Coronary Artery Calcium for Risk Stratification of Heart Failure Mortality: The Coronary Artery Calcium Consortium

Abstract

Background:

There is increasing interest in predicting heart failure (HF), a major cause of morbidity and mortality with a significant financial burden. The role of coronary artery calcium (CAC), an accessible and inexpensive test, in predicting long-term HF mortality amongst asymptomatic adults remains unknown. We aim to determine if CAC burden is associated with HF-related mortality in the CAC Consortium.

Methods and Results:

The study included 66,636 primary prevention patients from the CAC Consortium. Multivariable competing risks regression was used to assess the association between CAC and HF-related mortality adjusting for demographics and traditional risk factors. The mean age was 54.4 years, 67% male, 89% white, and 55% had CAC >0. 260 HF-related mortality events were observed during a median follow up of 12.5 years, 75.3% occurred among those with a baseline CAC score >100. Compared with CAC = 0, there was a stepwise higher risk (P < 0.005) of HF mortality for CAC 1–100 (subdistribution hazard ratio [SHR]: 2.27; 95% CI: 1.3–3.99), 100–400 (SHR: 3.68; 95% CI 2.1–6.43), and >400 (SHR: 7.05; 95% CI 4.05–12.29). This increasing risk of HF mortality across higher CAC scores persisted across age groups, sex, and in the intermediate and high-risk groups as calculated by the pooled cohort (PCE) and PREVENT equations.

Conclusions:

Higher CAC is associated with increasing incidence of long-term HF-related mortality in the primary prevention population, particularly intermediate and high-risk patients. Early preventive approaches in patients with high CAC must focus on preventing heart failure and ASCVD with lifestyle changes and medications.

CAC HF Mortality- Lay Summary

• CAC scoring is becoming increasingly valuable in the primary prevention setting as it enhances risk stratification and helps to guide more personalized management in the prevention of cardiovascular disease.

• The incidence of HF is rising, and we have shown that CAC is a strong predictor of HF mortality, even in patients free of symptoms or heart failure at baseline.

• The identification of individuals at higher risk of HF-related death may allow for earlier use of more tailored therapies aimed at preventing HF.

Lay Summary:

Coronary artery calcium (CAC) scoring has become increasingly useful in preventing cardiovascular disease by improving risk assessment and guiding personalized care. CAC can help predict which primary prevention patients are at higher risk of HF-related death and may allow for more tailored therapies to prevent HF.

Introduction

Heart failure (HF) is a major cause of morbidity and mortality globally. From 2017 to 2020, 6.7 million Americans >19 years of age had HF, a stark increase from the roughly 6 million in 2015 to 2018.¹ The prevalence of HF is projected to increase by 46% from 2012 to 2030, eventually affecting >8 million people >17 years of age, with the total prevalence of HF projected to rise from 2.05% in 2012 to 3.0% in 2030.¹ HF also exerts a significant financial burden, costing the US healthcare system nearly $39.2 to $60 billion annually, and is expected to exceed $70 billion by 2030.^2,3

Despite significant advances in medical management of HF, prediction of HF in otherwise asymptomatic individuals remains elusive, thus limiting the ability to engage in targeted HF prevention.

Coronary artery calcium (CAC) is a measure of plaque burden that is non-invasively measured on cardiac-gated non-contrast computed tomography (CT) via the Agatston method.⁴ It is routinely used to detect subclinical atherosclerosis and predict ASCVD risk, particularly in individuals classified as intermediate risk through the Pooled Cohort Equations (PCE).⁵ Serial measurement of CAC has been introduced as a tool to monitor the progression of subclinical atherosclerosis, and has a good predictive value for incident CAD and all-cause mortality.^6,7 However, CAC has now been recognized as a marker not only of atherosclerosis, but also of overall cardiovascular health as well as non-cardiovascular aging, suggesting its potential role in assessing a more general set of health outcomes beyond ASCVD.⁸ This is thought to be because CAC serves as an “integrator” of cumulative risk exposures over a patient’s lifetime and directly displays the effect of this risk on cardiac tissue, taking into account susceptibility and resilience factors.⁹

The most common cause of HF in the United States is coronary artery disease (CAD)¹⁰, and CAC is associated with both silent ischemic myocardial scar as well as incident ASCVD events. CAD is also an independent predictor of risk for both HF with reduced and preserved ejection fraction.¹¹ However, there is data suggesting that CAC and CAC progression are associated with HF independent of CAD, and CAC scores have also been shown to be associated with all-cause mortality amongst patients with HF.¹² Indeed, many cases of HF are not preceded by symptomatic CAD, but rather, structural and functional changes that have been linked to aging and subclinical CAD.^13,14

Given these observations, the rising prevalence and financial burden of HF, and the emergence of medical therapies that might prevent HF, we sought to investigate CAC as a predictor of long-term mortality due to HF in a general clinical primary prevention population.

Methods

Study Design and Population

The Coronary Artery Calcium Consortium is a multicenter, retrospective cohort study comprising 66,636 patients >18 years without CAD or cardiovascular symptoms from four centers in the United States: Cedars-Sinai Medical Center, PrevaHealth Wellness Diagnostic Center, Harbor-UCLA Medical Center, and Minneapolis Heart Institute. The study aimed to assess the association between CAC and long-term, disease-specific mortality and has been previously described in detail.¹⁵ The patients were referred to undergo CAC testing from 1991–2010 due to the presence of risk factor(s) for atherosclerotic cardiovascular disease (ASCVD) and uncertainty in their long-term risk. Written informed consent was collected at each center at the time of CAC testing, and Institutional Review Board approval for coordinating center actions was by the Johns Hopkins University School of Medicine.

Coronary Artery Calcium Measurement

A standard protocol was used to quantify CAC using non-contrast, ECG-gated cardiac CT at all participating centers. 93% of the patients underwent electron beam CT testing and 7% underwent multidetector CT, and assessments have demonstrated no clinically significant difference in CAC ascertainment between the two methods.¹⁶ Scans were read using the Agatston method, and participants were subsequently categorized into 4 groups: CAC=0, 1 to 99, 100 to 399, ≥400. There have been previous studies published that have used the cutoff of 400.^17–20

Outcome Ascertainment

Mortality was determined by linking to the Social Security Administration Death Master File using a previously validated algorithm, and individual-level cause of death was ascertained through ICD coded death certificates from the National Death Index (NDI).²¹ There was up to 90% specificity and sensitivity for the identification of deaths when compared to known deaths identified through the electronic medical record in a subset of the CAC Consortium.¹⁵ For this study, HF mortality included any death where HF was listed as the underlying cause of death or in another location within the cause of death section of the death certificate. For sensitivity analyses, cases were excluded if the underlying cause of death was non-cardiovascular (for example, underlying cause of death of pneumonia or kidney failure with co-morbid heart failure). Participants had follow-up data until June 2014 and the mean follow-up was 12.3 +−3 years.

Evaluation of ASCVD Risk Factors

Participants’ baseline characteristics, risk factors, and laboratory data (as available) were gathered either during their CAC scan or as part of their routine clinical visit. Race information was only available for a subset of the study population (42,964 patients, 64%). Hypertension was defined as current use of antihypertensive medications or a previous diagnosis. Dyslipidemia was defined as having elevated triglycerides and/or low HDL-C, a previous diagnosis of hyperlipidemia, or receiving lipid-lowering medications. For participants with available lab results, dyslipidemia was determined by HDL-C levels <50 mg/dL in women or <40 mg/dL in men, LDL-C levels >160 mg/dL, or fasting triglyceride levels >150 mg/dL. Smoking status was categorized as current smokers or not. Diabetes was defined as a prior diagnosis or treatment with anti-diabetic agents. Family history of CAD was assessed differently across centers, with most centers considering a first-degree relative with a history of CAD, while the Columbus, OH site used a more stringent criterion (<55 years old in male relatives and <65 years old in female relatives).

To address partially missing data for risk factors (affecting up to 28% of the cohort), multiple imputation was used following a previously validated algorithm.¹⁵ The 10-year risk of ASCVD was calculated using the PCE and the PREVENT equations.^22,23

Statistical Analysis

The study population characteristics were categorized based on CAC burden, including CAC=0, 1–99, 100–399, and ≥400. Continuous variables were expressed as means and standard deviations, while categorical variables were presented as percentages. Given the non-normal distribution of CAC scores, the median was used to represent the central tendency. Statistical tests used include the Student’s t-test and Wilcoxon signed-rank test for normally and non-normally distributed continuous variables, respectively, and the Chi-square test for categorical variables.

For associative analyses, we compared HF mortality events across CAC burden categories. We grouped individuals into guideline-based ASCVD risk categories (<7.5%, 7.5–20%, and >20%) calculated by both the PCE and PREVENT equations. The low and borderline groups were combined to enhance statistical power due to a limited number of HF death events. The events were reported as absolute numbers and proportions for the overall sample and by CAC category, with event rates calculated per 1,000 patient-years of follow-up. Cumulative incidence (CI) plots were constructed for HF death events across CAC groups, and differences were assessed using the log-rank test. Receiver operator curves (ROC) were also constructed based on 10-year logistic regression models taking into account age, sex, study site, ASCVD risk score calculated by the PCE and PREVENT equations, and the incremental addition of CAC on the area under the curve (AUC) modeled as a log-transformed continuous variable. Differences in discriminatory ability was calculated using the Delong test.

The association between CAC burden and HF death was evaluated using competing risks regression with Fine and Gray hazard subdistribution models, and non-CVD mortality as the competing risk. Models were adjusted for age, sex, diabetes, current smoking, hypertension, hyperlipidemia, and family history of CAD.

Sensitivity analyses were conducted to explore the association between CAC and incident HF death after excluding individuals with a non-cardiovascular underlying cause of death. Statistical analyses were performed using STATA 17, with significance set at p<0.05.

Results

The average age of participants was 54.4 years, with 33% women, and 11% being non-white. A total of 36,879 individuals (55.3%) had a CAC score >0. More than half of the cohort (55%) had a 10-year ASCVD risk less than 5%. However, hypertension (31%), dyslipidemia (54.4%), and a family history of CAD (46.1%) were common among the cohort. With the exception of current smoking and a family history of CAD, the prevalence of ASCVD risk factors consistently increased with higher CAC burden groups (Table 1).

Table 1.

Baseline characteristics of participants in the CAC Consortium by CAC score categories.

Variable	Overall N=66,636	CAC 0 n=29,757	CAC 1–99 n=20,534	CAC 100–399 n=9,067	CAC ≥400 n=7,278
Age, mean +− SD, years	54.4 +− 10.6	49.9 +− 9.21	54.3 +− 9.5	59.3 +− 9.5	64.3 +− 9.7
Women, %	33	44.5	27.2	22.4	16.0
Race, %
White	89.1	88.7	88.9	90.4	89.5
Asian	3.8	4.2	3.5	3.4	3.5
Black	2.3	2.3	2.4	2.1	2.2
Hispanic	3.1	3.3	3.3	2.6	3.1
Other	1.7	1.6	1.9	1.6	1.8
CAC Score, mean +− SD, AU	164 +− 480	0 +− 0	28 +− 27	211 +− 85	1156 +− 976
CAC Score, median (Q1, Q3), AU	3 (0, 95)	0 (0, 0)	18 (5, 45)	193 (138, 275)	833 (566, 1367)
10-year ASCVD risk (PCE), mean +− SD	7.4 +− 8.9	3.8 +− 4.7	7.3 +− 7.5	11.5 +− 10.3	17.1 +− 13.5
10-year ASCVD risk category (PCE)
<7.5%	68.6	87.6	67.2	44.5	25.3
7.5–20%	23.5	11.0	26.9	40.5	44.1
>20%	7.9	1.4	5.9	15.1	30.6
10-year ASCVD risk (PREVENT), mean +− SD	6.0 +− 5.8	3.7 +− 3.5	5.9 +− 5.0	8.8 +− 6.5	12.3 +− 8.1
10-year ASCVD risk category (PREVENT)
<7.5%	74.2	89.7	74.5	54.7	34.3
7.5–20%	22.1	9.7	23.1	38.5	48.9
>20%	3.8	0.6	2.4	6.8	16.9
Hypertension, %	31	22.8	31.9	40.2	50.0
Dyslipidemia, %	54.4	48	56.6	60.9	66.0
Current smoker, %	9.6	8.9	9.7	11	10.7
Family history of CAD, %	46.1	45.6	46.4	46.2	47.3
Diabetes mellitus, %	6.8	3.9	6.6	9.6	15.4

There were 260 HF deaths over a median follow-up of 12.5 (10.6 – 14.1) years, and a total of 75.3% of all HF deaths occurred in patients with a baseline CAC>100. Furthermore, over half (53.1%) of all HF deaths occurred in individuals with CAC≥400. 67.7% of the deaths were in females, and 184 in those who were above the age of 65 at baseline (Table 1). The HF death rate was less than 0.6 per 1,000 patient-years in those with CAC<400 and greater than 1.6 per 1,000 patient-years in those with CAC≥400 (Figure 1).

Figure 1.

Absolute HF death event rate according to CAC burden.

The figure above displays the incidence of HF death rate. It was less than 0.6 per 1,000 patient-years in those with CAC<400 and greater than 1.6 per 1,000 patient-years in those with CAC≥400, indicating that asymptomatic adults with CAC>400 have a three times higher incidence of HF-related mortality.

Compared with a CAC=0, there was a stepwise higher risk (p<0.005) of HF mortality for CAC 1–99 (subdistribution hazard ratio [SHR]: 2.27; 95% CI: 1.3–3.99), 100–399 (SHR: 3.68; 95% CI 2.1–6.43), and ≥400 (SHR: 7.05; 95% CI 4.05–12.29) (Table 2). Females and younger adults at baseline in higher CAC burden groups had a steeper risk gradient as compared to males and adults above the age of 65 at baseline, respectively (Table 3).

Table 2.

Multivariable-adjusted SHRs (95% CIs) for CAC burden with HF mortality.

CAC Score Group	Events (n = 260)	Unadjusted Rate (95% CI)	Model 1 SHR (95% CI)A	p	Model 2 SHR (95% CI)B	p
0	17	0.05 (0.03 – 0.07)	Ref	Ref	Ref	Ref
1 to 99	47	0.18 (0.14 – 0.24)	2.34 (1.36 – 4.19)	0.002	2.27 (1.30 – 3.99)	0.004
100 to 399	58	0.53 (0.41 – 0.68)	4.07 (2.32 – 7.15)	<0.001	3.68 (2.10 – 6.43)	<0.001
≥400	138	1.67 (1.41 – 1.97)	8.24 (4.70 – 14.43)	<0.001	7.05 (4.05 – 12.29)	<0.001
Log (CAC+1)			1.41 (1.30 – 1.53)	<0.001	1.37 (1.27 – 1.48)	<0.001

^A:Adjusted for age, sex, and study site.

^B:Adjusted for age, sex, study site, HTN, HLD, family history, DM, smoker.

Table 3.

Multivariable-adjusted SHRs (95% CIs) for CAC burden with HF mortality stratified by sex and age.

CAC Score Group	N	Model 1 SHR (95% CI)A	p	Model 2 SHR (95% CI)B	p2
Female	176
0		Ref	Ref	Ref	Ref
1 to 99		3.6 (1.62 – 7.99)	0.002	3.23 (1.45 – 7.2)	0.004
100 to 399		5.52 (2.42 – 12.62)	<0.001	4.59 (2.04 – 10.33)	<0.001
≥400		14.2 (6.13 – 32.85)	<0.001	10.46 (4.57 – 23.96)	<0.001
Log (CAC+1)		1.52 (1.33 – 1.73)	<0.001	1.45 (1.28 – 1.64)	<0.001
Male	84
0		Ref	Ref	Ref	Ref
1 to 100		1.64 (0.76 – 3.54)	0.209	1.61 (0.74 – 3.47)	0.228
100 to 400		3.08 (1.45 – 6.51)	0.003	2.89 (1.37 – 6.11)	0.005
>400		5.53 (2.66 – 11.52)	<0.001	5.05 (2.43 – 10.48)	<0.001
Log (CAC+1)		1.35 (1.22 – 1.49)	<0.001	1.32 (1.20 – 1.46)	<0.001
Age <65	76
0		Ref	Ref	Ref	Ref
1 to 99		3.52 (1.65 – 7.49)	0.001	3.39 (1.59 – 7.24)	0.002
100 to 399		6.97 (3.25 – 14.95)	<0.001	6.45 (3.00 – 13.89)	<0.001
≥400		17.91 (7.70 – 41.67)	<0.001	15.4 (6.51 – 36.40)	<0.001
Log (CAC+1)		1.50 (1.33 – 1.70)	<0.001	1.47 (1.29 – 1.66)	<0.001
Age >65	184
0		Ref	Ref	Ref	Ref
1 to 99		1.87 (0.80 – 4.35)	0.146	1.84 (0.79 – 4.29)	0.159
100 to 399		3.08 (1.36 – 7.00)	0.007	2.88 (1.27 – 6.53)	0.011
≥400		5.71 (2.56 – 12.73)	<0.001	5.04 (2.27 – 11.19)	<0.001
Log (CAC+1)		1.39 (1.24 – 1.55)	<0.001	1.35 (1.21 – 1.50)	<0.001

^A:Adjusted for age, sex, and study site.

^B:Adjusted for age, sex, study site, HTN, HLD, family history, DM, smoker.

Amongst the low/borderline and intermediate-risk groups of individuals as calculated by the PCE, there was a stepwise increasing risk for HF death across CAC burden categories. In the high-risk group, there was a higher risk of HF death in those with CAC≥400.

Risk of HF-related mortality stratified by risk group according to the PREVENT equation showed a similar stepwise increase across higher CAC scores. The difference in HF-related mortality risk between individuals with CAC 100–399 and those with CAC ≥400 was most appreciated amongst those at low ASCVD risk calculated by the PCE and PREVENT equations (Table 4).

Table 4.

Multivariable-adjusted SHRs (95% CIs) for CAC burden with HF mortality stratified by ASCVD and PREVENT risk groups.

CAC Score Group	N	Model 1 SHR (95% CI)A	p	Model 2 SHR (95% CI)B	p2
ASCVD Low/Borderline Risk (PCE)	36
0		Ref	Ref	Ref	Ref
1 to 99		3.31 (1.40 – 7.82)	0.006	3.14 (1.32 – 7.49)	0.010
100 to 399		3.57 (1.12 – 11.37)	0.031	3.35 (1.02 – 10.96)	0.046
≥400		20.97 (7.15 – 61.44)	<0.001	18.69 (6.36 – 54.95)	<0.001
Log (CAC+1)		1.43 (1.21 – 1.69)	<0.001	1.41 (1.20 – 1.67)	<0.001
ASCVD Intermediate Risk (PCE)	76
0		Ref	Ref	Ref	Ref
1 to 99		3.56 (1.03 – 12.34)	0.045	3.74 (1.08 – 12.95)	0.037
100 to 399		8.17 (2.44 – 27.37)	0.001	8.46 (2.54 – 8.20)	0.001
≥400		11.33 (3.27 – 39.18)	<0.001	11.54 (3.36 – 39.65)	<0.001
Log (CAC+1)		1.40 (1.22 – 1.60)	<0.001	1.40 (1.22 – 1.60)	<0.001
ASCVD High Risk (PCE)	148
0		Ref	Ref	Ref	Ref
1 to 99		1.29 (0.48 – 3.50)	0.611	1.33 (0.49 – 3.65)	0.576
100 to 399		1.78 (0.68 – 4.61)	0.237	1.79 (0.69 – 4.66)	0.234
≥400		3.64 (1.48 – 9.02)	0.005	3.55 (1.43 – 8.83)	0.006
Log (CAC+1)		1.39 (1.21 – 1.60)	<0.001	1.37 (1.22 – 1.56)	<0.001
ASCVD Low/Borderline Risk (PREVENT)	50
0		Ref	Ref	Ref	Ref
1 to 99		2.56 (1.11 – 5.90)	0.027	2.46 (1.07 – 5.65)	0.035
100 to 399		4.52 (1.86 – 11.01)	0.001	4.20 (1.69 – 10.44)	0.002
≥400		15.51 (6.01 – 40.05)	<0.001	14.53 (5.56 – 37.96)	<0.001
Log (CAC+1)		1.42 (1.22 – 1.64)	<0.001	1.40 (1.20 – 1.63)	<0.001
ASCVD Intermediate Risk (PREVENT)	105
0		Ref	Ref	Ref	Ref
1 to 99		3.73 (1.09 – 12.75)	0.036	3.64 (1.07 – 12.41)	0.039
100 to 399		7.55 (2.27 – 25.07)	0.001	6.98 (2.12 – 23.01)	0.001
>400		12.92 (3.88 – 43.06)	<0.001	11.53 (3.51 – 37.91)	<0.001
Log (CAC+1)		1.46 (1.28 – 1.66)	<0.001	1.43 (1.26 – 1.62)	<0.001
ASCVD High Risk (PREVENT)	105
0		Ref	Ref	Ref	Ref
1 to 99		1.33 (0.43 – 4.12)	0.623	1.41 (0.45 – 4.45)	0.561
100 to 399		1.49 (0.50 – 4.43)	0.477	1.53 (0.51 – 4.62)	0.449
≥400		2.86 (1.01 – 8.08)	0.047	2.89 (1.01 – 8.28)	0.048
Log (CAC+1)		1.32 (1.13 – 1.55)	<0.001	1.31 (1.12 – 1.52)	0.001

^A:Adjusted for age, sex, and study site.

^B:Adjusted for age, sex, study site, HTN, HLD, family history, DM, smoker.

Modeling CAC as a log-transformed continuous variable also showed an increased risk of HF death across all ages, genders, and risk groups. A sensitivity analysis was performed excluding 71 patients with non-cardiovascular underlying cause of death (N=189), and these results showed a similar stepwise increasing risk of death across the CAC burden categories (Supplemental Table 5).

CI plots showed divergence in HF-related mortality event rates in the overall sample as early as less than one year of follow-up in those with CAC >400, and at four years in the remaining CAC groups (Figure 2A). Amongst those in the low/borderline risk group, divergence in CI did occur in those with CAC >400 as early as less than one year. (Figures 2B, C). In the intermediate risk groups calculated using the PCE, those with CAC 100–399 and >400 diverged from CAC 0 and 1–99 about four years after follow-up. When calculated by the PREVENT equations, the curve shows divergence across all CAC groups starting at six years.

Figure 2:

Cumulative incidence plots for survival probability in the overall sample (A), individuals with a 10-year ASCVD risk <7.5% calculated by PCE and PREVENT (B,C), 7.5–20% (D,E), >20% (F,G). p <0.001 for all figures.

Figure 2 A is a Cumulative incidence curve which shows divergence in HF-related mortality event rates in the overall sample as early as less than one year of follow-up in those with CAC >400, and at four years in the remaining CAC groups.

Figures 2 B,C are Cumulative incidence curves of HF-related mortality event rates in the low/borderline risk groups calculated by both the PCE AND PREVENT risk calculators. They show minimal divergence in event rates in patients with CAC<400. However, divergence in cumulative incidence did occur in those with CAC >400 as early as less than one year.

Figures 2 D,E are Cumulative Incidence curves of HF-related mortality event rates in the intermediate-risk groups calculated by both the PREVENT AND PCE risk calculators. When risk is calculated using the PCE, those with CAC 100–399 and >400 diverged from CAC 0 and 1–99 about four years after follow-up. When calculated by the PREVENT equations, the curve shows divergence across all CAC groups starting six years.

Figures 2 F,G are Cumulative Incidence curves of HF-related mortality event rates in the high-risk groups calculated by the PREVENT AND PCE risk calculators. Those with CAC >400 diverged from those with CAC<400 as early as four years after follow-up, and had a higher divergence at ten years when compared to high-risk patients in lower CAC score groups.

In the high-risk groups calculated by both risk calculators, those with CAC >400 diverged as early as four years after follow-up and had a higher divergence at ten years then all other risk groups (Figure 2 F,G).

The addition of CAC score improved the discrimination of HF-related mortality when added to base models, which accounted for age, sex, study site, and risk score calculated by either the PCE or PREVENT equations (AUC=0.86 versus AUC=0.883, Delta+0.023, p=0.001 for PCE and AUC0.86 versus AUC=0.89, Delta+0.027, p=0.005 for PREVENT) (Figure 3).

Figure 3.

Receiver operating characteristics (ROC) curves displaying the base model accounting for age, sex, study site, and ASCVD risk score (calculated by the PCE), with addition of total calcium score (A). ROC displaying the base model accounting for age, sex, study site, and ASCVD risk score (calculated by PREVENT), with addition of total calcium score (B).

The ROC curves above display the base model which accounts for age, sex, study site, and risk score calculated by the PCE, and the base model with the addition of CAC score. It shows that the addition of CAC score improves the discrimination of HF-related mortality (AUC=0.86 versus AUC=0.883, Delta+0.023, p=0.001).

The ROC curves above display the base model which accounts for age, sex, study site, and risk score calculated by the PREVENT equations, and the base model with the addition of CAC score. It shows that the addition of CAC score improves the discrimination of HF-related mortality (AUC=0.86 versus AUC=0.89, Delta+0.027, p=0.005).

Discussion

In this large sample of asymptomatic adults undergoing CAC testing for risk stratification, we observed a stepwise increase in the event rate of death due to HF across increasing CAC burden groups. Individuals with CAC 1–99, 100–399, and ≥400 had between a 2.3 to 8.2-fold higher risk for death due to HF compared to those with CAC 0, independent of traditional risk factors and adjusted for the competing risk of non-cardiovascular disease death.

CAC has been previously shown to be associated with HF in several studies. The Heinz Nixdorf Recall study examined the relationship between CAC and HF in the general population and found that individuals with CAC had a higher prevalence of HF compared to those without CAC.²⁴ The Rotterdam Study similarly found that higher levels of CAC amongst the elderly were associated with an increased risk of HF.²⁵ Another from the Multi-Ethnic Study of Atherosclerosis (MESA) showed that over a median follow-up of 9.6 years, 3.2% of subjects developed incident HF.²⁶

CAC has been traditionally used as a surrogate of plaque burden, however, it is increasingly recognized as a marker of overall cardiovascular aging.⁹ The novelty of this study is that it was conducted in a clinical sample of primary prevention patients undergoing CAC testing for ASCVD risk prediction, with a long-term outcome of HF-related mortality as defined by death certificates.

Although this study cannot distinguish ischemic versus nonischemic causes of HF, or HF with preserved ejection fraction (HFpEF) versus HF with reduced ejection fraction (HFrEF), CAD is the most common cause of HF in the United States. It exerts its pathophysiology through mechanisms beyond clinically overt CAD, including through hypoperfusion and endothelial dysfunction which can cause subclinical myocardial dysfunction, eventually progressing to symptomatic HF.^11,27 Subclinical atherosclerosis has also been shown to be associated with cardiac and aortic remodeling in individuals free of HF and MI, and with alterations of myocardial strain parameters suggesting that development of coronary vascular and myocardial disease may be a common phenomenon occurring in tandem amongst aging individuals.²⁸ Higher CAC has also been shown to be associated with higher LV mass and volumes, and worse LV systolic and diastolic function, indicating potential adverse cardiac remodeling.^29,30

The findings in our study are novel in that all patients were free of clinical ASCVD and HF at baseline. Interestingly, subjects below the age of 65 and with CAC ≥400 at baseline were found to have SHR of HF mortality of 17.91 (95% CI: 7.70–41.67) after adjusting for confounders. This is a marked increase from their respective counterparts above the age of 65, which can be explained by the increasing prevalence of higher CAC with increasing age. Another group in this study that had a disproportionately higher risk of HF related death is those in the low/borderline and intermediate risk groups calculated by the PCE with CAC ≥400. This compared similarly to those with CAC ≥400 in the low/borderline risk groups calculated by the PREVENT equations. Generally, the risk gradient was less marked in patients already deemed to be high risk by the PCE or PREVENT equations.

Because of the progressive nature of HF, it is imperative to detect asymptomatic individuals at risk of suffering HF-related mortality. Notably, two of the four categories in the stages of HF, A and B, include patients who are asymptomatic.³¹ Stage A represents patients with risk factors for HF including traditional risk factors for CAD and CAD itself, while stage B represents those with structural heart disease such as left ventricular hypertrophy or systolic dysfunction, but who are asymptomatic from an HF perspective. Those with elevated CAC may also benefit from being included amongst those with stage B heart failure, prompting further investigation with wither historical or quantitative assessment of functional capacity, with results likely attributable predominantly to ischemic events.

HF prevention has historically been emphasized in the secondary prevention population, and management of these patients generally focused on reducing hospitalizations. There is growing evidence supporting a paradigm shift towards primary HF prevention. Although there is overlap in lifestyle management for prevention of ASCVD and HF, there are differences pertaining to medical prevention of HF. For example, in patients with type 2 diabetes who had or were at risk for ASCVD, dapagliflozin, and liraglutide and dulaglutide, a sodium-glucose transport protein 2 (SGLT2) inhibitor and glucagon-like peptide-1 (GLP-1) receptor agonists, respectively, lowered the rate of cardiovascular death or hospitalization for HF.^32–34 Furthermore, Finerenone, a novel non-steroidal mineralocorticoid receptor antagonist has also been shown to be effective in reducing the risk of HF in patients with chronic kidney disease (CKD) and/or type 2 diabetes.³⁵ The results of our study suggest that higher risk populations may benefit from initiation of medical management in addition to lifestyle modifications to curtail their risk of developing heart failure, and in doing so, decrease their risk of HF-associated mortality. This would however, need to be corroborated by further studies.

This study’s key strengths include assessing CAC in over 66,000 individuals free of known cardiovascular disease with an average prevalence of ASCVD risk factors. Some limitations include the overall low number of HF death events, which is due to the fact that the CAC Consortium is a primary prevention population. Furthermore, the CAC Consortium is unfortunately underpowered for less represented risk groups such as, African Americans, Asiatic populations, and non-whites in general. Previous studies have shown that CAC strongly predicts CAD and all-cause mortality amongst Asian-Americans and in all studied races/ethnic groups.^19,36–38 Lastly, although it is a major limitation that the inclusion of African American and Asiatic populations is just 11%, national health data indicates that their population together comprises about 18% of the US population and 11% therefore aligns reasonably and may reflect the broader demographic trends observed in healthcare utilization for primary prevention in the United States. Unfortunately, the CAC Consortium did not collect echocardiogram, stress testing, or catheterization data which makes it impossible to definitively distinguish between patients who developed HF with reduced versus preserved ejection fraction, and those with ischemic versus nonischemic cardiomyopathies. The CAC Consortium also did not collect any interim non-fatal events prior to death. For example, we do not have data on which patients might have suffered myocardial infarctions or strokes antecedent to their HF-related death.

Conclusion

In conclusion, this study found that higher CAC was associated with increasing incidence of HF-related mortality, particularly among primary prevention patients at intermediate and high estimated 10-year cardiovascular risk. Risk stratification and more intensive focus on lifestyle therapies and medication initiation are likely to have a role in these groups to prevent HF in asymptomatic adults in the pre-HF stages. Future studies are needed assessing the role of CAC scoring to guide primary prevention populations for risk assessment of death due to HF, and in doing so, help drive personalized discussions regarding further lifestyle and medication management.

Abbreviations:

HF

Heart failure

CAC

Coronary artery calcium

CT

Computed tomography

ASCVD

Atherosclerotic cardiovascular disease

PCE

Pooled cohort equation

CAD

Coronary artery disease

HFpEF

Heart failure with preserved ejection fraction

HFrEF

Heart failure with reduced ejection fraction

GLP-1

Glucagon-like peptide-1

SGLT2

Sodium-glucose transport protein 2

CKD

Chronic kidney disease