI, and cardiac death in patients with volume scores above the 75th percentile were significantly higher compared with the total study group, and no cardiovascular events occurred in patients with scores of zero. Receiver operating characteristic (ROC) analysis demonstrated that it outperformed both PROCAM and Framingham models ( P <0.0001), in which 36% and 34% of MIs occurred in the high-risk cohorts, respectively.
The utility of CAC testing was also recently described in CAD-equivalent individuals, that is, those with diabetes mellitus. Risk factors and CAC scores were prospectively measured in 510 asymptomatic type 2 diabetic subjects (mean age, 53 ± 8 years; 61% men) without prior CVD, with a median follow-up of 2.2 years. In the multivariable model, the CAC score and extent of myocardial ischemia were the only independent predictors of outcome. ROC analysis demonstrated that CAC predicted cardiovascular events with the best area under the curve (0.92), significantly better than the United Kingdom Prospective Diabetes Study risk score (0.74) and Framingham score (0.60). The relative risk to predict a cardiovascular event for a CAC score of 101 to 400 was 10.13, and it increased to 58.05 for scores >1000 ( P <0.0001). No cardiac events or perfusion abnormalities occurred in subjects with CAC ≤10 Agatston units up until 2 years of follow-up.
These findings were nicely summarized by the AHA/ACC expert consensus document on coronary artery calcium scoring, which took into account many of these studies. Compared with patients with no detectable coronary calcium, the relative risk ratio for CAC 100-400, 401-1000, and >1000 was 4.3 (95% CI, 3.5-5.2; P <0.0001), 7.2 (95% CI, 5.2-9.9; P <0.0001), and 10.8 (95% CI, 4.2-27.7; P <0.0001), respectively. Importantly, patients with CAC score of zero have a very low rate of CAD death or MI (0.4%) during 3 to 5 years of observation ( Fig. 27-3 ).
At the same time, critics tend to point to limitations such as potential study generalizability of self-referral cohorts, validity of the risk factor measures and resultant multivariable models used in the studies, and risk of test-induced bias. However, these concerns have been addressed by a report from the Multi-Ethnic Study of Atherosclerosis (MESA), a population-based cohort, which reported the utility of CAC scores in predicting future events. According to Detrano and coworkers, among nearly 6800 asymptomatic individuals observed for a median of 41 months, the hazard ratio for future hard CAD events (MI or MI-related death) with CAC 1-100 versus CAC = 0 was 5.3 (95% CI = 2.4-11.7; P <0.0001). The respective hazard ratios with CAC 101-300 and >300 were 10.8 (4.8-24.2; P <0.0001) and 12.0 (5.4-26.5; P <0.0001), with a 5-year cumulative incidence of CAD events directly associated with higher CAC scores, exceeding 10% in those with scores >300 ( Fig. 27-4 ). These risk ratios are very much similar to those of published studies and confirm the pooled summary findings previously reported and lay to rest any concern about the prognostic value of CAC testing.
What Is the Value of Testing for Coronary Artery Calcium Across Ethnic and Racial Groups?
Most of the published data to date have related to white populations; however, two studies have addressed the value of CAC in other ethnic groups. First, Nasir and coworkers, in nearly 15,000 ethnically diverse self-referred patients, assessed the role of CAC for the prediction of all-cause mortality. In comparison of prognosis by CAC scores in ethnic minorities and non-Hispanic whites, relative risk ratios were highest for African Americans, with scores ≥400 exceeding 16.1 ( P < 0.0001). Hispanics with CAC scores ≥400 had relative risk ratios from 7.9 to 9.0; Asians with CAC scores ≥1000 had relative risk ratios 6.6-fold higher than those of non-Hispanic whites ( P <0.0001). Second, the utility of CAC testing has also been reported in the prospective MESA study. According to Detrano and coworkers, the risk associated with a doubling of the CAC score (a 1-unit increase in log 2 [CAC + 1]) for a hard CAD event was 1.3 (1.2-1.4) in whites, 1.5 (1.3-1.7) in African Americans, 1.3 (1.1-1.5) in Hispanics, and 1.4 (1.1-1.8) in Chinese. These findings firmly establish that CAC scores provide significant information in all four major ethnic groups in the United States.
Does Coronary Artery Calcium Add Incremental Value to Global Risk Estimates?
The extent of CAC has been shown in several studies to predict cardiac events in symptomatic and asymptomatic individuals. However, decisions about the predictive utility of new tests should be based on the additional utility of a new test for risk prediction. The most important question about use for primary CAD risk stratification is whether it is predictive above and beyond the current standard risk assessment method of choice, the FRS, which is an inexpensive, easily available, and office-based tool. One way to determine additive utility of a new test is through the use of ROC curve analyses. The ROC curve is a plot of true-positive rate versus false-positive rate over the entire range of possible cutoff values. The area under the ROC curve (AUC) ranges between 1.0 for the perfect test and 0.5 for the useless test.
Studies comparing predictive capacity of conventional and newer biomarkers for prediction of cardiovascular events consistently demonstrate that adding a number of newer biomarkers (such as C-reactive protein, interleukins, and other proposed risk stratifiers) changes the C-statistic by only 0.009 ( P = 0.08). Small changes such as these in the C-statistic suggest limited or modest improvement in risk discrimination with additional risk markers. However, CAC scanning has been shown to markedly improve the C-statistic, suggesting robust improvement in risk discrimination ( Table 27-2 ).
| Study | C-Statistic with Risk Factors, FRS | C-Statistic with Risk Factors, FRS plus CAC | P Value |
|---|---|---|---|
| Arad et al (St. Francis Heart Study) | 0.69 | 0.79 | 0.0006 |
| Budoff et al | 0.611 | 0.813 | <0.0001 |
| Anand et al | 0.60 | 0.92 | <0.0001 |
| Becker et al | 0.68 | 0.77 | <0.01 |
| Detrano et al (MESA) | 0.77 | 0.82 | <0.001 |
Raggi and colleagues were among the first to assess the added contribution of CAC over and above the FRS. In a study of more than 10,000 asymptomatic individuals observed nearly 5 years, the C-statistic (from ROC curve analyses) for FRS in estimating risk of all-cause death was 0.67 (95% CI, 0.62-0.72; P <0.0001) for women and 0.68 (95% CI, 0.64-0.73; P <0.0001) for men. When CAC was added to this analysis, the C-statistic increased to 0.75 (95% CI, 0.70-0.80) for women ( P <0.0001) and 0.72 (95% CI, 0.68-0.77) for men ( P <0.0001), indicating a significant improvement in mortality prediction.
In a similar fashion, Greenland and coworkers found that the ROC curve for prediction of CAD death or nonfatal MI was 0.68 for FRS plus CAC, which was significantly greater than that of the FRS alone (0.63; P <0.001), with increasing levels of CAC associated with greater risk within each FRS group. Importantly, those in the intermediate-risk FRS group with high CAC scores had event rates as high as or higher than those of persons within the high-risk FRS group with lower CAC scores ( Fig. 27-5 ). The recent population-based St. Francis Heart Study of 5585 asymptomatic individuals confirmed the findings of previous reports. The CAC score predicted CAD events independently of standard risk factors and C-reactive protein ( P = 0.004), was superior to the FRS in the prediction of events (area under ROC curve of 0.79 ± 0.03 versus 0.69 ± 0.03; P = 0.0006), and enhanced stratification of those falling into the Framingham categories of low, intermediate, and high risk ( P <0.0001). Similarly, an improvement in AUC from 0.77 to 0.82 was noted in the landmark MESA study.
