Intersecting Genetics of Frailty and Cardiovascular Disease

Abstract

Objectives

To determine the genetic correlates of physical frailty and sarcopenia, focusing on single nucleotide polymorphisms (SNPs) in genome-wide association studies (GWAS), and to explore the genetic overlap of frailty with cardiovascular disease (CVD) and its risk factors.

Methods

PubMed was systematically searched for GWAS studies investigating the association between SNPs and objective measures of physical frailty or sarcopenia. SNPs were retained if they were associated with one of the phenotypes of interest by a p-value of 5.0×10−8 or less.

Results

Ten studies were included, with a total of 237 SNPs in 181 genes being associated with physical frailty or sarcopenia; as measured by handgrip strength or lean (muscle) mass. These genes were cross-referenced in the GWAS Catalog, and many of them were found to be associated with CVD or metabolic syndrome.

Conclusions

Evidence from GWAS has shown that frailty is associated with common genetic polymorphisms. Many of these polymorphisms have been implicated in CVD, supporting the hypothesis of a shared pathophysiology between these entities. Future studies are eagerly anticipated to map out the mechanistic links and discover therapeutic targets and novel biomarkers for frailty.

Introduction

Frailty is a geriatric syndrome characterized by accelerated biological aging and subclinical impairments in multiple organ systems (1). Skeletal muscle is one of the prototypical organs affected, with the age-relative decline in muscle mass and strength being known as sarcopenia (2, 3). The accumulation of impairments leads to a state of decreased resiliency and increased vulnerability to physiological stressors, leading to a heightened risk of morbidity and mortality (4). This has been demonstrated in older adults exposed to the stress of cardiovascular disease (CVD) and procedures such as cardiac surgery and transcatheter intervention, wherein frailty is a major risk factor for fatal and nonfatal adverse outcomes (5, 6). Frail patients have a well-defined phenotype that includes slowness, weakness, fatigue, and inactivity (7). Using tools designed to objectively measure this phenotype, it is estimated that 25% of community dwelling older adults above the age of 80 are frail (7, 8), with a 2-fold increase in those with CVD (6). Much has been speculated about the reasons for the inter-relationship between frailty and CVD, including common mechanistic links to low-grade inflammation and insulin resistance (9, 10). Despite the elevated prevalence and negative effects of frailty, little is known about its pathobiology (11) or its genetic underpinnings.

A single nucleotide polymorphism (SNPs) is a commonly occurring DNA sequence variation that involves substitution, deletion, or insertion of one nucleotide in the genome. When SNPs occur within a gene or regulatory region, they may play a causal role in disease by affecting the gene's function. SNPs may also play a role in predicting an individual's response to environmental and iatrogenic stressors, naturally leading to the hypothesis that SNPs may be involved in the pathophysiology of frailty. Studies on the genetic determinants of frailty have used diverse methodological approaches, initially focusing on selected candidate SNPs in isolation, and in the last decade shifting to another approach — genome-wide association studies (GWAS). GWAS compare hundreds of thousands of SNPs across the genomes of large cohorts of individuals to detect in unbiased fashion loci that contribute to a phenotype (12). As of yet, no definitive genetic picture of frailty has been drawn. We therefore sought to conduct a systematized review on the existing body of GWAS evidence on genetic determinants of physical frailty and sarcopenia, in order to draw informative conclusions and potentially shed light on the reported relationship with CVD.

Methods

To review the evidence for GWAS in the field of frailty in systematized fashion, we searched PubMed until October 20, 2020 (and references of retrieved publications) for original GWAS publications that sought to associate SNPs with objective measures of physical frailty or sarcopenia such as Fried's scale, lean (muscle) mass, gait speed, handgrip strength, or other physical performance measures. We excluded publications that focused on non-SNP genetic parameters (e.g. mitochondrial DNA variants, epigenetic modifications, telomere length) and those that focused on longevity rather than frailty or sarcopenia (Supplemental Figure 1). We also excluded publications that focused on younger individuals, highly trained athletes, and animal models. Our search identified 237 publications, 10 of which met the inclusion criteria (Table 1).

Figure 1.

Overlap Between Frailty, Cardiovascular Disease, and Metabolic Syndrome

Table 1.

GWAS Publications on Physical Frailty and Sarcopenia

First Author	Year	Sample Size	Frailty Measure	Number of SNPs with p<5×10−08
Tikkanen	2018	334,925	Handgrip strength	139
Willems	2017	195,180	Handgrip strength	16
Matteini	2016	34,910	Handgrip strength	1
Hernandez Cordero	2019	181,862	Lean mass appendicular	4
Zillikens	2017	73,420 / 101,767	Lean mass appendicular / total	3 / 5
Karasik	2019	85,519	Lean mass	4
Ran	2020	46,503	Lean mass	29
Liu	2009	7,415	Lean mass	2
Pei	2019	6,587	Lean mass legs	2
Guo	2013	3,913	Lean mass appendicular	1

Results

The studies reviewed described 237 different SNPs affecting 181 individual genes (including non-coding RNA genes) (Supplemental Table 1), which were correlated with measures of frailty by significant p-values below the recommended threshold of 5.0×10−8 or less (13, 14). We then used the GWAS Catalog (https://www.ebi.ac.uk/gwas/) to search for significant associations between these genes and CVD, as well as other conditions. Among these, 68 of the frailty-associated genes were also found to have been associated by GWAS with phenotypes related to CVD and 146 with the metabolic syndrome (MetS) or its components. Sixty-one of these genes were found to have been associated with all three phenotypes: frailty, CVD, and MetS (Supplemental Table 2). Most notably, polymorphisms of the FTO, SLC8A1, MC4R, SLC39A8, and GLYAT genes were associated with frailty in GWAS and elsewhere implicated with the development of CVD and its risk factors as well as MetS. The function of the aforementioned selected genes and their potential role in the pathogenesis of both frailty and CVD, as well as their connection with MetS, are explored below.

The FTO gene encodes the “fat mass and obesity-associated protein” (15), a 2-oxoglutarate Fe(II) dependent nucleic acid demethylase that regulates growth and mRNA translation (16, 17). Thirty-three SNPs in FTO were found by a total of 4 GWAS to be significantly associated with either lean mass (16, 18, 19) or handgrip strength (20). One of the SNPs in FTO was previously found by GWAS to be associated with higher fasting insulin levels, higher rates of insulin resistance, higher rates of developing diabetes, higher levels of serum triglycerides, lower levels of HDL cholesterol, and higher risks of incident coronary artery disease (19). FTO has been strongly linked by GWAS to diabetes mellitus type 2 (15), as well as increased body mass index (BMI) (17, 18, 21) and increased hip and waist circumference (18).

The SLC8A1 gene encodes the solute carrier family 8 member 1, a sodium/calcium exchanger that extrudes calcium ions from cardiomyocytes during the relaxation phase (15). Two SNPs in SLC8A1 were found by 2 GWAS to be associated with handgrip strength (20, 22). SLC8A1 has been linked by GWAS to systolic and diastolic blood pressure (BP) (15), and elevated BP is a known risk factor for the development of CVD and MetS, and also for frailty (23, 24). Moreover, 4 SNPs in SLC8A1 have been linked by GWAS to PR interval and 4 more SNPs lined to QT interval (15, 25), the elongation of which are predisposing factors for adverse CVD events.

The MC4R gene encodes the melanocortin 4 receptor, a receptor for adrenocorticotropic hormone and melanocyte stimulating hormone (MSH) that plays an important role in obesity (26). SNPs in MC4R were found by 2 GWAS to be associated with handgrip strength (20) or lean mass (19). SNPs in MC4R have also been linked by GWAS to BMI, waist-hip ratio, diabetes mellitus type 2, HDL cholesterol, or systolic BP (15), all of which figure in the MetS profile.

The SLC39A8 gene encodes a solute-carrier protein that is thought to transport zinc and possibly cadmium across the cell membrane. Two SNPs in SLC39A8 were found by GWAS to be significantly associated with handgrip strength (20). SNPs in SLC39A8 have also been linked by GWAS to HDL cholesterol levels (15, 27), BMI (15, 21) and increased systolic BP (15), which we have already mentioned to be important features of MetS. One of the SNPs that correlated with decreased handgrip strength was also found to correlate with increased systolic and diastolic BP (28), and seven studies have confirmed the effect of this SNP on systolic BP (15). A SNP in SLC39A8 has also been linked by GWAS with increased levels of NT-proBNP and CVD mortality (29).

The GLYAT gene encodes the metabolic enzyme glycine N-acyltransferase that is responsible for the detoxification of certain compounds and for regulating glucose metabolism (30, 31). One SNP in GLYAT was found by GWAS to correlate with appendicular lean mass (32). Likely because it regulates glucose metabolism, GLYAT expression has been associated with hyperglycemic states (32). GLYAT has also been linked by GWAS to systolic BP (15) and to CVD (33).

Discussion

The connection between frailty and CVD has been supported by data from multiple epidemiological studies with a previous review (34) of 9 such studies highlighting a bi-directional relationship. In cross-sectional (prevalence) and longitudinal (incidence) studies, frailty was shown to be a risk factor for CVD, and conversely, CVD was shown to be a risk factor for frailty. A seminal report from the Cardiovascular Health Study revealed that the relationship extended beyond frailty and clinical diagnoses of CVD to subclinical manifestations of CVD (e.g. reduced ankle-brachial index, increased left ventricular mass, carotid plaque, infarct-like brain lesions) (35). Furthermore, a report from the English Longitudinal Study of Aging revealed that the Framingham risk score was predictive of incident frailty, and that the most predictive components were older age, diabetes mellitus type 2, and systolic hypertension (24).

The connection between frailty and MetS is similarly supported by epidemiological data, although the evidence is somewhat hazy (36). MetS refers to a constellation of three or more of: central obesity, low HDL cholesterol, high triglycerides, hyperglycemia, and hypertension (37). One earlier study found that MetS was not directly associated with frailty, but that insulin resistance and inflammation were (9) — an association supported by others as well (10, 38). A subsequent study found that MetS was associated with prevalent sarcopenia but only in men aged <75 years of age (36). MetS was not associated with non-sarcopenia-related constructs of frailty such as Rockwood's Frailty Index (23). Another study found that the presence of MetS (but not its severity) was strongly associated with frailty, and that reduced gait speed and handgrip strength, along with easy fatiguability, progressed more rapidly in individuals with MetS (39, 40). Yet another study found that older adults with MetS were nearly 2-fold more likely to develop incident frailty over 3.5 years of follow-up, with reduced handgrip strength being most predictive of MetS (41).

Finally, the connection between frailty, CVD, and MetS is well highlighted by the tendency toward sarcopenic obesity, where it has been demonstrated that the loss of muscle strength occurring simultaneously with obesity is associated with a 23% increase in CVD (42). This relationship is thought to be explained in part by decreased physical activity and common risk markers between sarcopenic obesity and CVD (42), and obesity itself is a driving factor for frailty and CVD (43, 44). This overlap at the epidemiological and genetic levels adds compelling weight to the case for a shared pathophysiology between these entities and supplies grounds for a putative biological mechanism (Figure 1). However, the mechanism is unlikely to be a simple one since these genes are also involved in cellular activities such as metabolism (FTO, SH2B1, ADCY3, SEC16B, MC4R, ACMSD, GNPDA2, and GLYAT), growth (FTO, SH2B1, FAM150B, TFAP2B, UHRF1BP1, and TGFA), gene transcription (BPTF, MLLT10, TCF4, POU6F2, TRPS1, POLD3, and TFAP2D), and other basal functions. It is therefore envisaged that multiple polymorphic hits to these finely tuned genetic underpinnings of basic cell function may upset the homeostatic balance over time and predispose to MetS, CVD, and frailty phenotypes, all to varying proportions depending on the particular constellation of hits in the individual.

Limitations

While one-third of the frailty-associated genes were remarked to have been implicated in CVD and MetS phenotypes, a smaller number (N=4) of these were pinpointed to a specific common SNP. Different GWAS methods may ascribe certain SNPs to different genes depending on the method used. A portion of the SNPs from the publications we identified were not included in the GWAS Catalog, hampering our efforts to explore their potential associations with CVD and MetS. The majority of studies reported only a handful of SNPs, but there was one study which reported thousands of SNPs in the supplemental materials of which we included only the four mentioned in the article itself in order not to skew our paper (45). Most SNPs were noted to either raise or lower a particular risk factor or components of a disease, although the directionality was not always reported in the GWAS Catalog. Confirming the directionality of a given SNP would be important for future efforts seeking to develop diagnostic kits or therapeutic targets. Only 5 out of 237 SNPs (0.02%) displayed an inconsistency in the directionality of effects across these areas, while the directional effects of the rest of the SNPs affecting more than one area tended to be concordant (i.e. SNPs associated with higher levels of frailty were also associated with higher levels of CVD or MetS). This concordance reinforces the hypothesis of a shared genetic basis, at least in part, for physical frailty, CVD, and MetS.

Future Research

There is currently no biomarker that has shown specificity for the diagnosis of frailty nor is there a drug on the horizon that has shown meaningful therapeutic effects (46). This unmet need is of great clinical relevance for the health care of older adults; including those with CVD to refine estimates of predicted risk, guide decision making, and potentially offer ancillary therapies to slow down frailty while treating the cardiac condition to improve patient outcomes (47). Over the past decade, GWAS has evolved from early publications identifying 5 or 10 SNPs with p-values around 10–3, to now identifying tens or even hundreds of SNPs with p-values as low as 10–40. As the evidence for genetic correlates of frailty from GWAS continues to accrue and become more robust, Mendelian randomization (MR) studies will be instrumental to identify protein biomarkers that can serve as prognostic indicators and therapeutic targets for frailty (48).

In MR, SNPs are used as proxies for lifetime exposures to gain insights into the causal nature of the association between a modifiable protein level and a condition of interest. Since alleles are randomly assigned at conception, MR studies are similar to randomized controlled trials in that the known and unknown confounders are randomly distributed between genotype groups, thus mitigating the risk of confounding. Furthermore, genotype assignments necessarily predate the condition of interest, thus mitigating the risk of reverse causation, another common bias. To arrive at causal conclusions, MR studies depend on establishing a strong and direct relationship between the SNP and protein level (rather than pleotropic effects) by measuring the expression of the protein or focusing on SNPs close to the encoded gene. MR studies have recently been undertaken for frailty (49), already substantiating a link between 2 SNPs associated by GWAS to handgrip strength with coronary artery disease and myocardial infarction (50).

Often there exist relatively weak associations between a multitude of SNPs and the condition of interest — particularly for complex conditions such as frailty. While each individual SNP may have a small if not trivial effect on its own, overlooked by low resolution candidate gene approaches, the aggregate count of these SNPs — known as a Genetic Risk Score (GRS) — may have a highly predictive effect. GRS is a weighted summary of the variation among the many loci identified for a particular person, which given a sufficient number of informative loci, provides a sensitive indicator of the target condition's genetic landscape (51). In light of the growing database of informative loci for frailty, and availability of genetic sequencing options, GRS represent a promising area of investigation (20) with clinical applicability to identify patients at risk for frailty and intervene early on with preventative cardiovascular health interventions (52, 53).

Conclusions

While the phenotypic construct of frailty has been advanced and refined over the past 20 years, the biological construct has remained difficult to pinpoint. In the absence of a clearly defined pathophysiology, the field has struggled to produce specific biomarkers and targeted therapies. Contemporary GWAS, reviewed here, has made significant progress in addressing this knowledge gap and identifying SNPs associated with various components of the frailty phenotype. Of particular interest to the cardiovascular community, many of these SNPs were remarked to also be associated with cardiovascular risk factors, CVD, and MetS, supporting the epidemiological links between frailty and CVD that have been consistently reported in the literature. Future studies are eagerly anticipated and will leverage MR and GRS methodologies to map out the mechanistic links and, in turn, discover drug targets and novel biomarkers for frailty.

Conflict of Interest

None declared.

Ethical statement

This project complies with the current laws and ethical standards of the country in which it was performed.

Electronic supplementary material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-021-1673-8 and is accessible for authorized users.

Supplementary material, approximately 100 KB.