Relationship Between Aortic Valve and Mitral Annular Calcification With Coronary Artery Calcification in Asymptomatic Individuals: Cross-Sectional and Longitudinal Analyses

Abstract

Background

Aortic valve calcification (AVC), mitral annular calcification (MAC), and coronary artery calcification (CAC) all share common atherosclerotic origins. However, the relationship between these entities is not fully understood.

Methods

A total of 722 asymptomatic individuals who underwent health screening with serial cardiac computed tomography (CT) were retrospectively selected for analysis. AVC, MAC, and CAC were identified on CT, and the severity was quantified using Agatston units (AU). Multivariable regression models were used to identify the association between the severity of CAC and the probability of prevalent AVC and MAC, and the relation between annualized progression rates of AVC, MAC and CAC.

Results

On initial CT, the prevalence of AVC, MAC, and CAC was 11.4%, 6.5%, and 46.3%, respectively. Increasing baseline CAC severity was associated with a higher probability of both prevalent AVC (odds ratio [OR] per 100 AU increase, 1.03; 95% confidence interval [CI], 1.02–1.04; P < 0.001) and MAC (OR per 100 AU increase, 1.06; 95% CI, 1.05–1.07; P < 0.001), even after correction for other risk factors. On follow-up CT, the interval changes of MAC and CAC scores were also highly correlated (P < 0.001). However, no significant relationship was found between the interval changes of AVC and MAC scores, or between AVC and CAC.

Conclusion

We observed a close correlation between AVC, MAC, and CAC, which is in accord with their common atherosclerotic origin. However, the correlation between MAC and CAC progression but not with AVC suggests that other factors such as hemodynamics may have an important role in the further development of calcification.

Graphical Abstract

INTRODUCTION

Valvular heart diseases caused by aortic valve calcification (AVC) and mitral annular calcification (MAC) are a significant cause of cardiovascular mortality and morbidity worldwide.1 AVC and MAC are known to share features similar to atherosclerosis.2,3 Previous studies have found that factors traditionally associated with atherosclerosis, such as old age, hypertension, diabetes, obesity, and smoking, are responsible for the development of both AVC and MAC.1,2,3,4,5,6 Further, the prevalence of both AVC and MAC has been found to be directly related to the severity of coronary artery calcification (CAC), supporting a common development mechanism.7,8,9 Computed tomography (CT) is a sensitive and reproducible method to detect and quantify the disease burden in patients with these conditions.10,11,12 However, it is known that once after initiation, the progression of cardiovascular calcification is heavily influenced by local factors related to bone-mineral metabolism,13,14,15,16,17 and it has not been investigated whether the accelerated progression of calcification in one location is associated with a higher progression rate in another. Also, the findings linking AVC, MAC and CAC have not been investigated in the East Asian population, which is known to have a lower burden of atherosclerosis compared to other ethnic groups.6,18 Therefore, we aimed to assess the relation between AVC, MAC and CAC, and identify factors associated with calcification progression, in a cohort of ethnic Koreans who underwent cardiac CT as part of a self-referred health examination.

METHODS

Study design and population

This study used data from the KOrea Initiatives on Coronary Artery calcification (KOICA) registry, a retrospective, observational, multicenter registry of 93,914 individuals undergoing self-referred health examination with cardiac CT at six high-volume healthcare centers in Korea. Further details regarding the registry can be found in previous reports.18,19 From the registry, we selected those who were examined at Seoul St. Mary’s Hospital, and with at least two CT scans during the study period (April 2009 to July 2016). All CT scans including follow-up CT scans were obtained as part of a health examination using the same protocol as detailed below. Patients with previous percutaneous coronary intervention (PCI) were excluded from analysis.

Using a questionnaire form, participants self-reported details on demographics, current symptoms, and medical history. Laboratory samples were obtained after a 12-hour fast, and the lipid levels including low-density lipoprotein (LDL) cholesterol were measured using a direct enzymatic method (Hitachi Medical Corp., Tokyo, Japan). The estimated glomerular filtration function (eGFR) was calculated using the CKD-EPI equation.

Image acquisition

Cardiac CT scans were obtained using a 64-slice, dual-source CT scanner (SOMATOM Definition; Siemens, Forchheim, Germany). A non-contrast calcium scan was first acquired using prospective electrocardiogram-triggering at 70% of the RR interval (tube voltage 120 kVp, gantry rotation time 330 ms, and maximum tube current 400 mA∙s). Next, enhanced CT angiography scans were obtained after the administration of 80–110 mL of iodinated contrast using a retrospective electrocardiogram-gated protocol. CT scans were transferred and reconstructed using a computerized workstation (Advantage Windows Workstation 4.3; GE Healthcare, Milwaukee, WI, USA) with a slice thickness of 3 mm.

AVC and MAC were assessed using CT images with a commercially available CT processing program (3mensio Structural Heart 10.0; Pie Medical Imaging, Maastricht, The Netherlands). The severity of AVC and MAC was quantified according to the Agatston method on non-contrast axial images.12,20 The AVC score was defined as the sum of the values for each calcified lesion between the base and tip of the aortic valve leaflets, as recommended in the evaluation of patients with aortic stenosis.12 Likewise, the MAC score was defined as the sum of the values for each calcified lesion on the mitral annulus, as validated in previous studies.8,13,21 Calcifications of the coronary arteries, aortic root, and left ventricular outflow tract were carefully excluded. All analyses were done by the consensus of two cardiologists (K.A.K and S.-Y.L.), and a third cardiologist (H.-O.J) with more than 20 years of experience in cardiac imaging was consulted if necessary.

The severity of CAC was analyzed by radiologists specializing in cardiac imaging using dedicated software (syngo.CT CaScoring; Siemens Healthcare, Forcheim, Germany) in accord with standard guidelines.22 Only segments with a diameter larger than 1.5 mm were included for analysis, and the sum of all lesion scores for each major coronary artery was used to generate the total CAC score using the Agatston method. Further details on the coronary CT angiography protocol and reporting system at our institution can be found in previous publications.23,24

Statistical analysis

Categorical data are presented as numbers and frequencies and compared using the χ² test or Fisher’s exact test. Continuous variables are expressed as mean ± standard deviation and compared using the Student’s t-test or Mann–Whitney U test as appropriate.

AVC and MAC were defined as prevalent when AVC score > 0 or MAC score > 0 on initial CT, and the association between the severity of CAC and prevalent AVC and MAC and was analyzed with adjustment for confounders using multivariable logistic regression. Variables considered clinically significant or with P < 0.1 in univariable analysis were selected for multivariable analysis, and a stepwise backward elimination process was used while retaining age, sex, and the use of antihypertensive and glucose-lowering medications until arrival at the final number of variables according to the Akaike information criterion. The association between the CAC scores and prevalent AVC and MAC was investigated with the CAC score analyzed both as an ordinal variable in four groups based on commonly reported cutoff points (CAC score 0, 1–99, 100–399, and ≥ 400, with CAC score=0 as the reference group)25 and as a continuous variable. Restricted cubic splines were constructed to visualize the relationship between CAC severity and the probability of AVC and MAC, with knots placed at CAC scores 10, 100, and 400.

For longitudinal analysis, the correlations between the change in AVC, MAC, and CAC scores were explored using Pearson’s correlation coefficients. Our primary approach was to use the scores in their original forms; however, due to the skewed distribution of the variables we also performed sensitivity analysis using square-root transformed scores. Second, factors influencing the progression rate of AVC, MAC, and CAC were compared using a multivariable linear mixed-effects model with individual intercepts. This method has been used in previous studies on cardiovascular calcifications and has the advantage of using all available data, and without the need to analyze de novo calcification and progression from baseline calcification separately.26 Adjustment was made for baseline age, sex, body mass index (BMI), smoking, hypertension, diabetes, antihypertensive medications, glucose-lowering medications, lipid-lowering medications, systolic blood pressure, leukocyte count, C-reactive protein, glycated hemoglobin (HbA1c), high-density lipoprotein (HDL) cholesterol, LDL cholesterol, triglycerides, eGFR, calcium, and phosphate, as well as the corresponding baseline calcium score (e.g. baseline AVC score for AVC progression analysis).

Statistical analyses were performed using R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria), and a two-sided P value < 0.05 was considered statistically significant. Further details of the statistical methods used in this study can be found in the Supplementary Material (Supplementary Data 1).

Ethics statement

The Institutional Review Board (IRB) of Seoul St. Mary’s Hospital granted approval for the study (IRB KC23RISI0357) and waived the need for written informed consent due to its retrospective nature. This study is in accord with the Declaration of Helsinki.

RESULTS

Baseline characteristics

The flowchart of the study population is shown in Fig. 1. From the KOICA registry, 722 individuals who underwent examination at Seoul St. Mary’s Hospital with at least two CT scans and without previous PCI were selected for baseline cross-sectional analysis. In longitudinal analysis, 30 patients who underwent PCI during the interval and one individual with incomplete data were further excluded. The median follow-up duration was 3.2 (interquartile range [IQR], 2.0–7.2) years (Supplementary Fig. 1), and 1,625 CT scans from 691 individuals were included.

Fig. 1. Selection process of the study population.

KOICA = KOrea Initiatives on Coronary Artery calcification, CT = computed tomography, PCI = percutaneous coronary intervention.

The baseline characteristics of the study population are shown in Table 1. The mean age of the participants was 54.9 years, and 80.3% were male. The individuals with CAC were more likely to be older, male, and have hypertension, diabetes, and dyslipidemia. They also had significantly higher systolic and diastolic blood pressures, higher HbA1c, lower HDL-cholesterol and eGFR, and were also more likely to have prevalent AVC and MAC. Prevalent AVC was identified in 82 (11.4%) of the participants, while prevalent MAC was identified in 47 (6.5%). A high proportion of the individuals with AVC and MAC had calcifications in other locations, and isolated AVC or MAC was uncommon (Fig. 2). The median AVC score was 27.1 (IQR, 10.3–56.7) AU, the median MAC score 13.4 (IQR, 2.8–76.7) AU, and the median CAC score 38.6 (IQR, 8.3–135.0) AU in those with prevalent AVC, MAC, and CAC, respectively (Supplementary Fig. 2). The echocardiographic characteristics of the study population are shown in Supplementary Table 1. The individuals with AVC were more likely to have aortic regurgitation and aortic stenosis, and two participants had bicuspid aortic valves, both of whom had prevalent AVC. Individuals with AVC also had higher left ventricular outflow tract velocities, higher left ventricular mass index, and worse diastolic profiles. Similarly, the individuals with MAC were more likely to have mitral regurgitation, and had higher left ventricular mass index and worse diastolic profiles.

Table 1. Baseline characteristics of the study population stratified according to the presence of coronary artery calcification.

Characteristics		CAC score = 0 (n = 370)	CAC score > 0 (n = 352)	P value
Age, yr		52.6 ± 8.1	57.2 ± 7.9	< 0.001
Sex				< 0.001
	Male	266 (71.9)	314 (89.2)
	Female	104 (28.1)	38 (10.8)
BMI, kg/m2		24.6 ± 3.6	25.0 ± 2.7	0.052
Smoking		135 (36.5)	153 (43.5)	0.066
Hypertension		96 (25.9)	154 (43.8)	< 0.001
Diabetes		31 (8.4)	65 (18.5)	< 0.001
Atrial fibrillation		0 (0.0)	4 (1.1)	0.056
Antihypertensive medications		82 (22.2)	141 (40.1)	< 0.001
Glucose-lowering medications		25 (6.8)	62 (17.6)	< 0.001
Lipid-lowering medications		32 (8.6)	60 (17.0)	0.001
Systolic blood pressure, mmHg		123 ± 14	126 ± 12	0.005
Diastolic blood pressure, mmHg		74 ± 10	76 ± 9	0.008
WBC count, 109/L		5.82 ± 1.67	5.89 ± 1.63	0.586
hs-CRP, mg/dL		0.14 ± 0.29	0.18 ± 0.42	0.191
HbA1c		5.6 ± 0.5	5.9 ± 0.8	< 0.001
HDL-cholesterol, mg/dL		50.5 ± 11.6	48.7 ± 10.7	0.033
LDL-cholesterol, mg/dL		119 ± 29	122 ± 35	0.261
Triglycerides, mg/dL		122 ± 79	135 ± 78	0.048
eGFR, mL/min/1.73 m2		85.7 ± 9.8	82.0 ± 10.2	< 0.001
Calcium, mg/dL		9.2 ± 0.4	9.2 ± 0.3	0.214
Phosphate, mg/dL		3.4 ± 0.5	3.5 ± 0.5	0.662
CAC severity grade				< 0.001
	None (0 AU)	370 (100.0)	0 (0.0)
	Mild (1–99 AU)	0 (0.0)	243 (69.0)
	Moderate (100–399 AU)	0 (0.0)	80 (22.7)
	Severe (≥ 400 AU)	0 (0.0)	29 (8.2)
Prevalent AVC		20 (5.4)	62 (17.7)	< 0.001
	AVC score, AU	5.5 ± 55.7	9.2 ± 35.1	0.291
Prevalent MAC		9 (2.4)	38 (10.8)	< 0.001
	MAC score, AU	0.9 ± 9.1	8.6 ± 58.9	0.014

CAC = coronary artery calcification, BMI = body mass index, WBC = white blood cell, hs-CRP = high-sensitivity C-reactive protein, HbA1c = glycated hemoglobin, HDL = high density lipoprotein, LDL = low-density lipoprotein, eGFR = estimated glomerular filtration rate, AU = Agatston unit, AVC = aortic valve calcification, MAC = mitral annular calcification.

Fig. 2. Distribution of prevalent aortic valve calcification, mitral annular calcification, and coronary artery calcification at baseline.

AVC = aortic valve calcification, MAC = mitral annular calcification, CAC = coronary artery calcification.

Association of CAC severity with the probability of AVC and MAC

In baseline cross-sectional analysis, the prevalence of AVC increased across CAC categories, from 5.4% in those with no CAC to 37.9% in those with severe CAC (≥ 400 AU) (Supplementary Table 2). On multivariable logistic regression, increasing CAC severity was associated with a significantly higher probability of prevalent AVC (P < 0.001) (Supplementary Table 3). Compared to those without CAC, the odds ratio (OR) for the detection of prevalent AVC was 1.06 (95% confidence interval [CI], 1.01–1.12) for mild CAC (1–99 AU), 1.02 (95% CI, 0.95–1.10) for moderate CAC (100–399 AU), and 1.26 (95% CI, 1.12–1.42) in those with severe CAC (≥ 400 AU). Similar results were found when the CAC score was analyzed as a continuous variable (OR per 100 AU, 1.03; 95% CI, 1.02–1.04; P < 0.001) (Supplementary Table 4). Other factors associated with prevalent AVC were older age (OR per 10 years, 1.13; 95% CI, 1.10–1.17; P < 0.001) and higher BMI (OR per 5 kg/m², 1.05; 95% CI, 1.01–1.08; P = 0.012) (Table 2).

Table 2. Relationship between baseline aortic valve and mitral annular calcification with coronary artery calcification severity.

Characteristics		Aortic valve calcificationa		Mitral annular calcificationb
		OR (95% CI)	P value	OR (95% CI)	P value
CAC severity grade			< 0.001		< 0.001
	None (0 AU)	Reference		Reference
	Mild (1–99 AU)	1.06 (1.01–1.12)		1.01 (0.97–1.04)
	Moderate (100–399 AU)	1.02 (0.95–1.10)		1.07 (1.01–1.13)
	Severe (≥ 400 AU)	1.26 (1.12–1.42)		1.60 (1.46–1.74)
Age (per 10 yr)		1.13 (1.10–1.17)	< 0.001	1.04 (1.02–1.07)	< 0.001
Female sex		1.01 (0.96–1.07)	0.691	1.05 (1.00–1.09)	0.034
BMI (per 5 kg/m2)		1.05 (1.01–1.08)	0.012	1.05 (1.00–1.09)	0.015
Antihypertensive medications		1.03 (0.98–1.08)	0.255	1.02 (0.98–1.05)	0.430
Glucose-lowering medications		1.02 (0.95–1.09)	0.541	1.08 (1.03–1.14)	0.002
eGFR (per 10 mL/min/1.73 m2)		1.02 (1.00–1.05)	0.126	0.99 (0.97–1.01)	0.237
Bicuspid aortic valve		2.35 (1.60–3.44)	< 0.001

The area under the curve for the models are 0.85 and 0.89, respectively.

OR = odds ratio, CI = confidence interval, BMI = body mass index, eGFR = estimated glomerular filtration rate.

^aAdjusted for coronary artery calcification severity (in categories) + age, sex, BMI, antihypertensive medications, glucose-lowering medications, estimated glomerular filtration index, and bicuspid aortic valve. ^bAdjusted for coronary artery calcification severity (in categories) + age, sex, BMI, antihypertensive medications, glucose-lowering medications, and estimated glomerular filtration index.

The prevalence of MAC also increased across CAC categories, from 2.3% in those with no CAC to 55.2% in those with severe CAC (≥ 400 AU) (Supplementary Table 2). On multivariable logistic regression, increasing CAC severity was associated with a significantly higher probability of prevalent MAC (P < 0.001) (Table 2). Compared to those without CAC, the OR for the detection of prevalent AVC was 1.01 (95% CI, 0.97–1.04) for mild CAC (1–99 AU), 1.07 (95% CI, 1.01–1.13) for moderate CAC (100–399 AU), and 1.60 (95% CI, 1.46–1.74) in those with severe CAC (≥ 400 AU). Similar results were found when the CAC score was analyzed as a continuous variable (OR per 100 AU, 1.06; 95% CI, 1.05–1.07; P < 0.001) (Supplementary Table 4). Other factors associated with prevalent AVC were older age (OR per 10 years, 1.08; 95% CI, 1.05–1.10; P = 0.001), female sex (OR, 1.05; 95% CI, 1.01–1.10; P = 0.012), higher BMI (OR per 5 kg/m², 1.043; 95% CI, 1.01–1.06; P = 0.011), and the use of glucose–lowering medications (OR, 1.08; 95% CI, 1.02–1.13; P = 0.004) (Table 2).

Analysis using restricted cubic splines also found that increasing CAC severity was associated with increasing probability of AVC and MAC across the entire range of CAC scores (Fig. 3). Notably, the relation between CAC severity and MAC probability was more pronounced than the relation found for AVC probability.

Fig. 3. Restricted cubic splines demonstrating the relationship between the severity of coronary artery calcification and the probability of (A) aortic valve calcification (B) mitral annular calcification. Adjusted for coronary artery calcification severity, age, sex, body mass index, estimated glomerular filtration rate, antihypertensive medications, glucose-lowering medications, and bicuspid aortic valve (for aortic valve calcification).

OR = odds ratio, AVC = aortic valve calcification, MAC = mitral annular calcification, CAC = coronary artery calcification, AU = Agatston unit.

Relationship between longitudinal changes in AVC, MAC, and CAC scores

On the last CT scan, AVC scores increased by 11.8 ± 62.2 AU, MAC scores by 14.3 ± 142.7 AU, and CAC scores by 90.0 ± 241.4 AU compared to the baseline CT (Supplementary Fig. 3). There was no significant correlation between the changes in AVC and CAC scores, or the changes in AVC and MAC scores; however, there was a weak correlation between the changes in MAC and CAC scores (Pearson correlation coefficient, 0.23; 95% CI, 0.18–0.28; P < 0.001) (Supplementary Table 5). This was also true in sensitivity analysis using square-root transformed scores (Supplementary Table 6).

When multivariable linear mixed effects models were used to identify factors associated with the rate of AVC, MAC, and CAC progression, only the baseline AVC score had a significant association with AVC progression (Table 3). Factors associated with MAC progression were the baseline MAC score, smoking (β-coefficient, 0.99; 95% CI, 0.18–1.80; P = 0.015), and phosphate (β-coefficient per 1 mg/dL, 0.82; 95% CI, 0.07–1.51; P = 0.031). Factors associated with CAC progression were the baseline CAC score, HbA1c (β-coefficient per 1%, 2.44; 95% CI, 0.03–4.84, P = 0.039), LDL-cholesterol (β-coefficient per 10 mg/dL, 0.58; 95% CI, 0.16–0.98; P = 0.008), and phosphate (β-coefficient per 1 mg/dL, 6.17; 95% CI, 3.67–8.90; P < 0.001).

Table 3. Multivariable linear mixed-effects models for identification of factors influencing the progression of coronary artery, aortic valve, and mitral annular calcifications.

Variables	Δ AVC scorea		Δ MAC scoreb		Δ CAC scorec
	β-coefficient (95% CI)	P value	β-coefficient (95% CI)	P value	β-coefficient (95% CI)	P value
Baseline AVC score (per 1 AU)	0.28 (0.27–0.29)	< 0.001	-		-
Baseline MAC score (per 1 AU)	-		0.24 (0.23–0.25)	< 0.001	-
Baseline CAC score (per 1 AU)	-		-		0.19 (0.17–0.20)	< 0.001
Age (per 10 yr)	0.01 (−0.68–0.66)	0.974	0.38 (−0.14–0.93)	0.154	1.24 (−0.60–2.99)	0.187
Female sex	−0.36 (−1.62–1.11)	0.566	−0.62 (−1.64–0.39)	0.226	−0.77 (−4.37–2.84)	0.672
BMI (per 5 kg/m2)	−0.06 (−0.87–0.71)	0.881	0.38 (−0.26–1.07)	0.238	−0.85 (−3.15–1.30)	0.451
Smoking	0.43 (−0.19–1.03)	0.159	0.99 (0.18–1.80)	0.015	1.11 (−3.15–1.30)	0.181
Antihypertensive medications	0.93 (−0.82–1.31)	0.468	0.52 (−0.40–1.42)	0.204	0.38 (−2.48–3.37)	0.791
Glucose-lowering medications	0.33 (−0.67–1.35)	0.538	0.07 (−1.44–1.50)	0.927	10.00 (4.70–14.89)	< 0.001
Lipid-lowering medications	−1.11 (−2.44–0.46)	0.107	0.07 (−0.80–0.88)	0.889	−1.44 (−5.19–2.24)	0.447
SBP (per 10 mmHg)	0.04 (−0.31–0.43)	0.836	0.06 (−0.22–0.33)	0.691	0.85 (−0.17–1.85)	0.104
WBC count (per 109/L)	0.12 (−0.19–0.41)	0.451	−0.15 (−0.39–0.09)	0.219	−0.53 (−1.33–0.22)	0.215
hs-CRP (per 1 mg/dL)	−0.12 (−1.35–1.07)	0.835	−0.02 (−0.92–0.90)	0.933	−0.11 (−3.58–3.19)	0.945
HbA1c (per 1%)	0.23 (−0.71–1.08)	0.594	−0.32 (−0.93–0.31)	0.350	2.44 (0.03–4.84)	0.039
HDL-cholesterol (per 10 mg/dL)	0.06 (−0.34–0.46)	0.900	−0.08 (−0.41–0.25)	0.621	−0.64 (−1.90–0.63)	0.274
LDL-cholesterol (per 10 mg/dL)	−0.13 (−0.27–0.02)	0.097	0.07 (−0.04–0.17)	0.264	0.58 (0.16–0.98)	0.008
Triglycerides (per 10 mg/dL)	−0.06 (−0.13–0.01)	0.087	−0.03 (−0.09–0.02)	0.212	−0.01 (−0.20–0.18)	0.912
eGFR (per 10 mL/min/1.73 m2)	−0.13 (−0.63–0.35)	0.627	0.11 (−0.29–0.57)	0.597	0.73 (−0.77–2.28)	0.312
Calcium (per 1 mg/dL)	1.33 (−0.08–2.82)	0.071	0.45 (−0.80–1.65)	0.433	−1.42 (−5.38–3.12)	0.485
Phosphate (per 1 mg/dL)	0.04 (−0.58–1.27)	0.415	0.82 (0.07–1.51)	0.031	6.17 (3.67–8.90)	< 0.001

The R² for the regression models are 0.83, 0.82, and 0.76, respectively.

AVC = aortic valve calcification, MAC = mitral annular calcification, CAC = coronary artery calcification, CI = confidence interval, BMI = body mass index, SBP = systolic blood pressure, WBC = white blood cell, hs-CRP = high-sensitivity C-reactive protein, HbA1c = glycated hemoglobin, HDL = high-density lipoprotein, LDL = low-density lipoprotein, eGFR = estimated glomerular filtration rate.

^aAdjusted for common risk factors + baseline AVC score, ^bAdjusted for common risk factors + baseline MAC score, ^cAdjusted for common risk factors + baseline CAC score.

In an exploratory analysis to determine the association between changes in coronary artery, aortic valve, and MACs using multivariable linear mixed effects models to correct for common risk factors, the change in AVC scores were not correlated with changes in CAC scores, but the interval change in MAC scores was significantly correlated with the change in CAC scores (β-coefficient, 0.04; 95% CI, 0.03–0.05; P < 0.001) (Supplementary Table 7). Conversely, the change in CAC scores was significantly correlated with the change in MAC scores (β-coefficient, 0.29; 95% CI, 0.19–0.39; P < 0.001), but not with the change in AVC scores. Meanwhile, changes in AVC and MAC scores only showed a borderline association.

DISCUSSION

In this study, we analyzed the relationship between AVC, MAC, and CAC in a retrospective cohort of asymptomatic individuals undergoing serial cardiac CT scans as part of a self-referred health examination. Increasing CAC severity was associated with a higher probability of prevalent AVC and MAC, and this trend was evident across the entire spectrum of CAC scores. Longitudinal changes in MAC and CAC scores were significantly correlated, but changes in AVC and MAC scores or AVC and CAC scores were not. Atherosclerotic risk factors were associated with CAC progression, but not for AVC or MAC progression.

Previous large cohort studies have shown that factors traditionally associated with atherosclerosis, such as old age, hypertension, diabetes, obesity, and smoking, are also responsible for the development of both AVC and MAC.2,3,4,5,6 Our findings are consistent with prior studies demonstrating that CAC is closely correlated with AVC7,9,27 and MAC,8,28,29,30 which support the concept that AVC and MAC are indicators of systemic atherosclerotic burden. The prevalence and severity of cardiac calcifications were lower compared to the Multi-Ethnic Study of Atherosclerosis, which reported prevalences of 13% and 9% and median scores of 56 AU and 98 AU, for AVC and MAC, respectively.7,8 The East Asian population is known to have lower cardiovascular risk and atherosclerotic burden.5,6,18 Despite potential ethnic disparities, our results confirm the relationship of AVC and MAC with CAC and extend the insights of previous studies to the Korean population.

However, despite the close correlation with CAC, both AVC and MAC had associations with age and BMI which remained after adjustment for CAC severity, while MAC was also associated with female sex and diabetes. These findings are in accord with previous reports and underline the importance of site-specific mechanisms for calcification development.2,3

In contrast to the established correlation between AVC and MAC with CAC in cross-sectional studies, there is limited data regarding the relationship between longitudinal changes in the severity of cardiac calcifications at each location. After initial calcium deposition following lipid infiltration and inflammation, the propagation of cardiac calcification is known to be largely dependent on local signaling factors in a self-perpetuating cycle where calcium begets calcium.2,14,17,31 This is especially true for AVC and MAC, the progression of which has been found to be associated with no traditional cardiovascular risk factor except the baseline calcification severity itself.5,13,32 Our study also found that CAC progression had a close association with HbA1c and LDL-cholesterol, but AVC and MAC progression did not. Thus, in the absence of shared cardiovascular risk factors implying a common mechanism, it may be difficult to conceive a relationship between the progression of cardiac calcifications at each site.

To the best of our knowledge, this is the first study to demonstrate a correlation between longitudinal changes in MAC and CAC progression. One explanation for this unexpected finding may be found in factors related to calcium-phosphate metabolism, as higher phosphate levels were associated with a higher progression rate of both MAC and CAC. A number of studies including ours has noted that higher serum phosphate has a role in the development and progression of CAC.33,34,35,36 Elevated serum phosphate upregulates fibroblast growth factor 23, which is known to promote the development of CAC in patients with CKD,37,38 and also MAC progression in the MESA cohort.26 Although we did not measure markers of mineral metabolism other than calcium and phosphate, it seems plausible that factors involving calcium-phosphate metabolism may be responsible for the correlation between MAC and CAC progression observed in our study. In contrast, mechanical stress is known to be the main driving factor in the later stages of AVC development,2,39,40,41,42 which is supported by the correlation between hemodynamic severity and disease progression in aortic stenosis.43,44 The turbulent, high velocity flow across the calcified aortic valve may be one reason for the differences observed between the progression of AVC compared to those of MAC and CAC.42 Further studies will be needed to confirm these results, and possibly to identify common therapeutic targets to slow MAC and CAC progression.

We acknowledge a number of limitations in our study. First, the proportion of individuals with AVC and MAC as well as the severity of the calcifications in our study population was generally low, meaning that care should be taken in interpreting our results. Also, it remains uncertain whether our results can be applied to patients with more severe disease. Second, AVC and MAC were assessed on separate occasions from CAC assessments; however, as the investigators were blinded to the CAC scores during measurements of AVC and MAC scores, this should strengthen, rather than weaken, the association between AVC, MAC, and CAC found in our analysis. Third, we only adjusted for clinical and laboratory data obtained at the baseline examination, which may have been subject to change during the follow-up period. However, these changes are unlikely to have affected our main finding demonstrating an association between MAC and CAC progression rates, as no factor other than the initial severity of calcification has been found to affect the progression rate of AVC and MAC.5,13 Fourth, although the follow-up duration of 3.2 years is similar to previous studies including the Multi-Ethnic Study of Atherosclerosis,5,6 our study may have lacked power to detect changes which occur over a longer time period. Fifth, information regarding medical history or current medications were collected using a self-reported questionnaire, and there is a possibility that the responses may not have been completely accurate. We also did not collect detailed information on medications or their doses. Lastly, there was significant heterogeneity in follow-up patterns, and those who underwent PCI were excluded, which may have resulted in selection bias. As a retrospective study we cannot rule out the possibility of residual confounders, although we used multivariable regression methods to correct for risk factors.

In a retrospective cohort of asymptomatic, ethnic Koreans undergoing serial cardiac CT scans as part of a self-referred health examination, increasing CAC severity was associated with a higher probability of prevalent AVC and MAC across the entire spectrum of CAC scores. In longitudinal analysis, atherosclerotic risk factors were associated with CAC progression, but not for AVC or MAC progression.

Progression in MAC and CAC scores were highly correlated, but changes in AVC and MAC scores had only a borderline association, and changes in AVC scores were not correlated with changes in CAC scores. Further studies will be needed to confirm these results, and possibly to identify common therapeutic targets to slow MAC and CAC progression.

SUPPLEMENTARY MATERIALS

Supplementary Data 1

Methods: statistical analysis

Supplementary Table 1

Baseline echocardiographic characteristics of the study population according to the presence of aortic valve and mitral annular calcification

Supplementary Table 2

Prevalence of aortic valve and mitral annular calcifications according to coronary artery calcification severity

Supplementary Table 3

Relationship between coronary artery calcification severity and prevalent aortic valve calcification

Supplementary Table 4

Relationship between coronary artery calcification severity and prevalent mitral annular calcification

Supplementary Table 5

Correlation coefficients between changes in coronary artery, aortic valve, and mitral annular calcifications

Supplementary Table 6

Correlation coefficients between changes in coronary artery, aortic valve, and mitral annular calcifications (square-root transformed)

Supplementary Table 7

Multivariable linear mixed-effects models for association between changes in coronary artery, aortic valve, and mitral annular calcifications

Supplementary Fig. 1

Distribution of follow-up durations between the first and last examinations.

Supplementary Fig. 2

Distribution of baseline (A) aortic valve calcification (B) mitral annular calcification (C) coronary artery calcification scores.

Supplementary Fig. 3

Distribution of changes in coronary artery calcification scores plotted with changes in (A) aortic valve calcification scores (B) mitral annular calcification scores.