Comparison of Vascular Aging Models on the Basis of Different Vascular Indicators for Cardiovascular Events Prediction: A Chinese Community‐Based Cohort Study

Abstract

Background

Vascular aging (VA) is a promising concept for cardiovascular risk assessment, but the optimal vascular indicator for vascular age calculation remains uncertain. This study aimed to compare VA indices on the basis of different vascular indicators for predicting cardiovascular events.

Methods

This prospective study included 3561 participants from a Chinese community‐based cohort. Four distinct VA indices including brachial–ankle pulse wave velocity (baPWV), carotid intima‐media thickness, carotid plaque, or ankle–brachial index, and conventional risk factors were calculated to predict age. ΔAge was defined as the residual from a linear regression of vascular age on chronological age. Supernormal VA and early VA were defined by the 10th and 90th percentiles of ΔAge, respectively. Cox proportional hazards models were used to investigate associations with major adverse cardiovascular events (MACEs).

Results

During a mean 7.2‐year follow‐up, 294 participants experienced MACEs. After multivariable adjustment, only baPWV‐based ΔAge was associated with MACE (hazard ratio [HR] per SD, 1.14 [95% CI, 1.02–1.27]). Early VA phenotype based on baPWV was associated with a higher risk of MACEs (HR, 1.44 [95% CI, 1.05–1.97]), whereas supernormal VA phenotype was not protective. VA indices based on carotid intima‐media thickness, carotid plaque, or ankle–brachial index showed no significant associations with MACEs.

Conclusions

In this Chinese community‐based cohort, baPWV was the superior indicator for calculating VA indices, compared with carotid intima‐media thickness, carotid plaque, and ankle–brachial index. baPWV‐based ΔAge and early VA phenotype were independently associated with cardiovascular events, while other indicators were not. baPWV‐based VA demonstrates clinical utility for improving cardiovascular risk stratification.

Nonstandard Abbreviations and Acronyms

baPWV

brachial–ankle pulse wave velocity

cfPWV

carotid–femoral pulse wave velocity

cIMT

carotid intima‐media thickness

EVA

early vascular aging

MACE

major adverse cardiovascular event

NVA

normal vascular aging

SuperNoVA

supernormal vascular aging

VA

vascular aging

Clinical Perspective.

What Is New?

• This large community‐based cohort study systematically compared various vascular aging indices derived from different noninvasive vascular indicators to determine their predictive value for incident cardiovascular events in the Chinese population.

• We found that the vascular aging index based on brachial–ankle pulse wave velocity provides the highest predictive accuracy and significantly improves cardiovascular risk stratification beyond chronological age and traditional risk models.

What Are the Clinical Implications?

• Incorporating this optimal, noninvasive vascular aging index into routine screening protocols offers a practical tool for clinicians to better identify high‐risk individuals, thereby facilitating early and targeted cardiovascular prevention strategies.

Cardiovascular disease (CVD) remains the leading cause of morbidity and death worldwide, creating a significant global health challenge. ¹ For decades, preventive strategies have relied on traditional risk assessment tools, such as the Framingham Risk Score, which estimate the probability of future cardiovascular events on the basis of established risk factors. ² While these assessment tools are foundational to clinical practice, their heavy reliance on chronological age can paradoxically underestimate the long‐term risk in younger individuals and fail to capture the cumulative burden of risk factors on vascular health. ³ This limitation has spurred the development of more intuitive and integrated concepts for risk communication and stratification, among which vascular age has emerged as a particularly promising tool. ⁴ By translating a patient's multifactorial risk profile into an easily understood age of their arteries, it offers a more tangible representation of cumulative vascular damage. ⁵

The concept of vascular age is operationalized through the measurement of subclinical vascular damage using various noninvasive biomarkers. ⁶ These markers can be broadly categorized as assessing either functional or structural aspects of the vasculature. For instance, brachial–ankle pulse wave velocity (baPWV) is a robust measure of arterial stiffness, reflecting the functional health of the arterial tree. ⁷ In contrast, carotid intima‐media thickness (cIMT) and the presence of carotid plaque are direct structural indicators of the atherosclerotic disease burden. ⁸ The ankle–brachial index (ABI) serves as another functional measure, primarily identifying advanced, flow‐limiting peripheral artery disease. ⁹ Each of these indices captures a distinct dimension of the complex vascular aging (VA) process.

Building upon this framework, recent research has introduced the concept of VA phenotypes by quantifying the difference between vascular age and chronological age (termed ΔAge). ¹⁰ Individuals with a significantly higher vascular age than chronological age are classified as having early VA (EVA), representing an accelerated risk trajectory. ¹¹ Conversely, those with a substantially lower vascular age are deemed to have supernormal VA (SuperNoVA), suggesting a resistance to vascular damage. ¹² Previous studies have demonstrated that these phenotypes, particularly EVA, are associated with an increased risk of future cardiovascular events. ¹⁰ , ¹³ , ¹⁴ However, a critical limitation in the current literature is that these prognostic assessment tools have largely been developed and validated using a single type of biomarker, most notably pulse wave velocity, a functional measure of arterial stiffness. Consequently, while the concept of EVA is prognostically validated, the optimal method for its identification remains unresolved. Therefore, it is currently unknown which indicators that reflect the different dimensions of VA provide the most robust and clinically useful foundation for constructing these prognostic phenotypes.

Therefore, the primary aim of this study was to construct and directly compare the prognostic value of 4 distinct vascular age indices, based on baPWV, cIMT, carotid plaque, and ABI, for predicting major adverse cardiovascular events (MACEs) in a large, community‐based Chinese cohort. We hypothesized that the optimal vascular age index would be the one that best captures the global, functional consequences of VA, thereby providing superior risk prediction.

Methods

A technical appendix, statistical code, and data set will be made available from the corresponding author at drzhy1108@163.com (Y.Z.) or gongyanjun111@163.com (Y.G.).

Study Design and Population

The study population was a community‐based cohort that was established in the Shougang community‐based Shijingshan District in Beijing, China, between December 2011 and April 2012 and has been described previously. ¹⁵ In brief, a baseline survey was conducted on participants aged ≥40 years. A total of 3881 participants underwent unified cross‐sectional follow‐up from September to December 2014, during which standardized institutional questionnaires and physical examination data, and morning fasting blood samples were collected. Notably, this follow‐up phase incorporated the most comprehensive assessment of multiple indicators reflecting both vascular function and structure. Finally, a total of 3561 eligible study participants were included in this analysis (Figure 1). This study protocol was reviewed and approved by the Ethics Committee of Peking University First Hospital. All participants provided written informed consent.

Figure 1. Flowchart of the study design, population, and follow‐up.

ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; and cIMT, carotid intima‐media thickness.

Collection and Definition of Variables

All participants completed a standardized baseline questionnaire, which included demographic information, lifestyle‐related risk factors, comorbidities, and detailed medication history. Anthropometric measurements, including height, weight, and waist circumference, were collected by trained staff following standardized procedures. Peripheral blood pressure was measured 3 times on the right arm in a seated position after resting for 5 minutes using an HEM‐7200 electronic sphygmomanometer (Omron, Kyoto, Japan). The averages of the 3 readings for systolic blood pressure and diastolic blood pressure were recorded.

Current smoking was defined as smoking ≥1 cigarette per day for at least half of the previous year. Current drinking was defined as drinking alcohol at least once per week for at least half of the previous year. Body mass index was calculated as weight (kg) divided by the square of height (m). Hypertension was defined as any self‐reported history of hypertension, systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg, or taking antihypertensive medication. Diabetes was defined as any self‐reported history of diabetes, treatment with hypoglycemic medication, or fasting blood glucose ≥7.0 mmol/L. Dyslipidemia was defined as any self‐reported history of dyslipidemia; treatment with lipid‐lowering medication; or total cholesterol ≥5.18 mmol/L, triglycerides ≥1.7 mmol/L, low‐density lipoprotein cholesterol ≥3.37 mmol/L, or high‐density lipoprotein cholesterol <1.04 mmol/L. Obesity was defined as a body mass index ≥28 kg/m² according to the Chinese guideline. ¹⁶

baPWV and ABI were measured in all participants after resting for at least 5 minutes using the BP‐203RPE III device (Omron Colin, Tokyo, Japan) in a supine position in a quiet and temperature‐controlled examination room. Trained research staff placed cuffs around both upper arms and ankles to monitor brachial and tibial artery pulse waveforms. The device automatically calculated the value of baPWV and ABI. Following the examination, the higher baPWV and lower ABI readings from the left or right side were selected for subsequent analysis.

Certified sonographers performed standardized bilateral carotid ultrasonography on all participants using a Terason Echo ultrasound system (Terason, Burlington, MA). Following Mannheim Consensus protocols, the presence of carotid plaques (defined as focal lesions ≥0.5 mm thick, exceeding 50% of adjacent cIMT, or with absolute thickness >1.5 mm) was documented across all carotid segments. ¹⁷ Simultaneously, the cIMT value was measured on the far wall during systole within plaque‐free 10‐mm segments of the distal common carotid artery near the bulb bilaterally. Both carotid plaque status (yes/no) and maximum cIMT values were recorded for subsequent analyses.

The CVD risk score was calculated using the China‐PAR prediction tool. ¹⁸ Specifically, this sex‐specific model uses Cox proportional hazards regression and incorporates the following covariates: age, waist circumference, geographic region (north/south of Yangtze River), urbanization level (urban/rural), family history of CVD (≥1 parent/sibling with myocardial infarction [MI] or stroke), systolic blood pressure, antihypertensive treatment within the previous 2 weeks, total cholesterol, high‐density lipoprotein cholesterol, current smoking status, and diabetes history. Individual risk scores were derived by multiplying each covariate value by its sex‐specific coefficient, summing these products, and applying the resultant value to the sex‐specific risk equation.

Construction of a Vascular Age Index and Definition of Vascular Aging Phenotypes

Four distinct VA indices were developed integrating specific vascular indicators (baPWV, cIMT, carotid plaque, or ABI), conventional cardiovascular risk factors, and relevant medications. Within each multivariate regression model, VA was defined as the predicted value of chronological age. Each model used a standardized framework as follows. Variable selection was performed using a backward stepwise approach based on the Akaike information criterion, with exclusion of collinear variables (variance inflation factor >4). Continuous variables demonstrating nonlinear relationships with chronological age were modeled using smoothing splines within a generalized additive model framework. The optimal number of knots for each spline term was determined through 10‐fold cross‐validation to minimize root mean square error. The final variables included in each of the 4 models and their respective explanatory power (R ² ) are summarized in Table S1.

ΔAge was defined as the residual obtained from a linear regression of VA on chronological age, rather than the simple arithmetic difference. SuperNoVA and EVA phenotypes were classified using the 10th and 90th percentiles of ΔAge as cutoffs, respectively, while normal VA (NVA) was defined as values between these thresholds. ¹⁰

Ascertainment of End Point

For the current analysis, end point event collection was terminated on December 31, 2021. Data on deaths from any cause were collected from the Chinese Center for Disease Control and Prevention (National Mortality Surveillance System). ¹⁹ Data on in‐hospital stroke and MI events were obtained from the Beijing Municipal Health Commission (Inpatient Medical Record Home Page System). Death coding was based on the 4‐digit (sublevel) extended codes of the International Classification of Diseases, Tenth Revision (ICD‐10), and disease coding was based on the 6‐digit extended codes of the ICD‐10 (Table S2). The primary end point was MACEs, which was a composite of cardiovascular death, nonfatal stroke, and nonfatal MI. The secondary end points included cardiovascular death, stroke, and MI. For the current analysis, end point event collection was terminated on December 31, 2021.

Statistical Analysis

Continuous variables that were normally distributed are shown as the mean±SD and were compared between groups using ANOVA. Nonnormally distributed continuous variables are expressed as the median (interquartile range) and were compared using the Kruskal–Wallis test. Categorical variables are reported as the count (percentage) and were compared using the χ² test.

Restricted cubic spline analyses examined the associations between ΔAge (based on each vascular indicator) and MACEs, which were adjusted for sex, chronological age, and CVD risk score, with knot positions automatically optimized by minimizing the Akaike information criterion. Kaplan–Meier survival curves were used to depict the temporal distribution of MACEs across VA phenotype groups on the basis of each vascular indicator and were analyzed with the log‐rank test. Univariable and multivariable Cox proportional hazards regression evaluated associations of ΔAge and VA phenotypes on the basis of each vascular indicator with primary and secondary end points. Multivariable model 1 was adjusted for sex and chronological age; model 2 was adjusted for sex, chronological age, and CVD risk score.

Exploratory subgroup analyses of the multivariable Cox proportional hazards regression for the association of ΔAge and MACE were performed for key characteristics stratification including sex (male or female), chronological age (<65 or ≥65 years), obesity (yes or no), current smoking (yes or no), current drinking (yes or no), hypertension (yes or no), diabetes (yes or no), dyslipidemia (yes or no), antihypertensive medications (yes or no), antidiabetic medications (yes or no), and lipid‐lowering medications (yes or no). In these stratified analyses, the multivariable models were adjusted for the same covariates as the main analysis, except for the stratification variable itself. All statistical analyses were conducted using MSTATA (https://www.mstata.com/), and R (version 4.1.0, http://www.R‐project.org). A 2‐sided P value of <0.05 was considered statistically significant.

Results

Vascular Age Indices and Baseline Characteristics

A total of 3561 participants were included, with a mean chronological age of 58.67±8.45 years, and 2318 were women (65.1%). We first constructed 4 distinct vascular age indices, each incorporating traditional risk factors and 1 of the 4 vascular indices. When comparing their ability to explain the variance in chronological age, the VA index based on baPWV demonstrated the highest explanatory power (R ²=0.505), followed by the VA indices based on cIMT (R ²=0.438), carotid plaque (R ²=0.419), and ABI (R ²=0.393) (Table S1).

Given the superior performance of the baPWV‐based VA index, we used its derived phenotypes (EVA, NVA, and SuperNoVA) to stratify and describe the baseline characteristics of the cohort (Table 1). For the baPWV‐based model, EVA was defined when the vascular age was greater than the chronological age of ≥5.42 years (ΔAge >5.42 years), and SuperNoVA was defined when the vascular age was lower than the chronological age of ≥5.15 years (ΔAge <5.15 years).

Table 1.

Baseline Characteristics of Different VA Phenotypes on the Basis of baPWV

Characteristic	Overall	VA phenotype on the basis of baPWV			P value
		EVA	NVA	SuperNoVA
	N=3561	n=356	n=2849	n=356
Female sex	2318 (65.1)	199 (55.9)	1861 (65.3)	258 (72.5)	<0.001
Chronological age, y	58.67±8.45	60.25±9.08	58.16±8.40	61.12±7.57	<0.001
Vascular age, y	58.67±6.01	67.41±5.34	58.29±5.11	52.96±3.71	<0.001
ΔAge, y	0.00±4.21	7.95±2.28	−0.12±2.70	−6.95±1.57	<0.001
Body mass index, kg/m2	25.96±3.53	25.81±3.44	25.95±3.50	26.13±3.80	0.461
Waist circumference, cm	84.79±9.66	87.71±9.21	84.70±9.63	82.60±9.71	<0.001
Systolic blood pressure, mm Hg	126.75±16.50	137.99±16.39	126.13±16.13	120.53±14.37	<0.001
Diastolic blood pressure, mm Hg	74.65±9.65	72.70±9.64	74.60±9.61	77.00±9.51	<0.001
Pulse, beats/min	75.36±10.35	75.73±11.37	75.17±10.25	76.52±10.02	0.052
baPWV, cm/s	1481.00 (1315.00–1687.00)	1894.00 (1693.50–2109.00)	1475.00 (1318.00–1653.00)	1304.00 (1193.50–1438.00)	<0.001
cIMT, mm	0.90±0.14	0.95±0.14	0.90±0.14	0.89±0.13	<0.001
Carotid plaque	1966 (55.2)	248 (69.7)	1544 (54.2)	174 (48.9)	<0.001
ABI	1.07 (1.01–1.12)	1.08 (1.02–1.14)	1.07 (1.01–1.12)	1.06 (1.00–1.11)	0.001
Cardiovascular risk score, %	5.22 (2.67–10.00)	8.71 (5.14–13.51)	4.95 (2.49–9.62)	4.57 (2.67–8.67)	<0.001
Current smoking	568 (16.0)	34 (9.6)	459 (16.1)	75 (21.1)	<0.001
Current drinking	497 (14.0)	60 (16.9)	386 (13.5)	51 (14.3)	0.232
Hypertension	1587 (44.6)	259 (72.8)	1223 (42.9)	105 (29.5)	<0.001
Diabetes	679 (19.1)	120 (33.7)	518 (18.2)	41 (11.5)	<0.001
Dyslipidemia	2695 (75.7)	276 (77.5)	2157 (75.7)	262 (73.6)	0.472
Antihypertensive medications	868 (24.4)	151 (42.4)	658 (23.1)	59 (16.6)	<0.001
Antidiabetic medications	296 (8.3)	64 (18.0)	218 (7.7)	14 (3.9)	<0.001
Lipid‐lowering medications	303 (8.5)	44 (12.4)	244 (8.6)	15 (4.2)	<0.001
Estimated glomerular filtration rate, mL/min per 1.73 m2	74.02±11.50	73.69±12.17	74.22±11.47	72.69±11.02	0.050
Total cholesterol, mmol/L	5.03±0.97	4.88±1.08	5.04±0.96	5.14±0.91	0.002
Triglycerides, mmol/L	1.43 (1.02–2.05)	1.40 (1.00–2.00)	1.42 (1.01–2.04)	1.50 (1.05–2.26)	0.069
High‐density lipoprotein cholesterol, mmol/L	1.23±0.28	1.23±0.29	1.22±0.28	1.23±0.28	0.851
Low‐density lipoprotein cholesterol, mmol/L	2.95±0.80	2.86±0.87	2.96±0.79	2.98±0.77	0.046
Fasting blood glucose, mmol/L	5.48 (5.10–6.08)	5.79 (5.31–6.65)	5.46 (5.09–6.04)	5.32 (4.98–5.83)	<0.001

All values are reported as n (%), mean±SD or median (interquartile range). ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; cIMT, carotid intima‐media thickness; EVA, early vascular aging; NVA, normal vascular aging; SuperNoVA, supernormal vascular aging; and VA, vascular aging.

The EVA group exhibited the highest vascular age (67.4±5.3 years) and median baPWV (1894.0 cm/s), whereas the SuperNoVA group demonstrated the lowest (53.0±3.7 years and 1304.0 cm/s, respectively). The EVA group exhibited a higher‐risk cardiometabolic profile, characterized by higher systolic blood pressure, larger waist circumference, and elevated fasting blood glucose. The prevalence of hypertension, diabetes, and carotid plaques was also highest in the EVA group, consistent with their higher cardiovascular risk scores and greater use of cardiovascular medications. Conversely, the SuperNoVA group was chronologically older and had a higher proportion of women. This group also showed a higher prevalence of current smoking, higher diastolic blood pressure, and elevated total and low‐density lipoprotein cholesterol. No significant differences were observed among the groups for body mass index, pulse rate, current alcohol consumption, triglycerides, or high‐density lipoprotein cholesterol.

Association Between VA and MACEs

During a mean follow‐up of 7.2±1.3 years, a total of 294 (8.26%) participants experienced MACEs, of which 49 (1.37%) experienced cardiovascular death, 228 (6.40%) experienced stroke, and 56 (1.57%) experienced MI.

No significant nonlinear relationship was observed between baPWV‐based ΔAge and MACEs (P for nonlinearity >0.05) (Figure 2A). As baPWV‐based ΔAge increased, the risk of MACEs gradually increased (P‐overall=0.015; Figure 2A). However, there were no significant associations between cIMT‐based ΔAge, carotid plaque–based ΔAge, ABI‐based ΔAge, and the risk of MACEs (all P‐overall >0.05; Figure 2B–2D).

Figure 2. Restricted cubic spline of ΔAge on the basis of the different indices and MACEs.

A, baPWV; (B) cIMT; (C) carotid plaque; (D) ABI. ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; cIMT, carotid intima‐media thickness; and VA, vascular aging.

After adjusting for chronological age, sex, and cardiovascular risk score, with each 1‐SD increase in baPWV‐based ΔAge, the risk of MACEs was significantly increased by 14% (hazard ratio [HR], 1.14 [95% CI, 1.02–1.27]; P=0.017; Table 2). In contrast, these associations were not observed between cIMT‐based ΔAge (HR, 1.03 [95% CI, 0.92–1.15]; P=0.602), carotid plaque–based ΔAge (HR, 1.02 [95% CI, 0.92–1.14]; P=0.667), ABI‐based ΔAge (HR, 1.00 [95% CI, 0.90–1.11]; P=0.995), and the risk of MACEs (Table 2).

Table 2.

Association of ΔAge on the Basis of the Different Indices With the Risk of MACEs

	Events/n (%)	Model 1		Model 2		Model 3
		HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
ΔAge (baPWV)	294/3561 (8.26)	1.28 (1.14–1.43)	<0.001	1.19 (1.07–1.33)	0.001	1.14 (1.02–1.27)	0.017
ΔAge (cIMT)	294/3561 (8.26)	1.17 (1.04–1.31)	0.007	1.09 (0.98–1.21)	0.112	1.03 (0.92–1.15)	0.602
ΔAge (carotid plaque)	294/3561 (8.26)	1.17 (1.05–1.32)	0.005	1.1 (0.99–1.22)	0.091	1.02 (0.92–1.14)	0.667
ΔAge (ABI)	294/3561 (8.26)	1.18 (1.05–1.32)	0.004	1.08 (0.98–1.20)	0.135	1.00 (0.90–1.11)	0.995

Model 1: unadjusted; model 2: adjusted for chronological age and sex; model 3: adjusted for chronological age, sex, and cardiovascular risk score. ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; HR, hazard ratio; cIMT, carotid intima‐media thickness; and MACE, major adverse cardiovascular event.

The Kaplan–Meier survival curve indicated that there were significant differences across different VA phenotypes on the basis of baPWV (log‐rank P<0.001), carotid plaque (log‐rank P=0.022), and ABI (log‐rank P=0.002). However, there were no significant differences for cIMT‐based VA phenotypes (log‐rank P>0.05; Figure 3).

Figure 3. Kaplan–Meier survival curves of VA phenotypes on the basis of the different indices and MACEs.

A, baPWV; (B) cIMT; (C) carotid plaque; (D) ABI. ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; cIMT, carotid intima‐media thickness; EVA, early vascular aging; NVA, normal vascular aging; SuperNoVA, supernormal vascular aging; and VA, vascular aging.

For baPWV‐based VA phenotypes, after adjusting for chronological age, sex, and cardiovascular risk score, the EVA group had a significantly increased risk of MACEs compared with the NVA group (HR, 1.44 [95% CI, 1.05–1.97]; P=0.023; Table 3). Specifically, the EVA group exhibited a significantly increased risk of stroke (HR, 1.48 [95% CI, 1.04–2.09]; P=0.029), whereas the associations with cardiovascular death and MI were not statistically significant (both P>0.05; Table S3). In contrast, the SuperNoVA group showed no significant association with MACE risk compared with the NVA group (HR, 1.02 [95% CI, 0.68–1.52]; P=0.932). While the EVA group based on carotid plaque and ABI were associated with increased MACE risk compared with the NVA group in unadjusted models, these associations became nonsignificant after adjusting for chronological age and sex (Table 3). In the fully adjusted models, no significant associations with MACEs were observed for VA phenotypes based on cIMT, carotid plaque, or ABI (all P>0.05).

Table 3.

Association of VA Phenotypes on the Basis of the Different Indices With the Risk of MACEs

VA phenotype	Events/n (%)	Model 1		Model 2		Model 3
		HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Based on baPWV
EVA	52/356 (14.61)	2.01 (1.48–2.72)	<0.001	1.61 (1.19–2.18)	0.002	1.44 (1.05–1.97)	0.023
NVA	213/2849 (7.48)	1.00 (Reference)	—	1.00 (Reference)	—	1.00 (Reference)	—
SuperNoVA	29/356 (8.15)	1.09 (0.74–1.61)	0.653	0.95 (0.65–1.41)	0.816	1.02 (0.68–1.52)	0.932
Based on clMT
EVA	39/356 (10.96)	1.36 (0.97–1.91)	0.074	1.05 (0.74–1.47)	0.93	0.93 (0.66–1.32)	0.690
NVA	230/2849 (8.07)	1.00 (Reference)	—	1.00 (Reference)	—	1.00 (Reference)	—
SuperNoVA	25/356 (7.02)	0.87 (0.57–1.31)	0.496	0.76 (0.50–1.15)	0.196	0.92 (0.60–1.40)	0.684
Based on carotid plaque
EVA	42/356 (11.80%)	1.50 (1.08, 2.08)	0.016	1.13 (0.81, 1.58)	0.456	0.96 (0.68, 1.35)	0.806
NVA	229/2849 (8.04%)	1.00 (Reference)	—	1.00 (Reference)	—	1.00 (Reference)	—
SuperNoVA	23/356 (6.46%)	0.80 (0.52, 1.23)	0.312	0.76 (0.50, 1.17)	0.22	0.91 (0.59, 1.42)	0.680
Based on ABI
EVA	46/356 (12.92)	1.75 (1.27–2.41)	<0.001	1.27 (0.92–1.75)	0.148	1.06 (0.76–1.48)	0.736
NVA	215/2849 (7.55)	1.00 (Reference)	—	1.00 (Reference)	—	1.00 (Reference)	—
SuperNoVA	33/356 (9.27)	1.24 (0.86–1.79)	0.253	1.08 (0.75–1.56)	0.688	1.27 (0.86–1.86)	0.225

Model 1: unadjusted; model 2: adjusted for chronological age and sex; model 3: adjusted for chronological age, sex, and cardiovascular risk score. ABI indicates ankle–brachial index; baPWV, brachial–ankle pulse wave velocity; cIMT, carotid intima‐media thickness; EVA, early vascular aging; HR, hazard ratio; MACE, major adverse cardiac event; NVA, normal vascular aging; SuperNoVA, supernormal vascular aging; and VA, vascular aging.

Subgroup Analyses

In exploratory subgroup analyses, the association between baPWV‐based ΔAge and the risk of MACEs was generally consistent across various key clinical factors (all P for interaction >0.05; Table S4). However, the association was statistically significant among women (HR, 1.22 [95% CI, 1.05–1.43]; P=0.011), individuals aged <65 years (HR, 1.20 [95% CI, 1.03–1.40]; P=0.020), and those without dyslipidemia (HR, 1.28 [95% CI, 1.02–1.61]; P=0.034), whereas no significant association was found in men, individuals aged ≥65 years, and those with dyslipidemia (all P>0.05).

For cIMT‐based ΔAge, dyslipidemia may be a potential modifying factor for the association with MACEs (P for interaction=0.006). In individuals without dyslipidemia, higher cIMT‐based ΔAge was associated with increased MACE risk (HR, 1.28 [95% CI, 1.03–1.59]; P=0.028), while no such association was found in those with dyslipidemia (HR, 0.96 [95% CI, 0.85–1.08]; P=0.480). No significant main associations were found between ΔAge based on carotid plaque or ABI and MACE risk across all subgroups (all P>0.05).

Discussion

In this community‐based prospective cohort, we conducted the first direct comparison of the prognostic utility of VA indices derived from distinct functional or structural vascular indicators. Our main finding is that the VA index based on baPWV demonstrated the best prognostic value for MACEs. After full multivariable adjustment, higher baPWV‐based ΔAge remained significantly associated with a higher risk of MACEs. Consistent with this, the baPWV‐based EVA group was also independently associated with an increased risk of MACEs compared with the NVA group. However, the baPWV‐based SuperNoVA group was not associated with a significantly lower risk of MACEs. In striking contrast, after the same adjustments, neither the ΔAge nor VA phenotypes based on cIMT, carotid plaque, or ABI showed a significant association with the risk of MACEs.

Vascular Aging and Cardiovascular Risk

VA refers to the pathophysiological process in which the structure and function of blood vessels undergo degenerative changes with increasing age. ²⁰ The biological age of the vasculature, which reflects the burden of these changes, is a more informative predictor of cardiovascular risk than chronological age. ⁵ Multiple approaches have been proposed to operationalize this concept, ranging from risk score–based estimations to the direct use of individual vascular biomarkers. ⁴ The prevailing contemporary method is the construction of vascular age indices in which chronological age is estimated from vascular indicators and conventional risk factors. ²¹ Evidence from some prospective studies has demonstrated the prognostic value of VA. Bruno et al first showed that vascular age phenotypes derived from carotid–femoral pulse wave velocity (cfPWV), particularly the EVA phenotype, were strongly associated with incident cardiovascular events in a European cohort. ¹⁰ More recently, studies conducted in Chinese populations using baPWV have further validated these associations, confirming that individuals classified as EVA have an increased risk of cardiovascular events and death. ¹³ , ¹⁴ Our findings were consistent with previous studies, confirming that vascular age phenotypes based on baPWV were significantly associated with cardiovascular risk. Moreover, a greater ΔAge, indicative of an advanced vascular age relative to chronological age, was significantly associated with an increased risk of incident cardiovascular events.

In contrast with prior studies that identified a protective effect of the SuperNoVA phenotype, ¹⁰ , ¹³ , ¹⁴ , ²² we did not observe a significant association between SuperNoVA and reduced cardiovascular risk. This discrepancy may be explained by several reasons. First, differences in baseline risk profiles may play a role. Our SuperNoVA group, despite showing more favorable vascular indicators, may still have carried a higher burden of residual risk factors not fully captured by the VA index. Second, the relatively small number of cardiovascular events within the SuperNoVA group may have limited our ability to detect the protective effect, which is a common challenge when evaluating low‐risk populations. Third, methodological variations in VA index construction, including the selection of covariates and characteristics of the reference population, may have influenced the classification of individuals at the extremes of the ΔAge spectrum. Notably, our findings are consistent with a recent study, which similarly reported no significant association between SuperNoVA and incident cardiovascular events. ²³ Overall, these findings suggest that the prognostic value of the SuperNoVA phenotype may vary across populations and study designs, underscoring the need for further large‐scale investigations using diverse cohorts and multiple vascular indicators to clarify its clinical relevance. Importantly, the lack of a statistically significant protective association for the SuperNoVA phenotype observed in our study should not be overgeneralized. This finding may be specific to the characteristics of our cohort, particularly the relatively low absolute event rate in this healthy group that limited statistical power. Consequently, the potential benefit of delayed vascular aging might still be evident in higher‐risk populations or over extended follow‐up periods.

Comparative Performance of Different Vascular Indicators

A variety of clinical indicators have been applied to evaluate vascular function and structure. In our analysis, 4 widely used indicators, including baPWV, cIMT, carotid plaque, and ABI, were compared within the same framework. Prior studies have shown that both baPWV and cfPWV, as functional markers of arterial stiffness, are associated with cardiovascular outcomes. ¹⁰ , ¹³ , ¹⁴ Notably, a recent head‐to‐head comparison in a Chinese cohort found that a baPWV‐based VA index provided stronger prognostic value than cfPWV‐based VA index. ²³ In addition, Xu et al proposed a VA index integrating baPWV and cIMT, which demonstrated predictive capacity for adverse cardiovascular events. ²² However, it did not compare the independent contributions of functional and structural indicators. Our study performs a direct comparison between VA indices derived from distinct categories of indicators. We observed that only baPWV‐based ΔAge and EVA phenotypes were independently associated with cardiovascular risk. In contrast, models derived from cIMT, carotid plaque, or ABI did not retain significance after adjustment for conventional risk factors. Our findings indicate that baPWV may be the most valuable indicator for constructing a VA index and reflecting the process of VA in the Chinese community‐based population.

A plausible explanation is that baPWV reflects systemic arterial stiffness and thus integrates the cumulative effects of long‐term risk exposures and multiple biological pathways across the entire arterial tree. ²⁴ Experimental and translational studies have shown that oxidative stress, impaired vascular repair, chronic inflammation, and endothelial dysfunction are central mechanisms driving VA, all of which contribute to progressive stiffening of arteries. ²⁰ By capturing these global functional alterations, baPWV provides a comprehensive measure of vascular health and a sensitive marker of early vascular injury. In contrast, cIMT and carotid plaque represent localized structural changes that reflect atherosclerosis at a specific vascular site. Specifically, cIMT measured at the distal common carotid artery primarily reflects wall hypertrophy rather than atherosclerotic lesions. ²⁵ Their predictive value may be limited because they do not account for systemic arterial function and may be influenced by focal or anatomic variability. ABI, on the other hand, is mainly a marker of advanced peripheral artery disease and therefore reflects late‐stage vascular damage rather than early functional decline. ²⁶ As a result, these indicators may fail to capture subclinical VA in the general population.

Despite the statistical significance, we acknowledge that the predictive value of baPWV‐based vascular age was relatively modest after full adjustment for traditional risk factors. Therefore, we do not propose it as a stand‐alone replacement for established risk prediction models. Instead, its primary clinical value lies in risk communication and screening. The concept of vascular age translates complex stiffness data into an understandable metric to motivate lifestyle modifications. ²⁷ Furthermore, it serves as a complementary tool to identify individuals with early vascular aging who might otherwise be overlooked by conventional scoring systems.

Potential Effects of Sex, Age, and Dyslipidemia

Although we did not observe statistically significant interactions in exploratory subgroup analyses of the baPWV‐based VA index, there were several noteworthy patterns. First, the association between ΔAge and cardiovascular risk appeared stronger in women, which is consistent with results from previous Chinese studies. ¹³ , ²² These findings suggest that sex hormones and sex‐specific VA trajectories, particularly the impact of menopause‐related estrogen decline, may underlie the stronger associations observed in women. ²⁸ Second, we found that vascular age was more predictive in nonelderly participants. While previous studies have not directly examined age‐specific interactions, Zuo et al focused on a strictly middle‐aged population and demonstrated that vascular age categories had significant prognostic value. ¹⁴ Our findings, therefore, extend this observation, suggesting that VA indices may be particularly informative before the onset of advanced age, when competing comorbidities may dilute the relative contribution of VA to clinical outcomes. Finally, to our knowledge, our study is the first to report a potential modifying role of dyslipidemia. For cIMT‐based vascular age, we observed that the association with the risk of cardiovascular events was significant only in participants without dyslipidemia. One possible explanation is that in individuals with dyslipidemia, conventional risk models already capture much of the risk attributable to lipid‐related vascular injury, thereby diminishing the incremental predictive value of vascular age. ²⁹ In contrast, among those without dyslipidemia, vascular age may reveal subclinical vascular injury independent of lipid metabolism, thus identifying individuals at hidden risk. ³⁰ This novel observation warrants confirmation in future studies and may provide new insight into how VA interacts with metabolic health in determining cardiovascular outcomes.

Strengths and Limitations

This study has several strengths. First, we conducted a longitudinal cohort analysis in a large Chinese community‐based population, providing further evidence that VA phenotypes are closely associated with cardiovascular risk. Second, we performed the first systematic comparison of VA indices derived from multiple functional and structural indicators, demonstrating that baPWV‐based vascular age was the most valuable for cardiovascular risk prediction. Third, we carried out extensive exploratory subgroup analyses. These analyses highlighted that sex differences, age groups, and lipid metabolism may play an important role in modifying the prognostic value of vascular age, adding new insights beyond existing literature.

However, this study also has several limitations. First, due to sample size constraints, the number of cardiovascular events within extreme phenotypes such as SuperNoVA remained limited, which may have reduced statistical power to detect modest protective associations. Second, although we adjusted for conventional cardiovascular risk factors, residual confounding cannot be excluded, particularly from lifestyle and genetic factors not fully captured in our VA indices. Third, our findings were derived from a Chinese community‐based adult population, and the lack of external validation in independent cohorts limits the broader generalizability of our findings. Notably, baPWV is extensively validated and widely applied in East Asian populations, whereas cfPWV is more commonly used in Western cohorts. ³¹ This distinction supports the implementation of region‐specific vascular aging strategies. Fourth, our analysis relied on baseline measurements, and we could not fully account for time‐varying changes in medication use or other risk factors, which might have influenced the results. Finally, although our study included both functional and structural vascular indicators, cfPWV was not available, and other potentially informative indices, such as augmentation index, reactive hyperemia index, and arterial stiffness index, were not assessed.

Conclusions

In this community‐based cohort of Chinese adults, we comprehensively evaluated VA indices derived from both functional and structural indicators and demonstrated that baPWV‐based vascular age was the most informative predictor of future cardiovascular events. Exploratory subgroup analyses suggested that the prognostic value of VA may be particularly relevant in women, nonelderly populations, and individuals without dyslipidemia. These findings emphasize the clinical utility of baPWV‐derived vascular age for improving cardiovascular risk stratification. Future studies in more diverse populations, with larger event numbers in extreme phenotypes and with additional vascular biomarkers, are needed to confirm our observations and refine the clinical application of VA phenotypes.

Sources of Funding

Beijing Natural Science Foundation (7252190, L246054) and National Key Research and Development Program of China (2021YFC2500503).

Disclosures

None.

Supporting information

Tables S1–S4

STROBE Statement