Role of coronary artery calcium score in the primary prevention of cardiovascular diseaseBMJ 2021; 373 doi: https://doi.org/10.1136/bmj.n776 (Published 04 May 2021) Cite this as: BMJ 2021;373:n776
- 1Division of Cardiovascular Prevention and Wellness, Department of Cardiology, Houston Methodist DeBakey Heart & Vascular Center, Houston, TX, USA
- 2Center for Outcomes Research, Houston Methodist, Houston, TX, USA
- Correspondence to K Nasir
First developed in 1990, the Agatston coronary artery calcium (CAC) score is an international guideline-endorsed decision aid for further risk assessment and personalized management in the primary prevention of atherosclerotic cardiovascular disease. This review discusses key international studies that have informed this 30 year journey, from an initial coronary plaque screening paradigm to its current role informing personalized shared decision making. Special attention is paid to the prognostic value of a CAC score of zero (the so called “power of zero”), which, in a context of low estimated risk thresholds for the consideration of preventive therapy with statins in current guidelines, may be used to de-risk individuals and thereby inform the safe delay or avoidance of certain preventive therapies. We also evaluate current recommendations for CAC scoring in clinical practice guidelines around the world, and past and prevailing barriers for its use in routine patient care. Finally, we discuss emerging approaches in this field, with a focus on the potential role of CAC informing not only the personalized allocation of statins and aspirin in the general population, but also of other risk-reduction therapies in special populations, such as individuals with diabetes and people with severe hypercholesterolemia.
ACC: American College of Cardiology
AHA: American Heart Association
ASCVD: atherosclerotic cardiovascular disease
CAC: coronary artery calcium
CCTA: coronary computed tomography angiography
CHD: coronary heart disease
CT: computed tomography
ESC: European Society of Cardiology
hsCRP: high sensitivity C reactive protein
LDL-C: low density lipoprotein cholesterol
MESA: Multi-Ethnic Study of Atherosclerosis
NNT: number needed to treat
RCT: randomized controlled trial
SCCT: Society of Cardiovascular Computed Tomography
Atherosclerotic cardiovascular disease (ASCVD) remains a major cause of death globally.1234 Clinical decision making for ASCVD risk assessment and subsequent allocation of interventions in primary prevention aim at matching intensity to absolute risk. Traditionally, this has been guided by an estimate of the impact of demographic characteristics, clinical risk factors, and laboratory measures as they relate to the risk of a coronary or ASCVD event.56789
In the past two decades, extensive evidence has shown that clinical risk scores lack sufficient accuracy in predicting ASCVD risk among asymptomatic individuals.101112131415 In the past the challenge posed by this limited performance was under-detection of high risk individuals who could potentially benefit from early preventive interventions101112; however, with broadening of the eligibility criteria for statins in the 2013 American College of Cardiology/American Heart Association (ACC/AHA) and 2016 European Society of Cardiology (ESC) prevention guidelines,1617 a substantial proportion of adults became eligible for therapy, raising concerns of overtreatment.1819 Considering the significant impact of either over- or underestimation of ASCVD risk on lifelong management decisions and healthcare costs, improved risk assessment methods were needed to accurately identify not only those at higher risk, but also those at very low risk, for appropriate allocation of finite resources in a cost effective manner.
Since coronary atherosclerotic plaque is the main underlying substrate for coronary events,2021 for more than two decades debate has centered on whether the coronary artery calcium (CAC) score—a marker of coronary atherosclerosis—can aid in supplementing ASCVD risk assessment approaches.2223 However, to establish the role of CAC testing in primary prevention the evidence needs to be critically assessed. It is important to clarify whether the information gained with CAC testing can guide shared decision making at an individual level, as well as societal allocation of finite resources to reduce the burden of ASCVD.
In this review we discuss key, high quality studies published in the past 30 years that have influenced the transition of the clinical role of CAC, from an initial screening-like paradigm to a guideline endorsed decision aid, to be used by individuals at borderline or intermediate risk for a more definitive risk stratification and personalized management. We summarize current guideline recommendations relevant to CAC testing, and evaluate prevailing barriers to adoption. Finally, we discuss emerging approaches in this ever innovative field.
Sources and selection criteria
We conducted independent literature searches using PubMed and Embase of items published between 1 January 1990 and 1 December 2020. The following keyword terms were used: “coronary artery calcium”, “coronary calcium”, “coronary artery calcification”, “coronary calcification” ,“coronary artery calcifications”, “coronary calcifications”, “coronary calcium score”, “calcium score”, “Agatston score”, and “CAC score.” The reference lists cited in relevant articles were also reviewed. Given the wealth of studies published in this area, the current review is not exhaustive but summarizes key, representative studies in the field. The final list of references was selected based on relevance, with a precedence given to studies that have informed international risk assessment guidelines. Studies were selected according to their quality, which was defined subjectively as a combination of sample size, generalizability (with a precedence given to multiethnic populations), study design (with a precedence given to cohort studies), clear identification of a primary prevention population at baseline, quantification of CAC using the Agatston scoring method in study participants, length of follow-up, and quality of the statistical analyses. Basic science studies describing the biology and significance of coronary wall calcification were cited when deemed relevant.
Coronary calcification: basic features and significance for risk prediction
Detection of calcification of the coronary artery walls, a highly specific marker of atherosclerosis,2425262728293031 is currently performed using non-contrast multi-detector computed tomography (CT) scanning (fig 1). Since the 1990s, the most widely used approach for the quantification of CAC in both clinical and research settings is the method described by Agatston and Janowitz, which has become the standard CAC scoring technique.32 Using this approach, each lesion with area ≥1mm2 and radiological attenuation >130 Hounsfield units is assigned a score that integrates volume and radiological attenuation (density). Both are weighted upward,32 although it has been proposed that plaque density should be inversely weighted to account for the stability and lower risk of rupture of heavily calcified plaques.33 An overall score is calculated for each patient as a sum of the individual lesions, ranging from zero (indicative of no detectable calcified plaque) to infinity (the higher the score, the higher the burden).
Several features of the CAC score yield its value as an optimal tool for ASCVD risk stratification. First, CT is a highly sensitive imaging test for the detection of calcium dense structures, including small calcifications in the coronary walls. Second, detection of CAC is almost pathognomonic of atherosclerotic plaque, with only a few, uncommon diseases also leading to this finding.2425 Third, although non-contrast CT is unable to detect non-calcified plaque, a landmark 1995 histopathologic study showed that CAC burden correlates strongly with total coronary plaque, the area of coronary plaque being ~5 times that of the calcium area.34 This strong correlation was confirmed in subsequent studies.3536 Fourth, while non-calcified plaque may be present in the absence of calcification, severe coronary stenoses are rare.3738 Fifth, as a marker of coronary plaque burden, CAC functions as a summary measure of individual lifetime exposure to coronary risk factors and vascular susceptibility versus resilience to those.39 This explains the added prognostic value of CAC beyond single-time measures of traditional risk factors.
The Agatston CAC score is relatively simple to quantify, requires little test time, and interpretation is done in a semi-automated manner. CT parameters for CAC quantification are standardized, and this together with the availability of widely researched, clinically relevant CAC thresholds facilitates its use as a risk communication and decision making tool in routine preventive care.40
Initial evidence supporting CAC testing until 2013
With emergence of statin therapy for primary prevention of ASCVD in the late 1990s,4142 the potential for CAC to inform their allocation gained attention. Specifically, CAC was envisioned by some as a screening-like test that could aid in the detection of subclinical disease among apparently healthy individuals, who could benefit from preventive interventions to reduce their risk of ASCVD. This led to the initial notion of CAC testing as “the mammogram of the heart.”43
In the early 2000s, several retrospective clinical cohort studies portrayed CAC as a strong independent predictor of adverse cardiovascular outcomes and all-cause mortality.4445464748495051525354555657585960 Subsequently, two large prospective population based cohorts, the Multi-Ethnic Study of Atherosclerosis (MESA) in the US and the Heinz Nixdorf Recall Study in Germany generated further, high quality evidence on the prognostic value of CAC.5759 Both showed a 9- to 16-fold higher risk of coronary heart disease (CHD) and ASCVD for individuals with severe CAC compared with those with no CAC, and this was confirmed in other cohorts in Europe and the US.616263646566 These strong associations remained statistically significant adjusting for demographics, metabolic risk factors, and other features.5759616263646566 Importantly, CAC resulted in substantial improvement in risk discrimination and reclassification when added to traditional risk factors and clinical scores, and this was particularly notable among individuals at intermediate estimated risk (table 1).5759616263646566676869 In contrast, studies comparing conventional risk factors and several other newer biomarkers (high sensitivity C reactive protein (hsCRP), the ankle-brachial index, and carotid intima-media thickness, among others) for the prediction of CHD/ASCVD events consistently yielded minimal or no improvements.67687071
CAC as a tool to “de-risk”
In 2013, the ACC/AHA ASCVD risk management guidelines broadened substantially the eligibility criteria of the adult population for statin therapy.16 Apart from the concern of widespread use of statins potentially resulting in overtreatment of individuals unlikely to benefit at least in the short term (nearly half of US adults became potential candidates for statins), subsequent studies found that the Pooled Cohort Equations and other clinical risk scores overestimated CHD/ASCVD risk.13727374 Although the authors of the 2013 guideline considered that the need for further “screening” with CAC to identify additional candidates for therapy became less compelling,75 the novel context of widespread consideration of statin therapy shifted the aim of further risk assessment toward identifying individuals at sufficiently low risk in whom pharmacotherapies could be safely avoided or delayed. In this setting, CAC would perform robustly as a decision aid.
Indeed, many studies have shown that the absence of CAC (CAC=0) confers a very low risk for future cardiovascular events and mortality.7677 In a 2009 meta-analysis of 13 studies comprising 71 595 asymptomatic individuals—of whom 41% had no CAC at baseline—after a median follow-up of 3-5 years only 154 individuals with CAC=0 at baseline experienced a cardiovascular event (incidence 0.5%), compared with 4.1% among the participants with detectable CAC.77 Similar findings were reported in the CAC Consortium, a large clinical/self-referral cohort study,78 as well as in MESA and other population based cohorts.79 In all, the risk of events with CAC=0 in asymptomatic individuals from the general population was very small, even after more than 10 years of follow-up.
CAC=0 among high risk individuals and guideline recommended statin candidates
A series of studies from 2010 to 2014 showed that among individuals deemed “high risk”—ie, with elevated hsCRP levels,80 multiple traditional risk factors,8182 dyslipidemia,83 tobacco use,84 family history of premature CHD,85 candidates for preventive polypill,86 and patients with diabetes87—a large proportion had zero CAC. This was associated with either low or markedly reduced risk of incident ASCVD, and the estimated number needed to treat (NNT) to prevent one event was high. For example, in MESA the NNT to prevent one CHD event over five years among individuals with hsCRP ≥2 mg/L and CAC=0 was 549, while this was 42 for those with hsCRP ≥2 mg/L and CAC >0.80
Importantly, a 2015 analysis in MESA showed that nearly half of participants considered statin candidates (as defined in the 2013 ACC/AHA guidelines) had no detectable CAC, especially among those in the 5% to 20% 10 year risk range, with CAC=0 reclassifying risk to a category in which guidelines no longer recommend treatment.72 Among participants with 10 year ASCVD risk 7.5-20%, CAC=0 was associated with event rates (~4.5%) below the guideline defined threshold for statin benefit, whereas CAC >0 was associated with events consistent with net benefit from statin therapy (~10.5%) (fig 2). Similar findings were reported in other cohorts.88 Subsequent analyses in the Walter Reed cohort and the St Francis trial suggested lack of benefit with statins in participants with CAC=0.8990 Together, these studies led to the concept of “power of zero,” stressing that the absence of CAC confers a sufficiently low risk of future cardiovascular events that may result in a reclassified lower risk, allowing for more flexible treatment goals including deferring specific pharmacotherapies and focusing on lifestyle interventions in certain individuals.919293
CAC zero compared with other negative markers
Analyses in MESA also showed that the prognostic power of CAC=0 as a “negative risk factor” for ASCVD was superior to that of other tools and features often used in routine care.9495 These included a carotid intima-media thickness <25th centile, no carotid plaque, brachial flow mediated dilation >5% change, ankle-brachial index >0.9 and <1.3, hsCRP <2 mg/L, homocysteine <10 µmol/L, N-terminal pro-brain natriuretic peptide <100 pg/mL, no microalbuminuria, no family history of CHD, absence of metabolic syndrome, and a healthy lifestyle. Similar results in BioImage95 expanded the notion of CAC=0 for “de-risking” to older individuals.
Cost effectiveness of CAC in shared decision making
In the setting of statins being available in generic forms, cheap, and even effective in lower risk individuals, it remained uncertain whether a strategy involving CAC testing to guide decision making would be cost effective compared with treat-all approaches. To evaluate this, a 2017 study compared the cost effectiveness of a selective approach using CAC testing to guide statin therapy versus treating individuals according to the 2013 ACC/AHA guideline criteria for statin therapy (estimated 10 year ASCVD risk ≥5%).96 Using data from MESA, both strategies were found to be equally cost effective in the overall study population. In a simulation of 10 000 individuals, a treat-all strategy only averted 21 ASCVD events, but increased the person-years of statin therapy by more than 47 000. However, in situations assuming CAC cost <$100 and higher cost and/or disutility associated with statin therapy, the CAC strategy was favored. In the Jackson Heart Study, avoidance of statin therapy among those with CAC=0 was cost effective compared with initiating statin therapy in all African Americans at intermediate risk for ASCVD, potentially resulting in greater quality-adjusted life expectancy at a lower cost when the patient has a strong preference to avoid pharmacotherapies.97
Implications of CAC=0 for shared decision making
This compelling body of evidence pointed to the need for more deliberation on the potential value of CAC testing to de-risk individuals who would be considered potential candidates for preventive therapies (according to clinical risk scores) but have few events in the mid-term; and on whether in a context of limited resources CAC=0 could help in prioritizing the societal allocation of therapies. The approach of separating out the subset of individuals with absent CAC and focusing on individuals with underlying subclinical atherosclerosis would help direct intensive treatments toward highest risk individuals likely to derive the most benefit, for whom the NNT would be small enough to make prophylactic pharmacotherapy cost effective from a societal standpoint.
It is important to note, however, that this approach does not dictate that all individuals should be considered for CAC testing (fig 3). The purpose is to facilitate shared decision making for appropriate risk estimation that can guide decisions for consideration of pharmacotherapies such as statins. In those willing to pursue or avoid therapy irrespective of CAC burden, testing would have minimal impact and should not be pursued. On the other hand, for individuals uncertain about their absolute risk and optimal management, and willing to forgo certain pharmacotherapies and focus on lifestyle interventions, CAC can provide significant value.
CAC for the allocation of other preventive therapies
Besides statins, CAC scoring may guide a personalized, safe allocation of other preventive pharmacotherapies. An example is low dose aspirin. A 2014 analysis in MESA suggested that CAC could be a valuable aid in identifying optimal primary prevention candidates for low dose aspirin, as well as individuals with the lowest CHD risk in whom therapy could lead to net harm.98 This led to a mention, in the 2019 ACC/AHA primary ASCVD prevention guidelines, to consider CAC to inform a personalized allocation of aspirin, most specifically, avoidance of aspirin in the presence of CAC=0.7 An updated analysis of 3540 MESA participants at low risk for bleeding, which incorporated observed bleeding event data and the most recent meta-analytic estimates of efficacy and safety, reported consistent findings.99100 In persons with CAC >100, the NNT with aspirin to prevent one ASCVD event was markedly lower than the number needed to harm (ie, cause a bleeding event), consistent with net benefit. Conversely, CAC=0 identified a group of individuals in which the number needed to cause a bleeding event (567) was much lower than the NNT to prevent an ASCVD event (1190), indicative of net harm. This was true even among individuals at >20% estimated 10 year risk using the Pooled Cohort Equations (256 v 541). Qualitatively consistent findings have been reported recently among individuals at low risk of bleeding from the younger Dallas Heart Study cohort (mean age 44).101102
Besides aspirin, CAC has been proposed as a helpful tool to inform the intensity of blood pressure management,103 and to identify sufficiently high risk primary prevention individuals who could potentially benefit from oral thromboprophylaxis with rivaroxaban, as evaluated in a modeling analysis using observational MESA data.104
The wealth of epidemiological evidence portraying the prognostic value of CAC in primary prevention has resulted in a prominent role in current clinical practice guidelines around the world. Below we discuss key guideline recommendations from North America, Europe, Asia, and Oceania, and table 2 summarizes current ACC/AHA and ESC documents.
The 2018 AHA/ACC/Multi-Society Cholesterol Guidelines and the 2019 ACC/AHA Primary Prevention Guidelines formally endorsed CAC as a shared decision making aid for personalized risk management in primary prevention.67 Compared with the 2013 guidelines,16 the 2018/2019 documents not only expanded the estimated 10 year ASCVD risk range in which CAC could be considered for personalized risk stratification (from 5-7.5% to 5-20%) but also upgraded most CAC related recommendations to IIa. Moreover, extensive guidance was included regarding the utility of CAC=0 to inform the avoidance of statin therapy among individuals not at high risk and without diabetes, family history of premature CHD, or cigarette smoking. Of particular importance, the 2018 cholesterol guidelines included a IIb recommendation to consider CAC testing in adults aged 76 to 80 and avoid statin therapy if CAC=0.6 In addition, as discussed above, the 2019 document briefly outlined the utility of CAC to inform preventive therapy decisions involving aspirin.7
The guidelines also endorsed the use of age and sex specific CAC centiles as an alternative to absolute CAC score thresholds.67 This allows for a more nuanced interpretation of CAC burden by age and sex. For example, a CAC score of 22 in a 45 year old white American man entails a similar relative risk increase in ASCVD events as a CAC score of 100-200 in a 60 year old individual, and can be used to identify candidates for early aggressive interventions, despite a relatively low absolute score.106 In women, CAC >0 is rare at age 45, and can inform the early consideration of preventive pharmacotherapies as well.106
Adoption of CAC as a decision aid has been historically slower in ESC guidelines, and the endorsement more tepid. However, in the 2019 ESC/European Atherosclerosis Society Guideline for the Management of Dyslipidaemias, CAC was recommended as a risk modifier to be used to upgrade risk estimations and strengthen the case for preventive statin therapy. Nonetheless, this recommendation also listed other tools with more limited supporting evidence, such as the ankle-brachial index or hsCRP.5 Moreover, the role of CAC=0 to de-risk was not addressed. However, a subsequent position paper by the European Association of Preventive Cardiology (EAPC) explicitly addressed the utility of CAC=0 to reclassify risk downward.105
In the UK, although the National Institute for Health and Care Excellence (NICE) has provided recommendations for the use of coronary CT angiography (CCTA) for the assessment of chest pain,107 NICE guidance on the use of CAC testing for risk assessment in asymptomatic individuals is, to the best of our knowledge, currently not available.
Guidance on the use of CAC for ASCVD risk assessment from local scientific societies is limited in many Asian countries. CT is described mostly as a “screening tool” in Japanese prevention guidelines,9 and CAC and CCTA are currently widely used for clinical plaque screening in South Korea, typically as part of health checkups for workers.108109 It remains unclear, however, whether an aggressive use of CAC/CT angiography for screening in populations at low risk of CHD yields improved clinical outcomes.
The Society of Cardiovascular Computed Tomography (SCCT) Expert Consensus and other guidelines
Several recommendations included in the ACC/AHA 2018 and 2019 documents mirrored those outlined in the 2017 SCCT Expert Consensus Statement on CAC assessment in asymptomatic adults.110 In addition, the SCCT proposed a selective consideration of CAC scoring among individuals at low estimated risk with a family history of premature CHD, aimed at enhancing the prevention of ASCVD events in select younger adults at increased relative risk.110 A similar recommendation had been included in the 2016 Canadian Cardiovascular Society Guidelines for the Management of Dyslipidemia, although the authors considered the quality of the supporting evidence to be low at that time.8 Of note, the 2016 Canadian guidelines recommended against the use of CAC=0 to defer or delay preventive pharmacotherapies. Nonetheless, it must be noted that this document was prepared before the publication of several key studies on CAC as a negative risk marker, such as landmark analyses in the Walter Reed cohort and the St Francis trial.89909495
The SCCT Expert Consensus also provided guidance for a more personalized preventive pharmacological management, under which higher doses of statins and aspirin would be considered with higher CAC scores.110 Of note, the 2017 position statement on CAC scoring from the Cardiac Society of Australia and New Zealand mirrored several of these recommendations, and included an explicit recognition of individuals with CAC=0 as very low risk, who “do not benefit from treatment.”111
Overcoming barriers to adoption
Despite a large, comprehensive body of supporting evidence, several barriers have slowed the adoption of CAC testing in routine care. Cost initially represented an important barrier, and a study in Ohio (US) demonstrated that availability of CAC testing at no charge markedly increased its use for ASCVD risk assessment overall, as well as among minorities and other vulnerable subgroups.112 The cost of non-contrast CT for CAC quantification is now ≤$150 (£109; €127) in most US centers, however, lack of reimbursement in many health systems remains a barrier.113 Radiation exposure represented another initial concern; however, with current technology this is now ~1 millisievert (similar to that of a bilateral mammogram or a transatlantic flight), and reducing the associated radiation exposure further is an active area of research.40114
Another prevailing barrier is the lack of experimental evidence to show improved hard clinical outcomes of CAC based ASCVD risk management.115 The ongoing ROBINSCA (Risk Or Benefit In Screening for Cardiovascular Disease) trial aims at generating such evidence by comparing a risk factor based approach with a CAC based strategy.116 The inclusion criteria used, which enriched the study population with individuals with specific risk factors such as obesity, tobacco use, or a family history of premature CVD, limits generalizability to the broad general population; however, the study can shed light on the potential value of CAC in intermediate and high risk groups. At baseline, CAC substantially reduced the number of individuals indicated for preventive treatment compared with clinical scoring, although CAC <100 rather than CAC=0 was used by the investigators to define “low risk.” Results in terms of incident events at five years comparing traditional risk factor based and CAC based management strategies are expected to become available in the coming years. It should be noted that lack of experimental evidence in terms of impact on hard clinical outcomes also applies to the use of clinical CHD/ASCVD risk scores, which is the current standard first step for risk assessment.56789
It had been proposed that CAC testing could increase costs through greater downstream testing. Nonetheless, a four year follow-up analysis in the EISNER (Early Identification of Subclinical Atherosclerosis by Noninvasive Imaging Research) study did not show statistically significant differences in downstream cost between CAC and non-CAC study arms.117 Additional prevailing barriers to adoption of CAC in routine care include tepid or lack of endorsement in some local guidelines, competing priorities for the use of available CT scanning resources in many centers, and limited experience in the clinical interpretation of the test, among others.
Besides the already widely accepted guideline endorsed uses of CAC in risk assessment and management, novel approaches will likely further expand its role in personalized primary prevention in the coming years.
CAC for personalized management of patients with diabetes
While the role of CAC and especially the prognostic power of CAC=0 are now well established among those at borderline or intermediate risk, current guidelines suggest to exercise caution in potentially higher risk groups (eg, individuals with diabetes).67 Nonetheless, the 2018 ACC/AHA cholesterol guidelines acknowledged the role of CAC as an aid to guide the personalized management of individuals aged 76 to 80.6118119 Similarly, CAC may have a role guiding tailored management of other groups at increased average risk but in which ASCVD risk is heterogeneous (fig 4).
An example is the rapidly expanding population of patients with diabetes, for whom there are now multiple atherosclerotic risk reduction pharmacotherapies available beyond statins. These include aspirin, glucagon-like peptide-1 receptor agonists (GLP-1RA) and icosapent ethyl.120121122 Analyses in MESA and other cohorts have shown that CAC stratifies risk also among individuals with diabetes, with ASCVD event rates being proportional to baseline CAC burden.87123124 Of particular importance was the 2018 Walter Reed Medical Center study, which included 13 644 individuals (mean age 50, 71% men) free of ASCVD, followed for a median of 9.4 years.89 Preventive statin therapy was associated with a markedly reduced risk of incident ASCVD events in multivariable adjusted analyses across all CAC burden strata, except for individuals with baseline CAC=0, in whom event rates with and without statin therapy were almost identical. Although limited by its observational design and a relatively young, healthy study population with low event rates, this landmark study triggered calls to expand the role of CAC=0 as a key decision aid in ACC/AHA guidelines. Ongoing studies in MESA and other cohorts will shed further light on the potential value of CAC and particularly CAC=0, aiding a more cost effective allocation of ASCVD risk reduction therapies in patients with diabetes. This may help direct currently costly therapeutic options such as GLP-1RA to individuals at highest risk and most likely to benefit.125158
CAC for personalized management of patients with severe hypercholesterolemia
A similar personalized CAC guided management paradigm may apply to individuals with severe hypercholesterolemia, including patients with heterozygous familial hypercholesterolemia. Besides statins, other novel lipid lowering therapies such as proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors are currently recommended for these individuals on the basis of their levels of low density lipoprotein cholesterol (LDL-C).567126 Although these recommendations assume homogeneously high ASCVD risk among individuals with increased LDL-C, studies in small cohorts suggest otherwise, with a non-negligible proportion of persons with very high levels of LDL-C being resilient to developing calcified coronary plaque. In a study of 206 individuals with familial hypercholesterolemia (mostly middle aged individuals with average on-treatment LDL-C 150 mg/dL), Miname and colleagues showed that close to 50% had CAC=0, and baseline CAC burden was associated with incident events at a median of 3.7 years of follow-up.127 In MESA, 37% of participants with baseline LDL-C levels ≥190 mg/dL had CAC=0, and this was associated with markedly lower 10 year event rates than individuals with detectable calcified plaque.128
The findings from these and other studies129 highlight the multifactorial nature of atherosclerosis, which is not only heterogenous among individuals with severe hypercholesterolemia but is also the main driver of future clinical risk. While we do not advocate cessation of, or refraining from statins in those with severely elevated levels of LDL-C (indeed, a large proportion of participants in the aforementioned studies were treated with statins, including many of those with CAC=0), CAC can provide valuable insight into risk stratification strategies for a more cost effective allocation of novel, costly lipid lowering treatments.130131132 As in patients with diabetes, this may help funnel expensive therapies such as PCSK9 inhibitors to patients most likely to benefit, and away from individuals more resilient to atherosclerosis and events who are therefore less likely to benefit. Such an approach could have important benefits for patients, who would avoid the out-of-pocket cost of these medications, and for societies, particularly in environments with most limited health resources.
Nevertheless, this is an area of intense debate and active research, and further studies are needed.130131132133 Specifically, a better understanding of the “warranty period” of CAC=0 in patients with severe hypercholesterolemia is needed. Also, large studies with long follow-up are needed to define the long term implications of a restrictive use of novel add-on lipid-lowering medications in patients with CAC=0. It is important to note that the real-world uptake of PCSK9 inhibitors has so far been limited.134135136 This further highlights the potential importance of tools that can help direct limited resources to those most likely to benefit.
CAC to enrich RCTs in primary prevention
The availability of novel add-on therapies for secondary prevention, as well as among asymptomatic patient subgroups at higher average risk (eg, individuals with diabetes) has increased markedly in the past five years.137 Conversely, RCT evidence generation for novel add-on therapies in broad higher risk primary prevention populations has been slower. Average event rates among individuals at high estimated risk using clinical scores such as the Pooled Cohort Equations are markedly lower than in secondary prevention populations, particularly in a context of expanded use of statins, and absolute risk is highly heterogeneous in this group, with noticeable differences in event rates by baseline plaque burden. These features may result in RCTs of add-on therapies that require very large study populations to be sufficiently powered, and very high, unaffordable study costs, ultimately harming feasibility and precluding evidence generation.
In this context, CAC could help increase trial efficiency by informing participant inclusion in primary prevention RCTs of novel add-on therapies.138 Specifically, study populations could be enriched with participants with high CAC scores, increasing the overall event rates and statistical power, while potential participants unlikely to have events (those with CAC=0) could be excluded (fig 5). This could result in RCTs with smaller sample sizes and therefore lower cost, as suggested in a recent modeling analysis in MESA.139 A plaque based enrichment approach blurs the line between primary and secondary prevention, and has been used for other biomarkers such as hsCRP in landmark RCTs such as JUPITER (Justification for the Use of Statins in Prevention: an Intervention Trial Evaluating Rosuvastatin).140 A multi-biomarker (including CAC) enrichment approach is being used as part of the study entry criteria of VESALIUS-CV (Effect of Evolocumab in Patients at High Cardiovascular Risk Without Prior Myocardial Infarction or Stroke; NCT03872401), an ongoing international RCT of evolocumab in primary prevention.141 Upcoming studies will provide further insights on this potential use of CAC.
Timing of repeat CAC scanning
With expanding use of CAC as a guideline endorsed decision aid and increasing recognition of the prognostic power of CAC=0, there is interest to define the time frame during which these individuals remain at low or lower risk of events, and when to repeat the scan to detect potential conversion to CAC >0—the so-called “warranty period” of CAC=0. In a 2010 study of 422 individuals with baseline CAC=0 undergoing yearly CAC scans, conversion to CAC >0 occurred in 25% participants after five years.142 Key predictors of CAC conversion were age, tobacco use, and diabetes. Subsequent analyses in larger US and Korean populations confirmed that a high baseline burden of risk factors was associated with a shorter warranty.143144 More recently, among 3116 MESA participants with CAC=0, the average warranty period among those at borderline or intermediate estimated risk was 3-5 years, and three years in patients with diabetes.145 These studies can help tailor clinical follow-up in patients who choose to prioritize lifestyle interventions and generic drug options. Importantly, the critical role of traditional risk factors as modifiers of the warranty of CAC=0 stresses the importance of preventing their onset also among individuals with CAC=0.
In population based cohorts, most asymptomatic adults <55 have CAC=0.39106 Even in clinical cohorts of individuals undergoing CAC scoring, which typically include participants with multiple risk factors, the prevalence of CAC=0 is ~70% among individuals aged 30 to 49.146 Event rates are very low in young individuals with CAC=0 (eg, 1.0% 13 year CVD mortality in CAC Consortium participants aged 30-49 with CAC=0), however, those rates are very similar, in absolute terms, to those of the general population of the same age (1.3%).146 Therefore, a key question is whether CAC=0 is sufficiently reassuring in younger adults,133 and whether additional measurements or tests could be used to identify individuals at increased risk of events within the large population of young-to-middle aged adults with CAC=0. This is currently an active area of research.
Measurements that can be performed as part of the same cardiac CT scan used for CAC scoring, such as aortic valve calcium and mitral annular calcium, have shown to be poor independent predictors of CHD/CVD events.147 For thoracic aortic calcification, in a 2017 analysis in MESA participants with CAC=0, the frequency of any thoracic aortic calcification was low, and this finding did not improve risk estimation beyond traditional risk factors.148 Importantly, the prevalence of these extra-coronary calcifications in younger individuals is very low, further limiting their relevance in risk assessment specifically in this age group. Ongoing analyses in MESA will provide further insights into whether carotid plaque imaging has independent prognostic value among individuals with CAC=0. It has also been proposed that plaque screening using CCTA could be valuable for this purpose, however, compelling studies are lacking.
Whether CAC=0 is sufficiently reassuring in younger adults, and whether further imaging or other (eg, genetic) testing is justified need to be evaluated in the context of societal priorities, available prevention resources, and willingness among the general population to undergo additional testing or use lifelong preventive medications despite a low 10 year absolute risk. An alternative paradigm to further curtail events among younger adults could integrate a stronger focus on population based cardiovascular health protection policies together with enhanced detection of prevalent risk factors.
Time for a CAC 2.0 score?
Enhanced incorporation of coronary plaque density as a protective rather than a higher risk feature represents a promising potential improvement of the CAC score moving forward.149150 Studies have shown that for a given coronary plaque volume, plaque density is inversely associated with risk of events during follow-up.33151152 An updated CAC score using inverse density weighting would be particularly relevant in older individuals as well as in users of statins, both of whom have a higher prevalence of highly calcified plaques. Nonetheless, and until such scores are eventually developed, it is important to note that the Agatston CAC score is strongly associated with incident events in those groups, as demonstrated in cohorts of statin users and in studies with a high use of statins at baseline.127128151153154 In contrast, clinical risk estimators such as the Pooled Cohort Equations or SCORE classify almost all older individuals as high risk, and their use is not recommended in the presence of statin therapy.56774 Therefore, the current Agatston score can be a valuable aid in shared decision making and personalized management of these large, highly relevant groups.
In a context of lower cost of CCTA testing and reduced radiation dose, an alternative future paradigm could involve an expanded use of CCTA for risk assessment in primary prevention. This would allow for a more detailed evaluation of plaque morphology and vulnerability characteristics, early forms of (non-calcified) plaque, and quantification of low-attenuation plaque,155 none of which can be assessed with non-contrast CT for CAC scoring. Nonetheless, in asymptomatic participants from CONFIRM (coronary CT angiography evaluation for clinical outcomes: an international multicenter registry), CCTA did not significantly add to prediction models including CAC.156 Moreover, the need for intravenous contrast, the more complex patient preparation, and the longer testing and interpretation times still represent important barriers that prevent a generalized use of CCTA in asymptomatic individuals. As of 2021, no international primary prevention guideline endorses CCTA for this purpose in asymptomatic individuals. The ongoing Miami Heart Study (NCT02508454) will provide further insights on the interplay between CAC and CCTA based plaque features in a large, population based, young-to-middle aged primary prevention cohort.157
First defined in 1990, the Agatston CAC score is a powerful marker of calcified coronary plaque, strongly correlates with total coronary plaque burden, and is independently associated with incident CHD/ASCVD events. Over the years CAC has evolved from an initial plaque screening paradigm to its current guideline endorsed role as a decision aid for individuals at borderline or intermediate risk who seek more definitive risk assessment as part of a clinician-patient discussion. This journey has been informed by a wealth of studies in American and European cohorts, which have produced solid evidence of marked risk heterogeneity within strata defined by clinical risk scores, and of CAC being able to improve risk prediction when added on top of traditional risk factors and scores—particularly in the borderline or intermediate risk range. In a context of progressively low estimated ASCVD risk thresholds for consideration of preventive therapy with statins in international primary prevention guidelines, CAC=0, which is associated with low event rates in several populations, has become a particularly powerful prognostic tool and may be incorporated in clinician-patient discussions to guide the safe avoidance of preventive therapy with statins and aspirin. Moving forward, ongoing studies will help define the potential role of CAC informing the personalized allocation of other risk reduction therapies among special populations, such as individuals with diabetes and severe hypercholesterolemia. Finally, whether CAC can be used to enrich the study population of primary prevention RCTs, and how to better account for plaque density in CAC scoring, represent promising areas of innovation. Together, the available evidence suggests that CAC will remain a key tool in personalized primary ASCVD preventive care in the coming years.
Does risk management informed by clinical risk scoring plus CAC in certain individuals improve hard clinical outcomes compared with clinical risk scores alone?
Can CAC have a value in the allocation of novel add-on risk reduction therapies in patients with diabetes and severe hypercholesterolemia?
Can CAC be used to identify higher risk study populations, reduce study size, and reduce cost of RCTs of novel add-on therapies in primary ASCVD prevention?
Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: Khurram Nasir is on the advisory board of Amgen, Novartis, Medicine Company, and his research is partly supported by the Jerold B. Katz Academy of Translational Research. Miguel Cainzos-Achirica is fully supported by the Jerold B Katz Academy of Translational Research and has no other disclosures to report.
Contributorship statement and guarantors: KN and MC-A contributed to the planning, drafting, and critical review of the manuscript. Both are responsible for the overall content as guarantors.
No patients were asked for input in the creation of this article.
Provenance and peer review: commissioned; externally peer reviewed.