Abstract
Background
Accelerated aging is a dynamic process, yet few studies examined the association of changes in accelerated aging with cardiovascular disease (CVD) and mortality. This study aims to evaluate this association in three prospective cohorts from China and the UK.
Methods
Data were drawn from the Kailuan cohort (n = 107,830), the Dongfeng-Tongji (DFTJ) cohort (n = 14,032), and the UK Biobank (n = 316,087). Accelerated aging was assessed by PhenoAge and Klemera-Doubal method (KDM) age, measured at baseline (Kailuan cohort: 2006–2009; DFTJ cohort: 2008–2010; UK Biobank: 2006–2010) and the first follow-up (Kailuan cohort: 2010–2013; DFTJ cohort: 2013; UK Biobank: 2012–2013). Changes in accelerated aging were classified as persistent accelerated aging, recovery from accelerated aging, delayed accelerated aging, and stable non-accelerated aging. Cox proportional hazard models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Meta-analysis was performed to summarize estimates across three cohorts.
Results
Median follow-up periods were 10.3–15.9 years across three cohorts. When defining accelerated aging by PhenoAge, baseline accelerated aging was significantly associated with increased risks of CVD (pooled HR: 1.41, 95% CI: 1.25, 1.60) and mortality (pooled HR: 1.47, 95% CI: 1.33, 1.63). Compared to participants with persistent accelerated aging, participants recovering from accelerated aging (pooled HR of CVD: 0.76, 95% CI: 0.72, 0.81; pooled HR of mortality: 0.84, 95% CI: 0.78, 0.89), delaying accelerated aging (pooled HR of CVD: 0.75, 95% CI: 0.70, 0.79; pooled HR of mortality: 0.77, 95% CI: 0.72, 0.83) or maintaining non-accelerated aging (pooled HR of CVD: 0.59, 95% CI: 0.48, 0.71; pooled HR of mortality: 0.58, 95% CI: 0.55, 0.62) exhibited decreased risks of both CVD and mortality. When defining accelerated aging by KDM age, the results remained consistent with those of PhenoAge.
Conclusions
Accelerated aging is a significant risk factor for CVD and mortality. Recovering from or delaying accelerated aging, or maintaining non-accelerated aging, was associated with reduced risks of CVD and mortality.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at 10.1186/s12916-025-04365-8.
Keywords: Accelerated aging, Biological age, Cardiovascular disease, Mortality, Cohort
Background
In recent decades, global population aging has become a significant challenge, with the proportion of individuals aged 65 and older projected to increase from 10% in 2022 to 16% in 2050 [1, 2]. This rapid demographic shift is intensifying the burden of age-related health outcomes, especially cardiovascular disease (CVD) and premature death [3].
Aging is a complex process marked by physiological dysregulation across multiple systems, and key mechanisms, such as chronic inflammations and telomere attrition, are central to the progressive deterioration [4, 5]. While aging is a universal phenomenon, the rate of aging varies widely among individuals, even among those of the same chronological age. Biological age, which reflects an individual’s physiological condition, is a useful measure for assessing the aging rate and detecting accelerated aging [6]. Recently, there has been growing interest in understanding the associations of accelerated aging with CVD and mortality [7]. Studies have shown that accelerated aging was associated with increased risks of CVD and mortality [8–10]. Moreover, a multistate analysis demonstrated that individuals with accelerated aging were at higher risks of progressing from initial cardiometabolic disease to cardiometabolic multimorbidity and eventually to death [11]. Clarifying these associations is crucial for informing strategies to prevent CVD and premature death.
However, previous studies primarily focused on a single assessment of accelerated aging, without accounting for the dynamic nature of aging [8–15]. Existing evidence has suggested that accelerated aging can be delayed or reversed through interventions such as a healthy lifestyle [16], caloric restriction [17], and high-intensity interval training [18]. Compared to a single assessment of accelerated aging, examining changes in accelerated aging over time could provide insights into different aging patterns, such as delayed or reversed accelerated aging [19]. In addition, previous studies were mainly based on a single cohort, which limited the generalizability of their findings [8–11, 14, 15]. Therefore, research integrating multiple cohorts is needed to demonstrate the associations between changes in accelerated aging and the risks of CVD and mortality.
This study aimed to assess the associations of changes in accelerated aging with the risks of CVD and mortality using data from three cohorts in China and the UK.
Methods
Populations
Participants were recruited from three cohorts: the Kailuan study and the Dongfeng-Tongji (DFTJ) cohort in China, and the UK Biobank. The Kailuan cohort enrolled 101,510 participants from the Kailuan Company in Tangshan, China, between 2006 and 2007. Biennial follow-up surveys have been conducted thereafter to update information and recruit new participants. The DFTJ cohort recruited 27,009 retired employees from Dongfeng Motor Corporation in Shiyan, China, between 2008 and 2010. During the first follow-up in 2013, a total of 24,175 participants were re-assessed. Detailed information about the DFTJ cohort is available in the previous publication [20]. The UK Biobank is a large-scale cohort study that collected data from over 500,000 participants across the UK between 2006 and 2010. A subgroup of ~ 20,000 participants underwent repeated assessments during the first follow-up in 2012–2013. Comprehensive details about the UK Biobank can be found elsewhere [21].
Detailed inclusion and exclusion criteria for each cohort are provided elsewhere (Additional file 1: Text S1 and Fig. S1). Briefly, in the Kailuan cohort, we included 147,552 participants aged 20 to 79 years, who were newly enrolled between 2006 and 2021 to construct biological ages for the Chinese population. When examining the associations between baseline accelerated aging and the risks of CVD and mortality, we included 107,830 participants from the Kailuan cohort, 14,032 from the DFTJ cohort, and 316,087 from the UK Biobank. For analyses of changes in accelerated aging, the included participants were 76,812 from the Kailuan cohort, 9480 from the DFTJ cohort, and 6186 from the UK Biobank.
The Kailuan study was approved by the Medical Ethics Committee of Kailuan General Hospital (no. 2006–5). The DFTJ cohort was approved by the Medical Ethics Committee of the School of Public Health, Tongji Medical College, Huazhong University of Science and Technology (no. 2008–3). The UK Biobank study was approved by the North West Multi-centre Research Ethics Committee (REC reference numbers 11/NW/0382, 16/NW/0274, and 21/NW/0157). All participants provided written informed consent.
Assessment of biological age and accelerated aging
In this study, we assessed two biological ages, PhenoAge and Klemera-Doubal method (KDM) age, both of which were derived from clinical indicators. PhenoAge was initially proposed by Levine et al., and it was calculated from a Gompertz proportional hazards model of mortality, and represents the age at which a participant’s predicted mortality score aligns with the average mortality hazard [22]. KDM age was calculated through a series of regressions that related individual clinical indicators to chronological age within a reference population. It reflects the age at which an individual’s physiology would be approximately normal [23].
In the Kailuan cohort, we initially used 30 clinical indicators (Additional file 1: Table S1) as candidate variables to develop biological ages for the Chinese population, based on data from participants newly enrolled between 2006 and 2021. Following procedures from previous studies [24, 25], we finally selected nine clinical indicators (fasting blood glucose, waist circumference, systolic blood pressure, high-density lipoprotein cholesterol, total cholesterol, blood urea nitrogen, mean corpuscular hemoglobin, platelets, red blood cell count) to develop PhenoAge and KDM age. The detailed process for selecting clinical indicators and the formula for calculating PhenoAge and KDM age are provided in Additional file 1: Text S1.
In the DFTJ cohort, we used the same nine clinical indicators and the same formula as in the Kailuan cohort to calculate PhenoAge and KDM age.
In the UK Biobank, we used nine clinical indicators (white blood cell count, red blood cell distribution width, lymphocyte percentage, mean corpuscular volume, albumin, blood glucose, C-reactive protein, alkaline phosphatase, and blood creatinine) to calculate PhenoAge [13–15], and another set of nine indicators (forced expiratory volume, systolic blood pressure, alkaline phosphatase, total cholesterol, albumin, blood creatinine, C-reactive protein, glycated hemoglobin, and blood urea) to calculate KDM age [11, 13]. Both PhenoAge and KDM age, derived from these indicators, have been shown to be associated with the risks of age-related diseases in the UK Biobank [11, 13–15]. The formula for calculating PhenoAge and KDM age in the UK Biobank is detailed in Additional file 1: Text S1.
Biological age acceleration was determined by the residuals from regressing biological age on chronological age. A residual > 0 indicates accelerated aging (biologically older), while ≤ 0 indicates non-accelerated aging (biologically younger). Participants were divided into four groups based on their status at baseline and first follow-up: persistent accelerated aging (accelerated aging at both time points), recovery from accelerated aging (accelerated to non-accelerated), delayed accelerated aging (non-accelerated to accelerated), and stable non-accelerated aging (non-accelerated aging at both time points).
Outcome ascertainment
The primary outcomes of this study were incident CVD, all-cause mortality, and CVD mortality. CVD was defined as coronary heart disease (CHD) or myocardial infarction (MI), and stroke. In the Kailuan cohort, incident CVD and all-cause mortality were tracked until December 31, 2022, while CVD mortality, which needed extra time for adjudication, was updated to December 31, 2016. In the DFTJ cohort, incident CVD and mortality were followed until December 31, 2018. In the UK Biobank, incident CVD was identified until October 31, 2022, for England, September 24, 2021, for Scotland, and May 28, 2021, for Wales, and mortality was tracked until December 19, 2022. Further details on outcome ascertainment are provided in Additional file 1: Text S1.
Statistical analyses
Multivariable Cox proportional hazards models were used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs). The proportional hazards assumption was evaluated using Schoenfeld residuals, and no evident violation was found. A detailed description of the covariates is provided in Additional file 1: Table S2 [26–28]. In model 1, no covariates were adjusted. In model 2, we adjusted for age, sex, and self-reported race (UK Biobank only). In model 3, we further adjusted for education levels, Townsend deprivation index (UK Biobank only), smoking status, drinking status, diet, physical activity levels, body mass index, sleep duration, and number of comorbidities. When assessing the associations of changes in accelerated aging and accelerated aging at the first follow-up with outcomes, we adjusted for covariates collected at the first follow-up. Person-time was calculated from the first follow-up date to the earliest date of death/incident CVD, loss to follow-up, or end of follow-up. For the associations between baseline accelerated aging and outcomes, we adjusted for baseline covariates, with person-time calculated from the baseline recruitment date to the earliest date of the aforementioned events.
Stratified analyses were conducted based on baseline chronological age and sex. We used the likelihood ratio tests to evaluate multiplicative interactions. To assess the predictive performance of PhenoAge and KDM age for mortality and CVD events, we replaced chronological age with PhenoAge and KDM age in model 3, respectively, and calculated the Harrell’s C-index for each model.
In sensitivity analyses, we first excluded cases of CVD or death within the first year to reduce potential reverse causality bias. Second, we redefined accelerated aging by the age gap between chronological age and biological age (biological age - chronological age). A positive age gap indicated accelerated aging, while a non-positive age gap indicated non-accelerated aging. Third, we further adjusted for individual comorbidities in the Cox proportional hazards models, given that participants with the same number of comorbidities may not be comparable. In addition, for the Kailuan cohort, we additionally adjusted for chronic kidney disease. Fourth, we adjusted for baseline covariates rather than those collected at the first follow-up when examining the associations between changes in accelerated aging and the risks of CVD and mortality, with person-time calculated from baseline to the earliest date of death/incident CVD, loss to follow-up, or end of follow-up. Fifth, we assessed the associations of changes in accelerated aging with CVD risks and conducted predictive performance assessments based on competing risk models, accounting for the competing risk between incident CVD and non-CVD mortality. Sixth, because our study populations with repeated measures data were subsets of the original cohorts, we constructed inverse probability weighted regression models to reduce potential selection bias [29]. Finally, we calculated E-values for the associations between changes in accelerated aging and the risks of CVD and mortality to account for the unmeasured confounding [30].
Missing values of covariates were imputed using median values for continuous variables and mode values for categorical variables when the missing rate was ≤ 5%, and a missing indicator was applied for variables with a missing rate > 5% [31].
Statistical analyses were performed using the R software (version 4.1.3). Analyses were conducted independently in each cohort, and meta-analysis was performed to summarize estimates across three cohorts.
Results
Characteristics of participants
A total of 107,830 (the Kailuan cohort), 14,032 (the DFTJ cohort), and 316,087 (UK Biobank) participants were included in the baseline accelerated aging analyses. Additionally, 76,812 (the Kailuan cohort), 9480 (the DFTJ cohort), and 6186 (the UK Biobank) individuals were included for analyzing changes in accelerated aging. Baseline and first follow-up characteristics of participants are presented in Table 1 and Additional file 1: Table S3. Baseline characteristics of participants lost at the first follow-up and those not lost are shown in Additional file 1: Table S4.
Table 1.
Baseline characteristics of the participants a
| Kailuan cohort | Dongfeng-Tongji cohort | UK Biobank | |
|---|---|---|---|
| Participants, n | 107,830 | 14,032 | 316,087 |
| Age (years) | 50.3 ± 12.0 | 62.4 ± 7.4 | 56.1 ± 8.1 |
| Males, n (%) | 85,800 (79.6) | 6140 (43.8) | 141,191 (44.7) |
| White, n (%) | - | - | 300,103 (94.9) |
| College/university degree, n (%) | - | - | 104,939 (33.2) |
| Education levels | |||
| Primary school or below | 9400 (8.7) | 4114 (29.3) | - |
| Middle school | 75,131 (69.7) | 5305 (37.8) | - |
| High school or beyond | 23,299 (21.6) | 4613 (32.9) | - |
| Adequate physical activity, n (%) | 16,215 (15.0) | 10,343 (73.7) | 244,748 (77.4) |
| Never smokers, n (%) | 64,974 (60.3) | 10,009 (71.3) | 178,139 (56.4) |
| Never drinkers, n (%) | 63,886 (59.2) | 10,255 (73.1) | 13,228 (4.2) |
| Healthy diet, n (%) | 90,394 (83.8) | 7652 (54.5) | 45,960 (14.5) |
| Number of comorbidities, n (%) | |||
| 0–1 | 103,839 (96.3) | 11,293 (80.5) | 274,567 (86.9) |
| 2 | 3386 (3.1) | 1921 (13.7) | 32,529 (10.3) |
| ≥ 3 | 605 (0.6) | 818 (5.8) | 8991 (2.8) |
| Accelerated aging defined by PhenoAge, n (%) | 50,276 (46.6) | 6577 (46.9) | 141,499 (44.8) |
| Accelerated aging defined by KDM age, n (%) | 49,049 (45.5) | 6439 (45.9) | 152,455 (48.2) |
| Townsend deprivation index | - | - | −1.4 ± 3.0 |
| Sleep duration (h/day) | 7.2 ± 1.1 | 8.1 ± 1.3 | 7.2 ± 1.1 |
| Body mass index (kg/m2) | 25.0 ± 3.5 | 24.3 ± 3.3 | 27.2 ± 4.7 |
a Continuous variables were expressed as mean ± standard deviation, and categorical variables were described as number (percentage). KDM, Klemera-Doubal method
In both the Kailuan cohort and the DFTJ cohort, PhenoAge and KDM age were highly correlated with chronological age, while the correlations were slightly weaker in the UK Biobank (Additional file 1: Table S5). In three cohorts, approximately half of the participants were at accelerated aging at both baseline and the first follow-up (Additional file 1: Table S6).
The median time interval from baseline to the first follow-up was 4.0 years in the Kailuan cohort, 4.6 years in the DFTJ cohort, and 4.5 years in the UK Biobank. In the baseline accelerated aging analyses, during median follow-ups of 15.9 years (Kailuan cohort), 10.3 years (DFTJ cohort), and 13.7 years (UK Biobank), we identified 14,691 deaths, 1375 cardiovascular deaths, 10,688 CVD cases, 2283 MI cases, and 8710 stroke cases in the Kailuan cohort; 1472 deaths, 502 cardiovascular deaths, 4758 CVD cases, 3732 CHD cases, and 1026 stroke cases in the DFTJ cohort; and 22,810 deaths, 2673 cardiovascular deaths, 30,470 CVD cases, 25,055 CHD cases, and 6814 stroke cases in the UK Biobank. In analyses assessing changes in accelerated aging, median follow-up periods were 11.8 years (Kailuan cohort), 5.7 years (DFTJ cohort), and 9.8 years (UK Biobank).
Association of accelerated aging with CVD and mortality
In three cohorts, baseline accelerated aging defined by PhenoAge was significantly associated with increased risks of CVD (pooled HR: 1.41, 95% CI:1.25, 1.60), CHD/MI (pooled HR: 1.33, 95% CI:1.22, 1.46), stroke (pooled HR: 1.53, 95% CI:1.35, 1.73), all-cause mortality (pooled HR: 1.47, 95% CI:1.33, 1.63), and CVD mortality (pooled HR: 1.74, 95% CI:1.64, 1.85). Comparable associations were observed when defining accelerated aging by KDM age. Similarly, accelerated aging observed at the first follow-up was significantly associated with elevated risks of CVD, CHD/MI, stroke, all-cause mortality, and CVD mortality (Table 2).
Table 2.
Associations of accelerated aging at a single time point with the risks of cardiovascular disease and mortality a
| Baseline | The first follow-up | |||||
|---|---|---|---|---|---|---|
| HR (95% CI) b | HR (95% CI) c | HR (95% CI) d | HR (95% CI) b | HR (95% CI) c | HR (95% CI) d | |
| Accelerated aging assessed by PhenoAge | ||||||
| Kailuan cohort | ||||||
| CVD | 1.75 (1.68, 1.82) | 1.69 (1.63, 1.76) | 1.60 (1.53, 1.66) | 1.71 (1.63, 1.80) | 1.63 (1.55, 1.72) | 1.55 (1.47, 1.63) |
| MI | 1.64 (1.51, 1.78) | 1.56 (1.44, 1.70) | 1.47 (1.35, 1.60) | 1.58 (1.41, 1.77) | 1.47 (1.31, 1.65) | 1.39 (1.24, 1.57) |
| Stroke | 1.79 (1.71, 1.86) | 1.73 (1.66, 1.81) | 1.63 (1.56, 1.71) | 1.75 (1.65, 1.85) | 1.68 (1.59, 1.78) | 1.60 (1.50, 1.69) |
| Mortality | 1.51 (1.46, 1.56) | 1.46 (1.41, 1.51) | 1.46 (1.41, 1.50) | 1.44 (1.37, 1.51) | 1.39 (1.32, 1.46) | 1.38 (1.31, 1.45) |
| CVD mortality | 2.03 (1.81, 2.26) | 1.93 (1.73, 2.16) | 1.90 (1.70, 2.13) | 1.95 (1.60, 2.36) | 1.85 (1.52, 2.24) | 1.80 (1.47, 2.19) |
| Dongfeng-Tongji cohort | ||||||
| CVD | 1.45 (1.37, 1.53) | 1.44 (1.36, 1.52) | 1.34 (1.26, 1.42) | 1.46 (1.34, 1.58) | 1.45 (1.34, 1.58) | 1.36 (1.24, 1.48) |
| CHD | 1.33 (1.25, 1.42) | 1.33 (1.25, 1.42) | 1.24 (1.16, 1.32) | 1.33 (1.22, 1.46) | 1.34 (1.22, 1.46) | 1.24 (1.13, 1.37) |
| Stroke | 1.76 (1.56, 2.00) | 1.70 (1.50, 1.92) | 1.63 (1.44, 1.86) | 2.00 (1.63, 2.46) | 1.95 (1.59, 2.39) | 1.88 (1.52, 2.32) |
| Mortality | 1.38 (1.25, 1.53) | 1.33 (1.20, 1.48) | 1.33 (1.19, 1.47) | 1.32 (1.10, 1.58) | 1.30 (1.09, 1.57) | 1.32 (1.10, 1.60) |
| CVD mortality | 1.76 (1.47, 2.10) | 1.72 (1.43, 2.05) | 1.69 (1.41, 2.03) | 1.49 (1.07, 2.09) | 1.48 (1.06, 2.08) | 1.53 (1.08, 2.16) |
| UK Biobank | ||||||
| CVD | 1.62 (1.59, 1.66) | 1.45 (1.42, 1.49) | 1.31 (1.28, 1.34) | 1.63 (1.35, 1.95) | 1.43 (1.19, 1.73) | 1.37 (1.13, 1.65) |
| CHD | 1.66 (1.62, 1.70) | 1.47 (1.43, 1.51) | 1.31 (1.28, 1.35) | 1.54 (1.26, 1.89) | 1.33 (1.08, 1.64) | 1.25 (1.01, 1.55) |
| Stroke | 1.57 (1.49, 1.64) | 1.46 (1.40, 1.54) | 1.36 (1.30, 1.43) | 1.75 (1.22, 2.51) | 1.62 (1.12, 2.35) | 1.65 (1.13, 2.40) |
| Mortality | 1.89 (1.84, 1.94) | 1.75 (1.71, 1.80) | 1.60 (1.56, 1.64) | 1.95 (1.55, 2.45) | 1.78 (1.41, 2.25) | 1.69 (1.34, 2.15) |
| CVD mortality | 2.25 (2.08, 2.44) | 1.92 (1.77, 2.08) | 1.68 (1.55, 1.82) | 1.41 (0.76, 2.60) | 1.21 (0.65, 2.26) | 1.14 (0.61, 2.16) |
| Pooled results e | ||||||
| CVD | 1.60 (1.44, 1.78) | 1.52 (1.37, 1.69) | 1.41 (1.25, 1.60) | 1.60 (1.43, 1.78) | 1.53 (1.39, 1.67) | 1.45 (1.31, 1.60) |
| CHD/MI | 1.54 (1.33, 1.77) | 1.45 (1.33, 1.58) | 1.33 (1.22, 1.46) | 1.43 (1.34, 1.53) | 1.38 (1.29, 1.48) | 1.29 (1.20, 1.39) |
| Stroke | 1.70 (1.55, 1.86) | 1.62 (1.45, 1.81) | 1.53 (1.35, 1.73) | 1.77 (1.67, 1.87) | 1.70 (1.61, 1.79) | 1.62 (1.53, 1.71) |
| Mortality | 1.59 (1.32, 1.90) | 1.51 (1.29, 1.77) | 1.47 (1.33, 1.63) | 1.52 (1.24, 1.87) | 1.40 (1.33, 1.47) | 1.39 (1.32, 1.46) |
| CVD mortality | 2.04 (1.79, 2.32) | 1.90 (1.79, 2.02) | 1.74 (1.64, 1.85) | 1.79 (1.52, 2.10) | 1.71 (1.45, 2.01) | 1.68 (1.42, 1.98) |
| Accelerated aging assessed by Klemera-Doubal method age | ||||||
| Kailuan cohort | ||||||
| CVD | 1.88 (1.81, 1.95) | 1.87 (1.80, 1.94) | 1.76 (1.69, 1.83) | 1.81 (1.72, 1.90) | 1.78 (1.69, 1.87) | 1.68 (1.60, 1.77) |
| MI | 1.99 (1.83, 2.17) | 1.97 (1.81, 2.14) | 1.86 (1.70, 2.03) | 1.89 (1.69, 2.13) | 1.84 (1.64, 2.07) | 1.76 (1.56, 1.98) |
| Stroke | 1.86 (1.78, 1.94) | 1.85 (1.77, 1.93) | 1.74 (1.66, 1.82) | 1.78 (1.68, 1.88) | 1.75 (1.66, 1.86) | 1.66 (1.56, 1.76) |
| Mortality | 1.48 (1.43, 1.53) | 1.55 (1.50, 1.60) | 1.55 (1.50, 1.60) | 1.39 (1.33, 1.46) | 1.43 (1.37, 1.50) | 1.43 (1.36, 1.50) |
| CVD mortality | 2.03 (1.82, 2.27) | 2.14 (1.92, 2.39) | 2.10 (1.88, 2.36) | 1.87 (1.55, 2.27) | 1.91 (1.58, 2.31) | 1.87 (1.53, 2.28) |
| Dongfeng-Tongji cohort | ||||||
| CVD | 1.54 (1.46, 1.63) | 1.53 (1.45, 1.62) | 1.42 (1.34, 1.50) | 1.51 (1.39, 1.64) | 1.51 (1.39, 1.64) | 1.39 (1.27, 1.51) |
| CHD | 1.45 (1.36, 1.55) | 1.44 (1.35, 1.53) | 1.31 (1.23, 1.40) | 1.41 (1.29, 1.55) | 1.40 (1.27, 1.53) | 1.28 (1.16, 1.41) |
| Stroke | 1.72 (1.52, 1.94) | 1.73 (1.53, 1.96) | 1.67 (1.47, 1.90) | 1.89 (1.54, 2.31) | 1.91 (1.56, 2.34) | 1.83 (1.49, 2.26) |
| Mortality | 1.33 (1.20, 1.47) | 1.35 (1.22, 1.49) | 1.35 (1.21, 1.50) | 1.29 (1.08, 1.55) | 1.32 (1.10, 1.58) | 1.30 (1.08, 1.57) |
| CVD mortality | 1.87 (1.57, 2.24) | 1.91 (1.60, 2.29) | 1.90 (1.58, 2.28) | 1.49 (1.07, 2.08) | 1.52 (1.09, 2.12) | 1.53 (1.08, 2.16) |
| UK Biobank | ||||||
| CVD | 1.40 (1.37, 1.43) | 1.60 (1.57, 1.64) | 1.46 (1.42, 1.49) | 1.20 (1.00, 1.43) | 1.52 (1.25, 1.83) | 1.45 (1.20, 1.76) |
| CHD | 1.40 (1.37, 1.44) | 1.62 (1.58, 1.66) | 1.46 (1.42, 1.50) | 1.10 (0.90, 1.36) | 1.41 (1.14, 1.74) | 1.33 (1.08, 1.65) |
| Stroke | 1.43 (1.37, 1.50) | 1.57 (1.49, 1.65) | 1.48 (1.41, 1.56) | 1.47 (1.03, 2.10) | 1.76 (1.21, 2.56) | 1.75 (1.20, 2.56) |
| Mortality | 1.27 (1.24, 1.30) | 1.40 (1.37, 1.44) | 1.30 (1.26, 1.33) | 1.16 (0.93, 1.44) | 1.37 (1.09, 1.73) | 1.30 (1.03, 1.65) |
| CVD mortality | 1.66 (1.53, 1.79) | 2.02 (1.87, 2.19) | 1.82 (1.68, 1.97) | 1.25 (0.67, 2.30) | 1.52 (0.80, 2.88) | 1.37 (0.72, 2.64) |
| Pooled results e | ||||||
| CVD | 1.59 (1.34, 1.89) | 1.66 (1.48, 1.87) | 1.54 (1.35, 1.76) | 1.50 (1.20, 1.89) | 1.62 (1.44, 1.83) | 1.52 (1.33, 1.73) |
| CHD/MI | 1.59 (1.28, 1.97) | 1.66 (1.39, 1.98) | 1.52 (1.25, 1.86) | 1.44 (1.07, 1.95) | 1.55 (1.29, 1.86) | 1.45 (1.18, 1.78) |
| Stroke | 1.66 (1.41, 1.95) | 1.71 (1.54, 1.90) | 1.62 (1.46, 1.80) | 1.78 (1.69, 1.88) | 1.76 (1.67, 1.86) | 1.67 (1.58, 1.77) |
| Mortality | 1.36 (1.23, 1.50) | 1.44 (1.33, 1.56) | 1.40 (1.25, 1.57) | 1.37 (1.31, 1.44) | 1.42 (1.36, 1.48) | 1.42 (1.35, 1.48) |
| CVD mortality | 1.83 (1.61, 2.09) | 2.04 (1.92, 2.17) | 1.91 (1.79, 2.03) | 1.73 (1.47, 2.03) | 1.79 (1.52, 2.10) | 1.75 (1.48, 2.07) |
a The reference group consisted of participants without accelerated aging
b Models were not adjusted for covariates
c Models were adjusted for age, sex, and self-reported race (UK Biobank only)
d Models were further adjusted for education levels, Townsend deprivation index (UK Biobank only), smoking status, drinking status, diet, physical activity levels, body mass index, sleep duration, and number of comorbidities
e Results from each cohort were combined using a fixed-effects model when the Cochran's Q test for heterogeneity was not significant, and a random-effects model was used when the Cochran's Q test for heterogeneity was significant
CHD coronary heart disease, CI confidence interval, CVD cardiovascular disease, HR hazard ratio, MI myocardial infarction
In the Kailuan cohort and the DFTJ cohort, both PhenoAge and KDM age significantly outperformed chronological age in predicting incident CVD and mortality, achieving a maximum C-index of 0.814. In the UK Biobank, PhenoAge significantly improved predictions of incident CVD and mortality compared to chronological age, yielding a maximum C-index of 0.777 (Additional file 1: Tables S7–8). After accounting for the competing risk between incident CVD and non-CVD mortality, the predictive performance of biological ages for incident CVD remained largely unchanged, although the improvement in CVD prediction provided by biological ages over chronological age did not reach statistical significance in the UK Biobank (Additional file 1: Tables S9).
Association of changes in accelerated aging with CVD and mortality
When defining accelerated aging by PhenoAge, compared with participants who remained persistent accelerated aging, those who recovered from accelerated aging (pooled HR for CVD: 0.76, 95% CI: 0.72, 0.81; CHD/MI: 0.84, 95% CI: 0.76, 0.93; stroke: 0.73, 95% CI: 0.67, 0.79; all-cause mortality: 0.84, 95% CI: 0.78, 0.89; CVD mortality: 0.80, 95% CI: 0.64, 1.00), delayed accelerated aging (pooled HR for CVD: 0.75, 95% CI: 0.70, 0.79; CHD/MI: 0.80, 95% CI: 0.73, 0.89; stroke: 0.72, 95% CI: 0.67, 0.78; all-cause mortality: 0.77, 95% CI: 0.72, 0.83; CVD mortality: 0.71, 95% CI: 0.57, 0.89), and maintained non-accelerated aging (pooled HR for CVD: 0.59, 95% CI: 0.48, 0.71; CHD/MI: 0.66, 95% CI: 0.60, 0.72; Stroke: 0.47, 95% CI: 0.44, 0.50; all-cause mortality: 0.58, 95% CI: 0.55, 0.62; CVD mortality: 0.49, 95% CI: 0.29, 0.85) exhibited significantly lower risks of CVD, CHD/MI, stroke, all-cause and CVD mortality (Fig. 1; Additional file 1: Table S10). Comparable associations were observed when defining accelerated aging by KDM age (Fig. 2; Additional file 1: Table S10).
Fig. 1.
Associations of changes in accelerated aging assessed by PhenoAge with cardiovascular disease and mortality. Model 1 included no covariates. Model 2 was adjusted for age, sex, and self-reported race (UK Biobank only). Model 3 was further adjusted for education levels, Townsend deprivation index (UK Biobank only), smoking status, drinking status, diet, physical activity levels, body mass index, sleep duration, and number of comorbidities. CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; DFTJ, Dongfeng-Tongji; HR, hazard ratio; MI, myocardial infarction
Fig. 2.
Associations of changes in accelerated aging assessed by KDM age with cardiovascular disease and mortality. Model 1 included no covariates. Model 2 was adjusted for age, sex, and self-reported race (UK Biobank only). Model 3 was further adjusted for education levels, Townsend deprivation index (UK Biobank only), smoking status, drinking status, diet, physical activity levels, body mass index, sleep duration, and number of comorbidities. CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; DFTJ, Dongfeng-Tongji; HR, hazard ratio; KDM, Klemera-Doubal method; MI, myocardial infarction
Subgroup and sensitivity analyses
In the Kailuan cohort, individuals chronologically aged < 60 years at baseline exhibited stronger protective associations of recovery from accelerated aging, delayed accelerated aging, and stable non-accelerated aging with CVD, MI, stroke, and all-cause mortality, compared to those chronologically aged ≥ 60 years at baseline (all P-interaction values ≤ 0.031). In the UK Biobank, when defined by KDM age, the protective associations of recovery from accelerated aging, delayed accelerated aging, and stable non-accelerated aging with CVD, CHD, and stroke were stronger in males than females (all P-interaction values ≤ 0.033) (Additional file 1: Tables S11–12).
Sensitivity analyses confirmed the robustness of observed associations, with E-values ranging from 1.46 to 5.91 (Additional file 1: Tables S13–20).
Discussion
In this study of three prospective cohorts, we found that accelerated aging was a significant risk factor for CVD and mortality. Importantly, compared to participants with persistent accelerated aging, participants who recovered from accelerated aging, delayed accelerated aging, or maintained non-accelerated aging had reduced risks of CVD and mortality.
PhenoAge and KDM age, both derived from clinical indicators, may be more cost-effective for applications in large-scale populations compared to aging clocks based on omics or imaging data [32–36]. In this study, the PhenoAge and KDM age used in the Chinese populations exhibited robust predictive performance for CVD and mortality. Conversely, although the KDM age calculated in the UK Biobank have been widely used in western populations [11, 13, 14, 37], it did not outperform chronological age in predicting CVD and mortality risks in our study. This finding may partly reflect a suboptimal fit of KDM age within the UK Biobank, particularly among the smaller subset of participants with repeated measures data. Moreover, since KDM age is calibrated to mirror chronological age, in theory it may capture less aging-related information independent of chronological age than other measures like PhenoAge, which quantifies survival differences among individuals of the same chronological age. Theoretically, its predictive performance for diseases and mortality may be limited, particularly in smaller samples [38].
Previous studies have found that accelerated aging, as defined by clinical indicators, was associated with increased risks of CVD and mortality [8, 9, 11], consistent with our findings. In the China Kadoorie Biobank study, Lu Chen et al. defined accelerated aging based on clinical indicators and found it was associated with long-term risks of mortality [8]. Both their study and ours used blood glucose, waist circumference, systolic blood pressure, and total cholesterol to calculate biological age. In the Coronary Artery Risk Development in Young Adults cohort study, KDM age acceleration was associated with incident CVD [9]. The clinical indicators used to calculate KDM age in that study partially overlapped with the indicators used to calculate KDM age in the UK Biobank in our study; for example, both included total cholesterol, C-reactive protein, and forced expiratory volume. Additionally, a multistate analysis using UK Biobank data demonstrated that individuals with accelerated aging, defined by clinical indicators consistent with those used in the UK Biobank in our study, were more likely to progress from cardiometabolic disease to multimorbidity and ultimately to death [11]. Previous studies and ours commonly utilized clinical indicators reflecting glucolipid metabolism, inflammation, vascular resistance, and cardiorespiratory function to calculate biological ages and assess accelerated aging, suggesting these domains as key drivers of accelerated aging [4, 5, 39, 40]. Moreover, these clinical indicators are readily accessible and amenable to interventions in clinical settings, underscoring their significant potential for clinical applications.
While prior research has focused on a single assessment of accelerated aging, our study expanded this perspective by evaluating changes in accelerated aging over time. Across three prospective cohorts, we consistently found that participants who recovered from or delayed accelerated aging, or maintained non-accelerated aging had reduced risks of both CVD and mortality compared to those with persistent accelerated aging. In the UK Biobank, associations between changes in accelerated aging and CVD/mortality were less significant than those in the other two cohorts, likely due to the smaller sample size of participants with repeated measures data. Recovery from accelerated aging exhibited slightly stronger protective associations with incident CVD and all-cause mortality than delayed accelerated aging, opposite to patterns observed in the other two cohorts, possibly reflecting differences in cohort demographics and biological age calculations. In the DFTJ cohort, when defining accelerated aging by PhenoAge, recovery from accelerated aging showed a slightly stronger protective association with incident stroke than delayed accelerated aging, whereas the opposite pattern was observed when defining accelerated aging by KDM age. This finding warrants further validation in other studies.
Evidence indicated that interventions such as a healthy lifestyle [16], caloric restriction [17], and high-intensity interval training [18] had certain effects on delaying or reversing accelerated aging. To prevent CVD and premature deaths, individuals with non-accelerated aging could modify risk factors to prevent or delay accelerated aging, coupled with regular monitoring for early identification and timely intervention of high-risk individuals. Additionally, individuals with accelerated aging should be prioritized for effective, safe interventions aimed at reversing accelerated aging. In addition, significant interactions between accelerated aging and age or sex highlight the need for tailored strategies in different demographic subgroups to prevent, delay, or reverse accelerated aging.
In this study, the robustness of the findings is enhanced by the prospective study design, large sample sizes, and extended follow-up periods. However, several limitations should be noted. First, the clinical indicators used to calculate PhenoAge and KDM age were not exactly the same across multiple cohorts, which may limit the generalizability of the biological ages. Nevertheless, across multiple cohorts, PhenoAge and KDM age estimated using different sets of clinical indicators resulted in largely consistent associations of changes in accelerated aging with the risks of CVD and mortality, suggesting the robustness of our findings. Second, due to data unavailability, changes in accelerated aging were assessed by only two surveys, restricting longitudinal tracking across multiple follow-ups. Third, the populations participating in the first follow-up were subsets of the baseline cohorts, which may potentially introduce selection bias. Analyses showed that participants lost at the first follow-up were generally older, more likely to smoke, had lower educational levels, and more comorbidities at baseline. Given these disparities, assuming higher CVD and mortality risks among those lost to first follow-up, we would infer this selection bias to lead to an underestimation of estimates. We addressed potential selection bias by constructing inverse probability weighted regression models in sensitivity analyses, yielding results consistent with the main findings. Fourth, although multiple confounders were adjusted, residual and unmeasured confounding may remain. Nevertheless, the relatively high E-values indicated that the unmeasured confounding was unlikely to alter the observed associations. Finally, comorbidity histories were self-reported; thus, the possibility of misclassification cannot be excluded.
Conclusions
Our study highlights accelerated aging as a significant risk factor for CVD and mortality. Incorporating PhenoAge and KDM age into routine health screenings may be helpful to identify individuals at higher risks for CVD and premature death. Participants who recovered from accelerated aging, delayed accelerated aging, or maintained non-accelerated aging had reduced risks of CVD and mortality, compared to those with persistent accelerated aging. These findings support interventions targeting the prevention, delay, or reversal of accelerated aging to prevent cardiovascular events and premature death.
Supplementary Information
Additional file 1:
Text S1. Populations, outcome ascertainment, and calculation of PhenoAge and KDM age.
Table S1. Biomarkers used to develop biological age in the Kailuan cohort.
Table S2. Definitions of covariates.
Table S3. Characteristics of the participants at the first follow-up.
Table S4. Baseline characteristics of participants lost at the first follow-up and those not lost at the first follow-up.
Table S5. Spearman correlation coefficients and mean absolute errors between biological age and chronological age.
Table S6. Number (percentage) of participants with accelerated aging at baseline and the first follow-up.
Table S7. Predictive performance of PhenoAge and KDM age for mortality.
Table S8. Predictive performance of PhenoAge and KDM age for cardiovascular events.
Table S9. Predictive performance of PhenoAge and KDM age for cardiovascular events when employing competing risk models to account for the competing risk between incident cardiovascular disease and non-cardiovascular deaths.
Table S10. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality.
Table S11. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality by baseline chronological age groups.
Table S12. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality by sex.
Table S13. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality after excluding cases of cardiovascular disease or death within the first year of follow-up.
Table S14. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when using age gap between chronological age and biological age to define accelerated aging.
Table S15. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when adjusting for individual comorbidities.
Table S16. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when adjusting for baseline covariates rather than those collected at the first follow-up.
Table S17. Associations of changes in accelerated aging with the risks of cardiovascular disease when employing competing risk models to account for the competing risk between incident cardiovascular disease and non-cardiovascular deaths.
Table S18. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when constructing the inverse probability weighted regression models.
Table S19. Associations of accelerated aging at the first follow-up with the risks of cardiovascular disease and mortality when constructing the inverse probability weighted regression models.
Table S20. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality, and E-values for these associations.
Figure S1. Flow chart of the study.
Acknowledgements
The authors thank the investigators and participants involved in the Kailuan cohort, the Dongfeng-Tongji cohort, and the UK Biobank for their contributions.
Abbreviations
- CHD
Coronary heart disease
- CI
Confidence interval
- CVD
Cardiovascular disease
- DFTJ
Dongfeng-Tongji
- HR
Hazard ratio
- KDM
Klemera-Doubal method
- MI
Myocardial infarction
Authors’ contribution
AP, S-LW, Y-FL, J-JZ, H-CY: study conception and design; J-JZ, H-CY: data analyses and drafting of the manuscript; J-JZ, H-CY, T-TG, S-HC, Y-XW, GL, HG, X-MZ, M-AH, J-LG: interpretation of the results; AP, S-LW, Y-FL are the guarantors of this work and have full access to all the data and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.
Funding
AP was supported by the National Key Research and Development Program of China (2023YFC3606300 and 2022YFC3600600) and the National Natural Science Foundation of China (82325043 and 82021005). The sponsors or funders play no roles in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Data availability
The UK Biobank is an open-access database. The datasets analyzed in the current study will be made available for researchers who apply to use the UK Biobank data by registering and applying at http://www.ukbiobank.ac.uk/register-apply. This study has been conducted using the UK Biobank resource under application number 109546.
Declarations
Ethics approval and consent to participate
The Kailuan study was approved by the Medical Ethics Committee of Kailuan General Hospital (no. 2006–5). The DFTJ cohort was approved by the Medical Ethics Committee of the School of Public Health, Tongji Medical College, Huazhong University of Science and Technology (no. 2008–3). The UK Biobank study was approved by the North West Multi-centre Research Ethics Committee (REC reference numbers 11/NW/0382, 16/NW/0274, and 21/NW/0157). All participants provided written informed consent.
Consent for publication
All authors agreed and approved the publication of the current manuscript.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Ji-Juan Zhang and Han-Cheng Yu contributed equally as the first authors.
Yun-Fei Liao, Shou-Ling Wu, and An Pan contributed equally as the corresponding authors.
Contributor Information
Yun-Fei Liao, Email: yunfeiliao2012@163.com.
Shou-Ling Wu, Email: drwusl@163.com.
An Pan, Email: panan@hust.edu.cn.
References
- 1.United Nations, Department of Economic and Social Affairs, Population Division. World Population Prospects 2024. Summary of Results. 2024. Accessed 2024 https://population.un.org/wpp/assets/Files/WPP2024_Summary-of-Results.pdf.
- 2.United Nations, Department of Economic and Social Affairs, Population Division. World Population Prospects 2022: Summary of Results. 2022. Accessed 2022 https://reliefweb.int/report/world/world-population-prospects-2022-summary-results.
- 3.Global incidence, prevalence, years lived with disability (YLDs), disability-adjusted life-years (DALYs), and healthy life expectancy (HALE) for 371 diseases and injuries in 204 countries and territories and 811 subnational locations, 1990–2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2024;403:2133–61. [DOI] [PMC free article] [PubMed]
- 4.López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. The hallmarks of aging. Cell. 2013;153:1194–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. Hallmarks of aging: an expanding universe. Cell. 2023;186:243–78. [DOI] [PubMed] [Google Scholar]
- 6.Félix J, Martínez de Toda I, Díaz-Del Cerro E, González-Sánchez M, De la Fuente M. Frailty and biological age. Which best describes our aging and longevity? Mol Aspects Med. 2024;98:101291. [DOI] [PubMed] [Google Scholar]
- 7.Zhao D, Wang Y, Wong ND, Wang J. Impact of Aging on Cardiovascular Diseases: From Chronological Observation to Biological Insights: JACC Family Series. JACC Asia. 2024;4:345–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chen L, Zhang Y, Yu C, Guo Y, Sun D, Pang Y, et al. Modeling biological age using blood biomarkers and physical measurements in Chinese adults. EBioMedicine. 2023;89:104458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Forrester SN, Baek J, Hou L, Roger V, Kiefe CI. A comparison of 5 measures of accelerated biological aging and their association with incident cardiovascular disease: the CARDIA study. J Am Heart Assoc. 2024;13:e032847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mao R, Wang F, Zhong Y, Meng X, Zhang T, Li J. Association of biological age acceleration with cardiac morphology, function, and incident heart failure: insights from UK Biobank participants. Eur Heart J Cardiovasc Imaging. 2024;25:1315–23. [DOI] [PubMed] [Google Scholar]
- 11.Jiang M, Tian S, Liu S, Wang Y, Guo X, Huang T, et al. Accelerated biological aging elevates the risk of cardiometabolic multimorbidity and mortality. Nat Cardiovasc Res. 2024;3:332–42. [DOI] [PubMed] [Google Scholar]
- 12.Levine ME, Lu AT, Quach A, Chen BH, Assimes TL, Bandinelli S, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY). 2018;10:573–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chen L, Wu B, Mo L, Chen H, Zhao Y, Tan T, et al. Associations between biological ageing and the risk of, genetic susceptibility to, and life expectancy associated with rheumatoid arthritis: a secondary analysis of two observational studies. Lancet Healthy Longev. 2024;5:e45–55. [DOI] [PubMed] [Google Scholar]
- 14.Yang Z, Shen Y, Zhang T, Tang X, Mao R. Associations of biological age accelerations and genetic risk with incident endometrial cancer: a prospective analysis in UK Biobank. Int J Surg. 2025;111:512-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang T, Duan W, Jia X, Huang X, Liu Y, Meng F, et al. Associations of combined phenotypic ageing and genetic risk with incidence of chronic respiratory diseases in the UK biobank: a prospective cohort study. Eur Respir J. 2024;63:2301720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fiorito G, Caini S, Palli D, Bendinelli B, Saieva C, Ermini I, et al. DNA methylation-based biomarkers of aging were slowed down in a two-year diet and physical activity intervention trial: the DAMA study. Aging Cell. 2021;20:e13439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Belsky DW, Huffman KM, Pieper CF, Shalev I, Kraus WE. Change in the rate of biological aging in response to caloric restriction: CALERIE biobank analysis. J Gerontol A Biol Sci Med Sci. 2017;73:4–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lohman T, Bains G, Cole S, Gharibvand L, Berk L, Lohman E. High-intensity interval training reduces transcriptomic age: a randomized controlled trial. Aging Cell. 2023;22:e13841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Poganik JR, Zhang B, Baht GS, Tyshkovskiy A, Deik A, Kerepesi C, et al. Biological age is increased by stress and restored upon recovery. Cell Metab. 2023;35:807-20.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wang F, Zhu J, Yao P, Li X, He M, Liu Y, et al. Cohort profile: the Dongfeng-Tongji cohort study of retired workers. Int J Epidemiol. 2013;42:731–40. [DOI] [PubMed] [Google Scholar]
- 21.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kwon D, Belsky DW. A toolkit for quantification of biological age from blood chemistry and organ function test data: BioAge. Geroscience. 2021;43:2795–808. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Klemera P, Doubal S. A new approach to the concept and computation of biological age. Mech Ageing Dev. 2006;127:240–8. [DOI] [PubMed] [Google Scholar]
- 24.Wang C, Hong S, Guan X, Xiao Y, Fu M, Meng H, et al. Associations between multiple metals exposure and biological aging: evidence from the Dongfeng-Tongji cohort. Sci Total Environ. 2023;861:160596. [DOI] [PubMed] [Google Scholar]
- 25.Liu Z. Development and validation of 2 composite aging measures using routine clinical biomarkers in the Chinese population: analyses from 2 prospective cohort studies. J Gerontol A Biol Sci Med Sci. 2021;76:1627–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Yuan Y, Liu K, Zheng M, Chen S, Wang H, Jiang Q, et al. Analysis of changes in weight, waist circumference, or both, and all-cause mortality in Chinese adults. JAMA Netw Open. 2022;5:e2225876. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Said MA, Verweij N, van der Harst P. Associations of combined genetic and lifestyle risks with incident cardiovascular disease and diabetes in the UK biobank study. JAMA Cardiol. 2018;3:693–702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gilthorpe MS. The importance of normalisation in the construction of deprivation indices. J Epidemiol Community Health. 1995;49(Suppl 2):S45-50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Goldstein ND, Kahal D, Testa K, Burstyn I. Inverse probability weighting for selection bias in a Delaware community health center electronic medical record study of community deprivation and hepatitis C prevalence. Ann Epidemiol. 2021;60:1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the e-value. Ann Intern Med. 2017;167:268–74. [DOI] [PubMed] [Google Scholar]
- 31.Zhang JJ, Yu HC, Geng TT, Zhang JJ, Zhou XT, Wang YX, et al. Serum 25-hydroxyvitamin D concentrations, vitamin D receptor polymorphisms, and risk of infections among individuals with type 2 diabetes: a prospective cohort study. Am J Clin Nutr. 2024;120:398–406. [DOI] [PubMed] [Google Scholar]
- 32.Argentieri MA, Xiao S, Bennett D, Winchester L, Nevado-Holgado AJ, Ghose U, et al. Proteomic aging clock predicts mortality and risk of common age-related diseases in diverse populations. Nat Med. 2024;30:2450–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Zheng Z, Li J, Liu T, Fan Y, Zhai QC, Xiong M, et al. DNA methylation clocks for estimating biological age in Chinese cohorts. Protein Cell. 2024;15:575–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lau CE, Manou M, Markozannes G, Ala-Korpela M, Ben-Shlomo Y, Chaturvedi N, et al. NMR metabolomic modeling of age and lifespan: a multicohort analysis. Aging Cell. 2024;23:e14164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Gopu V, Camacho FR, Toma R, Torres PJ, Cai Y, Krishnan S, et al. An accurate aging clock developed from large-scale gut microbiome and human gene expression data. iScience. 2024;27:108538. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Dove A, Wang J, Huang H, Dunk MM, Sakakibara S, Guitart-Masip M, et al. Diabetes, prediabetes, and brain aging: the role of healthy lifestyle. Diabetes Care. 2024;47:1794–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Levine ME. Modeling the rate of senescence: can estimated biological age predict mortality more accurately than chronological age? J Gerontol A Biol Sci Med Sci. 2013;68:667–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Galkin F, Mamoshina P, Aliper A, de Magalhães JP, Gladyshev VN, Zhavoronkov A. Biohorology and biomarkers of aging: current state-of-the-art, challenges and opportunities. Ageing Res Rev. 2020;60:101050. [DOI] [PubMed] [Google Scholar]
- 39.Sen I, Trzaskalski NA, Hsiao YT, Liu PP, Shimizu I, Derumeaux GA. Aging at the crossroads of organ interactions: implications for the heart. Circ Res. 2025;136:1286–305. [DOI] [PubMed] [Google Scholar]
- 40.Mutlu AS, Duffy J, Wang MC. Lipid metabolism and lipid signals in aging and longevity. Dev Cell. 2021;56:1394–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Additional file 1:
Text S1. Populations, outcome ascertainment, and calculation of PhenoAge and KDM age.
Table S1. Biomarkers used to develop biological age in the Kailuan cohort.
Table S2. Definitions of covariates.
Table S3. Characteristics of the participants at the first follow-up.
Table S4. Baseline characteristics of participants lost at the first follow-up and those not lost at the first follow-up.
Table S5. Spearman correlation coefficients and mean absolute errors between biological age and chronological age.
Table S6. Number (percentage) of participants with accelerated aging at baseline and the first follow-up.
Table S7. Predictive performance of PhenoAge and KDM age for mortality.
Table S8. Predictive performance of PhenoAge and KDM age for cardiovascular events.
Table S9. Predictive performance of PhenoAge and KDM age for cardiovascular events when employing competing risk models to account for the competing risk between incident cardiovascular disease and non-cardiovascular deaths.
Table S10. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality.
Table S11. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality by baseline chronological age groups.
Table S12. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality by sex.
Table S13. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality after excluding cases of cardiovascular disease or death within the first year of follow-up.
Table S14. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when using age gap between chronological age and biological age to define accelerated aging.
Table S15. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when adjusting for individual comorbidities.
Table S16. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when adjusting for baseline covariates rather than those collected at the first follow-up.
Table S17. Associations of changes in accelerated aging with the risks of cardiovascular disease when employing competing risk models to account for the competing risk between incident cardiovascular disease and non-cardiovascular deaths.
Table S18. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality when constructing the inverse probability weighted regression models.
Table S19. Associations of accelerated aging at the first follow-up with the risks of cardiovascular disease and mortality when constructing the inverse probability weighted regression models.
Table S20. Associations of changes in accelerated aging with the risks of cardiovascular disease and mortality, and E-values for these associations.
Figure S1. Flow chart of the study.
Data Availability Statement
The UK Biobank is an open-access database. The datasets analyzed in the current study will be made available for researchers who apply to use the UK Biobank data by registering and applying at http://www.ukbiobank.ac.uk/register-apply. This study has been conducted using the UK Biobank resource under application number 109546.