coronary artery calcium score pretty inaccurate

 two recent articles just came out that further questioned the value of coronary artery calcium (CAC) scoring as an accurate indicator of clinically significant coronary artery disease. The first one assessed the accuracy of CAC and the important role of LDL cholesterol in cardiovascular risk in individuals without calcifications on CAC testing (see cad coronary art calc high LDL poor correl with plaque EurHeartJ2025in dropbox, or doi.org/10.1093/eurheartj/ehaf497). these articles extend my concern about CAC as being a valuable intervention: https://gmodestmedblogs.blogspot.com/2025/04/coronary-artery-calcium-dec-plaque.html

 

Details:

-- patients were enrolled in the Western Denmark Heart Registry of symptomatic individuals for coronary artery disease who had had a coronary computed tomography angiography (CTA), from 2008 to 2021

    -- coronary CTA can accurately assess both calcified and noncalcified coronary artery plaques as well as how occlusive those plaques are. those people with non-calcified plaques would have a coronary artery calcium score (CAC score) of zero (CAC₀), even if the plaque were 100% obstructive. this current study assessed the likelihood of an atherosclerotic cardiovascular outcome (though not including heart failure) in these symptomatic individuals with CAC₀. 

-- 23,777 individuals at least 18 years old had a CAC evaluation done with no detectable calcium, CAC₀

-- median age 54, 61% women, LDL 116 mg/dL, 18% current smoking/72% no smoking, 6% diabetes, 32% hypertension, Charlson Comorbidity Index zero (ie, a healthy population), statin use before LDL cholesterol measurement 16%

 

-- clinical outcomes assessed:

    -- myocardial infarction and coronary heart disease, the latter including MI plus late coronary revascularization with PCI or bypass grafting at least 90 days after the CAC assessment that occurred, until May 2024

    -- the role of LDL cholesterol in individuals with CAC₀ on future cardiovascular events across different age groups, from age 18

 

Results:

-- prevalence of noncalcified plaques as determined by coronary computed tomography angiography (coronary CTA) was highest in individuals aged 18 to 45 (84%), followed by age 46 to 60 (56%) and >60y (28%)

-- for each 1 mmol/L higher LDL value (38.67 mg/dL), the adjusted odds ratio (aOR) for noncalcified plaques found on CTA:

    -- overall: 21% higher, aOR 1.21 (1.16-1.27)

    -- age <=45: 39% higher, aOR 1.39 (1.23-1.56)

    -- age 46-60: 22% higher, aOR 1.22 (1.14-1.31)

    -- age >60: 26% higher, aOR 1.26 (1.00-1.60)

        -- similar results were found for men and women

 

-- comparing individuals with high LDL levels (>4.3 mmol/L, >166.3 mg/dL) versus those with low LDL levels (<2.5 mmol/L, <96.7 mg/dL), there was an 81% higher risk, aOR 1.81 (1.54-2.13) of any plaque, an 88% higher risk of a non-obstructive plaque (aOR 1.88 (1.54-2.30), and a 55% higher risk for obstructive plaques (aOR 1.55 (1.21-1.99). ie,  the higher the LDL, the higher the likelihood of a nonobstructive or obstructive plaque, despite a CAC₀

    -- Individuals in the 18–45-year age group with high vs low LDL (>4.3 mmol/L, 166.3 mg/dL vs <2.5 mmol/L, 96.7 mg/dL) had an aOR for any plaque of 2.69 (1.76–4.11), whereas in the >60-year age group the corresponding aOR was 1.33 (0.99–1.80) for high vs low LDL-C. ie, the predictive value for individuals for having an atherosclerotic plaque was actually higher in the young age group with higher LDL levels (presumably because they had more cumulative exposure to high LDL levels over the years)

 

-- Medication use and noncalcified plaque in individuals with CAC₀

    -- individuals with any plaque were more likely to receive statin and aspirin treatment prior to the CTA

    -- and they were more likely to receive new onset statin and aspirin within six months after the CTA (statins 36% versus 7%, aspirin 13% versus 1%, and this was true for both obstructive and nonobstructive plaques)

 

-- LDL and coronary heart disease (CHD) events in individuals with CAC₀, with median follow-up 7.1 years

    -- overall incidence rate 1.17 per 1000 person-years (1.02-1.34) for MI, and 1.72 per 1000 person-years (1.54-1.93) for CHD

    -- prevalence of all types of noncalcified plaque was higher in individuals experiencing CHD events compared to others (any plaque 30% versus 11%, obstructive plaque 22% versus 4%)

        -- most CHD events, however, occurred in individuals without noncalcified plaque (for MI there were 154 (75%), for CHD there were 108 (70%))

    --the 5-year cumulative incidence of MI and CHD was small, 0.51% (95% CI 0.43–0.61) and 0.79% (0.68–0.91), respectively

        -- higher LDL was associated with subsequent increased risk of MI and CHD in those with CAC₀

            -- for 1 mmol/ L higher LDL there was a multivariable-adjusted rate of:

                -- MI: 26% higher risk, aHR of 1.26 (1.08– 1.48)

                -- CHD: 28% higher risk, aHR of 1.28 (1.13–1.46)

            -- this was observed in both men and women

            -- compared to individuals with low LDL (<2.5 mmol/L), individuals with high LDL (>4.3 mmol/L) vs low LDL, there was an aHR of 1.84 (1.03–3.26) for MI and 2.04 (1.29– 3.22) for CHD

            -- incidence of MI or CHD per 1 mmol/L (38.67 mg/dL) higher LDL level by age:

                -- at age 18–45-years, the aHR was 1.38 (1.00–1.91) for MI and 1.37 (1.04-1.82) for CHD

                -- at age 46–60 year, the aHR was 1.14 (0.91–1.43) for MI and 1.24 (1.04–1.49) for CHD

                -- at age >60 year, the aHR was 1.38 (1.04–1.82) for MI and 1.26 (1.00–1.60) for CHD

 

Commentary:

-- it is clear from several studies that atherosclerotic cardiovascular disease (ASCVD) is a cumulative disease that often begins very early on, increases over the years, and leads to bad cardiovascular events in older age; the primary intervention needs to be early prevention, since ASCVD is the leading cause of death in the US and increasingly so internationally, and the variety of ASCVD risk factors such as smoking are responsible for the highest morbidity:

    -- fatty streaks are found in pretty much everyone aged 15-34; advanced atherosclerotic lesions in 2% men and 0% women aged 15-19; and advanced atherosclerotic lesions in 20% of men(!) and 8% of women aged 30-34.

    -- studies have found that 78-97% of lesions in patients with acute coronary syndromes (ACS) were less than 75% stenotic, half less than 50%. plaques with lipid cores >40% of volume are most vulnerable to rupture

       -- we do know that early plaques tend to have a very active lipid core with lots of inflammation and a thin cap that is quite vulnerable given that the internal inflammation is associated with the elaboration of metalloproteinases that weaken the fibrous cap. the shear stress of blood passing through the coronary artery leads to disruption of the cap, exposure of the lipid contents to the blood, thrombosis, and acute coronary syndromes (and these early plaques are not calcified, CAC₀, though they are visible on coronary CTA)

      -- as an aside, there was a recent article that found that taking statins actually increases the CAC scores, presumably by stabilizing the lipid core and decreasing the inflammation in the newer, lipid-rich, fragile atheromas that are particularly susceptible to thrombosis and ACS; the higher CAC score was likely from the statins transforming these early (CAC₀ plaques) into stable, more fibrotic, calcified atheromas with a much smaller lipid core and less inflammation. And this happens within 6-8 weeks of statin use, thereby being associated with a significant decrease in adverse cardiovascular events in the next 6-12 months : https://gmodestmedblogs.blogspot.com/2024/05/statins-increase-coronary-artery-calcium.html. 

     -- it was notable in the current study that despite being symptomatic with an average LDL of 116 mg/dL, only 16% of the people were on statins. perhaps if more were on statins, the CAC scores would have been higher??

-- the 10-year ASCVD risk calculators miss the boat for younger people, since even people under 50-60 years old are still quite likely to die from an atherosclerotic event yet their longevity usually exceeds that 10-year horizon.

    -- and we know that a first heart attack has a significant mortality: an older study found that 2/3 of patients with cardiac arrest outside of the hospital die (https://pmc.ncbi.nlm.nih.gov/articles/PMC1112895/); a more recent study noted that about 18% of patients with an MI died withing 28 days of their heart attack and 62% of them died instantly. And many people who survive do suffer from considerable lifelong morbidity

        -- it is unfortunate that many studies focus only on mortality. morbidity including from cerebrovascular events, chronic kidney disease, other manifestations of heart disease such as heart failure may profoundly affect subsequent quality of life

-- all of this fits in with several studies finding that CAC scores are pretty useless in younger patients who are symptomatic for coronary artery disease and who had both coronary CTA done as well as CAC. one recent study found that overall 14% had CAC₀ score, with 58% in those younger than 40yo, 34% in those 40-49yo, 18% in those 50-59yo, 9% in those 60-69yo and 5% in those at least 70yo: see cad calcium score less useful in younger pts JAMACardiol2022 in dropbox, or doi:10.1001/jamacardio.2021.4406

 

-- this trial adds the following to the discussion of the utility of CAC testing:

    -- in the 23,777 individuals without any detectable CAC, high LDL levels increase the risk of noncalcified plaques, which are invisible to CAC testing (LDL was found to be independently associated with all types of noncalcified plaques, both obstructive and nonobstructive ones, as well future MI and other CHD events)

    -- the risk of a future cardiovascular event is highest in those under 45 years old, a group with a very high incidence of noncalcified plaques even as the LDL level increases, and almost always have a very low 10-year ASCVD risk by the calculators as found in this study, again strongly reinforcing the need to aggressively treat the ASCVD risk factors at a young age with nonpharmacologic therapies (diet, exercise, etc.) as well as meds when indicated. 

        -- I would add to that that there are many ASCVD risk factors that are not included in the typical array: see https://gmodestmedblogs.blogspot.com/2025/04/prevent-cardiovasc-risk-calculator.html , which is a critique of the new PREVENT cardiovascular risk calculator as well as referring to a former blog elucidating many other risk factors which clearly affect cardiovascular outcomes, including environmental ones (e.g. air pollution, micro-plastics), personal ones (e.g. the role of stress, depression, glucose intolerance), other clinical ones (e.g. controlled HIV infection), and pretty much anything that causes a state of increased systemic inflammation (e.g. visceral but not peripheral obesity, rheumatologic diseases, chronic infections, etc.). This leads me to the conclusion that we need more of a gestalt approach to ASCVD prevention, since we do not have a reliable multi-risk calculator for ASCVD risk, and therefore being more aggressive in those with more of the known risk factors: eg, a patient who has hypertension who has controlled HIV who happens to live in an area of the city with more pollution, etc should be bumped up to a higher level of aggressive treatment (and we do know that with more aggressive treatment in lowering the LDL level, there is benefit down to an LDL <21, with minimal increased risk: https://gmodestmedblogs.blogspot.com/2018/08/very-low-ldl-levels-benefit-without-harm.html)

    -- in the UK, they use the QRISK3 risk calculator that includes more of these "nontraditional" cardiac risk factors (https://heart.bmj.com/content/109/22/1690 as well as in my prior blog delineating several of these "nontraditional" ones: https://gmodestmedblogs.blogspot.com/2017/08/a-new-atherosclerosis-risk-calculator_2.html)

-- a second article just came out finding that about ¼ of patients with acute coronary syndromes (ACS) had zero CAC scores: see cad coronary art calc in ACS not accurate AmJRoent2025 in dropbox, or doi.org/10.2214/AJR.24.31476

    -- this study evaluated baseline coronary CTA in symptomatic patients without known CAD but who subsequently developed an acute coronary syndrome (ACS), and then compared them to matched patients who did not subsequently have ACS, in a substudy of the ICONIC study

-- 216 patients were assessed, mean age 55.6 years, 91 women/125 men; 108 patients were identified in each of the ACS and control groups

-- in the ACS group, culprit lesions (the areas of the coronary artery that was associated with the ACS) in the subsets of patient with CAC₀ and CAC_>0 showed no significant differences in fibrous, fibrofatty, or necrotic core plaque volumes

-- after propensity-score matching, 23% of patients with ACS had CAC₀

-- also, they compared culprit lesions in the ACS group who had CAC₀ and compared them with non-culprit lesions in the CAC_>0 subset, finding no greater fibrofatty plaques or necrotic core

-- and, in the CAC₀ group who had ACS, there were greater volumes of noncalcified plaques compared with their control groups with no ACS

-- so, in 23% of people who had ACS, there was no detectable calcium in the coronary arteries, making the CAC evaluation relatively useless in these high-risk patients who were symptomatic and subsequently had an acute coronary syndrome. and, as per above, ACS is associated with significant mortality. and those who had an MI remain at much higher absolute risk of another one, even with similar lipid control with statins

-- this brings up the issue of what to do especially with younger people with potentially symptomatic CAD.

    -- we do know they are at very high risk of subsequent ASCVD events later in life (presumably related to the fact that they have had more ASCVD at a younger age than others, and therefore more later in life in this cumulative process)

        -- and, per the above studies, i would also add individuals >60yo, those with CAC₀ still had a 26% increased risk for these noncalcified plaques for each 1 mmol/L higher LDL value (increase of 38.67 mg/dL), reflecting the insufficient sensitivity of CAC assessment even in older people

    -- of course, the most important initial therapy in ASCVD prevention is nonpharmacologic, specifically reinforcing a healthy diet, exercise, and maintaining a normal BMI in all-comers

    -- medication should only be used when necessary, and in general meds should be for males only in the younger age group, given that females who might become pregnant should not be getting cholesterol lowering medications, given the potential adverse effects on children when the mother is pregnant

-- I would add that there are other lipid markers of atherosclerotic risk beyond LDL cholesterol that we should consider:

    -- lipoprotein(a) is a potent cardiovascular risk factor (more so than LDL in several studies), is genetically determined, and at this point there are no medications that clearly provide clinical benefit for decreasing Lp(a). this has led US and European cardiology societies to emphasize the importance of aggressive lipid control with more aggressive treatments to lower the LDL in those with high Lp(a) levels (i have found that even very elevated Lp(a) is much more common than is in the literature, in my  >50 patient sample) in hopes of decreasing cardiovascular events. There are several agents that do lower Lp(a) levels very well. the best of the current meds are the PCSK9 inhibitors that do lower Lp(a) levels, though less than the newer Lp(a)-lowering agents being developed, though for these new agents there is zero information as to whether they decrease cardiovascular events. but there is a pretty strong suggestion that PCSK9s do decrease cardiovascular events: see https://gmodestmedblogs.blogspot.com/2025/06/high-lpa-increases-risk-of-recurrent.html

    -- high apolipoprotein B levels also are more predictive of atherosclerotic events since they reflect the small, dense LDL particles that are more atherogenic

    -- and, i would add, that all individuals at any age who have cardiovascular risk factors be treated as early as possible and with increased vigilance: telling someone with hypertension at age 25 that they should lose weight, exercise more, and decrease salt intake with followup in 1 year undercuts the message (and likely the result of lowering the blood pressure) in someone who is at a much higher risk of having clinically significant cardiovascular disease in the future....

 

So,

-- this study does reveal the poor utility of assessing CAC values in people under age 45. It suggests that LDL cholesterol control is important despite a CAC score of zero, especially important given the anticpated longevity of people at that age

-- some studies/guidelines do suggest that patients with CAC levels of zero may need not need preventive statin therapy (eg https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.122.061010 , which found that there was no significant association between LDL and ASCVD in those with CAC score of zero. And the 2018 AHA/ACC guidelines on the management of blood cholesterol suggests that “If CAC is zero, treatment with statin therapy may be withheld or delayed, except in cigarette smokers, those with diabetes mellitus, and those with a strong family history of premature ASCVD" https://www.ahajournals.org/doi/10.1161/CIR.0000000000000624 )

     -- It was notable in this current study that even in those over 60yo with symptomatic ASCVD, 28% had CAC score of zero

     -- so these guideline statements are pretty clearly undercut by these studies above. and they undercut the concept that atherosclerotic disease is a cumulative process that needs aggressive preventative measures early on

-- maybe we should be doing screening coronary CTAs regularly?? As per many prior blogs, i am very concerned about the long-term effects of ionizing radiation (eg see https://gmodestmedblogs.blogspot.com/2018/06/low-dose-radiation-and-subsequent.html). BUT, given the prevalence of ASCVD and the array of bad outcomes, CTA does bubble up to being a reasonable screening, even if done every 5-10 years, when compared to some of our other radiation-associated preventive interventions such as CT screening for lung cancer in smokers (eg see https://gmodestmedblogs.blogspot.com/2014/10/uspstf-lung-cancer-screening-revisited.html), or even routine mammograms (https://gmodestmedblogs.blogspot.com/2024/05/new-uspstf-mammogram-screening.html ). which is not to say that we should not do these tests at all. perhaps at decreased frequency (as many of us are doing with mammogram screening)?

    -- it is true that several studies have found that a low CAC score is helpful in triaging patients (e.g. those who come to the emergency room with chest pain and a low CAC score have a low likelihood of having cardiovascular disease in some studies), but the findings of these current studies do suggest that CAC screening may be significantly less accurate than coronary CTA, though the latter has twice the radiation exposure. And we are likely to miss some people with bad coronary artery disease by relying on CAC scores...

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them