Introduction
Coronary artery calcification (CAC) is a strong predictor of coronary heart disease (CHD) and a well-validated risk stratification tool [1–3]. CAC can occur in the advanced process of atherosclerosis and reflects a linear estimate of the total plaque burden of coronary artery atherosclerosis. The presence of greater CAC is associated with increasingly higher risk of adverse cardiovascular events and all-cause death [1, 2, 4, 5] and thus patients with excessive CAC should be considered as high-risk patients, whereas a CAC score of zero is associated with a very low event rate [6]. CAC assessment in asymptomatic adults was incorporated in the American College of Cardiology Foundation/American Heart Association (AHA) guidelines with class IIa, level of evidence B status in 2010 and was recommended for patient risk assessment in asymptomatic adults at intermediate risk and for all patients older than 40 years with diabetes mellitus [7]. Clinically, the importance and use of CAC assessment has been increasing significantly. Not only the CAC score itself but also the CAC score in addition to traditional risk factors increases the diagnostic accuracy of the cardiovascular risk, especially in the asymptomatic population. One of the major areas of progress of research on CAC is more than 10 years of follow-up studies, which have been recently reported by several investigators. It is important for clinicians to know that the CAC score can provide a predictive value of coronary atherosclerosis and it could give us the long-term prognosis of individuals. In this review, we describe the key data supporting the use of CAC assessment based on publications from the past few years.
The Power of CAC in Risk Stratification
Can we predict the patient's cardiovascular events in advance? If we can do so, the number of patients with fatal cardiovascular disease (CVD) could decrease. The Framingham risk score (FRS) or the new atherosclerotic CVD (ASCVD) risk score of the American College of Cardiology (ACC)/AHA has to date been used as an established risk scoring method for primary prevention based on conventional risk factors for cardiovascular events in asymptomatic individuals. The FRS predicted the 10-year cardiovascular risk of individuals and categorized individual risk of developing CVD as low (10-year risk of less than 10%), intermediate (10-year risk of 10–20%) and high (10-year risk of more than 20%) [8]. Although the FRS is an important advance in the primary prevention of CVD, the risk stratification algorism is not perfect [9–11]. The limitations of the FRS include a substantial underestimation of lifetime risk (especially in women), misclassification of high-risk individuals as low-risk or intermediate-risk individuals, and misclassification of very low risk individuals into a higher risk stratum [10]. In a contemporary novel study of more than 307,000 participants, the new pooled cohort equation (ASCVD risk score) showed gross overestimation of risk, so virtually every group is fourfold less likely to experience an event than anticipated. This leads to significant overtreatment, and can contribute to low adherence [12]. CAC scoring can improve the risk stratification of CVD over the pooled cohort equation and the FRS (Figure 1A) [3, 13–16]. Recently, several studies have reported the incremental prognostic value of CAC over traditional risk factors after a long follow-up. Kelkar et al. [17] revealed that CAC could effectively identify high-risk women with low to intermediate FRS and improve risk detection algorisms based on traditional risk factors (net reclassification improvement of CAC 0.155, P=0.002). Hoffmann et al. [18] demonstrated that CAC improved discrimination and risk reclassification for CHD and CVD beyond traditional risk factors in an asymptomatic population in the Framingham Offspring Study. They found that addition of log CAC to the FRS significantly increased the discriminatory ability for major CHD over 8 years of follow-up (c-statistic for FRS alone 0.78 vs. 0.82 for FRS + log CAC, P<0.05) and CAC could reclassify more than half of the participants at intermediate 10-year risk of major CHD based on the FRS as having low risk (observed event rate 0%) or high risk (observed event rate 8%) [18]. Yeboah et al. [16] found that CAC in addition to the FRS provided superior discrimination especially in intermediate-risk individuals after 9 years of follow-up as opposed to thoracic aorta calcium, aortic valve calcification, mitral annular calcification, pericardial adipose tissue volume, or liver attenuation in addition to the FRS for incident CHD/CVD, and moreover, afforded the greatest decrease in the area under the curve (AUC) for detecting incident CHD as 0.712, 0.645, 0.651, 0.643, 0.643 and 0.641 respectively. McClelland et al. [19] reported a novel risk score to estimate 10-year CHD risk using CAC and traditional risk factors based on the Multi-Ethic Study of Atherosclerosis (MESA) data with validation in the Heinz Nixdorf Recall (HNR) study and the Dallas Heart Study (DHS). They demonstrated that CAC in addition to traditional risk factors could improve the risk prediction compared with traditional risk factors alone (c-statistic 0.80 vs. 0.75, P<0.0001). External validation in both the HNR study and the DHS provided evidence of very good discrimination and calibration, and Harrell's c-statistic was 0.779 in the HNR study and 0.816 in the DHS [19]. Moreover, Blaha et al. [20] reported the clinical implication of the location and distribution pattern of CAC in addition to the CAC score for the prediction of CHD/CVD events. They revealed that addition of the number of vessels with CAC significantly improved the capability to predict CHD and CVD events in survival analysis (hazard ratio [HR] 1.9–3.5 for four-vessel CAC vs. one-vessel CAC), AUC analysis (c-statistic improvement of 0.01–0.033), and net reclassification improvement analysis (categoryless net reclassification improvement of 0.10–0.45) [20].
Examples of the Utility of CAC Scoring in Patient's Risk Assessment.
(A) A 60-year-old Caucasian woman. She had a history of hypertension and dyslipidemia without antihypertensive drugs and antidyslipidemia drugs. The total cholesterol level was 248 mg/dL and the high-density lipoprotein (HDL) cholesterol level was 44 mg/dL. The systolic blood pressure was 150 mmHg. The coronary artery calcification (CAC) score was 867. CAC could be detected in all three vessels. The left anterior descending artery showed significant stenosis (white arrows). The estimated 10-year coronary heart disease (CHD) event risk for this patient with this risk factor profile including CAC was 17.8%. The estimated 10-year risk of having a CHD event for this patient without inclusion of CAC would be 6.3%. (B) A 62-year-old African American man. He had a history of hypertension and dyslipidemia with antihypertensive drugs and antidyslipidemia drugs. He also had a family history of myocardial infarction. The total cholesterol level was 179 mg/dL and the HDL cholesterol level was 54 mg/dL. The systolic blood pressure was 136 mmHg. The CAC score was zero. There was no CAC in all three vessels. The estimated 10-year CHD event risk for this patient with this risk factor profile including CAC was 4.2%. The estimated 10-year risk of having a CHD event for this patient without inclusion of CAC would be 11.0%.
These studies and others demonstrate that we can predict patient's cardiovascular events robustly if we use the CAC score for patient risk stratification. At present, the CAC score in addition to traditional risk factors is the best method for patient risk stratification and risk prediction for CHD and CVD. The new ACC/AHA prevention guidelines state “assessing CAC is likely to be the most useful of the current approaches to improving risk assessment among individuals found to be at intermediate risk after formal risk assessment” [21].
CAC Score Greater than Zero Equals Increased Risk of Atherosclerosis
A CAC score greater than zero indicates a greater risk of coronary artery disease (CAD) and CHD. As the CAC score increases, the risk of CAD and CHD increases [1, 22–24]. Long term follow-up data of more than 10 years in terms of CAC have recently been validated in multiple studies, demonstrating the significant association between increased CAC and higher incidence of outcomes among a population-based cohort [19, 25, 26], the general population [27], the elderly [27], women [17, 27], and in individuals with low risk factors [17, 28] and diabetes [29] or family history [30]. Shaw et al. [25] reported the ability of CAC to predict long-term mortality in 9715 asymptomatic patients. They revealed that overall mortality was 3, 6, 9, 14, 21 and 28% respectively for CAC subgroups with scores of 0, 1–10, 11–100, 101–399, 400–999, and 1000 or greater (P<0.001) at a mean follow-up of 14.6 years and the CAC score was highly predictive of all-cause death in Cox models after adjustment for CAD risk factors (P<0.001) [25]. Valenti et al. [26] reported that a CAC score greater than zero was the strongest predictor of death beyond cardiovascular risk factors, the FRS and the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) score, and was independently associated with nearly threefold risk of death (HR 2.67, 95% confidence interval [CI] 2.29–3.11). Compared with the base models of the FRS or the NCEP ATP III score alone, discrimination improved significantly when CAC score greater than zero was added to the FRS or the NCEP ATP score III (AUC 0.64 vs. 0.71 and 0.64 vs. 0.72 for the FRS and the NCEP ATP III score respectively; P<0.001 for both) [26]. Among 13,092 asymptomatic individuals undergoing CAC scan, our group demonstrated that CAC was strongly associated with risk of death during a median follow-up of 11 years in young and middle-aged groups, in both men and women, while the risk predictive value of CAC was slightly reduced in older individuals [27]. We reported that in 5584 asymptomatic and very low CVD risk individuals, mortality increased according to increasing CAC scores of 1–99 (HR 1.9, 95% CI 1.2–3.1), 100–399 (HR 2.1, 95% CI 1.2–3.6) and 400 or greater (HR 2.8, 95% CI 1.6–4.8) compared with that in patients with a CAC score of zero during a mean follow-up of 10.4 years [28]. In MESA, we also reported that a CAC score greater than zero was a strong and graded predictor of incident ASCVD at a median follow-up of 10.2 years [31]. In comparison to a CAC score of zero, CAC 1-100 (HR 2.1, 95% CI 1.6–2.6), 101-300 (HR 3.1, 95% CI 2.4–4.0) and more than 300 (HR 4.5, 95% CI 3.5–5.8) were all predictive of ASCVD events at 10 years [32]. Moreover, in a meta-analysis, Bavishi et al. [33] revealed a significant relationship between the magnitude of CAC and stress-induced myocardial ischemia. They showed a stepwise increase in the frequency of ischemia according to CAC abnormality. The average frequency of myocardial ischemia was 6.6% among individuals with a CAC score of zero and 23.6% among those with a CAC score of 400 or greater. However, there was a wide variance in the frequency of ischemia ranging from 0 to 24.1% among patients with a CAC score of zero, and ranging from 12.4 to 57.1% among patients with a CAC score of 400 or greater. Compared with those with a CAC score of zero, the pooled odds ratio (OR) for myocardial ischemia increased substantially for each increasing category of CAC abnormality: OR 1.7 (95% CI 1.04–2.2) for a for CAC score of 1–100; OR 3.3 (95% CI 1.4–8.2) for a CAC score of 101–399; and OR 6.9 (95% CI 3.5–13.4) for a CAC score of 400 or greater [33].
What are the risk factors for progression of CAC? Studies [34–39] have revealed the association between CAC progression and risk factors such as greater age, male sex, Caucasian race, higher body mass index, history of hypertension, dyslipidemia, diabetes, and metabolic syndrome, being a smoker, and a family history of myocardial infarction. Two studies have recently reported the clinical importance of family history of CHD for predicting the incidence and progression of CAC in asymptomatic individuals [30, 40]. Mulders et al. [40] reported that individuals with a family history of premature CAD had an increased risk of increased CAC (OR 2.23, 95% CI 1.48–3.36; P<0.05) compared with those without a family history, and the event rate was also low in those without a family history after 3.5 years of follow-up. Knapper et al. [30] revealed the prognostic utility of CAC scoring among cohorts of young and older patients with and without a family history of CAD. For patients older than 60 years with a family history of CAD, there was a significant decrease in the AUC with CAC over CAD risk factors (AUC 0.539 vs. 0.725, P<0.001); however, a decrease was not seen among patients younger than 60 years with a family history of CAD (AUC 0.636 vs. 0.626, P=0.67) [30]. Furthermore, a study of the relationship between air pollution and CAC has been reported from MESA. Kaufman et al. [41] revealed that fine particulate matter less than 2.5 µm in diameter (PM2.5) and nitrogen oxides (NOx) were associated with progression of CAC in the 10-year prospective cohort study. After adjustment for potential confounders, CAC progressed by 4.1 Agatston units per year (95% CI 1.4–6.8 Agatston units per year) for each 5 µg PM2.5/m3 increase and by 4.8 Agatston units per year (95% CI 0.9–8.7 Agatston units per year) for each 40-ppb NOx increase.
It is important to recognize that while statin therapy may induce slight progression of CAC (thought to be by stabilizing plaque), significant CAC progression remains a major risk factor for cardiovascular events. In a large cohort of 4609 asymptomatic patients, after adjustment for baseline score, age, sex, and time between scans, CAC progression was associated with a 3.34-fold risk of all-cause death (HR 3.34, 95% CI 2.65–4.21, P<0.0001) [42]. This study observed graded relationships of CAC progression and CHD event risk, strongly suggesting that the functions are linear, with greater CAC progression associated with greater risk. MESA studied 6778 people who had baseline and follow-up CAC scans approximately 2.5 years apart [24]. The median follow-up duration from the baseline was 7.6 years (maximum 9.0 years). Among participants with baseline CAC, those with annual progression of more than 300 units had adjusted HRs of 3.8 (1.5–9.6) for total CHD and 6.3 (1.9–21.5) for hard CHD compared with those without progression. This demonstrates that progression of CAC is associated with total and hard CHD risk; these relationships remained significant after adjustment for risk factors and baseline calcium level.
The Impact of Zero CAC Score
Nothing is more reassuring than a CAC score of zero for clinicians in patient risk stratification. Clinical evidence of the association between a CAC score of zero and CHD has been reported. A clinical asymptomatic population with a CAC score of zero can be considered as having very low risk of CHD (Figure 1B) [1, 22, 43, 44]. Recently, the long-term prognostic value of a CAC score of zero for asymptomatic individuals has been described in several studies. Valenti et al. [26] reported that a CAC score of zero conferred a 15-year warranty period against death for individuals at low to intermediate risk by the FRS and the NCEP ATP III score that was unaffected by age and sex. A CAC score of zero showed the lowest mortality rates among the low-risk categories such as FRS less than 10%, NCEP ATP III score less than 10%, no cardiovascular risk, and a CAC score of zero. The risk of all-cause death was greater in individuals with a CAC score greater than zero plus a low FRS or a low NCEP ATP III score (CAC score greater than zero plus low FRS, HR 3.3, 95% CI 2.49–4.32; CAC score greater than zero plus low NCEP ATP III score, HR 3.09, 95% CI 2.45–3.90) compared with those with a CAC score of zero plus a high FRS or a high NCEP ATP III score (CAC score of zero plus high FRS, HR 2.8, 95% CI 2.05–3.92; CAC score of zero plus high NCEP ATP III score, HR 2.94, 95% CI 2.15–4.01). They demonstrated that a CAC score of zero was associated with a vascular age that was 30 years less than chronological age for older individuals and was associated with a significantly lower annual mortality rate than for the equivalent chronological age category [26]. Similarly, our group demonstrated that the risk of death in patients without evidence of CAC was significantly lower compared with that in the general US population across various age groups regardless of sex [27]. A CAC score of zero was shown to be a stronger negative risk predictor for all CAD and CVD events after a mean follow-up of 10.3 years among negative atherosclerotic risk factors such as carotid intima-media thickness less than the 25th percentile, absence of carotid plaque, brachial flow-mediated dilation change of more than 5% change, ankle-brachial index greater than 0.9 and less than 1.3, high-sensitivity C-reactive protein level less than 2 mg/L, homocysteine level less than 10 µmol/L, N-terminal pro-brain natriuretic peptide level less than 100 pg/mL, no microalbuminuria, no family history of CHD (any/premature), absence of metabolic syndrome, and a healthy lifestyle [45]. A CAC score of zero resulted in the greatest reduction in posttest risk among all negative risk markers and had stable risk factor–adjusted diagnostic likelihood ratios across adverse clinical situations: 0.36 (SD, 0.09) in men and 0.46 (SD, 0.12) in women for CHD events; 0.49 (SD, 0.10) in men and 0.59 (SD, 0.12) in women for CVD events [45].
In contrast, a CAC score of zero does not always guarantee the long-term lowest mortality rate in individuals with diabetes mellitus [29]. Among 9715 asymptomatic individuals, including 810 diabetic individuals, the rate of all-cause death was similarly low between diabetic and nondiabetic individuals with a CAC score of zero during the initial 5 years (2.6 vs. 1.2%, P=0.06). However, at 15 years of follow-up, the concomitant existence of diabetes and a CAC score of zero was associated with an almost 2.5-fold increased risk of death compared with no diabetes with a CAC score of zero [29]. New recommendations from the ACC suggest CAC assessment in the setting of asymptomatic individuals with diabetes, as this affords the best risk stratification of the current assessment tools [46].
As described, a CAC score of zero implies the lowest risk of cardiovascular events, especially in nondiabetic individuals, for extended periods. Consequently, how can we maintain a CAC score of zero? In MESA, the proportion with a persistent CAC score of zero was examined during the median 9.6 years of follow-up in 1850 individuals with a CAC score of zero at the baseline. Participants with a CAC score of zero were significantly likelier to be younger, be female, and have fewer traditional risk factors; meanwhile, there was no single risk factor or specific low-risk factor phenotype that markedly improved the discrimination of a persistent CAC score of zero over demographic variables. A CAC score of zero may be predominantly influenced by the long-term maintenance of low risk factors for CVD or genetic factors rather than the absence of any specific risk factors in late adulthood [47]. Furthermore, as might be expected, unhealthy lifestyle habits could be a major contributor to CAC [48]. A combination of regular exercise, healthy diet, avoidance of smoking, and weight maintenance was associated with lower coronary calcium incidence, slower calcification progression, and lower all-cause mortality over 7.6 years [49]. As a specific solution, Imran et al. [50] recommended daily walking. They reported the inverse association between physical activity and CAC. The more than 15 to 22.5 metabolic equivalent hours per week group had a 46% lower prevalence of CAC compared with the reference group (<3.75 metabolic equivalent hours per week) after adjustment for age, sex, race, smoking, alcohol use, total physical activity, and familial clustering [50]. Habitual physical activity can prevent the development of CAD. Daily healthy food choices could also have a great impact on arterial health. Miedema et al. [51] reported that higher intake of fruit and vegetables during young adulthood was associated with lower odds of prevalent CAC after 20 years of follow-up. Spring et al. [52] reported that healthy lifestyle change during young age (18–30 years old) could be associated with decreased odds of detectable CAC and lower intima-media thickness in middle age.
The Utility of the CAC Score for Patient Treatment
In 2013 the ACC/AHA released updated CVD prevention guidelines [21, 53]. These guidelines changed the outcome from CHD to ASCVD including stroke. Moreover, the guidelines moved away from LDL cholesterol level and instead recommended a statin for individuals with a 10-year ASCVD risk of greater than 7.5%, which was lower than the former threshold, and the numbers of individuals eligible for statin therapy increased greatly. With the new guidelines, it is clear that a lot of future ASCVD events could be prevented; however, there could potentially be overestimation of patients at lower ASCVD risk [54–56]. DeFilippis et al. [55] showed the discriminative capability of the new 2013 guidelines in 4227 MESA participants aged 50–74 years and without diabetes at the baseline. They revealed the new guidelines overestimated cardiovascular events (predicted events 9.16% vs observed events 5.16%) and 78% discordance. Discordance between observed and expected risk was found throughout the risk continuum, including those at moderate risk. In men with an ASCVD risk score of 7.5–10%, the actual event rate was only 3.0% (predicted events 8.7% vs observed events 3.0%). Among women with an ASCVD risk score of 7.5–10%, the actual event rate was only 5.1% (predicted events 8.7% vs. observed events 5.1%) [55]. It is easy to image that risk overestimation could lead to increased use of preventive medications such as statin therapy, potentially exposing some patients to the unnecessary risks of these drugs and result in greater health care costs. Nasir et al. [56] evaluated the utility of the CAC score in reclassifying ASCVD populations by each risk stratum in which statins were recommended according to the guidelines in 4758 MESA participants. According to the guidelines, 2377 participants were recommended for moderate-intensity to high-intensity statin therapy. However, 41% of the 2377 participants had a CAC score of zero, with only 5.2 events per 1000 person-years. Among 589 participants considered for moderate-intensity statin, 338 (57%) had a CAC score of zero, with an ASCVD event rate of only 1.5 per 1000 person-years. From these results, almost 50% of the patients recommended for statin treatment had low event rates and actually had low risk (<7.5% 10-year risk). Thus a CAC score of zero could reclassify approximately half of candidates as not eligible for statin therapy [56].
A concern about CAC scoring is that it might lead to increased unnecessary downstream testing and intervention and increased health care costs for the public [57]. In the Early Identification of Subclinical Atherosclerosis by Noninvasive Imaging Research (EISNER) study [31], the rates of additional cardiovascular testing, invasive coronary angiography, and coronary revascularization were higher in participants with severely elevated CAC scores; however, CAC scoring showed no significant increase of overall estimated health care expenditures because a minimal CAC score (CAC score less than 10) or a CAC score of zero was associated with significantly lower rates of subsequent cardiovascular testing and costs. Individuals with a CAC score of zero showed lower rates of initiation of new lipid-lowering medication, in contrast, a progressive increase in new lipid-lowering medications occurred with increasing baseline CAC scores, with 19, 35, 43 and 65% respectively for CAC subgroups with scores of 0, 1–99, 100–399, and 400 or greater (P<0.001) [58].
In the 2013 guidelines, CAC scores of either greater than the 75th percentile for age and sex or 300 or greater were considered as high risk and warrant high-dose statin therapy. On the basis of studies from MESA, a CAC score of more than 100 was more predictive of events than a score greater than the 75th percentile, and indicates high cardiovascular risk, so we recommend use of a CAC score greater than 100 rather than either a CAC score of greater than 300 or greater than the 75th percentile [23]. Figure 2 shows our suggested flowchart for primary prevention of ASCVD. To summarize, individuals with a 10-year risk of 2.5–7.5% should be evaluated for the CAC score. If they have a CAC score greater than 100, they should be treated with statins as having a high risk of ASCVD. In addition, the CAC score may be promising for guiding aspirin use for the primary prevention of ASCVD. On the basis of an assessment of 4229 individuals from the MESA population without diabetes, Meiedema et al. have reported that individuals with a CAC score of 100 or greater had an estimated net treatment benefit from aspirin, whereas individuals with a CAC score of zero were estimated to be two to four times likelier to experience a major bleed from aspirin use.
Summary of the Proposed Protocol for Including Coronary Artery Calcification (CAC) Scores When Determining Atherosclerotic Cardiovascular Disease (ASCVD) Risk and Preventive Treatment. Individuals with Diabetes Mellitus (DM) or a Low-density Lipoprotein (LDL) Cholesterol Level of 190 mg/dL or Greater Should Be Treated with Statins.
Nondiabetic individuals with LDL cholesterol levels of 70–189 mg/dL first need to be assessed for ASCVD risk, and a treatment strategy should be decided on according to risk stratification.
CAC can robustly identify individuals who could be benefit from antiatherosclerotic therapies and aspirin and can also identify those who may not need treatment.
Conclusion
In this review we have described the clinical significance of CAC based on the current available data. In long-term follow-up studies, CAC in addition to traditional risk factors could adequately classify patients at risk compared with traditional risk factors alone. With increasing CAC scores, there is a greater likelihood of myocardial ischemia, and the risk of CVD events also increases significantly. In contrast, those with a CAC score of zero, especially without diabetes, could be considered as having very low risk with a 15-year warranty. For treatment, stratification according to the results of CAC scoring is robust to identify individuals who will benefit from antiatherosclerotic and aspirin therapy. The clinical role of the CAC score has been solidified as part of our 2010 screening and 2013 cholesterol guidelines. The CAC score will be likely play an increasingly important role in health care management, especially among the preventive community.