Atherosclerosis begins in a subclinical phase and slowly progresses over the years before the development of symptoms or clinical cardiovascular events. ¹

The ability to identify among asymptomatic individuals the subgroup that has a greater risk of developing cardiovascular events is a fundamental strategy in the prevention of cardiovascular events. ²

Clinical risk scores, such as the Framingham risk score (FRS), are widely used for stratifying individuals with greater cardiovascular risk. ³ However, it has been suggested that the risk in many individuals may be underestimated, considering that atherosclerosis is a heterogeneous disease that develops unevenly among individuals. In addition, a large percentage is classified as intermediate risk, where therapeutic approaches are controversial. ^{4 - 7}

Computed tomography makes it possible to measure the coronary calcium score (CCS), and, based on this data, it is possible to estimate arterial age, that is, based on the damage already present in the analyzed vessels, it is possible to estimate their age. Using arterial age in the Framingham risk score has shown to be more predictive of future events than chronological age. ^{8 - 11}

The Multi -Ethnic Study of Atherosclerosis (MESA) evaluated the impact of determining CCS on the prediction of coronary events in 6,722 men and women of diverse ethnicities. In comparison with individuals without coronary calcification, the risk of death or acute myocardial infarction increased by 7.7 times for individuals with CCS between 101 and 300 and 9.7 times for CCS > 300. ¹²

Using arterial age rather than chronological age reclassified 28% of MESA participants into risk categories other than the original. During outcome analysis, the score based on arterial age performed significantly better for risk determination (area under the receiver operating characteristic [ROC] curve 0.75 for FRS based on observed age versus 0.79 using arterial age). ¹¹

The objective of this study was to reclassify cardiovascular risk based on arterial age and to evaluate CCS progression during follow-up.

The sample consisted of 150 men with a mean age of 58.2 years ( Table 1 ).

Table 1

Variables	First stage (n = 150)	Second stage (n = 139)
Age (years)	58.2 ± 5.3	65.7 ± 5.2
Years of schooling – n (%)
≤ 8	86 (57.3)	79 (56.83)
9 to 12	39 (26.0)	37 (26.6)
> 12	25 (16.7)	23 (16.5)
Salary – n (%)
< 5 times minimum wage	103 (68.7)	93 (66.9)
5 to 10 times minimum wage	37 (24.7)	40 (28.8)
> 10 times minimum wage	10 (6.7)	6 (4.3)
BMI (kg/m2)	26.8 ± 2.5	27.3 ± 3.4
Waist circumference (cm)	96.3 ± 7.9	97.7 ± 9.0
SBP (mmHg)	139.8 ± 12.2	137.36 ± 17.7
DBP (mmHg)	83.6 ± 7.8	78.1 ± 9.7
HR (bpm)	65.5 ± 10.7	66.4 ± 9.6
Tobacco use – n (%)	18 (12.0)	5 (3.6)
Family history of CAD – n (%)	23 (15.3)	40 (28.8)
Weekly physical activity – n (%)
< 150 minutes	25 (16.7)	38 (27.3)
> 150 minutes	125 (83.3)	101 (72.7)
Altered physical examination – n (%)	14 (9.3)	17 (12.3)
Blood glucose (mg/dL)	106.0 ± 17.4	100.3 ± 19.5
Total cholesterol (mg/dL)	226.6 ± 38.0	195.7 ± 40.8
LDL cholesterol (mg/dL)	145.2 ± 32.8	120.6 ± 36.9
HDL cholesterol (mg/dL)	49.2 ± 13.2	50.2 ± 11.7
Triglycerides (mg/dL)	163.9 ± 141.2	124 ± 72.4
Coronary calcium score	94.4 ± 218.5 (0-83)	228.6 ± 468.5 (0-213)
No evidence	61 (40.7)	40 (28.7)
Mild (1 to 100)	57 (38.0)	52 (37.4)
Moderate (101 to 400)	21 (14.0)	25 (18.0)
High (401 to 1000)	10 (6.6)	14 (10.1)
Very high (> 1000)	1 (0.7)	8 (5.7)

* Quantitative data are described as mean ± standard deviation or median (25th to 75th percentiles) and categorical data are described as n (%). BMI: body mass index; CAD: coronary artery disease; DBP: diastolic blood pressure; HR: heart rate; SBP: systolic blood pressure.

During the study period, 11 follow-ups were lost: 6 of them died (no case of death from cardiovascular causes), and 5 declined to participate, leaving 139 individuals for the second stage.

Calculation of arterial age showed a difference of 1.7 years less in relation to chronological age, with an estimated mean arterial age of 56.5 years versus 58.2 years of chronological age. In the second stage, this difference increased to 3 years.

The restratification of cardiovascular risk using arterial age showed a statistically significant difference in relation to the stratification of cardiovascular risk by the FRS and ASCVD. We observed many individuals who were reclassified, as both higher and lower risk.

In relation to the FRS, 29% of individuals were reclassified to a higher risk category when using arterial age, while 34% remained in the same category, and 37% moved to a lower risk category. For the ASCVD, the findings were similar, with 33% remaining in the same category, while 31% moved to higher risk, and 36% moved to lower risk.

The greatest difference was observed in individuals with intermediate risk according to the FRS. Of the 107 individuals initially grouped in this category, only 27 (25.2%) remained in an intermediate risk category, while 52 (48.5%) were reclassified to low risk and 28 (26.2%) to high risk ( Figure 1 ).

Figure 1

When the risk categories were evaluated individually by the ASCVD classification, during the initial phase, there were 39 individuals classified as low risk, of whom 22 (56%) moved to a higher category. Of the 17 initially classified as high risk according to the ASCVD, only 5 remained (29%) as high risk after inclusion of the CCS.

During the second stage of the study, with an average time of 7.6 years after the first evaluation, a similar phenomenon occurred. In individuals classified as intermediate risk according to the FRS, the percentage who remained in the same classification was 30.5%, while 39.8% were reclassified to low risk and 29.6% to high risk ( Figure 2 ).

Figure 2

As an example, we cite the case of a 52-year-old individual with total cholesterol of 270 mg/dL, HDL-cholesterol of 43 mg/dL, systolic blood pressure of 129 mmHg, and diastolic blood pressure of 87 mmHg, with no history of smoking and without diabetes. When applying the FRS, the estimated risk was 8% (low risk), and ASCVD was 7.3 (borderline). However, the CCS was 195, which increased his risk according to arterial age to 30; therefore, he began to be considered as high risk for cardiovascular disease. In fact, during follow-up, his CCS increased to 631, a 323% increase, and only then was he considered high risk according to clinical criteria.

The agreement between the 3 scores (FRS, ASCVD, and arterial age) was also evaluated to classify risk, and it was observed that the kappa coefficient between arterial age and the FRS was 0.044 (p = 0.330); the kappa between arterial age and ASCVD was 0.073 (p = 0.080), and, between the FRS and ASCVD, it was 0.107 (p < 0.001), demonstrating significant agreement only between the FRS and ASCVD risk classifications, but with low, although significant agreement. In other words, the FRS and ASCVD classified risk groups in a similar manner, but both differed from the classification by arterial age.

There was a significant increase in the CCS in the period between the 2 assessments (first assessment: median =12.5; 25 to 75 percentiles: 0 to 81; second assessment: median = 33; 25 to 75 percentiles: 0 to 213; p < 0.001). The median variation in the CCS between the 2 assessments was 14 points (25 to 75 percentiles: 0 to 116).

When conducting progressive analysis of the change (delta) in the CCS, that is, when evaluating how much it varied over the follow-up period, we observed that individuals initially classified as low risk by arterial age had a very mild progression in the CCS, and this progression increased in proportion to higher risk estimated by the initial classification (p < 0.001). This relationship was not significant for the FRS (p = 0.862) or the ASCVD (p = 0.153). Differences between risk classifications with increasing CCS are displayed in Table 2 .

Table 2

Variables	Total sample n (%)	∆ calcium score Median (P25-P75)	p
Estimated risk classification by arterial age			< 0.001
< 10 (low risk)	72 (48.0)	0 (0-11.8) a
10 to 20 (moderate risk)	39 (26.0)	38.5 (6-246) b
> 20 (high risk)	39 (26.0)	148 (58.5-466) c
Framingham classification			0.862
< 10% (low risk)	34 (22.7)	12 (0-76.5)
10% to 20% (intermediate risk)	107 (71.3)	18.5 (0-126)
> 20% (high risk)	9 (6.0)	47 (1-542)
ASCVD			0.153
< 5% (low risk)	39 (26.0)	8.5 (0-35.0)
5% to 19.9% (intermediate risk)	94 (62.7)	25 (0-173)
≥ 20% (high risk)	17 (11.3)	30.6 (0.5-60.5)

^a,b,c Equal letters do not differ by Dunn’s test at 5% significance. The numbers presented are relative to the score according to the Agatston method.

Figure 3 illustrates the differences found regarding the change in the calcium score according to the increase in cardiovascular risk by arterial age.

Figure 3

The numerical classification (% cardiovascular risk) estimated in the first evaluation by different methods (risk classification by arterial age, FRS, and ASCVD) was compared using the ROC curve for the outcome CCS > 400 in the second evaluation, which is considered a high degree of coronary calcification. That is, we aimed to evaluate the sensitivity and specificity of the methods as a way of detecting individuals who would develop a greater degree of coronary calcification during the follow-up period (7.6 years).

We observed that the classification by arterial age was significantly higher than the other methods, with an area under the curve of 0.870, while the ASCVD presented an area under the curve of 0.629, and the FRS was practically at the baseline, with 0.544 (Central Figure).

Individualized evaluation of the variables analyzed in the study showed that the vast majority had no statistically significant relationship with the change in the CCS. This relationship was not observed for age, presence of systemic arterial hypertension, statin use, diabetes mellitus, high LDL, hypertriglyceridemia, smoking or ex-smoking, physical activity, alcohol consumption, level of schooling, marital status, income, BMI, and waist circumference. Accordingly, we reduced the possibility of biases due to confounding factors.

There was a statistically significant positive relationship between the systolic blood pressure levels and the change in the CCS (p = 0.041); the higher the systolic blood pressure levels at baseline, the greater the increase in the CCS ( Table 3 ).

Table 3

Quantitative variable	∆ calcium score Spearman’s correlation coefficient	p
Systolic blood pressure	0.174	0.041
Qualitative variable	Median (P25-P75)	p
Low HDL		0.041
No	13 (0-114)
Yes	457.5 (368-547)

SBP: systolic blood pressure.

Individuals with low HDL showed a significantly greater increase in the calcium score, as shown in Table 3 . Their average increase in the CCS during the study was 457, whereas, in the remaining individuals with normal HDL, the average increase in the calcium score was 13.

The CCS is an indicator of subclinical atherosclerosis, and elevated values indicate greater degree of atherosclerosis. ^{9 , 11 , 12}

The impact of traditional risk factors is not uniform among individuals, which is why traditional scores do not always have sufficient sensitivity and specificity for correct risk stratification. ^{4 - 6}

Studies have shown that incorporating arterial age into traditional risk scores increases the area under the curve, which improves the sensitivity and specificity of this score in determining cardiovascular risk. ^{11 , 25}

Our study has shown that a significant portion of individuals are restratified into different risk categories when evaluated using arterial age. The percentage whose risk category changed was greater than that observed in the PAAC study ²⁶ and the Heinz Nixdorf Recall Study, ²⁷ an assessment derived from the MESA study. The category that underwent the most changes is the one classified as intermediate by traditional risk scores, which is similar to the studies mentioned above, where between 1 in 4 and 1 in 6 individuals moved to a higher risk category. In our study, we observed a greater number of individuals who moved to a lower risk category in comparison to the cited studies.

These findings reinforce the indication that already exists in the guidelines for CCS to be determined, especially in individuals with intermediate cardiovascular risk according to the traditional risk scores. ^{22 , 28 , 29}

Furthermore, the most objective piece of data was that the classification of cardiovascular risk by arterial age showed a direct relationship with increased coronary calcification determined using CCS during follow-up. Individuals initially classified as low risk by arterial age had a very mild CCS progression, and this progression increased in proportion to higher risk estimated by the initial classification. This relationship had already been described in the literature for major cardiovascular outcomes; nonetheless, there is no description of the proportional increase in the CCS based on the initial classification by arterial age. These are the most relevant data from our study, and they may indicate performance of CCS, even in individuals with low risk according to traditional scores, keeping in mind that, even among these individuals, some had a higher arterial age and more significant CCS progression throughout the follow-up.

Among all the risk factors studied, high systolic blood pressure and low HDL were the ones that were significantly associated with the greatest CCS progression throughout the study.

HDL cholesterol has already shown to be an important cardiovascular risk factor when at low levels, regardless of other cholesterol levels. ^{30 , 31} In our study, it was the isolated factor with the greatest impact on the progression of coronary calcification.

Although a significant reclassification of cardiovascular risk occurred, it was not possible to evaluate the potential reduction in cardiovascular outcomes in this study population. This fact makes us question the applicability of the method to the general population, bearing in mind that the costs and potential risks associated with the test cannot be ignored.

The main study limitations are the relatively low number of participants, the absence of women in the sample, and the inability to show major cardiovascular outcomes (acute myocardial infarction, death of cardiovascular origin, or stroke). Furthermore, participants were volunteers and were selected by convenience.

One of the potential biases to be considered is the use of statins and their possible effect on coronary calcium. It has been established that the use of statins can increase the CCS; therefore, the progression of the CCS is not always something undesirable, given that it can only translate “stabilization” of the existing coronary plaques. We had information on the patients who used statins in both stages of the study, although we did not obtain information about the time and continuity of use.

Furthermore, the equipment used between the 2 study stages was different, which could cause a potential measurement bias.

An important limitation to be considered is the absence of women in the study. We are aware of the importance of including both sexes for more reliable analysis and so that the data can be generalized. Future research should be conducted including women to provide a more comprehensive understanding of the factors that influence vascular age and cardiovascular risk classifications.

Notwithstanding the potential biases, we consider that the factor of CCS progression was quite significant, and it had a direct relationship with the initial classification based on arterial age. Furthermore, elevated systolic blood pressure and low levels of HDL-cholesterol were associated with greater CCS progression.

Cardiovascular risk stratification using arterial age was more capable of identifying individuals with elevated risk of atherosclerosis progression measured by the CCS than traditional risk scores (FRS and ASCVD). The individual variables most associated with CCS progression were high systolic blood pressure and low HDL.