Optimal lifestyle behaviors and 10‐year progression of arterial stiffness: The Multi‐Ethnic Study of Atherosclerosis

Abstract Majority of previous studies showed no association between a single health behavior and arterial stiffness, but the benefit of simultaneously having multiple healthy behaviors (optimal lifestyle) on the progression of arterial stiffness is unknown. Among 2810 individuals (age 60.0 ± 9.4, 46.5% male), optimal lifestyle marker (yes/no) on four health behaviors (ie, BMI < 25 kg/m2, never or former smoker, never or moderate drinker, exercised > 500 METS min/week) across four visits (≈ 5 years) were summed to create an optimal lifestyle score. Carotid arterial stiffness was measured using distensibility coefficient (DC) and Young's elastic modulus (YEM) at visit 1 and after a mean of 9.5 years (visit 5). The association of optimal lifestyle with 10‐year percent change in DC and YEM was assessed using multiple linear regression. DC decreased by 5.3% and YEM increased by 24.4% over 10 years. Mean optimal lifestyle score was 9.4 ± 3.1 (range: 0–16). Individuals in quintiles 2–5 of optimal lifestyle score compared to quintile 1 (with the least optimal lifestyle score) did not show slower deceleration of DC [Q2, −0.3% (95% CI: −6.0, 5.4); Q3, −0.01% (−4.5, 4.5); Q4, −0.6% (−5.2, 3.9); Q5, −0.4% (−5.3, 4.4)], trend p‐value = .82] or slower progression of YEM [Q2, 0.1% (−7.1, 7.3); Q3, −0.8% (−8.0, 6.5); Q4, 4.5% (−2.3, 11.3); Q5, −0.2% (−8.3, 7.9)], trend p‐value = .49] after adjusting for risk factors. The association remained non‐significant when stratified by categories of age, sex, race, BP control, and diabetes. Our findings indicate that optimal score on multiple health behaviors may not independently slow arterial stiffness progression.

Healthy lifestyle habits such as regularly exercising, avoiding tobacco smoking and alcoholic beverages, and maintaining optimal body weight were associated with anti-oxidative and antiinflammatory effects, reduced pro-inflammatory cytokines, and production of nitric oxide within vascular endothelium. [9][10][11][12][13][14][15][16][17][18][19] These physiological signals may reduce structural changes within the arterial wall such as fragmentation of elastin, deposition of collagen, and smooth muscle proliferation which are precursors of arterial stiffness. 9 Despite these favorable effects on the artery, individual health behavior have not been consistently associated with arterial stiffness. [20][21][22][23] In a systematic review of 77 studies 20 and meta-analysis of 41 randomized controlled trials, 21 the majority of studies (≥ 80%) found no association of arterial stiffness with smoking, body mass index (BMI) and aerobic exercise. Whether simultaneously maintaining favorable scores on multiple health behaviors (optimal lifestyle) slow the progression of arterial stiffness is unknown.
The objective of this study was to investigate the association between long-term (≈ 5 years) optimal lifestyle (ie, BMI < 25 kg/m 2 , never smoker or quit smoking ≥12 months ago, never or moderate drinker, and exercised > 500 METS -/week) and 10-year progression of arterial stiffness.

Study design and participants
The Multi-Ethnic Study of Atherosclerosis (MESA) is a population based longitudinal study of risk factors for subclinical and clinical cardiovascular diseases (CVD). 24

Carotid artery stiffness
Ultrasound images of the right common carotid artery, approximately 1 cm below the carotid bulb, were obtained using a Logiq 700 ultra- DC and YEM are inversely related, and arterial stiffness corresponds to a higher score on YEM and a lower score on DC.

2.3
Optimal lifestyle BMI was calculated as weight (measured by a calibrated scale to the nearest 0.5 kg) in kilograms divided by height (measured by a stadiometer to the nearest 0.1 cm) in meters squared and optimal BMI was defined as a score of < 25 kg/m 2 . 27 Smoking status was classified as never (smoked < 100 cigarettes per lifetime), former (smoked ≥ 100 cigarettes, but stopped for ≥1 year prior to examination visit), or current smoker (smoked ≥ 100 cigarettes per lifetime and still smokes or quit < 1 years prior to examination visit). Participants who never smoked or quit smoking ≥12 months ago were considered as having optimal smoking status. 27 Self-reported frequency and duration of participation in nine different activities during a typical week in the past month were used to calculate total Metabolic Equivalent of Task (MET) minutes per week for moderate-vigorous activities and a score of ≥500 MET min/week was considered optimal physical activity. 28 Participants were asked, "Have you ever consumed alcoholic beverages?" "Do you presently drink alcoholic beverages?" and "number of drinks of drinks per week". Optimal alcohol consumption was defined as never or moderate alcohol consumption (ie, ≤14 drinks/week for men and ≤7 drinks/week for women). Number of optimal lifestyle indicator (yes/no) on four health behaviors (ie, BMI < 25 kg/m 2 , never or former smoker, never or moderate drinker, exercised > 500 METS min/week) across visits 1-4 were summed to create an overall optimal lifestyle score (ranged 0-16).

Statistical analysis
Multiple imputation using chained equations with 50 repetitions was used to impute missing values on adjusted covariates. In a multiple linear regression, we evaluated the association between quintiles of overall optimal lifestyle score and 10-year change in carotid artery DC and YEM using quintile 1 (least healthy) as a reference group. Trend tests were performed by including quintile of optimal lifestyle as a continuous ordinal variable. We also examined the association between optimal score of each individual health behavior across visits 1-4 and 10- year change in DC and YEM using participants with optimal score of zero (ie, no optimal score for individual health behavior across all visits 1-4) as a reference group. Our analysis was weighted by inverse probability of being a participant in this study because individuals included in our analysis were younger and less likely to be hypertensive and diabetic compared to all eligible participants. Age, hypertension and diabetes are known to be strong predictors of arterial stiffness. [6][7][8] In all the models, plots of the residuals against the fitted was checked to assess assumptions of linearity and homoscedasticity and to check outlier observations. 29 All analyses were performed using Stata 16.1 (Stat-aCorp. 2019). 30

Participant characteristics
In the MESA, there were 6814 participants at baseline. Only a sub-   (Table S2).

Optimal lifestyle and progression in arterial stiffness
In models adjusted for age, sex, race, study site, baseline distensibility coefficient or Young's elastic modulus, there was no significant association between optimal lifestyle score and 10-year arterial stiffness progression (Table S3: (Figure 1). The association of optimal lifestyle score with percent change in DC and YEM was also not significant when stratified by age, sex, race, blood pressure control, and diabetes status (trend p-value > = .05 for all) (Tables S4 and S5).
There was also no significant difference in the decline of DC or progression of YEM over 10 years by the individual health behaviors in the less adjusted model (Table S3:

DISCUSSION
In this study, using two measures of local arterial stiffness, we assessed whether long-term optimal lifestyle on multiple health behaviors (BMI < 25 kg/m 2 , never or former smoker, never or moderate drinker, and exercise > 500 METS min/week) was associated with a slower pro-  Mean difference in percent change of arterial stiffness F I G U R E 1 Linear regression association between quintiles of optimal lifestyle score (from four healthy behaviours -BMI <25 kg/m 2 , never or former smoker, never or moderate drinker, exercised >500 METS min/wk) across four visits (≈5 years) and ten-year percent change in distensibility coefficient (DC) and Young's elastic modules (YEM). Adjusted for baseline age, sex, race, study site, systolic and diastolic BP, diabetes mellitus, total cholesterol, HDL cholesterol, eGFR, anti-hypertensive and lipid lowering medications, baseline DC or YEM, and change in systolic and diastolic BP between visits 1 and 4. Circular dots denote estimates and horizontal lines indicate the corresponding 95% confidence intervals investigated predictors of arterial stiffness from 77 studies that performed a multiple regression analysis to identify independent associations and concluded that the contribution of risk factors other than age and BP to arterial stiffness is small or insignificant. In Cecelja's study, smoking and BMI were not associated with arterial stiffness in ≥ 86% of the reviewed studies while age and BP were consistently associated in 91% and 90% of the studies, respectively. 20 Likewise, a systematic review and meta-analysis of 41 randomized controlled trials by Ashor et al. 21 concluded no effect of a combination of aerobic and resistance exercise lasting ≥4 weeks on pulse wave velocity. Inconsistent association between alcohol consumption and arterial stiffness was also reported. 22,23,33 While cross-sectional studies reported J-shaped association between alcohol intake and arterial stiffness, 23,33 prospective cohort study reported no association. 22 Studies have shown that regularly exercising, not smoking and drinking, and maintaining optimal body weight have anti-oxidative effects, increase anti-inflammatory cytokines while reducing proinflammatory cytokines, and enhances production of nitric oxide. [9][10][11][12][13][14][15][16][17][18][19] These functional changes within vascular endothelium are believed to reduce the fragmentation of elastin, deposition of collagen, and smooth muscle proliferation which may result in slower progression for arterial stiffness. 9 However, the findings of our study and other studies [21][22][23][31][32][33] indicate that the physiologic change in the vascular wall due to healthy lifestyle may not have significant impact on slowing arterial wall stiffening.

Participant characteristics Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) p-value
Unhealthy lifestyle is known to be a risk factor for elevated BP-a major modifiable risk factor for arterial stiffness. However, even in a model unadjusted for baseline BP and other risk factors (diabetes mellitus, total cholesterol, HDL cholesterol, glomerular filtration rate, use of antihypertensive and lipid medications), we did not find an association between healthy lifestyle and lesser progression in arterial stiffness (Table S3). Aging and uncontrolled BP are major risk factors for arterial stiffness, 6-8 thus, we investigated if maintaining optimal lifestyle is associated with slower progression of arterial stiffness among younger or those with controlled BP. However, the association between optimal lifestyle and percent change in arterial stiffness remained nonsignificant when stratified by different categories of age and blood pressure control (Tables S4 and S5).
Our study is the first, to our knowledge, to show maintaining healthy lifestyle simultaneously on multiple health behaviors is not independently associated with a slower progression in arterial stiffness. Some Mean difference in percent change of Mean difference in percent change of F I G U R E 2 Linear regression association between number of visits with optimal health behaviours (BMI <25 kg/m 2 , never or former smoker, never or moderate drinker, exercised >500 METS min/wk) across four visits (≈ 5 years) and ten-year percent change in distensibility coefficient and Young's elastic modules. Adjusted for baseline age, sex, race, study site, systolic and diastolic BP, diabetes mellitus, total cholesterol, HDL cholesterol, eGFR, anti-hypertensive and lipid lowering medications, baseline DC or YEM, and change in systolic and diastolic BP between visits 1 and 4. Circular dots denote estimates and horizontal lines deonte 95% confidence intervals. Reference group are individuals with no optimal health behaviours at all visits any beneficial effect of improving healthy lifestyles on arterial stiffness, the effect might be more noticeable in individuals with unhealthier lifestyles at baseline. However, our analysis was weighted by inverse probability of being a participant in this study to improve generalizability our findings. Underestimating smoking and drinking habits while overestimating exercise habit are also common [35][36][37] and this nondifferential misclassification may have played a role in underestimating the true association between optimal lifestyle and arterial stiffness.
In conclusions, Prolonged (≈ 5 years) optimal lifestyle on multiple health behaviors (BMI < 25 kg/m 2 , never smoker or quit smoking ≥12 months ago, never drinker or current non-drinker, and exercised > 500 METS min/week), although known to help at reducing risk of cardiac events, 38,39 was not associated with slower progression in arterial stiffness over 10-years independent of age, systolic and diastolic BP, diabetes mellitus and other risk factors.

ACKNOWLEDGMENTS
The authors thank the other investigators, the staff, and the partici-

CONFLICT OF INTEREST
None to disclose.

AUTHOR CONTRIBUTIONS
Yacob G. Tedla conceptualized the study, performed data analysis, and drafted the manuscript. Adam Gepner, James H. Stein, and Philip Greenland were involved in the conceptualization of the study, data interpretation, and reviewed and edited the manuscript. Joseph A.
Delaney and Chia-Ying Liu reviewed and edited the manuscript and