https://orcid.org/0009-0004-5111-0936
Figures
Abstract
Background
Numerous observational studies have reported an association between frailty and atherosclerosis. However, the causal relationship between frailty and the occurrence of atherosclerosis in different anatomical sites remains unclear. we conducted a bidirectional Mendelian randomization (MR) study to evaluate the causal relationship between the frailty index (FI), and both systemic atherosclerosis and lipids.
Methods
We obtained summary statistics from large-scale genome-wide association studies (GWAS) of various phenotypes, including frailty (n = 175,226), coronary atherosclerosis (n = 56,685), cerebral atherosclerosis (n = 150,765), peripheral arterial disease (PAD) (n = 361,194), atherosclerosis at other sites (n = 17,832), LDL-C (n = 201,678), HDL-C (n = 77,409), and triglycerides (n = 78,700). The primary MR analysis employed the inverse variance weighted (IVW) method. Furthermore, to assess reverse causality, we employed inverse MR and multivariate MR analysis.
Results
Genetically predicted FI showed positive associations with the risk of coronary atherosclerosis (OR = 1.47, 95% CI 1.12–1.93) and cerebral atherosclerosis (OR = 1.99, 95% CI 1.05–3.78), with no significant association (p >0.05) applied to peripheral arterial disease and atherosclerosis at other sites. Genetically predicted FI was positively associated with the risk of triglycerides (OR = 1.31, 95% CI 1.08–1.59), negatively associated with the risk of LDL-C (OR = 0.87, 95% CI 0.78–0.97), and showed no significant association with the risk of HDL-C (p >0.05). Furthermore, both reverse MR and multivariate MR analyses demonstrated a correlation between systemic atherosclerosis, lipids, and increased FI.
Conclusion
Our study elucidated that genetically predicted FI is associated with the risk of coronary atherosclerosis and cerebral atherosclerosis by the MR analysis method, and they have a bidirectional causal relationship. Moreover, genetically predicted FI was causally associated with triglyceride and LDL-C levels. Further understanding of this association is crucial for optimizing medical practice and care models specifically tailored to frail populations.
Citation: Xu L, Wang Y, Ji H, Du W, You C, Chen J, et al. (2024) Frailty and risk of systemic atherosclerosis: A bidirectional Mendelian randomization study. PLoS ONE 19(5): e0304300. https://doi.org/10.1371/journal.pone.0304300
Editor: Amirmohammad Khalaji, Tehran University of Medical Sciences, ISLAMIC REPUBLIC OF IRAN
Received: October 25, 2023; Accepted: May 8, 2024; Published: May 23, 2024
Copyright: © 2024 Xu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: We were funded by the Zhangjiagang City Science and Technology Planning project (ZKS2125; ZKYL2343) for data collection and analysis. Previous manuscripts contained only one fund number, ZKS2125. We would like to attach ZKYL2343 to the financial disclosure. Updated financial disclosure: Youth Natural Science Foundation of Zhangjiagang TCM Hospital Affiliated to Nanjing University of Chinese Medicine (ZZYQ2009). Zhangjiagang City Science and Technology Planning project (ZKS2125; ZKYL2343). Suzhou Medical Health Science and technology innovation project (SKYD2022047). The funders have played a role in data collection and analysis.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: MR, Mendelian Randomization; FI, Frailty Index; GWAS, Genome-Wide Association Studies; PAD, Peripheral Arterial Disease; IVW, Inverse Variance Weighted; SNPs, Single Nucleotide Polymorphisms; IVs, Instrumental Variables; LDL-C, LDL Cholesterol; HDL-C, HDL Cholesterol; IEU, Integrative Epidemiology Unit; OR, Odds Ratio; CI, Confidence Interval
Introduction
Frailty, an age-related physical state, manifests as impaired physiological function in multiple organs or systems [1]. It contributes to the aging process across various bodily systems [2], leading to an enduring state that heightens the risk of cardiovascular events, falls, and disability in response to stressors [2–4]. Frail populations often exhibit a range of chronic conditions, such as coronary heart disease, heart failure, and ischemic stroke, which can result in varying degrees of physical disability. Observational studies have consistently linked frailty with coronary heart disease [5,6], ischemic stroke [7,8], and peripheral arterial disease [9] (PAD), all of which are associated with the underlying pathology of atherosclerosis. However, observational studies are susceptible to confounding bias and reverse causal associations, and whether a bidirectional association exists between frailty and systemic atherosclerosis with lipids remains uncertain.
Mendelian randomization (MR) is a genetic epidemiological approach that relies on data from genome-wide association studies (GWAS). It utilizes independent genetic variants as instrumental variables to investigate causal relationships between an exposure and an outcome [10]. The fundamental principle behind MR is that an individual’s genotype is randomly assigned during gamete formation. This helps to eliminate the possibility of confounding bias and reverse causal associations [10,11]. As a result, MR can effectively reduce the impact of reverse causation and other confounding factors, thereby providing valuable insights into the underlying relationship between frailty and systemic atherosclerosis. This is a significant finding, with important implications for both the public and clinical sectors.
The frailty index (FI) is widely recognized as the preferred instrument for evaluating frailty [1,12]. It assigns a continuous score ranging from 0 (no deficits) to 1 (all deficits) based on the cumulative presence of health deficits, encompassing somatic symptoms, psychological factors, comorbidities, and disability [1]. A higher FI score is associated with various clinical outcomes, including a range of chronic conditions, disability, and mortality [13]. In our study, we employed a bidirectional MR approach to examine the causal relationship between frailty, as measured by the FI, and systemic atherosclerosis. We investigated different sites of atherosclerosis, including coronary atherosclerosis, cerebral atherosclerosis, PAD, and atherosclerosis at other sites. Additionally, we accounted for the assessment of lipids because lipid-driven intravascular plaques are the pathological basis of atherosclerosis.
Materials and methods
We conducted a two-sample MR analysis using GWAS summary statistics to estimate the causal associations between FI and systemic atherosclerosis as well as lipids. We utilized GWAS summary statistics for frailty indices as the exposure variable and selected relevant single nucleotide polymorphisms (SNPs) as instrumental variables (IVs). Systemic atherosclerosis summary statistics were used as the outcome. Additionally, we investigated the causal association of FI with lipids, including LDL cholesterol (LDL-C), HDL cholesterol (HDL-C), and triglycerides. To explore the reverse direction of causality, we employed bidirectional MR analysis, and multivariate MR analysis was also used to estimate causal associations.
Exposure sources
Summary data for the frailty index were retrieved from the most recent GWAS meta-analysis of the UK Biobank and TwinGene, we identified the frailty index as an exposure factor [14]. They measured frailty by the FI, which was based on the accumulation of 49 health deficits during the life course and has been well-validated and widely used in clinical practice [1]. The UK Biobank included a total of 164,610 participants aged 60 to 70 with European ancestry, including 84,819 females, with data from 49 self-reported deficits. The TwinGene study included 10,616 Swedish population samples, with ages ranging from 41 to 87, and females accounted for 5,577 individuals (52.5%). The complete GWAS summary statistics can be found in the GWAS catalog under study number GCST90020053 (https://www.ebi.ac.uk/gwas/search?query=GCST90020053).
Outcome sources
The FinnGen study [15] integrated genomic data with healthcare information from the Finnish National Health Registry, utilizing the Finnish biobank that currently encompasses over 370,000 samples. Data on coronary atherosclerosis and atherosclerosis at other sites can be found (https://r9.finngen.fi/). Rainer Malik et al [7] conducted a study of over 520,000 European subjects in a GWAS meta-analysis examined the large artery atherosclerosis-related ischemic stroke subtype, which we defined as cerebral atherosclerosis, and relevant data are available in the Integrative Epidemiology Unit (IEU) OpenGWAS data infrastructure, study number: ebi-a-GCST005840. (https://gwas.mrcieu.ac.uk/datasets/ebi-a-GCST005840/).
GWAS data based on the UK Biobank study pooled information from over approximately 500,000 samples [16] and PAD data with 361,194 European samples are available in the Neale lab public database. The latest GWAS data for LDL-C, HDL-C, and Triglycerides are available in the MRC IEU OpenGWAS data infrastructure under study numbers: ieu-b-5089, ieu-b-4844, and ieu-b-4850. Data sources are summarized in Table 1.
Selection of instrumental variables
The requirements of MR analysis for instrumental variables must satisfy three assumptions (Fig 1): (i) Relevance assumption: IVs must be correlated with exposure factors (FI). (ii) Independence assumption: IVs must be independent of confounding factors. (iii) Exclusivity hypothesis: IVs affect outcome only through FI and are not directly associated with outcome. Those with a significant association with phenotype (p <5*10−8) were screened out first. A total of 14 SNPs associated with FI were selected (p <5*10−8). SNPs that are similar in the genome have similar genetic effects, referred to as linkage disequilibrium, which may affect outcome effect values. To address this, we applied a threshold of r2 = 0.001 and a window size of 10,000 kb to remove the potential influence of linkage disequilibrium. The online tool PhenoScanner V2 [17,18] was used to further remove the effects of confounding factors, and SNPs associated with confounding factors such as obesity, adiposity, and smoking were excluded from our analysis. The strength of the instrumental variables is often assessed using the F-statistic in the exposure regression [19]. In our study, instrumental variables with an F-statistic <10 were considered weak and were excluded from further analysis.
The MR analysis was based on three fundamental assumptions: The relevance assumption, independence assumption, and exclusivity hypothesis. Red represents the positive MR analysis pathway (FI-related). Blue represents the reverse MR analysis pathway (atherosclerosis-related and lipid-related). MR, Mendelian randomization; FI, Frailty index; SNP, Single nucleotide polymorphism.
Statistical analysis
The FI on systemic atherosclerosis and lipids was evaluated using a two-sample MR analysis. The primary method employed to assess the causal association between FI and systemic atherosclerosis and lipids was the inverse variance weighted (IVW) method. A significance threshold of p <0.05 was set, and the results of causal associations were reported as odds ratio (OR) with 95% confidence interval (CI). The IVW method utilized a meta-analysis approach, combining Wald ratios for each genetic variant to generate combined estimates of the exposure’s effect on the outcomes, thereby providing consistent causal estimates [20]. In addition to the IVW method, other estimation methods such as the weighted median, simple median, and penalized weighted median were also employed. The simple median estimator is equivalent to a weighted median estimator with equal weights. While the simple median provides consistent causal effect estimates when at least 50% of the IVs are valid, the weighted median requires at least 50% of the weights to come from valid IVs [21]. Heterogeneity was assessed using Cochran’s Q test, and to evaluate pleiotropy, MR-Egger regression intercept and MR-PRESSO methods were utilized. Leave-one-out analysis was performed in the IVW method to determine if the results were driven by any specific SNP [22]. Reverse MR was used to estimate the causal association between systemic atherosclerosis and FI in the opposite direction. Multivariable MR, an expanded method of MR, was applied in situations where it is challenging to find IVs that are exclusively related to only one exposure, allowing for directed pleiotropy when IVs are associated with multiple exposures [23].
All statistical analyses were conducted using the "TwoSampleMR" and "MRPRESSO" packages in R version 4.3.0. A two-sided threshold of p <0.05 in the MR analysis was considered statistically significant.
Results
After accounting for linkage disequilibrium effects, a total of 14 SNPs that showed an association with the FI at a significance level of p <5*10−8 were initially selected. However, further examination revealed that four of these SNPs (rs2071207, rs583514, rs1363103, and rs10891490) were associated with confounding factors such as weight, waist circumference, leg fat mass, smoking, and systolic blood pressure. Since this violated assumption (ii) of independence, these four SNPs were excluded from the IVs. As a result, a final set of 10 SNPs was defined as IVs to assess the causal relationship between the FI as the exposure factor and the outcome of interest (Table 2). These 10 SNPs were selected based on their relevance to the FI and their independence from confounding factors, meeting the requirements for conducting a robust Mendelian randomization analysis.
Genetically predicted FI and systemic atherosclerosis
We assessed the causal relationship between genetically predicted FI and systemic atherosclerosis, including coronary artery atherosclerosis, cerebral artery atherosclerosis, PAD, and atherosclerosis at other sites. The two-sample MR analysis using the IVW method revealed a positive association between genetically predicted FI and the risk of coronary artery atherosclerosis (OR = 1.47, 95% CI 1.12 to 1.93, p = 0.005) and cerebral artery atherosclerosis (OR = 1.99, 95% CI 1.05 to 3.78, p = 0.034). However, there was no significant causal relationship observed between genetically predicted FI and PAD (OR = 1, 95% CI 1 to 1.01, p = 0.242), or atherosclerosis at other sites (OR = 1.27, CI 0.79 to 2.07, p = 0.324) (Fig 2) (S1 File). Cochran’s Q test indicated that the estimates of IVs on the associations with coronary artery atherosclerosis and cerebral artery atherosclerosis did not exhibit heterogeneity (p >0.05). The MR-Egger regression intercept and MR-PRESSO methods provided no evidence of pleiotropy (p >0.05) (S2 File). Leave-one-out analysis indicated that the influence of FI on coronary artery atherosclerosis and cerebral artery atherosclerosis was not driven by a single SNP, indicating the overall stability of the results (Fig 3).
(a) Effect of genetically predicted FI on atherosclerosis, including coronary atherosclerosis, cerebral atherosclerosis, peripheral arterial disease, and atherosclerosis at other sites. (b) Effect of genetically predicted FI on lipids, including LDL-C, HDL-C, and triglycerides. MR, Mendelian randomization; FI, frailty index; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol; OR, odds ratio; CI, confidence interval.
(a) coronary atherosclerosis; (b) cerebral atherosclerosis; (c) peripheral arterial disease; (d) atherosclerosis at other sites; (e) LDL-C; (f) HDL-C; (g) triglycerides. LDL-C, LDL cholesterol; HDL-C, HDL cholesterol.
Genetically predicted FI and lipids
We further evaluated the causal relationship between genetically predicted FI and lipids, including LDL-C, HDL-C, and triglycerides. The two-sample MR analysis using the IVW method revealed a positive association between genetically predicted FI and triglyceride levels (OR = 1.31, CI 1.08 to 1.59, p = 0.005). There was a negative association between genetically predicted FI and LDL-C levels (OR = 0.87, CI 0.78 to 0.97, p = 0.011). However, there was no significant association observed between genetically predicted FI and HDL-C levels (OR = 0.96, CI 0.82 to 1.13, p = 0.658) (Fig 2) (S3 File). Cochran’s Q test indicated that the estimates of IVs on LDL-C, HDL-C, and triglyceride levels did not exhibit heterogeneity (p >0.05). The MR-Egger regression intercept and MR-PRESSO methods showed no evidence of pleiotropy (p >0.05) (S2 File). Leave-one-out analysis indicated that the influence of FI on LDL-C and triglyceride levels was not driven by a single SNP (Fig 3).
Reverse MR and multivariate MR
To investigate the causal relationship between systemic atherosclerosis and FI, we performed a reverse MR analysis. The results revealed a significant causal association between atherosclerosis at different anatomical locations and FI. Specifically, coronary artery atherosclerosis (OR = 1.06, 95% CI 1.04 to 1.07, p <0.001), cerebral artery atherosclerosis (OR = 1.03, 95% CI 1.01 to 1.05, p <0.001), PAD (p <0.001), and atherosclerosis at other sites (OR = 1.10, 95% CI 1.08 to 1.12, p <0.001) exhibited a clear causal association with FI (S4 File).
Furthermore, we employed multivariable MR analysis to assess the causal relationship between lipids and FI. The results indicated that LDL-C (OR = 1.08, CI 1.04 to 1.12, p <0.001), HDL-C (OR = 0.97, 95% CI 0.96 to 0.98, p <0.001), and triglycerides (OR = 0.96, CI 0.93 to 0.99, p = 0.008) had a significant causal association with FI (S5 File). Among these lipids, LDL-C and HDL-C showed a stronger correlation with FI than triglycerides.
Discussion
Our results indicate that the genetically predicted FI is associated with a higher risk of coronary artery atherosclerosis and cerebral artery atherosclerosis, while no significant associations were found with PAD and atherosclerosis at other sites. Moreover, genetically predicted FI is positively correlated with triglyceride levels, negatively correlated with LDL-C levels, and not significantly associated with HDL-C levels.
One of the pressing challenges in public health management is the global phenomenon of population aging [24]. As the number of frail individuals continues to rise, there are growing concerns about the decline in quality of life and the substantial economic burden associated with cardiovascular diseases. In the past decade, there has been increasing interest in understanding the relationship between frailty and cardiovascular diseases. A prospective cohort study [25] identified frailty as a significant risk factor for adverse cardiovascular events. Even after accounting for baseline factors such as age, sex, and ethnicity, frail individuals exhibited notably higher rates of cardiovascular events, including coronary heart disease, stroke, and PAD, than their non-frail counterparts. This highlights the importance of addressing frailty as a means of preventing and potentially reversing the occurrence of cardiovascular events. Priority should be given to strategies that target frail populations for screening for coronary atherosclerosis and cerebral atherosclerosis. By focusing on interventions that target frailty, it may be possible to have a positive impact on reducing the burden of cardiovascular diseases.
The FI is a well-established and validated instrument that relies on a questionnaire survey assessing functional deficits [1]. In frail individuals, the underlying pathological mechanisms for cardiovascular events are related to atherosclerosis and abnormal lipids. Observational studies have consistently indicated a connection between FI and cardiovascular events or subclinical atherosclerosis. Several cross-sectional analyses [26–28] have demonstrated that individuals with higher FI scores or in a prefrail state are more susceptible to cardiovascular diseases. Moreover, elevated FI scores have been associated with increased cardiovascular mortality rates [29]. A longitudinal study conducted over a 16-year period found that for every 0.1 increase in FI, the risk of all-cause mortality increased by 4% and the risk of cardiovascular disease-related mortality increased by 3–5% [30].
In patients with acute stroke, frailty can be exacerbated, as confirmed by a meta-analysis. Although certain studies may carry a risk of bias, frailty is commonly observed in individuals with acute stroke, regardless of the specific measurement method used. Two meta-analyses [9,31] have demonstrated a significantly higher prevalence of frailty among individuals with PAD. Additionally, research has shown an independent association between subclinical cardiovascular diseases, such as carotid artery atherosclerosis and femoral artery sclerosis, and frailty, irrespective of the presence of coronary heart disease, stroke, or myocardial infarction [32].
The relationship between PAD and frailty remains a topic of debate. While frailty is often observed in individuals with PAD, observational studies have not definitively established a bidirectional causal relationship between PAD and frailty. A study conducted by Brutto suggested that there is no independent correlation between frailty and large artery atherosclerosis in the peripheral vascular bed [33]. This could be due to the onset of frailty affecting various anatomical districts. In fact, frailty is a state of multi-organ fictional decline resulting from the disruption of the balance among multiple organs, which are unable to support each other [34]. Vital organs such as the heart and brain, which are supplied by major blood vessels, are more likely to be involved. In other words, individuals with a high FI have an increased risk of developing cardiovascular diseases affecting the heart and brain but not necessarily an increased risk of peripheral artery atherosclerosis or atherosclerosis in other arteries. This may explain the varying relationship between frailty and different types of atherosclerosis.
Blood lipids, such as triglycerides and LDL-C, play a significant role in the development of atherosclerosis, and research indicates that lowering these lipid levels can reduce the risk of cardiovascular diseases [35]. A Bayesian network analysis conducted on community-dwelling older people suggested that HDL-C levels may be associated with malnutrition and can have an impact on frailty status [36]. This finding aligns with the results of our reverse MR analysis. It is worth noting that HDL-C levels are more closely associated with age, with LDL-C levels declining further in older individuals [37]. However, frailty is not solely determined by age and does not exhibit a clear positive causal relationship with HDL-C levels. This discrepancy may be attributed to the complex interplay of multiple factors influencing frailty beyond chronological age.
Our study provides a systematic evaluation of the causal relationship between frailty and atherosclerosis occurring in different parts of the body, incorporating the association estimates of blood lipids. We utilized the latest GWAS data with an adequate sample size to ensure reliable results. However, our study still has limitations. First, the potential bias in MR analysis due to population and age structure should be considered. The GWAS data used in our study primarily originated from European, Swedish and Finnish populations, predominantly consisting of older adults and females [25], which may limit the generalizability of our findings to other populations. Second, despite our efforts to address pleiotropic effects, such as removing linkage disequilibrium effects and weak instrument variables and applying methods such as MR-Egger regression intercept and MR-PRESSO to account for horizontal pleiotropy, these methods have their inherent limitations. Although we utilized the PhenoScanner V2 tool to control for confounders, it is challenging to eliminate the influence of horizontal pleiotropy. Third, our assessment of blood lipids focused on LDL-C, HDL-C, and triglycerides, while other lipid measures were not included in our analysis. Future research should explore additional lipid indicators, such as apolipoprotein A1 and apolipoprotein B, to gain a more comprehensive understanding of lipid metabolism in relation to frailty and atherosclerosis.
Overall, our study highlights the bidirectional causal relationship between FI and systemic atherosclerosis, as well as the association between FI and lipids, taking into account the specific locations of arterial atherosclerosis. These findings contribute to a better understanding of the complex interplay between frailty, atherosclerosis, and lipid metabolism.
Conclusion
Our study elucidated that genetically predicted FI is associated with the risk of coronary atherosclerosis and cerebral atherosclerosis by the MR analysis method, and they have a bidirectional causal relationship. Moreover, genetically predicted FI was causally associated with triglyceride and LDL-C levels. Further understanding of this association is crucial for optimizing medical practice and care models specifically tailored to frail populations.
Supporting information
S1 File. Effect of genetically predicted FI on atherosclerosis.
FI, Frailty Index; MR, Mendelian randomization; OR, Odds ratio; CI, Confidence interval.
https://doi.org/10.1371/journal.pone.0304300.s001
(PDF)
S2 File. Heterogeneity and pleiotropy analysis.
FI, Frailty Index; MR, Mendelian randomization; IVW, Inverse variance weighted; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol.
https://doi.org/10.1371/journal.pone.0304300.s002
(PDF)
S3 File. Effect of genetically predicted FI on lipids.
FI, Frailty Index; MR, Mendelian randomization; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol; OR, Odds ratio; CI, Confidence interval.
https://doi.org/10.1371/journal.pone.0304300.s003
(PDF)
S4 File. Reverse MR of genetically predicted effects of atherosclerosis on FI.
FI, Frailty Index; MR, Mendelian randomization; IVs, Instrumental variables; OR, Odds ratio; CI, Confidence interval.
https://doi.org/10.1371/journal.pone.0304300.s004
(PDF)
S5 File. Multivariate MR of the effect of genetically predicted lipids on FI.
FI, Frailty Index; MR, Mendelian randomization; IVs, Instrumental variables; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol; OR, Odds ratio; CI, Confidence interval.
https://doi.org/10.1371/journal.pone.0304300.s005
(PDF)
Acknowledgments
The authors thank all the participants, investigators, and consortia who contributed to this study.
References
- 1. Dent E, Martin FC, Bergman H, Woo J, Romero-Ortuno R, Walston JD. Management of frailty: opportunities, challenges, and future directions. Lancet Lond Engl 2019;394:1376–86. https://doi.org/10.1016/S0140-6736(19)31785-4.
- 2. Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in elderly people. Lancet Lond Engl 2013;381:752–62. pmid:23395245
- 3. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:M146–156. pmid:11253156
- 4. Theou O, Rockwood MRH, Mitnitski A, Rockwood K. Disability and co-morbidity in relation to frailty: how much do they overlap? Arch Gerontol Geriatr 2012;55:e1–8. pmid:22459318
- 5. Tse G, Gong M, Nunez J, Sanchis J, Li G, Ali-Hasan-Al-Saegh S, et al. Frailty and Mortality Outcomes After Percutaneous Coronary Intervention: A Systematic Review and Meta-Analysis. J Am Med Dir Assoc 2017;18:1097.e1–1097.e10. pmid:29079033
- 6. Madhavan MV, Gersh BJ, Alexander KP, Granger CB, Stone GW. Coronary Artery Disease in Patients ≥80 Years of Age. J Am Coll Cardiol 2018;71:2015–40. https://doi.org/10.1016/j.jacc.2017.12.068.
- 7. Malik R, Chauhan G, Traylor M, Sargurupremraj M, Okada Y, Mishra A, et al. Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes. Nat Genet 2018;50:524–37. pmid:29531354
- 8. Burton JK, Stewart J, Blair M, Oxley S, Wass A, Taylor-Rowan M, et al. Prevalence and implications of frailty in acute stroke: systematic review & meta-analysis. Age Ageing 2022;51:afac064. https://doi.org/10.1093/ageing/afac064.
- 9. Wang Y, Wu X, Hu X, Yang Y. Prevalence of frailty in patients with lower extremity peripheral arterial disease: A systematic review and meta-analysis. Ageing Res Rev 2022;82:101748. pmid:36216291
- 10. Smith GD, Ebrahim S. “Mendelian randomization”: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32:1–22. pmid:12689998
- 11. Lawlor DA, Harbord RM, Sterne JAC, Timpson N, Davey Smith G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 2008;27:1133–63. pmid:17886233
- 12. Cesari M, Gambassi G, van Kan GA, Vellas B. The frailty phenotype and the frailty index: different instruments for different purposes. Age Ageing 2014;43:10–2. pmid:24132852
- 13. Williams DM, Jylhävä J, Pedersen NL, Hägg S. A Frailty Index for UK Biobank Participants. J Gerontol A Biol Sci Med Sci 2019;74:582–7. pmid:29924297
- 14. Atkins JL, Jylhävä J, Pedersen NL, Magnusson PK, Lu Y, Wang Y, et al. A genome‐wide association study of the frailty index highlights brain pathways in ageing. Aging Cell 2021;20:e13459. pmid:34431594
- 15. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023;613:508–18. pmid:36653562
- 16. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature 2018;562:203–9. pmid:30305743
- 17. Staley JR, Blackshaw J, Kamat MA, Ellis S, Surendran P, Sun BB, et al. PhenoScanner: a database of human genotype-phenotype associations. Bioinforma Oxf Engl 2016;32:3207–9. pmid:27318201
- 18. Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, et al. PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinforma Oxf Engl 2019;35:4851–3. pmid:31233103
- 19. Burgess S, Small DS, Thompson SG. A review of instrumental variable estimators for Mendelian randomization. Stat Methods Med Res 2017;26:2333–55. pmid:26282889
- 20. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife 2018;7:e34408. pmid:29846171
- 21. 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;40:304–14. pmid:27061298
- 22. Wu F, Huang Y, Hu J, Shao Z. Mendelian randomization study of telomere length and bone mineral density. Aging 2020;13:2015–30. pmid:33323545
- 23. Sanderson E. Multivariable Mendelian Randomization and Mediation. Cold Spring Harb Perspect Med 2021;11:a038984. pmid:32341063
- 24. Fang EF, Xie C, Schenkel JA, Wu C, Long Q, Cui H, et al. A research agenda for ageing in China in the 21st century (2nd edition): Focusing on basic and translational research, long-term care, policy and social networks. Ageing Res Rev 2020;64:101174. pmid:32971255
- 25. Damluji AA, Chung S-E, Xue Q-L, Hasan RK, Moscucci M, Forman DE, et al. Frailty and cardiovascular outcomes in the National Health and Aging Trends Study. Eur Heart J 2021;42:3856–65. pmid:34324648
- 26. Wong TY, Massa MS, O’Halloran AM, Kenny RA, Clarke R. Cardiovascular risk factors and frailty in a cross-sectional study of older people: implications for prevention. Age Ageing 2018;47:714–20. pmid:29796607
- 27. Boreskie KF, Rose AV, Hay JL, Kehler DS, Costa EC, Moffatt TL, et al. Frailty status and cardiovascular disease risk profile in middle-aged and older females. Exp Gerontol 2020;140:111061. pmid:32814098
- 28. Farooqi MAM, Gerstein H, Yusuf S, Leong DP. Accumulation of Deficits as a Key Risk Factor for Cardiovascular Morbidity and Mortality: A Pooled Analysis of 154 000 Individuals. J Am Heart Assoc 2020;9:e014686. https://doi.org/10.1161/JAHA.119.014686.
- 29. Jiang M, Foebel AD, Kuja-Halkola R, Karlsson I, Pedersen NL, Hägg S, et al. Frailty index as a predictor of all-cause and cause-specific mortality in a Swedish population-based cohort. Aging 2017;9:2629–46. pmid:29273703
- 30. Quach J, Theou O, Godin J, Rockwood K, Kehler DS. The impact of cardiovascular health and frailty on mortality for males and females across the life course. BMC Med 2022;20:394. pmid:36357932
- 31. Zhang H, Jie Y, Wang P, Sun Y, Wang X, Fan Y. Impact of frailty on all-cause mortality or major amputation in patients with lower extremity peripheral artery disease: A meta-analysis. Ageing Res Rev 2022;79:101656. pmid:35654353
- 32. McKechnie DGJ, Papacosta AO, Lennon LT, Ellins EA, Halcox JPJ, Ramsay SE, et al. Subclinical cardiovascular disease and risk of incident frailty: The British Regional Heart Study. Exp Gerontol 2021;154:111522. pmid:34428478
- 33. Del Brutto OH, Mera RM, Recalde BY, Costa AF, Sedler MJ. Mediation of age in the association between frailty and large artery atherosclerosis burden—A population study in community-dwelling older adults. J Stroke Cerebrovasc Dis Off J Natl Stroke Assoc 2020;29:104845. pmid:32389559
- 34. Guaraldi G, Raggi P. Atherosclerosis in frailty: Not frailty in atherosclerosis. Atherosclerosis 2017;266:226–7. pmid:28969855
- 35. Ference BA, Kastelein JJP, Ray KK, Ginsberg HN, Chapman MJ, Packard CJ, et al. Association of Triglyceride-Lowering LPL Variants and LDL-C-Lowering LDLR Variants With Risk of Coronary Heart Disease. JAMA 2019;321:364–73. pmid:30694319
- 36. Yuan Y, Lin S, Huang X, Li N, Zheng J, Huang F, et al. The identification and prediction of frailty based on Bayesian network analysis in a community-dwelling older population. BMC Geriatr 2022;22:847. pmid:36368951
- 37. Ferrara A, Barrett-Connor E, Shan J. Total, LDL, and HDL cholesterol decrease with age in older men and women. The Rancho Bernardo Study 1984–1994. Circulation 1997;96:37–43. pmid:9236414