Evolving Role of Calcium Density in Coronary Artery Calcium Scoring and Atherosclerotic Cardiovascular Disease Risk

Abstract

Coronary artery calcium (CAC) is a specific marker of coronary atherosclerosis that can be used to measure calcified subclinical atherosclerotic burden. The Agatston method is the most widely used scoring algorithm for quantifying CAC and is expressed as the product of total calcium area and a quantized peak calcium density weighting factor defined by the calcification attenuation in HU on noncontrast computed tomography. Calcium density has emerged as an important area of inquiry because the Agatston score is upweighted based on the assumption that peak calcium density and atherosclerotic cardiovascular disease (ASCVD) risk are positively correlated. However, recent evidence demonstrates that calcium density is inversely associated with lesion vulnerability and ASCVD risk in population-based cohorts when accounting for age and plaque area. Here, we review calcium density by focusing on 3 main areas: 1) CAC scan acquisition parameters; 2) pathophysiology of calcified plaques; and 3) epidemiologic evidence relating calcium density to ASCVD outcomes. Through this process, we hope to provide further insight into the evolution of CAC scoring on noncontrast computed tomography. (J Am Coll Cardiol Img 2022;15:1648–1662)

Coronary artery calcium (CAC) is a specific marker of coronary atherosclerosis that strongly associates with atherosclerotic cardiovascular disease (ASCVD) risk.¹ CAC evaluation from noncontrast computed tomography (CT) is a widespread and reproducible tool used to guide risk assessment and treatment, particularly when there is uncertainty regarding an individual’s 10-year risk for an ASCVD event.² The Agatston method is a successful and simple scoring algorithm to quantify CAC and estimate the extent of coronary artery disease (CAD) per atherosclerotic lesion, which is expressed as the product of total calcified plaque area in millimeters squared and a quantized peak calcium density weighting factor.³ Calcium density reflects plaque attenuation and is measured in HU, thus the CAC score is directly and positively related to calcium density.³ However, although the CAC score is weighted upward for higher peak calcium density on a per-plaque basis, it is well established that calcium density is inversely associated with lesion vulnerability and ASCVD risk in population-based cohorts when accounting for age and plaque area.⁴ These observations indicate that there are several opportunities for evolving the contribution of calcium density in CAC scoring to further enhance its role in ASCVD risk assessment, treatment, and prevention.

This review focuses on our growing understanding of calcium density and its role in CAC scoring for characterizing coronary atherosclerosis and ASCVD risk (Central Illustration). We first discuss imaging strategies for measuring calcium density to assess plaque stability and monitor plaque progression. We then summarize the association between demographics of increased calcium density and its association with age, sex, race/ethnicity, and upstream risk factors. Next, we examine the evidence on the association between calcium density and ASCVD outcomes as well as the outlook and implications of updating scoring algorithms for measuring coronary plaque burden. Through this process, we hope to advance the discussion of leveraging CAC density measurement for characterizing coronary atherosclerosis and improving precision in ASCVD risk assessment.

CENTRAL ILLUSTRATION. The Evolving Role of Calcium Density in CAC Scoring.

The current Agatston CAC scoring system could potentially be improved on by considering mean vs peak calcium density among individuals with low CAC burden, using different measures of calcium density variability, and further understanding the relationship between traditional ASCVD risk factors and calcium density phenotypes. ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; CAC = coronary artery calcium; HDL-C = high-density lipoprotein cholesterol; SBP = systolic blood pressure.

THE DEMOGRAPHICS OF INCREASED CALCIUM DENSITY

CAC DENSITY AND AGE.

Calcium density strongly and positively associates with age,⁵ which corresponds with the biological framework and natural progression of atherosclerosis. Approximately one-half of U.S. (Western) adults have CAC by middle-age, and the prevalence of CAC surpasses 90% by 80 years of age.^6–8 However, calcium density and/or CAC volume may both contribute to such age-dependent trends, therefore, use of the Agatston score to measure CAC progression through adulthood may become problematic for understanding the underlying pathobiology of atherosclerosis and calcification. Among more than 10,000 primary prevention patients with prevalent CAC in the CAC Consortium, peak calcium density and mean calcium density both positively associate with age (Figure 1). Furthermore, the positive relationship between biological age and calcium density has been shown in the MESA (Multi-Ethnic Study of Atherosclerosis) after adjustment for sex and CAC volume.⁹ Thus, calcium density independently associates with age and overall appears to reflect and measure the length of time that an individual has had coronary heart disease (CHD).

FIGURE 1. Association of Peak and Mean Calcium Density With Biological Age Among Primary Prevention Patients With Prevalent CAC.

Among more than 10,000 primary prevention patients with prevalent CAC in the CAC Consortium, peak calcium density and mean calcium density both positively associate age. Data from 10,373 primary prevention participants from the CAC Consortium with prevalent CAC. CAC = coronary artery calcium.

CAC DENSITY AND CLINICAL RISK FACTORS.

In contrast to age, several upstream ASCVD risk factors, including body mass index and diabetes, may share inverse associations with calcium (Figure 2). Diabetes appears to share the strongest inverse association with calcium density because individuals with diabetes on average have a 1.8% lower calcium density score compared with those without diabetes, whereas each 5 kg/m² increase in body mass index associates with a 1% lower calcium density score.⁹ The associations of systolic blood pressure and total cholesterol with calcium density are also negative in magnitude, but do not achieve statistical significance after adjusting for additional risk factors and medications. On the other hand, high-density lipoprotein cholesterol is positively associated with calcium density, consistent with the fact that higher high-density lipoprotein cholesterol has been established as a protective marker for CHD risk.⁹ These observations have all occurred within an Agatston calcium density scoring system that classifies atherosclerotic lesions with a maximum attenuation >400 HU in the same category³; therefore, the associations between ASCVD risk factors and calcium density may possibly be even larger in magnitude when calcium density is evaluated on a full continuous scale. Future studies that assess the prospective relationship between upstream ASCVD risk factors and calcium density incidence and progression can help to further delineate key differences among plaque area, volume, and density, which may have implications for risk assessment and preventive care.

FIGURE 2. Independent Associations Between Traditional ASCVD Risk Factors and Calcium Density.

In contrast to age, several upstream ASCVD risk factors, including body mass index and diabetes, may share inverse associations with calcium density. Diabetes appears to share the strongest inverse association with calcium density. Adapted with permission from Thomas et al.⁹ ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; HDL-C = high-density lipoprotein cholesterol; IL = interleukin; SBP = systolic blood pressure.

CAC DENSITY, ETHNICITY, AND SEX.

The associations of ethnicity with calcium density appear to be as relevant as the underlying ASCVD risk factor burden. Compared with Caucasians with CAC >0, persons of Chinese descent have a 10.3% higher calcium density score, followed by Hispanic (5.3%) and African American (4.5%) individuals with prevalent CAC, after adjusting for age, CAC volume, and risk factors.⁹ However, African American, Hispanic, and Chinese individuals are between 8% and 22% less likely to have prevalent CAC, after adjusting for age, education status, and ASCVD risk factors.¹⁰ Such ethnic differences in CAC have yet to be explained by traditional risk factor burden, and suggest that the pathobiology of atherosclerosis and calcification propensity may differ according to ethnicity and or factors closely associated with ethnicity. Although the current observations suggest that Chinese, Hispanic, and African American individuals have a lower probability for harboring prevalent CAC, the data on calcium density demonstrates that the latter 3 ethnicity groups may have more stable, calcified plaque compared with Caucasians if CAC is present. Similar findings have been observed in South Asians because South Asians have a higher calcium density but lower CAC volume on cross-sectional assessment compared with Caucasians.¹¹ The association of ethnicity with CAC density and volume and how these relationships influence downstream ASCVD risk and preventive care require further examination.

Sex-specific mechanisms in atherosclerosis, calcification, and ASCVD incidence are also important to consider. On average, men develop CAC earlier than women with recent modeling suggesting an approximate 15-year difference in the onset of CAC between the 2 sexes.¹² For a given CAC score group, CAC lesions in women have higher calcium density and plaque area, but a lower overall number of calcific lesions and vessels compared with men.¹³ These findings correspond with findings on coronary computed tomography angiography (CCTA) that show that male gender is more strongly associated with the mixed plaque morphology phenotype (OR: 1.39; 95% CI: 1.20–1.56), including a combination of noncalcified and calcified plaque, compared with strictly calcified plaque alone (OR: 1.23; 95% CI: 1.11–1.38).¹⁴ Furthermore, lower calcium density (131–171 HU vs >172 HU) appears to confer an approximate 4-fold higher risk of ASCVD mortality in men (P < 0.0001) but not women (P = 0.51) with prevalent CAC, after controlling for CAC volume, dyslipidemia, hypertension, and 10-year ASCVD risk.¹³

Although men may have a higher prevalence of both mixed and calcified plaques, women with mixed and/or calcified plaques have a higher mean peak calcium density compared with men, independent of CAC volume.^13,15,16 Younger women that have similar calcium density values compared with men are also likely to have preserved estrogen status,¹⁷ which could be a possible mechanism explaining the differences in the unique predictive utility of calcium density for men, but not in women. On the other hand, scattered calcification across multiple vessels and a higher CAC volume are more strongly associated with ASCVD mortality in women compared with men.^18,19 These latter findings may be mediated by impaired vascular and diastolic function that are specific to women without obstructive CHD.^20,21

RADIOLOGICAL APPROACHES FOR MEASURING CALCIUM DENSITY

MINIMUM AND MAXIMUM THRESHOLDS FOR CAC DENSITY DETECTION.

Calcium density represents the concentration of calcium within a given atherosclerotic plaque²² and as a radiological construct is measured in HU through the plaque attenuation on noncontrast CT. Using parameters set forth by Agatston et al³ in 1990, the image acquisition of calcium density on CT is performed using a minimum threshold of 130 HU. The sensitivity of CAC measurement may be improved when using lower plaque attenuation thresholds, but this appears to be dependent on the SD of HU in the center of the image, patient body habitus/size, and tube voltage.^23,24 In general, the detection of atherosclerotic plaques with a lower calcium density in larger individuals appears to benefit from lower HU thresholds (100–120 HU). Contrastingly, higher HU thresholds would be most specific for plaques with higher calcium density to prevent overestimation of CAC scores and ASCVD risk misclassification.²³ This suggests that a variety of patient risk factors and imaging parameters need to be considered when analyzing plaque density and conducting CAC scoring on a CT image.

In addition to the minimum calcium density detection threshold of 130 HU, the presence of a maximum peak density limit for weighting of CAC scores is also important to consider. The Agatston formula uses a quantized density weighting factor, assigning a weighting factor of 1 (130–199 HU), 2 (200–299 HU), 3 (300–399 HU), or 4 (>400 HU) depending on the peak density per atherosclerotic lesion (Figure 3).³ Although CAC scores are strongly predictive of downstream risk, this current method of scoring may result in ASCVD risk misclassification among persons with highly dense plaques²⁵ and could potentially be improved on by interpreting calcium density in the context of age, anatomical location, baseline risk, and statin therapy.²² For example, the presence of a calcium density >1,000 HU (1,000 plaque) on the per-lesion and per-patient levels is associated with a lower risk of acute coronary syndrome independent of age, sex, hypertension, diabetes, family history of CHD, and CHD severity.²⁵

FIGURE 3. CAC Scoring With the Agatston Algorithm.

The Agatston formula uses a quantized density weighting factor, assigning a weighting factor of 1 (130–199 HU), 2 (200–299 HU), 3 (300–399 HU), or 4 (>400 HU) depending on the peak density per atherosclerotic lesion. Abbreviation as in Figure 1.

Perhaps one of the biggest limitations of the Agatston CAC score is the assumption that higher calcium density is associated with higher ASCVD risk. The current Agatston CAC scoring system could potentially be improved on by considering mean/median vs peak calcium density among individuals with low CAC burden and/or using different measures of calcium density variability (Figure 4). Although the periphery of lesions usually has a lower attenuation compared with the core, peak calcium density may still be closely correlated with the range and SD of calcium density because peripheral plaque attenuation could be close to the 130 HU cutoff. The Agatston score was developed somewhat arbitrarily before any prognostic data on CAC was available. In the setting of decades of prognostic observations and data on CAC, scoring methodologies should continue to be considered to more accurately quantify subclinical atherosclerotic burden and ASCVD risk.

FIGURE 4. Different Approaches for Measuring and Quantifying CAC Density.

The current Agatston CAC scoring system could potentially be improved on by considering mean vs peak calcium density among individuals with low CAC burden and/or using different measures of calcium density variability. Abbreviations as in Figures 1 and 2.

The contribution of calcium density as a negative risk marker or risk enhancer thus appears to be dependent on whether plaque area is accounted for in risk calculators and multivariable modeling. When assessing the association between calcium density and ASCVD risk alone, calcium density is positively associated with clinical events, because individuals with higher calcium density are likely to have: 1) older biological age; 2) longer duration of CHD; and 3) higher total plaque area.²⁶ However, after formally adjusting/accounting for calcium area or volume, higher calcium density is associated with a lower risk of events,^4,27–30 most likely because higher calcium density confers a lower probability of plaque erosion and/or rupture for any given individual plaque, plaque area, and/or plaque volume. Thus, calcium density should not be assessed independent of calcium area and volume because cointerpretation of all 3 metrics is required to accurately assess plaque vulnerability and subclinical atherosclerosis-associated risk.

DENSITY OF THE CALCIFIED PLAQUE, PLAQUE VULNERABILITY, AND PLAQUE PROGRESSION

PLAQUE VULNERABILITY.

Plaque vulnerability is associated with several unique characteristics on CCTA, including spotty calcification, low calcium density, positive remodeling, and the napkin-ring sign (Figure 5). Imaging studies have demonstrated that patients suffering from acute myocardial infarction may have a mild degree of calcification with small calcium deposits in culprit lesions, also known as spotty calcification,^31,32 and that culprit lesions in acute coronary syndromes also tend be larger in volume and have a low calcium plaque density.³³ These small calcium deposits may be specifically defined by a calcium arc <90°³¹ and individuals with spotty calcification may also have the napkin-ring sign, which is a low-attenuation plaque encompassed by a hyperattenuating rim.³⁴ Contrastingly, those with stable angina pectoris are more likely to have extensive calcification with high-attenuation/density plaques that occupy a longer average length for each calcium deposit.³¹

FIGURE 5. Calcium Density and Plaque Vulnerability.

Plaque vulnerability is associated with several unique characteristics on cardiac CT and CCTA, including spotty calcification, low calcium density, positive remodeling, and the napkin-ring sign. *Can be detected on both coronary CT and CCTA. †Better detected on CCTA.

CT = computed tomography; CCTA = coronary computed tomography angiography.

Among several plaque parameters, low calcium density and the remodeling index (ratio between vessel diameter at the site of the lesion and vessel diameter at a reference segment proximal to the lesion) appear to have the highest predictive value for a future ASCVD event. For example, culprit lesions with a plaque attenuation <30 HU and <60 HU have an area under the curve (AUC) = 0.89 and 0.85, respectively, for discriminating non–ST-segment myocardial infarction vs unstable angina.³³ Leveraging CCTA (AUC = 0.93) to capture such information appears to offer incrementally higher prognostic information compared with traditional CAC scoring (AUC = 0.82) when both are added to ASCVD risk factors for predicting major adverse cardiovascular events, including cardiac death, nonfatal acute myocardial infarction, and coronary revascularization.³⁵ On the per plaque level, these findings may suggest that calcium density is potentially positively associated with plaque stability and that further studies are required to assess whether calcium density helps to risk stratify patients in both the acute and nonacute settings. Such observations may also be supported by data arising from dual-sector spiral CT showing that individuals with acute coronary events present with a mild degree of calcification, whereas those with chronic manifestations of CAD have high-density calcific plaques.³⁶

PLAQUE PROGRESSION IN THE SETTING OF CAC DENSITY.

New advancements in CCTA have made noninvasive plaque morphology characterization and calcium density detection more readily achievable.^37,38 Serial cardiac CT and CCTA imaging have been able to capture changes in plaque density composition over time in response to pharmacotherapy and disease progression for calcific atherosclerotic plaques. The efficacy of statins in stabilizing atherosclerotic lesions has been carefully demonstrated using the latter approach. Overall, statin therapy reduces the size and volume of the lipid-rich necrotic core in atherosclerotic plaques, subsequently leading to an increase in calcium density and plaque attenuation on CT imaging.^39–44 These findings have been reported in randomized controlled trials including short-term follow-up (7 weeks)⁴¹ as well as in observational studies incorporating longer follow-up periods (>2 years between imaging scans).⁴³

Among patients with multivessel CAD undergoing percutaneous coronary intervention and at least one severely obstructed nontarget lesion, high-intensity statin therapy resulted in a greater percent reduction in lipid-core burden index compared with standard therapy after 7 weeks (−32.2% vs −0.6%; P = 0.02).⁴¹ In the notable PARADIGM (Progression of Atherosclerotic Plaque Determined by Computed Angiography Imaging) study, statin therapy was associated with a 35% lower incidence of several characteristics, including high-risk plaque features, low-attenuation plaque, spotty calcification, and/or positive arterial remodeling, over an average interscan interval period of 3.8 years.⁴³ In particular, statin therapy is associated with a shift toward high-density calcium (700–999 HU) and 1,000 plaque volumes after adjusting for traditional risk factors among patients with suspected or known CAD.⁴⁵ Although there is a natural tendency for all plaques to calcify, the accelerated lesion calcification associated with statins may be attributable to macrophage Rac1-interleukin 1B signaling axis.⁴⁶

A reduction in lipid-rich necrotic core volume is associated with plaque stability, and statins have at times been noted to confer this benefit without significant changes in luminal stenosis.³⁹ More notably, the increase in calcium density that accompanies a diminution in necrotic core volume may paradoxically lead to a higher CAC score.^40,47,48 These observations suggest that statin-derived increases in plaque calcification are a marker of stability and may be one of the mechanisms by which statins confer a reduction in the future risk of ASCVD events. However, increases in CAC score progression in statin users over the natural course of disease reflects an increase in only calcified plaque volume change, whereas the increase in CAC scores among statinnaive individuals is driven by both calcified and noncalcified plaque volume change.⁴⁴ Thus, the results derived from repeat CAC scanning for ASCVD risk assessment should be interpreted according to statin therapy. Likewise, nonstatin therapies, including 5-lipoxygenase inhibitors⁴⁹ and icosapent ethyl,⁵⁰ also lead to reductions in low-attenuation plaque compared with placebo when added to statin therapy among persons with residual ASCVD risk. The 5-lipoxygenase inhibitor at a dose of 100 mg/d in particular resulted in a significantly higher dense calcification volume compared with placebo over 6 months.⁴⁹

The culmination of evidence on the role of statins and newly emerging novel lipid therapies in plaque characterization suggests that upweighting the Agatston CAC score for higher calcium density may incompletely explain the vascular biology concept that a higher degree of lesion density is associated with atherosclerotic plaque stability on the per plaque level. On the contrary, CAC density is likely a marker of overall total plaque age/duration of CHD⁵ on the per-person level, which is associated with an increased ASCVD risk and is likely a reason why the Agatston CAC score has managed to be such a successful tool in clinical risk assessment and stratification.

Serial calcium scoring has proven to be most clinically relevant and useful for those who have CAC = 0⁵¹ because the conversion to CAC >1 represents a milestone in the atherosclerotic process, whereby individuals may benefit from statin pharmacotherapy.⁵² On the other hand, serial calcium scoring for those with at least moderately elevated CAC has proven to be less useful for incremental ASCVD risk stratification. CAC progression confers a modestly higher risk for ASCVD events, although CAC progression does not independently associate with ASCVD events beyond consideration of only the follow-up Agatston CAC score.⁵³ Although measuring CAC progression among those with elevated CAC has limited clinical value, no previous studies have decompartmentalized the progression of calcium density vs calcium area for predicting ASCVD outcomes. This is particularly relevant for patients on therapies such as statins, which might change the natural history of atherosclerosis development. New CAC scores that incorporate the full range of calcium density may be particularly useful in this regard because they may be able to more successfully capture 2 potentially unique aspects of evolution of plaque progression and associations with ASCVD risk: 1) progression of CAC area or new CAC lesions; and 2) progression of CAC density in existing lesions.

ASSOCIATION BETWEEN CALCIUM DENSITY AND ASCVD RISK

CAC DENSITY, CAC VOLUME, AND ASCVD RISK.

The relationship between calcium density and incident ASCVD has been previously explored in several different community-based cohorts (Table 1). Findings from MESA have been among the first to show that higher calcium density was inversely associated with ASCVD and CHD events, after adjusting for age, sex, ethnicity, and the Framingham Risk Score during a 7.6-year follow-up period among nearly 3,500 participants with a mean age of 66 years.²⁷ In particular, each SD increase in calcium density conferred a 27% to 29% lower risk for incident ASCVD and CHD, and calcium density most notably improved event discrimination among individuals with intermediate 10-year risk when added to a model including CAC volume. In follow-up studies, the association between calcium density and incident ASCVD has remained robust after adjusting for the newer 10-year pooled cohort equations risk score and statin use.⁴

TABLE 1.

Association of CAC Density With ASCVD Outcomes

Sample	Cohort	Covariates	Measurement of Calcium Density	Outcome	HR (95% CI)
Not adjusted for area
63,215 patients13	CAC Consortium	Unadjusted	Peak calcium density quartiles	ASCVD mortality	Density value 130–171 HU 172–206 HU 207–242 HU >243 HU	Men Reference 2.25 (0.84–6.01) 10.36 (5.61–19.15) 9.37 (5.07–17.31)	Women Reference 2.22 (1.40–3.54) 2.29 (9.45–3.61) 4.59 (3.11–6.77)
379 symptomatic individuals referred to undergo CCTA74	CORE320	Age, sex, race, statin use, Agatston score	Mean calcium density,a per SD	MACE	1.63 (1.04–2.58)
296 chronic kidney disease patients54	NA	Framingham risk score, presence of ASCVD, albumin, hsCRP, handgrip strength, statin use	Calcium density weighting factor (categorical)	All-cause mortality	CAC = 0 Lowest calcium density tertile (1.21–3.09) Middle calcium density tertile (3.09–3.24) Highest calcium density tertile (3.24–4.00)		Reference 5.00 (0.84–29.90) 10.70 (2.00–57.30) 8.93 (1.54–51.80)
125 hemodialysis patients55	NA	Age, sex, diabetes, presence of ASCVD, HTN, CHF	Agatston/volume ratio (>1 vs <1)	All-cause mortality	2.24 (1.02–4.88)
Adjusted for calcium area or volume
3,398 primary prevention patients with prevalent CAC4	MESA	PCE risk score, race, statin use, volume score	Calcium density weighting factor (per-SD higher)	ASCVD events	0.75 (0.65–0.87)
3,398 primary prevention patients with prevalent CAC27	MESA	Framingham risk score, race, statin use, volume score	Calcium density weighting factor (per-SD higher)	ASCVD events	0.71 (0.60–0.85)
1,330 primary prevention patients28	Framingham Heart Study	Framingham risk score, CAC volume	Calcium density, per HU	ASCVD Events	0.84 (0.41–0.74)
28,025 primary prevention patients with at least one or more risk factors29	CAC Consortium	Age, sex, family history of CHD, smoking status, volume score	Calcium density weighting factor (per-SD higher)	ASCVD mortality	Not on statin therapy 0.78 (0.63–0.97)		On statin therapy 0.90 (0.60–1.30)
159 asymptomatic individuals, 78 individuals with stable angina, 41 acute coronary syndrome30	Kaohsiung Veterans General Hospital	Age, sex, body mass index, diabetes mellitus, hypertension, dyslipidemia, smoking, Agatston score, length of calcific plaque	Coefficient of variation of mean calcium	ACS	0.88 (0.81–0.96)

^aAgatston score/calcium area.

ACS = acute coronary syndrome; ASCVD = atherosclerotic cardiovascular disease; CAC = coronary artery calcium; CCTA = coronary computed tomography angiography; CHD = coronary heart disease; CHF = congestive heart failure; hsCRP = high-sensitive C-reactive protein; HTN = hypertension; MACE = major adverse cardiovascular events; MESA = Multi-Ethnic Study of Atherosclerosis; NA = not applicable; PCE = pooled cohort equations.

More recently, larger studies performed in the CAC Consortium have built on earlier MESA data to further show that the prognostic value of calcium density is dependent on plaque area.²⁶ Before adjusting for plaque area, mean calcium density has a significant positive association with ASCVD mortality, whereas higher mean calcium density is significantly associated with a lower risk of ASCVD mortality after adjusting for total plaque area.²⁶ These results suggest that calcium density is strongly colinear with both biological age and atherosclerotic plaque age (duration of CHD).⁵

CAC volume is positively associated with a 68%–83% higher risk of ASCVD and CHD after adjusting for demographics and upstream risk factors. Among studies conducted in MESA,^4,27 calcium density and volume did not share a formal interaction on ASCVD risk because the inverse association between calcium density and incident ASCVD is consistent across different CAC volume strata. However, the strongest associations between calcium density and incident ASCVD here were observed among individuals with intermediate (HR: 0.68; 95% CI: 0.53–0.88) or low 10-year risk (HR: 0.70; 95% CI: 0.50–0.97) compared with persons with a 10-year risk >20% (HR: 0.77; 95% CI: 0.63–0.95).⁴ Calcium density thus may be an important metric to consider early on in the atherosclerotic process for the prediction of long-term risk. On the other hand, sex, race, diabetes status, and kidney function have not been found to modify the association between calcium density and incident ASCVD events.⁴

In a recent study published by Foldyna et al²⁸ in 2019, investigators did not observe independent associations for calcium density with ASCVD mortality, myocardial infarction, and/or stroke after adjusting for CAC volume. In this analysis conducted in the Framingham Heart Study, CAC density and volume were robustly correlated (r = 0.75) per-person, and this association was even stronger on the per segment level (r = 0.86). Although calcium density was not significantly associated with ASCVD outcomes, morphologic features and the location of calcium deposition were significantly associated with incident ASCVD because spherical morphology and pericardial positioned calcium deposits independently conferred a 45% and 34% lower risk for ASCVD mortality, myocardial infarction, and/or stroke.²⁸ In addition, previous studies conducted in the CAC Consortium have found that the inverse association between calcium density and ASCVD events may only present among men,¹³ whereas those conducted in external chronic kidney disease populations have observed that higher calcium density has a positive linear or J-shaped relationship with all-cause mortality.^54,55 For example, an Agatston/volume ratio >1, one surrogate measure of calcium density, independently confers a 2.24-fold higher risk for all-cause mortality compared with an Agatston/volume ratio <1 among individual undergoing hemodialysis.⁵⁵ Overall, differences in the associations between calcium density and ASCVD outcomes are likely attributable to a combination of variables, including different study populations that have a unique history of ASCVD risk factor burden that has influenced plaque density phenotypes over time.

PEAK VS MEAN CALCIUM DENSITY.

Conflicting findings may be in part explained by the differences in the patient populations and/or in how calcium density was measured across studies. In the Framingham study,²⁸ calcium density was measured directly within each calcification in HU and was not limited to the categorical 1 to 4 U peak calcium density scale used by the Agatston algorithm,³ the latter of which was used in the previously discussed MESA analyses.^4,27 Differences between mean calcium density vs peak calcium density may also have prognostic implications for incident ASCVD. For example, there is likely a heterogenous mixture of individuals at both high- and low-risk within the highest Agatston calcium density factor quartile (>400 HU) when peak calcium density values are categorized, which may influence downstream prediction.^22,56,57

Mean calcium density and peak calcium density may differentially describe plaque composition early in the atherosclerotic process, thus they risk misclassification using peak calcium density as opposed to mean calcium density, which is an important concept to consider. In particular, mean calcium density may be able to more comprehensively capture the mixture of low-attenuation lipid-rich and higher-attenuation fibrous and calcified plaques across all lesions compared with peak calcium density. In a recent analysis performed among more than 10,000 primary prevention patients with at least 1 or more ASCVD risk factors, mean calcium density performed better than peak calcium density factor when combined with plaque area for ASCVD mortality prediction among persons with Agatston CAC 1–99.²⁶ Future studies could also strive to adjust for the size, morphology, and location of calcium deposits when assessing the relationship between calcium density and downstream ASCVD. Overall, the totality of evidence demonstrates that the association between calcium density and long-term ASCVD risk may depend not only on the underlying disease status, but also on how calcium density is measured and categorized.

In particular, calcium density likely should be interpreted differently on the per-plaque vs per-patient basis. Low-attenuation plaques may be the most prone to rupture on the per-lesion level (eg, high-risk noncalcified plaques). For example, culprit lesions are 3.4-fold more likely to have a plaque attenuation <30 HU compared with nonculprit lesions in persons with acute coronary syndrome.⁵⁸ Further information for risk stratification on the per-plaque level has derived from radiomic studies. A radiomic-based CAC score reflecting density, shape, and texture significantly improves the discrimination of ASCVD events when added to a model including, age, sex, and Framingham risk score.⁵⁹ Such work has supported related findings showing that calcium concentration within individual plaques is heterogeneous within the same person, and that evaluation of a calcium concentration metric, reflecting density, may more comprehensively evaluate ASCVD risk and lesion vulnerability compared with a global CAC score.⁶⁰

Studies that have observed a lower risk of ASCVD events with low calcium density on the per-patient level have not adjusted for plaque area and/or biological age, therefore, they may have a misleading interpretation.²⁶ Calcium area and density are highly correlated because of their strong associations with age. However, information can be gained by measuring the discordance between calcium area and density, which can provide information in regard to different plaque phenotypes and vulnerability. Therefore, although density should be adjusted for area in risk prediction models to observe the protective association between calcium density and ASCVD risk, there may also be utility to disaggregating the 2 markers to further predict lesion vulnerability. Other conceptual frameworks for approaching the assessment of calcium density on the per-plaque level include the measurement of calcium density range and/or SD across given lesions. Future studies that assess such intricacies in the setting of statin use are of paramount importance to further understanding the complex relationship between calcium density and ASCVD risk from the molecular to individual patient levels.

PHYSICAL ACTIVITY AND CALCIUM DENSITY.

The association between physical activity and calcium density may provide further insight into the need to further improve on the Agatston CAC score. As expected, higher cardiorespiratory fitness attenuates the positive association between CAC and ASCVD events⁶¹; however, individuals performing routine, long-term endurance physical activity have been found to have a high burden of CAC. For example, men performing >3,000 MET-minutes per week of physical activity are 11% more likely to have CAC >100 compared with those participating in <3,000 MET-minutes per week.⁶² Despite having a higher CAC burden, the men in the former higher physical activity group have not been found to experience a significantly increased risk for ASCVD mortality. The seemingly paradoxical relationship among physical activity, CAC, and ASCVD risk may in part be explained by calcium density. Compared with nonathletes, master athletes have a higher proportion of purely calcified plaque as opposed to mixed or noncalcified morphology plaques,^63,64 suggesting that long-term endurance training may confer a predisposition toward developing higher calcium density and more stable plaque.

THORACIC AORTIC CALCIUM AND CORONARY CALCIUM.

Calcium density has also been evaluated among persons with thoracic aorta calcification (TAC). Thoracic calcium density and CAC density are strongly correlated (r = 0.72) and collectively improve ASCVD event discrimination when added to a base model including the CAC and Framingham risk score.⁶⁵ After adjustment for the Agatston CAC score and the Framingham risk score, each SD increase in TAC density conferred a 51% lower risk of ASCVD events, whereas TAC volume positively associated with a near 2.5-fold higher risk for incident ASCVD among individuals with an intermediate 10-year risk.⁶⁵ These results suggest that calcium measurement in the thoracic aorta adds information beyond CAC.

Similar analyses in MESA have demonstrated that each SD increase in TAC density associates with a 44% and 52% lower risk of ASCVD and CHD events, respectively, independent of CAC and upstream risk factors.⁶⁶ Progression of TAC density (per 1-SD) also confers a 71% lower risk of incident CHD but not stroke, whereas each SD increase in TAC volume progression approximately doubles the risk for future CHD and stroke.⁶⁷ Thus, there is an overall analogous relationship for TAC density and volume compared with CAC density and volume for the prediction of downstream ASCVD risk. However, the prevalence of TAC has markedly varied across studies among persons without clinical ASCVD (5%–43%),^65,66 and TAC density and CAC density are not correlated (r = −0.04; P = 0.62) in this patient population.⁶⁶ Independent of CAC scores, the presence of TAC appears to provide additional predictive utility only in women⁶⁸; therefore, the measurement of TAC should be considered among women with an intermediate-risk factor profile when there is uncertainty in risk assessment and preventive clinical care.

OUTLOOK AND IMPLICATIONS OF ADJUSTING THE AGATSTON CAC SCORE

There are several potential methodological considerations while measuring calcium density to help refine CAC scoring. First, calcium volume and density scoring have been established to outperform Agatston CAC scoring for ASCVD risk prognostication; however, the Agatston score continues to be the most widely used CAC score clinically due to its simplicity and ease-of-use. For example, using a calcium density weighting factor, rather than a continuous density measure, has facilitated CAC scoring from existing reports and data without the barrier of harmonizing differences in CT scanners or software. The measurement of calcium density currently is limited to a quantized weighting scale, which assigns a value of 1 through 4 based on the measured peak calcium density attenuation value of the lesion (1: 130–199 HU; 2: 200–299 HU; 3: 300–399 HU; 4: >400 HU). However, leveraging the entire HU scale may help to improve ASCVD risk assessment. For example, the detection of microcalcification may improve precision in guiding the optimal CAC rescan interval for those with CAC = 0,⁶⁹ whereas removing the maximum detection threshold of 400 HU could help reduce risk misclassification among persons with higher CAC scores, particular those harboring 1,000 plaque.²⁵ On the other hand, thinner sliced segments on imaging may increase noise while calculating CAC scores, which would present certain challenges in preserving the predictive utility of CAC density for ASCVD risk assessment purposes.

Radiomics⁷⁰ may be a particularly useful approach to detect nuances in density per the HU scale while preserving the signal-to-noise ratio for calcium density detection. Extracting CAC features, including calcium density signal intensity and plaque shape, may be able to provide additive value during imaging-based ASCVD risk assessment. For example, a radiomic-based CAC score associates with major adverse cardiovascular events beyond traditional risk factors and significantly improves outcome discrimination when added to the Agatston score among individuals with a CAC score <300 AU.⁵⁹ However, radiomics can also be affected by acquisition and reconstruction parameters, with main issues involving interobserver and intraobserver reproducibility as well as reproducibility across different CT scanners.^71,72 These results suggest that the incorporation of complex textural and morphologic features of calcific plaques, beyond CAC volume, has the potential to facilitate the generation of more comprehensive CAC scores, which could improve clinical decision making for primary and secondary ASCVD prevention.

In addition to further improving on the quantized calcium density weighting factor scale, the differentiation between peak vs mean calcium density should also be considered for long-term ASCVD risk assessment. The Agatston method currently quantifies CAC as the product of total coronary plaque area and a quantized peak calcium density weighting factor for each lesion, which is then summed across all lesions to give a total CAC score.³ However, measurement of CAC is more reproducible when using mean calcium density as opposed to peak calcium density.⁷³ Thus, incorporating calcium density measurements from the entire atherosclerotic plaque vs a singular area of maximum attenuation appears to provide an advantage for reproducibility, which may translate into improved detection of subclinical atherosclerosis progression and broader risk assessment. Furthermore, although repeat CAC measurement beyond scores >100 AU is less useful for assessing the progression of CHD, recent analyses show that that measurement of the lipid core burden index could be prognostically important after the initiation of statin therapy among persons with multivessel CHD.⁴¹ Moving forward, it is possible that a SD of mean density across all calcified coronary plaques would be useful to consider CAC scoring.

CONCLUSIONS

Here, we summarize key concepts related to calcium density in the setting of CAC scoring for ASCVD risk assessment. We reviewed calcium density by focusing on 3 main areas, including the acquisition of CAC scans via computed tomography, pathophysiology of calcified plaques, and epidemiologic evidence involving CAC density and ASCVD outcomes. Calcium density has become an important area of inquiry because it upweights the Agatston score based on the concept that peak plaque density and ASCVD risk are directly correlated on the per-plaque basis; however, recent studies have demonstrated that calcium density shares an inverse association with plaque vulnerability and ASCVD risk on the per-patient level after adjusting for age and plaquearea. Further evolving our approach to the measurement of CAC density from the molecular to the individual patient levels in CAC scoring may help to refine risk assessment and enhance the primary and secondary prevention of ASCVD events in the general population.

HIGHLIGHTS.

• Coronary artery calcium is expressed as the product of calcium area and a quantized peak calcium density weighting factor.

• Calcium density strongly and positively associates with age, which may correspond with the biological framework and natural progression of atherosclerosis.

• Calcium density is inversely associated with lesion vulnerability and atherosclerotic cardiovascular disease risk in population-based cohorts when accounting for age and plaque area.

• The presence of a calcium density >1,000 HU (1,000 plaque) is associated with a lower risk of acute coronary syndrome independent of traditional risk factors.

• Mean calcium density performs better than peak calcium density factor when combined with plaque area for atherosclerotic cardiovascular disease mortality prediction among persons with limited coronary artery calcium burden.

ABBREVIATIONS AND ACRONYMS

ASCVD

atherosclerotic cardiovascular disease

AUC

area under the curve

CAC

coronary artery calcium

CAD

coronary artery disease

CHD

coronary heart disease

CCTA

coronary computed tomography angiography

CT

computed tomography

TAC

thoracic aorta calcification