DNA methylation and the epigenetic clock in relation to physical frailty in older people: the Lothian Birth Cohort 1936

Frailty is a clinical syndrome that becomes increasingly common at older ages [1]. Its core features are increased vulnerability to stressors due to impairments in multiple systems, decreased physiological reserves, and a decline in the ability to maintain homeostasis [2]. It increases the risk of adverse outcomes, including falls, disability, hospitalization, institutionalization, and death [2–4]. There are two principal models of frailty [2]. Fried’s phenotype model defines frailty on purely physical terms, based on three or more components (poor grip strength, slow walking speed, low physical activity, exhaustion, and unintentional weight loss) [3]. The frailty index, or cumulative deficit model, defines frailty much more broadly in terms of the accumulation of ‘deficits’ (symptoms, signs, diseases, and disabilities) [5]. The biological drivers of the multisystem dysregulation that underlies frailty remain unclear, particularly at the cellular and molecular levels.

Epigenetic changes affect all cells and tissues over the lifespan [6]. One such change involves alterations to deoxyribonucleic acid (DNA) methylation patterns. DNA methylation is involved in the regulation of gene expression and occurs at cytosine-phosphate-guanine (CpG) sites across the genome [7, 8]. The proportion of methylation at a particular CpG site is referred to as a beta value, which can change over the life-course [7, 8]. Methylation levels are affected by both genetic and environmental exposures [9]. The relationship between DNA methylation levels and aging is complex [10]. Early evidence showed that global DNA methylation level decreases with age, but subsequent studies revealed that aging is associated with differential methylation (mainly hypermethylation) of some genomic loci [11]. Multiple CpG sites have been identified where methylation is associated with age [12]. Several DNA methylation-based biomarkers are now used to estimate ‘epigenetic age’ [13, 14] or ‘epigenetic age acceleration,’ a measure of the difference between predicted epigenetic age and chronological age’ [15]. These indices—often referred to as the ‘epigenetic clock’—are associated with mortality independently of chronological age and other risk factors, supporting the notion that they capture some aspect of biological aging [15, 16]. Epigenetic alterations are thought to be one of the ‘hallmarks’ of aging [10] and hence may contribute to age-related pathologies such as frailty. That is, chronological age acts as a proxy for several biological changes associated with aging, of which DNA methylation is one.

Few studies have investigated the relationship between DNA methylation patterns and frailty in older people. One study, where frailty status was defined using cluster analysis, reported that global DNA methylation was lower in people who were frail compared to the non-frail, [17], but another study using the Fried phenotype of physical frailty found no such association [18]. This latter study also examined associations between promotor-specific CpG island methylation and frailty status, and found that lower levels of CpG island methylation were associated with a reduced likelihood of being frail [18]. Further indications that DNA methylation patterns might differ in people who are frail came in a recent study of 1820 older people which found that greater epigenetic age acceleration—the difference between predicted epigenetic age and chronological age—was associated with greater frailty as measured by a broadly defined frailty index, such that the frailty index increased by about 0.25% points per year of epigenetic age acceleration [19].

Here, we aimed to add to understanding of the relationship between epigenetic status and physical frailty in a large, narrow-age sample of 70-year olds. The limited chronological age range mitigates against cohort exposure effects on DNA methylation thus providing a more robust context to draw inferences about biological aging indices. First, we conducted an epigenome-wide association study (EWAS) to try to identify whether differential methylation at specific CpG sites was associated with current physical frailty status, as defined by the Fried phenotype. Secondly, we investigated whether people who were biologically older as indexed by epigenetic age acceleration measures were at increased risk of being physically frail.

The correlation between the two measures of epigenetic age acceleration was moderate (r = 0.38, p < 0.0001).

Manhattan plots showing the p values for the CpGs for physical frailty status (frail versus not frail, and pre-frail versus not frail) are presented in Fig. 1. There was a single significant association in the EWAS comparing frail versus not frail: cg18314882 on chromosome 8 in the MAF1 gene (p = 1.16 × 10–7): beta 0.0054 (SE 0.0010), indicating hypermethylation. Table 2 reports the local associations for CpGs within a 1601 base pair region (the rest of the CpG island). The significant site is an isolated result. Summary statistics for this CpG were mean 0.0118 (SD 0.007), minimum 0.001, maximum 0.105. Figure 2 shows a boxplot of this CpG by frailty status. A QQ plot of the p values for frail vs not frail is shown in the Additional file 1: Figure S1. We report the top 20 CpGs in the Additional file 1: Table S1. None of them are significant—the smallest p value is 0.012. The CpG that was significant in the EWAS comparing frail versus not frail was not significant in the EWAS comparing pre-frail versus not frail (p = 0.67).

Probe	Beta	SE	T	p
cg02621020	6.15E-04	0.001783	0.345243	7.30E-01
cg08825571	− 1.39E-03	0.000553	− 2.51254	1.22E-02
cg11538573	− 4.82E-04	0.000739	− 0.65216	5.15E-01
cg11883258	3.92E-03	0.004616	0.849065	3.96E-01
cg17170088	6.77E-04	0.000842	0.803668	4.22E-01
cg17176228	2.90E-04	0.000501	0.579033	5.63E-01
cg18314882	5.38E-03	0.001004	5.359905	1.16E-07
cg19517467	− 2.66E-04	0.0016	− 0.16609	8.68E-01
cg20573110	− 7.74E-06	0.001677	− 0.00461	9.96E-01
cg22260950	− 2.28E-04	0.001066	− 0.21425	8.30E-01
cg22861185	9.00E-04	0.000855	1.053655	2.92E-01
cg23119631	− 8.33E-04	0.001081	− 0.77111	4.41E-01

In this cross-sectional survey of 791 men and women aged 70 years, epigenome-wide association study analyses found no widespread differences in methylation patterns between those who were physically frail and those who were not frail, as defined by the Fried phenotype. The proportion of methylation at a single CpG site (cg18314882 on chromosome 8 in the MAF1 gene) was significantly different between those two groups. No such difference at any CpG site was found between those who were pre-frail and those who were not frail. These results suggest that blood DNA methylation is not a good biomarker for physical frailty. Older biological age as measured by extrinsic epigenetic age acceleration was associated with an increased risk of being physically frail, independent of potential confounding factors. Intrinsic epigenetic age acceleration was not associated with increased risk.

To our knowledge, there have been no previous EWAS of frailty status in later life. Here, we found that the proportion of methylation at one CpG methylation site (cg18314882 on chromosome 8 in the MAF1 gene) differed between people who were physically frail and those who were not frail. There were no similar effect sizes for other CpGs within in the same region which would have increased the validity of our finding. Previous evidence in online databases suggests that this locus is unmethylated in every tissue, such that a beta of 0.02 was found in the blood, brain, omentum; see for example, Slieker et al. 2013 where these data are first described [30]. This limits the implication of this locus given that we found little indication that methylation was changed in relation to frailty. MAF1 is a transcriptional repressor. Recent research has shown that it represses the expression of both pol III-dependent genes and certain RNA pol II-dependent genes that play a crucial part in oncogenesis, [31] and is important for the regulation of intracellular lipids [31, 32]. It is likely that MAF1 has a diversity of physiological functions, but knowledge of its roles is still limited [33]. It is possible that it could influence risk of physical frailty via its role in lipid regulation and hence obesity. Obesity is an established risk factor for physical frailty [34, 35].

Sarcopenia, an age-related syndrome characterized by loss of skeletal muscle mass and strength, is a major contributor to physical frailty [36]. There is evidence from studies that measured methylation either in muscle tissue or in whole blood that methylation levels at some loci may help to explain variations in aging skeletal muscle mass [37, 38]. In a study comparing DNA methylation dynamics in skeletal muscle tissue from 24 young male adults and 24 older male adults, 5963 CpG sites were reported to be differentially methylated between the two groups; there was predominantly hypermethylation throughout the genome in the older group compared to the young group [37]. In a study of 1550 female twins that set out to identify genomic regions that were associated with skeletal muscle mass using methylation levels measured in whole blood, seven associations between methylation at CpG loci and skeletal muscle mass were discovered and replicated with a false discovery rate of less than 0.1 [38]. No association between individual CpG methylation sites and grip strength was found in a EWAS on the 27 k methylation array based on a sample of 172 female twins [39].

Our finding that greater extrinsic epigenetic age acceleration was associated with an increased risk of being physically frail is consistent with findings from a cross-sectional study of 1820 men and women aged 50–75 years [19]. In that study, greater epigenetic age acceleration (defined as the difference between predicted methylation age and chronological age) was associated with higher scores on a frailty index made up of 34 potential ‘deficits,’ such that that the frailty index increased by about 0.25% points per year of epigenetic age acceleration. There tends to be a moderate correlation between scores on a frailty index and physical frailty status as defined by the Fried phenotype [40], but they differ in that while the latter describes a specific clinical syndrome [41], the cumulative deficit model describes the general state or condition of an individual. The fact that greater epigenetic age acceleration has been shown to be associated with greater risk of being physically frail and of scoring higher on a more broadly defined frailty index [19] adds to the evidence that people who are frail are likely to be biologically older.

Intrinsic epigenetic age acceleration was designed to estimate “pure” epigenetic aging effects that are not influenced by differences in blood cell counts [15]. In the current study, it was moderately correlated with extrinsic epigenetic age acceleration (r = 0.38), and in contrast to the latter, it was not significantly associated with risk of being physically frail. Although both these epigenetic age acceleration measures are predictive of mortality [15], findings that intrinsic epigenetic age acceleration but not extrinsic epigenetic age acceleration predicts lung cancer [42] and is associated with being a centenarian [43] have led to the suggestion that it may capture a cell-type independent component of the aging process [15].

One potential limitation of our study is that of the 1091 individuals who took part in the survey, 791 (73%) could be included in the current study. Some individuals were missing data on the physical activity component of the Fried phenotype of frailty and some had missing methylation data. Another limitation is that we were only able to look at methylation markers in blood rather than in any other tissues in relation to physical frailty. Our EWAS findings may be due to either a genuine null association between blood-based methylation markers and physical frailty or a lack of statistical power in our current analyses.

In this cross-sectional survey of 70-year-old men and women, we found evidence that those who were biologically older, as indexed by greater extrinsic epigenetic age acceleration, were more likely to be physically frail. Future research will need to investigate whether epigenetic age acceleration plays a causal role in the onset of physical frailty.