Causal associations of education, lifestyle behaviors, and cardiometabolic traits with epigenetic age acceleration: a Mendelian randomization study
===================================================================================================================================================

* Lijie Kong
* Chaojie Ye
* Yiying Wang
* Jie Zheng
* Shuangyuan Wang
* Hong Lin
* Zhiyun Zhao
* Mian Li
* Yu Xu
* Jieli Lu
* Yuhong Chen
* Min Xu
* Weiqing Wang
* Guang Ning
* Yufang Bi
* Tiange Wang

## Abstract

**Background** GrimAge acceleration (GrimAgeAccel) and PhenoAge acceleration (PhenoAgeAccel) are DNA methylation-based markers of accelerated biological ageing, standing out in predicting mortality and age-related cardiometabolic morbidities. Causal risk factors for GrimAgeAccel and PhenoAgeAccel are unclear.

**Objective** To evaluate causal associations of 18 common modifiable socioeconomic, lifestyle, and cardiometabolic factors with GrimAgeAccel and PhenoAgeAccel.

**Methods** We performed two-sample univariable and multivariable Mendelian randomization (MR), using summary-level data for GrimAgeAccel and PhenoAgeAccel derived from a genome-wide association study of 34,710 European participants. We used the inverse-variance weighted method as the main analysis, supplemented by three sensitivity analyses.

**Results** Eleven and eight factors were causally associated with GrimAgeAccel and PhenoAgeAccel, respectively. Smoking initiation was the strongest risk factor (β [SE]: 1.299 [0.107] years) for GrimAgeAccel, followed by higher alcohol intake, higher waist circumference, daytime napping, higher body fat percentage, higher BMI, higher C-reactive protein, higher triglycerides, childhood obesity, and type 2 diabetes; whereas education was the strongest protective factor (β [SE] per 1-SD increase in years of schooling: -1.143 [0.121] years). Higher waist circumference (β [SE]: 0.850 [0.269] years) and education (β [SE]: -0.718 [0.151] years) were the leading causal risk and protective factors for PhenoAgeAccel, respectively. Sensitivity analyses strengthened the robustness of these causal associations, and the multivariable MR analyses demonstrated independent direct effects of the strongest risk and protective factors on GrimAgeAccel and PhenoAgeAccel, respectively.

**Conclusion** Our findings provide novel quantitative evidence on modifiable causal risk factors for epigenetic ageing, and hint at underlying contributors and intervention targets to the ageing process.

**Key Points**

*   Genetically predicted higher educational attainment is the strongest protective factor for both GrimAgeAccel and PhenoAgeAccel, independent of causal lifestyle and cardiometabolic risk factors.

*   Smoking is the leading causal lifestyle risk factor for GrimAgeAccel or PhenoAgeAccel, followed by alcohol intake and daytime napping. Adiposity traits are the leading causal cardiometabolic risk factors for GrimAgeAccel and PhenoAgeAccel, followed by type 2 diabetes, triglycerides, and C-reactive protein.

*   This study provides novel quantitative evidence on modifiable causal risk factors for accelerated epigenetic ageing, indicating underlying contributors to the ageing process and promising intervention targets to promote healthy longevity.

Keywords
*   Epigenetic ageing
*   Education
*   Lifestyle behaviors
*   Cardiometabolic traits
*   Mendelian randomization

## Introduction

Ageing involves the gradual accumulation of a decline in multiple biological functions over time, leading to increased risks of developing age-related diseases and mortality [1,2]. Although chronological ageing is uniform and unchangeable, the rate of biological ageing is variable and modifiable depending on individual genetics, environmental exposures, and health-related behaviors [3]. Of several potential types of biological age predictors (e.g., epigenetic clock, leukocyte telomere length, and transcriptomic predictors), the epigenetic clock that composed of DNA methylation at multiple cytosine-phosphate-guanine (CpG) sites is currently the best one, as it correlates well with age and predicts mortality across populations [4]. The epigenetic age acceleration is the difference between chronological age and epigenetic age, and represents accelerated biological ageing [5]. The second-generation epigenetic age acceleration indicators, namely GrimAge acceleration (GrimAgeAccel) and PhenoAge acceleration (PhenoAgeAccel), have been evolved to incorporate ageing-related traits, and stand out in terms of predicting mortality and age-related cardiometabolic morbidities [6-9].

Limited observational evidence has suggested that certain socioeconomic, lifestyle behaviors, and cardiometabolic traits may be related to GrimAgeAccel and PhenoAgeAccel [6,7,10,11]. Whether and to what extent modifiable risk factors influence GrimAgeAccel and PhenoAgeAccel, if causally established, could shed light on potential contributors to the ageing process and elucidate promising targets for preventing age-related diseases and improving healthy longevity [12,13]. Thus far, such causal evidence is scarce.

To fill this knowledge gap, we applied Mendelian randomization (MR) to evaluate the causal associations of 18 common modifiable socioeconomic, lifestyle and cardiometabolic factors with GrimAgeAccel and PhenoAgeAccel. The MR method uses genetic variants as instrumental variables (IVs) to infer causality among correlated traits. Since genetic variants are randomly allocated at conception, the MR study is less susceptible to confounding and reverse causality than conventional observational studies [14]. We conducted both univariable and multivariable MR analyses to discern if modifiable factors have independent direct causal effects on GrimAgeAccel and PhenoAgeAccel.

## Methods

### Study design

This MR study design is presented in **Figure 1**. We performed two-sample univariable and multivariable MR strictly following the STROBE-MR guidelines (**Supplementary Table 1**) [15]. To obtain unbiased estimates of the causal effects, the MR analysis should adhere to three fundamental assumptions [16]: first, the IVs are truly associated with the exposures; second, the IVs are independent of confounders of the relationship between exposures and outcomes (GrimAgeAccel and PhenoAgeAccel); third, the IVs influence the outcomes only through the exposures, but not any direct or indirect pathways. All data used in this MR study are publicly available. Ethical approval and informed consent had been obtained in all original studies.

![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/07/14/2022.07.12.22277528/F1.medium.gif)

[Figure 1.](http://medrxiv.org/content/early/2022/07/14/2022.07.12.22277528/F1)

Figure 1. Study design and assumptions of the MR analysis
Assumption 1 indicates that the genetic variants proposed as instrumental variables should be robustly associated with the exposures; assumption 2 indicates that the used instrumental variables should not be associated with potential confounders of the relationship between exposures and outcomes; and assumption 3 indicates that the selected instrumental variables should influence the outcomes only through the exposures, not via alternative pathways.

Abbreviations: GrimAgeAccel=epigenetic-age acceleration obtained using the GrimAge clock; HDL=high-density lipoprotein; LDL=low-density lipoprotein; MR=mendelian randomization; PhenoAgeAccel=epigenetic-age acceleration obtained using the PhenoAge clock; PRESSO=pleiotropy residual sum and outlier.

### Selection rationale and data sources of genetic instruments

We selected 18 common modifiable factors, including educational attainment, lifestyle behaviors (smoking initiation, alcohol intake, coffee consumption, daytime napping, sleep duration, and moderate-to vigorous physical activity [MVPA]), and cardiometabolic traits (body mass index [BMI], waist circumference, body fat percentage [BF%], childhood obesity, type 2 diabetes, low-density lipoprotein [LDL] cholesterol, high-density lipoprotein [HDL] cholesterol, triglycerides, systolic blood pressure [SBP], diastolic blood pressure [DBP], and C-reactive protein [CRP]). Definitions of the 18 modifiable factors are shown in **Supplementary Table 2**.

We extracted genetic variants for each of the modifiable factors from the largest available genome-wide association studies (GWASs) of European ancestry [17-29], ensuring minimum sample overlap with the genetic variants for GrimAgeAccel and PhenoAgeAccel (**Supplementary Table 3**). We included single nucleotide polymorphisms (SNPs) robustly associated with the 18 modifiable factors at the genome-wide significance (P <5×10-8). To select independent genetic variants, a stringent condition (linkage disequilibrium threshold of r2 <0.01) was set to minimize the influence of linkage disequilibrium which may bias the results of randomized allele allocation. Where SNPs for the exposures were not available in the GWAS summary statistics of GrimAgeAccel or PhenoAgeAccel, we used proxies of SNPs with r2 >0.8 as substitutes, by using the LDproxy search on the online platform LDlink ([https://ldlink.nci.nih.gov/](https://ldlink.nci.nih.gov/)) [30].

### Data source for epigenetic age acceleration

The genetic associations with GrimAgeAccel and PhenoAgeAccel were extracted from a recent GWAS meta-analysis (summary statistics available at [https://datashare.is.ed.ac.uk/handle/10283/3645](https://datashare.is.ed.ac.uk/handle/10283/3645)), which included 34,710 European participants from 28 cohorts [31]. GrimAgeAccel and PhenoAgeAccel are two-generation epigenetic age acceleration indicators, expressing the biological ageing rate in years, of which GrimAgeAccel is more strongly associated with mortality than PhenoAgeAccel [11]. Detailed definitions of GrimAgeAccel and PhenoAgeAccel and data preparation in GWAS are shown in **Supplementary Table 2**.

### Statistical analyses

In the main analysis, we used the inverse-variance weighted (IVW) method to determine MR causal estimates (β coefficients with standard errors [SEs]) for associations of each modifiable factor with GrimAgeAccel and PhenoAgeAccel. The IVW combined the Wald ratio estimates of every single SNP in the set of IVs into one causal estimate using the random-effects meta-analysis approach [32]. To evaluate the robustness of the IVW estimates under different assumptions and to detect possible pleiotropy, we performed three sensitivity analyses, including the MR weighted median, the MR Egger, and the MR pleiotropy residual sum and outlier (MR-PRESSO) methods. The weighted median method selected the median MR estimate as the causal estimate, and provided a consistent causal estimate if over 50% of the weight in the analysis was derived from valid IVs [33]. The MR Egger method which allowed the intercept to be freely estimated as an indicator of pleiotropy was used to identify and adjust for the potential directional pleiotropic bias, but has limited precision [34]. The MR-PRESSO method was applied to detect and correct for any outlier SNP reflecting likely horizontal pleiotropic biases for MR causal estimates [35]. We evaluated the heterogeneity for the IVW estimates using the Cochran’s Q test [36], and identified the horizontal pleiotropy based on the p-value for the intercept in the MR-Egger model [34]. A false discovery rate (FDR) method was used to correct results for multiple testing, and FDR q-values were provided. In this study, strong causal evidence was defined as an association supported by the main analysis (FDR q-value <0.05) and at least one sensitivity analysis. Suggestive causal evidence was defined as a suggest association with P <0.05 and FDR q value ≥0.05 in the main analysis. Null causal evidence was defined as no statistically significant association revealed from the main analysis (P ≥0.05).

We further conducted multivariable MR (MVMR) analyses to assess whether the causal effects of the strongest protective factor and the strongest risk factor on GrimAgeAccel and PhenoAgeAccel were independent of other exposure factors [37]. Taking into account the effect size and significance of causal associations, we selected exposure factors with β coefficients >0.5 and P <0.05 in the main analysis as covariates in the adjustment models.

The two-sample MR analyses were conducted with the R packages ‘*TwoSampleMR’* and ‘*MRPRESSO*’ in R software (version 4.0.3). FDR q-values were estimated using the R package ‘*fdrtool*’.

## Results

### MR results for associations between 18 modifiable factors and GrimAgeAccel

Ten out of the 18 risk factors showed strong associations with increased GrimAgeAccel after FDR adjustment for multiple comparisons (**Table 1**). Smoking initiation was the strongest risk factor (β [SE]: 1.299 [0.107] years) for GrimAgeAccel, followed by higher alcohol intake (β [SE] per 1-SD increase: 0.899 [0.361] years), higher waist circumference (β [SE] per 1-SD increase: 0.815 [0.184] years), daytime napping (β [SE]: 0.805 [0.355] years), higher BF% (β [SE] per 1-SD increase: 0.748 [0.120] years), higher BMI (β [SE] per 1-SD increase: 0.592 [0.079] years), higher CRP (β [SE] per 1-SD increase: 0.345 [0.073] years), higher triglycerides (β [SE] per 1-SD increase: 0.249 [0.091] years), childhood obesity (β [SE]: 0.200 [0.075] years), and type 2 diabetes (β [SE]: 0.095 [0.041] years; **Figure 2a**). By contrast, educational attainment in years of schooling was the only one protective factor associated with decreased GrimAgeAccel (β [SE] per 1-SD increase: -1.143 [0.121] years). There was little evidence to support a causal association of other modifiable factors with GrimAgeAccel.

View this table:
[Table 1.](http://medrxiv.org/content/early/2022/07/14/2022.07.12.22277528/T1)

Table 1. MR results of the associations between 18 modifiable factors and GrimAgeAccel

![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/07/14/2022.07.12.22277528/F2.medium.gif)

[Figure 2.](http://medrxiv.org/content/early/2022/07/14/2022.07.12.22277528/F2)

Figure 2. Overview of the findings on associations of 18 modifiable factors with GrimAgeAccel and PhenoAgeAccel
a. Causal associations between 18 modifiable factors and GrimAgeAccel; b. Causal associations between 18 modifiable factors and PhenoAgeAccel.

GrimAgeAccel represents epigenetic-age acceleration obtained using the GrimAge clock; PhenoAgeAccel represents epigenetic-age acceleration obtained using the PhenoAge clock; β represents the effect of each modifiable factor on epigenetic-age acceleration. Red box indicates a strong association with p value <0.05 and FDR q value <0.05. Blue box indicates a suggestive association with p value <0.05 and FDR q value ≥0.05.

Grey box indicates null association with p value ≥0.05. The IVW method was used for the main analysis. Sensitivity analyses included the MR-PRESSO, the MR-WM, and the MR-Egger methods.

Abbreviations: BF%=body fat percentage; BMI=body mass index; CRP=C-reactive protein; DBP=diastolic blood pressure; HDL=high-density lipoprotein; IVW=inverse-variance weighted; LDL=low-density lipoprotein; MR=mendelian randomization; MVPA=moderate to vigorous physical activity; PRESSO=pleiotropy residual sum and outlier; SBP=systolic blood pressure; SD=standard deviation; WM=weighted median.

Associations of the above 11 modifiable factors with GrimAgeAccel were robust across sensitivity analyses with consistent effect directions and P <0.05 in at least one sensitivity analysis. The MR-Egger intercept tests indicated potential pleiotropy for CRP (Pintercept <0.05; **Supplementary Table 4**). Cochran’s Q-test showed possible heterogeneity for educational attainment, alcohol intake, BMI, BF%, type 2 diabetes, LDL cholesterol, HDL cholesterol, and CRP (Ph <0.05; **Supplementary Table 4**). However, with the exclusion of outlying SNPs, the MR-PRESSO analysis showed consistent results with the IVW analysis (**Table 1**).

### MR results for associations between 18 modifiable factors and PhenoAgeAccel

Eight modifiable factors showed strong associations with PhenoAgeAccel after FDR adjustment for multiple comparisons (**Table 2**). Higher waist circumference was the strongest risk factor (β [SE] per 1-SD increase: 0.850 [0.269] years) for PhenoAgeAccel, followed by higher BF% (β [SE] per 1-SD increase: 0.711 [0.152] years), higher BMI (β [SE] per 1-SD increase: 0.586 [0.102] years), smoking initiation (β [SE]: 0.519 [0.142] years), higher CRP (β [SE] per 1-SD increase: 0.349 [0.095] years), childhood obesity (β [SE]: 0.229 [0.095] years), and type 2 diabetes (β [SE]: 0.125 [0.051] years). Whereas the genetically predicted higher educational attainment in years of schooling was associated with decreased PhenoAgeAccel (β [SE] per 1-SD increase: -0.718 [0.151] years; **Figure 2b**). Suggestive causality was identified for the association between genetically predicted higher DBP and increased PhenoAgeAccel (β [SE] per 1-mmHg increase: 0.049 [0.024] years). No significant association was observed with PhenoAgeAccel for the other modifiable factors.

View this table:
[Table 2.](http://medrxiv.org/content/early/2022/07/14/2022.07.12.22277528/T2)

Table 2. MR results of the associations between 18 modifiable factors and PhenoAgeAccel

Associations for the above eight modifiable factors were robust in sensitivity analyses with consistent effect directions and P <0.05 in at least one sensitivity analysis. There was no evidence of pleiotropy for these risk factors except for CRP (Pintercept <0.05; **Supplementary Table 5**). Potential heterogeneity was observed for educational attainment, smoking initiation, BMI, waist circumference, BF%, type 2 diabetes, and CRP (Ph<0.05). One to four outliers were detected in the MR-PRESSO analyses; however, the associations remained consistent after removal of these outliers (**Table 2**). For SBP, with excluding the outlying SNP rs62523863, the MR-PRESSO analysis revealed a potentially positive association between genetically predicted higher SBP and PhenoAgeAccel (β [SE] per 1-mmHg increase: 0.038 [0.016] years).

### MVMR results for associations of the strongest protective and risk factors with GrimAgeAccel and PhenoAgeAccel

In MVMR analyses, the association between educational attainment (the strongest protective factor) and GrimAgeAccel remained significant with adjustment for smoking initiation, alcohol intake, waist circumference, daytime napping, BF%, or BMI (**Figure 3a**). Likewise, the significant association between smoking initiation (the strongest risk factor) and GrimAgeAccel persisted after adjustment for education, alcohol intake, waist circumference, daytime napping, BF%, or BMI. All covariates in MVMR analyses were selected based on their considerable effects on GrimAgeAccel in terms of size (β coefficients >0.5) and significance (P <0.05 in the main analysis).

![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2022/07/14/2022.07.12.22277528/F3.medium.gif)

[Figure 3.](http://medrxiv.org/content/early/2022/07/14/2022.07.12.22277528/F3)

Figure 3. Multivariable MR assessing the effects of the strongest protective factor and risk factor on GrimAgeAccel and PhenoAgeAccel
a. Effect of the strongest protective factor and risk factor on GrimAgeAccel; b. Effect of the strongest protective factor and risk factor on PhenoAgeAccel.

Causal estimates are β (95% CI) in years. Exposure factors with β coefficients >0.5 and P <0.05 in the main analysis as shown in Figure 2 were selected as covariates in the adjustment models.

Abbreviations: BF%=body fat percentage; BMI=body mass index; CI=confidence intervals; MR=mendelian randomization.

The associations of education (the strongest protective factor) and waist circumference (the strongest risk factor) with PhenoAgeAccel remained significant with adjustment for waist circumference, BF%, BMI, or smoking initiation, which had considerable causal effects on PhenoAgeAccel (β coefficients >0.5; **Figure 3b**).

## Discussion

This MR study for the first time delineated potential causal relationships of 18 common modifiable factors with GrimAgeAccel and PhenoAgeAccel, the robust second-generation epigenetic age acceleration indicators. We identified strong evidence for 11 and eight factors associated with GrimAgeAccel and PhenoAgeAccel, respectively. Smoking initiation exhibited the greatest effect on increased GrimAgeAccel (1.299 years), followed by higher alcohol intake, higher waist circumference, daytime napping, higher BF%, higher BMI, higher CRP, higher triglycerides, childhood obesity, and type 2 diabetes; whereas educational attainment showed the greatest effect on decreased GrimAgeAccel (−1.143 years per 1-SD increase in years of schooling). Higher waist circumference and educational attainment were the leading causal risk and protective factors associated with PhenoAgeAccel, respectively; BF%, BMI, smoking initiation, CRP, childhood obesity, and type 2 diabetes were also associated with increased PhenoAgeAccel. Multiple sensitivity analyses further strengthened the robustness of these causal relationships, and the MVMR analyses demonstrated the independent direct effects of the strongest risk and protective factors on GrimAgeAccel and PhenoAgeAccel, respectively. Suggestive causality was identified for the association between higher DBP and increased PhenoAgeAccel.

In this study, educational attainment was the major protective factor for both GrimAgeAccel and PhenoAgeAccel, and this causal effect was largely independent of other causal factors, such as smoking and adiposity-related traits. Educational attainment is a strong proxy for socioeconomic status and a more upstream determinant of health, with broad implications for a person’s life-long lifestyle behaviors and health-promoting resources [38]. A recent study using UK Biobank data has documented that each 1-year increase in genetically determined educational attainment was associated with equivalently 4.2 years of age-related increases in telomere length [39]. Telomere length and epigenetic age acceleration metrics (e.g., GrimAgeAccel and PhenoAgeAccel) point towards distinct mechanisms of the ageing process that are marked by telomeres and the DNA methylation-based epigenetic clocks, both of which are independently associated with chronological age and mortality risk [40]. Our findings, together with those of the UK Biobank, highlight the important impact of educational attainment on biological ageing rates from two different aspects of the ageing process. Therefore, public health strategies aimed at reducing educational inequalities and improving educational attainment may slow the biological ageing rate and help reduce age-related health burdens.

This MR study identified several common lifestyle behaviors, including smoking initiation, alcohol intake, and daytime napping, causally associated with increased GrimAgeAccel or PhenoAgeAccel (smoking initiation only), which were consistent with previous evidence from observational studies [6,7]. Smoking methylation proxy is a component of the GrimAge clock [7], thus it is not surprising that smoking initiation exhibited a large effect on increased GrimAgeAccel in this study, and the MVMR analysis further confirmed that the effect of smoking on GrimAgeAccel was independent of other causal lifestyle behaviors and adiposity-related traits. Similarly, genetically determined smoking has been associated with shorter telomere length in the UK Biobank [41], and evidence from the Danish Health Interview Survey suggested that the life expectancy of a heavy smoker was a little more than seven years shorter than that of a never smoker [42]. Our study also provided strong MR evidence that genetically predicted higher alcohol intake was associated with increased GrimAgeAccel. A study composed of the Hannum cohort and Family and Community Health Studies cohort found that the relationship between alcohol use and the first-generation epigenetic clocks seemed to be nonlinear [42]. However, given the dose-dependent relationship of alcohol intake with all-cause mortality and cancers [43], our findings suggest that reducing alcohol intake is necessary to decrease GrimAgeAccel and retard the overall health loss. Previous observational studies also reported positive or inverse correlations of other lifestyle behaviors such as MVPA, coffee, and sleep duration with GrimAgeAccel, PhenoAgeAccel, or mortality [11,44,45]. Nevertheless, in this study, there was little evidence supporting these associations were causal. The discrepancy between our findings and previous observations may result from the potential confounding or reverse causation in conventional observational studies. Moreover, the non-linear association patterns, as in the case of sleep duration and mortality, might also partially explain the inconsistent results [46]. Therefore, our null findings should be cautiously interpreted.

Interestingly, of all cardiometabolic traits (i.e., adiposity traits, type 2 diabetes, triglycerides, and CRP) which causally increased GrimAgeAccel or PhenoAgeAccel, adiposity traits were the most dominant traits. Emerging observational studies have pronounced the positive associations of BMI with GrimAgeAccel and PhenoAgeAccel [11], and a meta-analysis of 87 observational studies showed each 5-kg/m2 higher BMI corresponded to about 1 year of age-related decrease in telomere length [46]. Our study further provided strong evidence for causal associations of various adiposity phenotypes, including waist circumference, BF%, BMI, and childhood obesity, with increased GrimAgeAccel and PhenoAgeAccel. In this study, type 2 diabetes, triglycerides, and CRP showed strong but modest effects (β coefficients ≤0.5) on GrimAgeAccel or PhenoAgeAccel, which was consistent with the findings of the UK Biobank study on telomere length [39].

This MR study yields new insights into modifiable causal factors, including education, lifestyle behaviors, and cardiometabolic traits, associated with accelerating or decelerating epigenetic ageing. This study brings us one step closer to understanding the potential contributors to the ageing process and provides promising intervention targets for healthy ageing. Given the enormous burden induced by age-related morbidity and mortality, strategies to reduce educational inequalities, promote healthy lifestyles primarily through reducing smoking, alcohol intake, and daytime napping, and improve cardiometabolic traits, specifically adiposity, type 2 diabetes, triglycerides, and CRP, to slow biological ageing rate are imminent.

In this study, we included independent and genome-wide significant SNPs as instruments for each of the modifiable factors to ensure the first MR assumption was fulfilled. Moreover, we applied strict criteria strengthened by the FDR-corrected significance and the cross-validations by main and sensitivity analyses to draw robust causal conclusions. However, several limitations merited consideration. First, we found potential pleiotropy from the MR-Egger intercept test for CRP. However, we conducted MR-PRESSO analysis, and the association remained consistent after removal of outlying SNPs. Second, we could not rule out the possibility that the associations of certain modifiable risk factors such as alcohol intake and sleep duration with GrimAgeAccel or PhenoAgeAccel may be non-linear. Future studies with individual-level data are warranted to confirm the linear or non-linear relationships. Third, to ensure the consistency of genetic background, this MR study was performed only in European-ancestry participants, thus the generalization of our results to other ethnic groups should be cautious.

## Conclusions

This MR study provided novel quantitative evidence on modifiable causal socioeconomic, lifestyle, and cardiometabolic factors for accelerated epigenetic ageing. Our findings shed light on the underlying contributors to the biological ageing process and point toward promising intervention targets to slow the biological ageing rate and promote healthy longevity.

## Supporting information

Supplementary Table 1-5 [[supplements/277528_file03.pdf]](pending:yes)

## Data Availability

This study is based on publicly available data. The genetic association data of the 18 modifiable factors are available in Table 1. The GWAS summary statistics for GrimAgeAccel and PhenoAgeAccel are available at https://datashare.is.ed.ac.uk/handle/10283/3645.

[https://datashare.is.ed.ac.uk/handle/10283/3645](https://datashare.is.ed.ac.uk/handle/10283/3645). 

## Conflict-of-interest statement

The authors declare that they have no competing interests.

## Ethics committee approval

This study is based on publicly available summarized data. Ethical approval and informed consent had been obtained in all original studies.

## Availability of data and materials

The genetic association data of the 18 modifiable factors are available in Supplementary Table 3. The GWAS summary statistics for GrimAgeAccel and PhenoAgeAccel are available at [https://datashare.is.ed.ac.uk/handle/10283/3645](https://datashare.is.ed.ac.uk/handle/10283/3645). The analytical script of the MR analyses conducted in this study is available via the GitHub repository of the “TwoSampleMR” R package ([https://github.com/MRCIEU/TwoSampleMR/](https://github.com/MRCIEU/TwoSampleMR/)).

## Funding

This work was supported by the grants from the National Natural Science Foundation of China (82022011, 81970706, 81730023, 82088102, 81970728, 81941017), the Chinese Academy of Medical Sciences (2018PT32017, 2019PT330006), the “Shanghai Municipal Education Commission–Gaofeng Clinical Medicine Grant Support” from Shanghai Jiao Tong University School of Medicine (20171901 Round 2), the Shanghai Shenkang Hospital Development Center (SHDC12019101, SHDC2020CR1001A, SHDC2020CR3064B), the Shanghai Jiao Tong University School of Medicine (DLY201801), and the Ruijin Hospital (2018CR002).

## Authors’ contributions

LK and TW contributed to the conception and design of the study. LK and CY contributed to statistical analysis. LK contributed to drafting of the manuscript. TW guaranteed this work and take responsibility for the integrity of the data. All authors contributed to acquisition or interpretation of data, critical revision of the manuscript for important intellectual content, and final approval of the version to be published.

## Acknowledgements

We gratefully acknowledge the investigators and participants of all genome-wide association studies from which we used data.

## Footnotes

*   Figure 2 and Figure 3 revised;Supplemental files updated.

## List of abbreviations

BF%
:   Body fat percentage
BMI
:   Body mass index
CI
:   Confidence interval
CpG
:   Cytosine-phosphate-guanine
CRP
:   C-reactive protein
DBP
:   Diastolic blood pressure
FDR
:   False discovery rate
GrimAgeAccel
:   GrimAge acceleration
GWAS
:   Genome-wide association study
HDL
:   High-density lipoprotein
IV
:   Instrumental variable
IVW
:   Inverse variance weighted
LDL
:   Low-density lipoprotein
MVPA
:   Moderate-to vigorous physical activity
MR
:   Mendelian randomization
MVMR
:   Multivariable Mendelian randomization
PhenoAgeAccel
:   PhenoAge acceleration
PRESSO
:   Pleiotropy residual sum and outlier
SBP
:   Systolic blood pressure
SD
:   Standard deviation
SE
:   Standard errors
SNP
:   Single nucleotide polymorphism
WM
:   Weighted median

*   Received July 12, 2022.
*   Revision received July 14, 2022.
*   Accepted July 14, 2022.


*   © 2022, Posted by Cold Spring Harbor Laboratory

The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission.

## References

1.  1.Niccoli T, Partridge L. Ageing as a risk factor for disease. Curr Biol. 2012;22(17):R741–R52.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.cub.2012.07.024&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22975005&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

2.  2.Schumacher B, Pothof J, Vijg J, Hoeijmakers JHJ. The central role of DNA damage in the ageing process. Nature. 2021;592(7856):695–703.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41586-021-03307-7&link_type=DOI) 

3.  3.Warner HR. Current status of efforts to measure and modulate the biological rate of aging. J Gerontol A Biol Sci Med Sci. 2004;59(7):692–6.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/gerona/59.7.B692&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15304533&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000223239100006&link_type=ISI) 

4.  4.Jylhävä J, Pedersen NL, Hägg S. Biological Age Predictors. EBioMedicine. 2017;21:29–36.
    
    

5.  5.Horvath S, Raj K. DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nat Rev Genet. 2018;19(6):371–84.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41576-018-0004-3&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29643443&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

6.  6.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(4):573–91.
    
    

7.  7.Lu AT, Quach A, Wilson JG, Reiner AP, Aviv A, Raj K, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Albany NY). 2019;11(2):303–27.
    
    

8.  8.Wang C, Ni W, Yao Y, et al. DNA methylation-based biomarkers of age acceleration and all-cause death, myocardial infarction, stroke, and cancer in two cohorts: The NAS, and KORA F4. EBioMedicine. 2021;63:103151.
    
    

9.  9.Li A, Koch Z, Ideker T. Epigenetic aging: Biological age prediction and informing a mechanistic theory of aging [published online ahead of print, 2022 Jun 20]. J Intern Med.
    
    

10. 10.Quach A, Levine ME, Tanaka T, Lu AT, Chen BH, Ferrucci L, et al. Epigenetic clock analysis of diet, exercise, education, and lifestyle factors. Aging (Albany NY). 2017;9(2):419–46.
    
    

11. 11.Oblak L, van der Zaag J, Higgins-Chen AT, Levine ME, Boks MP. A systematic review of biological, social and environmental factors associated with epigenetic clock acceleration. Ageing Res Rev. 2021;69:101348.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.arr.2021.101348&link_type=DOI) 

12. 12.Sierra F, Caspi A, Fortinsky RH, Haynes L, Lithgow GJ, Moffitt TE, et al. Moving geroscience from the bench to clinical care and health policy. J Am Geriatr Soc. 2021;69(9):2455–63.
    
    

13. 13.Campisi J, Kapahi P, Lithgow GJ, Melov S, Newman JC, Verdin E. From discoveries in ageing research to therapeutics for healthy ageing. Nature. 2019;571(7764):183–92.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41586-019-1365-2&link_type=DOI) 

14. 14.Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601.
    
    [FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiRlVMTCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiYm1qIjtzOjU6InJlc2lkIjtzOjE2OiIzNjIvanVsMTJfMi9rNjAxIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMDcvMTQvMjAyMi4wNy4xMi4yMjI3NzUyOC5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

15. 15.Skrivankova VW, Richmond RC, Woolf BAR, Davies NM, Swanson SA, VanderWeele TJ, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ. 2021;375:2233.
    
    

16. 16.Zheng J, Baird D, Borges M-C, Bowden J, Hemani G, Haycock P, et al. Recent Developments in Mendelian Randomization Studies. Curr Epidemiol Rep. 2017;4(4):330–45.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29226067&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

17. 17.Lee JJ, Wedow R, Okbay A, et al. Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals. Nat Genet. 2018;50(8):1112–1121.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0147-3&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30038396&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

18. 18.Liu M, Jiang Y, Wedow R, et al. Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use. Nat Genet. 2019;51(2):237–244.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0307-5&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

19. 19.Zhong VW, Kuang A, Danning RD, et al. A genome-wide association study of bitter and sweet beverage consumption [published correction appears in Hum Mol Genet. 2019 May 02;:]. Hum Mol Genet. 2019;28(14):2449–2457.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/hmg/ddz061&link_type=DOI) 

20. 20.Dashti HS, Daghlas I, Lane JM, et al. Genetic determinants of daytime napping and effects on cardiometabolic health. Nat Commun. 2021;12(1):900.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41467-020-20585-3&link_type=DOI) 

21. 21.Dashti HS, Jones SE, Wood AR, et al. Genome-wide association study identifies genetic loci for self-reported habitual sleep duration supported by accelerometer-derived estimates. Nat Commun. 2019;10(1):1100.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41467-019-08917-4&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30846698&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

22. 22.Klimentidis YC, Raichlen DA, Bea J, et al. Genome-wide association study of habitual physical activity in over 377,000 UK Biobank participants identifies multiple variants including CADM2 and APOE. Int J Obes (Lond). 2018;42(6):1161–1176.
    
    

23. 23.Yengo L, Sidorenko J, Kemper KE, et al. Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum Mol Genet. 2018;27(20):3641–3649.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/hmg/ddy271&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30124842&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

24. 24.Shungin D, Winkler TW, Croteau-Chonka DC, et al. New genetic loci link adipose and insulin biology to body fat distribution. Nature. 2015;518(7538):187–196.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/nature14132&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25673412&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000349190300030&link_type=ISI) 

25. 25.Bradfield JP, Taal HR, Timpson NJ, et al. A genome-wide association meta-analysis identifies new childhood obesity loci. Nat Genet. 2012;44(5):526–531.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/ng.2247&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22484627&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

26. 26.Mahajan A, Taliun D, Thurner M, et al. Fine-mapping type 2 diabetes loci to single-variant resolution using high-density imputation and islet-specific epigenome maps. Nat Genet. 2018;50(11):1505–1513.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0241-6&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30297969&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

27. 27.Klarin D, Damrauer SM, Cho K, et al. Genetics of blood lipids among ∼300,000 multi-ethnic participants of the Million Veteran Program. Nat Genet. 2018;50(11):1514–1523.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0222-9&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30275531&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

28. 28.Evangelou E, Warren HR, Mosen-Ansorena D, et al. Genetic analysis of over 1 million people identifies 535 new loci associated with blood pressure traits [published correction appears in Nat Genet. 2018 Dec;50(12):1755]. Nat Genet. 2018;50(10):1412–1425.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0205-x&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30224653&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

29. 29.Han X, Ong JS, An J, Hewitt AW, Gharahkhani P, MacGregor S. Using Mendelian randomization to evaluate the causal relationship between serum C-reactive protein levels and age-related macular degeneration. Eur J Epidemiol. 2020;35(2):139–146.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s10654-019-00598-z&link_type=DOI) 

30. 30.Machiela MJ, Chanock SJ. LDlink: a web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants. Bioinformatics. 2015;31(21):3555–7.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/bioinformatics/btv402&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26139635&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

31. 31.McCartney DL, Min JL, Richmond RC, Lu AT, Sobczyk MK, Davies G, et al. Genome-wide association studies identify 137 genetic loci for DNA methylation biomarkers of aging. Genome Biol. 2021;22(1):194.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s13059-021-02398-9&link_type=DOI) 

32. 32.Lawlor DA, Harbord RM, Sterne JA, Timpson N, Davey Smith G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med. 2008;27(8):1133–1163.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/sim.3034&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17886233&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

33. 33.Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304–314.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/gepi.21965&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=27061298&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

34. 34.Burgess S, Bowden J, Fall T, Ingelsson E, Thompson SG. Sensitivity Analyses for Robust Causal Inference from Mendelian Randomization Analyses with Multiple Genetic Variants. Epidemiology (Cambridge, Mass). 2017;28(1):30–42.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=doi:10.1097/EDE.0000000000000559&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=27749700&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

35. 35.Verbanck M, Chen C-Y, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693–8.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41588-018-0099-7&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29686387&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

36. 36.Bowden J, Del Greco M F, Minelli C, Davey Smith G, Sheehan N, Thompson J. A framework for the investigation of pleiotropy in two-sample summary data Mendelian randomization. Stat Med. 2017;36(11):1783–802.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/sim.7221&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28114746&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

37. 37.Burgess S, Thompson SG. Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol. 2015;181(4):251–260.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/aje/kwu283&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25632051&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

38. 38.Figgins E, Pieruccini-Faria F, Speechley M, Montero-Odasso M. Potentially modifiable risk factors for slow gait in community-dwelling older adults: A systematic review. Ageing Res Rev. 2021;66:101253.
    
    

39. 39.Bountziouka V, Musicha C, Allara E, et al. Modifiable traits, healthy behaviours, and leukocyte telomere length: a population-based study in UK Biobank. Lancet Healthy Longev. 2022;3(5):e321–e331.
    
    

40. 40.Marioni RE, Harris SE, Shah S, et al. The epigenetic clock and telomere length are independently associated with chronological age and mortality [published correction appears in Int J Epidemiol. 2018 Feb 1;47(1):356]. Int J Epidemiol. 2018;45(2):424–432.
    
    

41. 41.Brønnum-Hansen H, Juel K. Abstention from smoking extends life and compresses morbidity: a population based study of health expectancy among smokers and never smokers in Denmark. Tob Control. 2001;10(3):273–278.
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MTQ6InRvYmFjY29jb250cm9sIjtzOjU6InJlc2lkIjtzOjg6IjEwLzMvMjczIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjIvMDcvMTQvMjAyMi4wNy4xMi4yMjI3NzUyOC5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

42. 42.Beach SR, Dogan MV, Lei MK, et al. Methylomic Aging as a Window onto the Influence of Lifestyle: Tobacco and Alcohol Use Alter the Rate of Biological Aging. J Am Geriatr Soc. 2015;63(12):2519–2525.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/jgs.13830&link_type=DOI) 

43. 43.GBD 2016 Alcohol and Drug Use Collaborators. The global burden of disease attributable to alcohol and drug use in 195 countries and territories, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016 [published correction appears in Lancet Psychiatry. 2019 Jan;6(1):e2]. Lancet Psychiatry. 2018;5(12):987–1012.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

44. 44.Loftfield E, Cornelis MC, Caporaso N, Yu K, Sinha R, Freedman N. Association of Coffee Drinking With Mortality by Genetic Variation in Caffeine Metabolism: Findings From the UK Biobank. JAMA Intern Med. 2018;178(8):1086–1097.
    
    

45. 45.Wang C, Bangdiwala SI, Rangarajan S, et al. Association of estimated sleep duration and naps with mortality and cardiovascular events: a study of 116 632 people from 21 countries. Eur Heart J. 2019;40(20):1620–1629.
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/eurheartj/ehy695&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom) 

46. 46.Gielen M, Hageman GJ, Antoniou EE, et al. Body mass index is negatively associated with telomere length: a collaborative cross-sectional meta-analysis of 87 observational studies. Am J Clin Nutr. 2018;108(3):453–475.
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2022%2F07%2F14%2F2022.07.12.22277528.atom)