Stress, diet, exercise: Common environmental factors and their impact on epigenetic age

Epigenetic aging clocks have gained significant attention as a tool for predicting age-related health conditions in clinical and research settings. They have enabled geroscientists to study the underlying mechanisms of aging and assess the effectiveness of anti-aging therapies, including diet, exercise and environmental exposures. This review explores the effects of modifiable lifestyle factors ’ on the global DNA methylation landscape, as seen by aging clocks. We also discuss the underlying mechanisms through which these factors contribute to biological aging and provide comments on what these findings mean for people willing to build an evidence-based pro-longevity lifestyle


Population aging as a socioeconomic problem
Aging populations are becoming a global phenomenon, with the number of people aged 65 years and older projected to reach 1.5 billion by 2050, causing collateral deficits in various socioeconomic areas, largely due to disabilities and the loss of performance in the last decades of the lifespan. These issues are best visible in developed countries, where decreasing birth rates are coupled with increasing lifespans. The list of countries with a negative demographic growth trend includes leading nations, and, as of 2022, China with its large population joined this worrisome trend. Further demographic shifts pose a serious challenge to national economies worldwide as increasing numbers of people retire and require social support and increased healthcare (Zhavoronkov, 2013).
The socioeconomic pressure has accelerated the efforts to develop and clinically implement diagnostic and therapeutic tools that would increase the healthy lifespan. The World Health Organization (WHO) has recognized the importance of addressing the challenges posed by aging populations and has launched a Global Action Plan on Healthy Aging. The plan seeks to promote healthy aging, increase access to healthcare services for older individuals, and support the development of age-friendly environments (World Health Organization, 2021).
Advancements in molecular and robotic technologies have enabled scientists to study the molecular causes of aging and form hypotheses on decelerating and even reversing this process. Aging clocks are the digital models used to quantify the aging processes and express their intensity as biological age, pace of aging, rate of aging, etc. Such models are used in (i) population studies exploring the risk factors of aging, (ii) experiments that measure the geroprotective effects of various interventions, and (iii) laboratory models in which the conventional definition of age is not applicable (e.g., in cell cultures).
The inconvenient and impractical alternative to aging clocks is observing subjects for decades until their aging pace can be derived from mortality data. In terms of testing geroprotectors as well as other interventions in longevity medicine, biological age calculation by aging clocks is the only feasible and validated solution. This type of technology is, therefore, a key piece in the search for a scientific solution to the global aging problem (Zhavoronkov et al., 2019b).

Types of aging clocks
The first aging clock using data from modern high-throughput technology was released in 2011 (Bocklandt et al., 2011). It was trained as a linear regression and validated on a collection of 128 DNA methylation (DNAm) profiles from an Illumina array, and it was able to measure a subject's chronological age with a mean absolute error (MAE) of 5.2 years. Shortly after this publication, more aging clocks followed which were trained in a similar fashion (Hannum et al., 2013;Koch and Wagner, 2011).
In 2013, the same approach was Steve Horvath applied this methodology to create a multi-tissue DNAm clock (Horvath, 2013). Horvath's clock is still being used today and has inspired many other scientists to develop their own tools to quantify aging. At the time, it featured the largest data set compiled from 82 studies and highlighted that tissues could age at different paces. The small number of 353 CpG sites used by the clock to produce accurate prediction across multiple tissues indicated that epigenetic regulation might be playing a key role in organismal aging.
Over the last 10 years, dozens of aging clocks have been developed . Almost every type of biophysiological data that changes with age has been utilized to build a predictor of chronological age with the help of machine learning . Currently, there are aging clocks that use clinical blood tests, facial photos, and omics data from both humans and other organisms (Bobrov et al., 2018;Fedor Galkin et al., 2020;Fleischer et al., 2018;Mamoshina et al., 2018;Meer et al., 2018;Putin et al., 2016;Zhavoronkov et al., 2020). Nonetheless, epigenetic models of aging remain popular among researchers and regularly see new developments.
While the first epigenetic aging clocks were trained with chronological age as the target variable, more recent models use alternative measures of age. PhenoAge, a 2018 epigenetic clock, used "phenotypic age" defined as a function of chronological age and a number of clinical biomarkers (Levine et al., 2018). A 2019 epigenetic clock GrimAge uses time-to-death as its target variable (Lu et al., 2019). A 2002 clock, DunedinPoAm, and its updated 2022 version, DunedinPACE, used the "pace of aging" derived from longitudinal data on the shifts in 19 clinical biomarkers as the model's target (Belsky et al., 2022(Belsky et al., , 2020Elliott et al., 2021). The models with such more elaborate definitions of the target are commonly called "second-generation" aging clocks to separate them from the models trained to predict chronological age obtained from cross-sectional studies. Nonetheless, first-generation clocks have not been replaced by these newer models and remain a popular choice among researchers. Moreover, second-generation clocks lack any shared core methodology and are based on DNAm array data, just as the first-generation clocks, which makes aging clock generations an arbitrary partition rather than a reflection of iterative improvement.
Aging clocks are a core technology in the development of geroprotective and senolytic therapies. However, despite the increasing levels of research and associated funding, longevity pharmacology has seen only limited progress (Zhavoronkov, 2020;Zhavoronkov and Bhullar, 2015, p. 11). Since 1990, governments worldwide have spent over $1 trillion on biomedical aging research (Zhavoronkov and Cantor, 2011). To secure further support for projects toward effective treatment in longevity medicine, robust criteria for the measurement of primary and further study objectives are necessary. Currently, the effectiveness of longevity therapeutics is evaluated by (i) longitudinal clinical trials focusing on mortality and (ii) trials using aging clocks (Zhavoronkov et al., 2019a). However, the latter allows much shorter iterations and, consequently, less expensive trials in the field of longevity pharmacology (de Magalhães, 2021). Some aging clocks have been patented which makes them ready for integration into existing commercial systems and further validation on a wide scale Galkin et al., 2022c;Horvath, 2021).

Epigenetic aging
Aging is a systemic process that impacts all levels of biological systems: organs, tissues, cells, and their molecular components. The hallmarks of aging, now numbered 12, capture the current main domains of aging-related mechanisms associated with aging as a comprehensive entity (López-Otín et al., 2023. The phenotypic markers of aging are easy to observe in the elderly, especially those with frailty and obvious somatic and neuropsychological deficits. However, defining an elderly or a youthful epigenetic profile is challenging. The epigenetic landscape determines gene expression intensity and is formed by an intracellular apparatus of chromatin remodelers and DNA modifiers such as DNA methyltransferases (DNMTs). Although DNMTs are not the only epigenetic factors, their impact on overall cellular phenotype cannot be underestimated (Fig. 1). For example, hypermethylation of the promoters of tumor suppressor genes has long been recognized as a driving force in multiple malignancies (Cheng et al., 2019;Herman et al., 1994;Yazici et al., 2020). Somatic mutations in blood cells' DNMT3A have been associated with a 3.0-3.8 increase in epigenetic age further validating the importance of epigenetic regulation in aging (Robertson et al., 2019).
Much research is being done on the role of epigenetic mechanisms in aging, particularly the role of DNAm. It is hypothesized that the environmental stress accumulated over a lifetime disrupts epigenetic profiles and causes noisy transcription, which contributes to the aging phenotype (Zhang et al., 2020). Although the loss of epigenetic signatures is a major contributor to aging, recent findings have confirmed that these changes can be modified in animal models to reverse aging (Yang et al., 2023).
Influence of the environmental factors was best determined so far in Fig. 1. The aging phenotype is effectuated by proteins that either fail to perform their original function or gain new adverse functions. The manifestations of the aging phenotype include systemic inflammation, low resilience to oxidative stress, and the accumulation of intracellular and extracellular waste, accompanied by a major disruption of signaling and protein-protein interactions (PPI). Epigenetic phenomena affect the proteome by regulating gene expression, but epigenetic profiles are also affected by proteins such as DNMTs. The full understanding of the aging process cannot be complete without the full knowledge of the role of the epigenome component in it. Aging clocks are perfect tools that allow us to study the links between the epigenome and the rate of aging.
twin studies, some of which reported 70-80% of epigenetic variance to be attributed to external influences (Cheung et al., 2018;van Dongen et al., 2016). Interestingly, epigenetic variance is more pronounced in older twin pairs, which is in line with the hypothesis of stress accumulation (van Dongen et al., 2016). Single-cell observations also show that intra-tissue epigenetic heterogeneity is a consistent indicator of aging, further supporting this hypothesis (Cheung et al., 2018). One study has illustrated the stochastic nature of age-related epigenetic changes as an increase in the Shannon entropy of methylation levels (Hannum et al., 2013). In other words, DNAm levels at most CpG sites tend to lose their initial hypo-or hyper-methylated identity and gravitate toward moderation (50%) with age.
Age-related epigenetic changes contribute to the aging phenotype by promoting genomic instability, carcinogenesis, and cardiovascular pathologies (Pagiatakis et al., 2021). While human studies on this topic are limited to epigenome-wide association studies, experiments on model organisms find much more conclusive evidence for the driving role of DNAm in aging (Gonzalo, 2010;Saul and Kosinsky, 2021). Recent research has focused on identifying specific epigenetic changes associated with aging as well as the molecular mechanisms that drive these changes. Besides the DNAm landscape alterations in aging individuals, histone modifications and noncoding RNA expression were reported to change as well.
In mice, negative aging-related traits can be transferred to offspring via epigenetic inheritance. More specifically, the offspring of older mice tend to have a 6.6% shorter life expectancy and develop a phenotype of an aged individual (heart fibrosis and muscle atrophy) earlier than the offspring of younger mice (Xie et al., 2018). The offspring of older murine fathers also have impaired learning and memory abilities. A similar observation has been made for humans: advanced paternal age results in an increased risk of autism, schizophrenia, and dyslexia (Saha et al., 2009). These phenotypes have been linked to hypomethylation and the consequent activation of mobile genomic elements in older sperm. Other effects of epigenetic aging include abnormal methylation patterns across the mTOR-signaling and immunological pathways.
The importance of epigenetics in aging is further emphasized by epigenetic rejuvenation studies, which have shown that modifying DNAm patterns can lead to tissue regeneration and prevent cellular senescence .
Thus, certain epigenetic patterns created by environmental factors can accelerate aging. Most commonly, such factors are studied in crosssectional settings. Longitudinal observations and twin studies may also provide more reliable validation of the epigenetic aging contributors, which can ultimately lead to clinical trials aiming to provoke organismal rejuvenation or treat aging-associated conditions, such as cancer (Nepali and Liou, 2021;Wang et al., 2022).

Epigenetic therapeutics
The interest is increasing toward the potential use of epigenetic therapies to delay or reverse aging-related changes. Some studies have shown that modifying the epigenome can improve age-related declines in cellular function and delay the onset of age-related diseases, suggesting that epigenetic changes may be a promising target for antiaging interventions (Cheng et al., 2019;Meiliana et al., 2022;Zhang et al., 2020).
Epigenetic clocks can be used to quantify the influence of specific stressors and protective factors on the pace of aging (Fig. 2). In certain cases, it may even be possible to extrapolate preexisting knowledge about the epigenetic effects of a compound or an activity to estimate its influence on aging mechanisms. In this way, epigenetic studies enable the discovery of novel therapies aimed at extending human healthspan (Zhang et al., 2020) and preventing the onset of noncommunicable diseases.
While epigenetic clocks have also been applied to validate the effect of such interventions as epigenetic reprogramming and heterochronic parabiosis, we limit the scope of this review only to the interventions that can be enacted by longevity enthusiasts today. Epigenetic reprogramming is a promising new technology that enables the cultivation of host-derived stem cells to be used for transplantation. The dedifferentiation to the pluripotent state has been associated with an erasure of epigenetic age, making the cells indistinguishable from embryonic stem cells. Some solutions involve epigenetic rejuvenation without full de-differentiation, which is considered safer due to lower malignancy risks (de Lima Camillo and Quinlan, 2021; Singh and Newman, 2018). Epigenetic reprogramming and rejuvenation are promising technologies currently tested in model organisms but their application in humans is not yet feasible. Similarly, heterochronic parabiosis is studied only in model organisms due to technical and legal difficulties (Pamplona et al., 2023;Zhang et al., 2021). Even less invasive procedures such as umbilical cord plasma transfusions, despite the possible anti-aging effects (− 0.82 years according to GrimAge), can be encountered only in research settings (Clement et al., 2022). For now, consumers are limited to dietary and lifestyle anti-aging therapies, whose effects will be discussed in the sections below.

Negative influence on the epigenetic age
Within the framework of longevity medicine, the benefit of interventions can be quantified with a decrease in the aging rate as measured by various aging clocks. For example, epigenetic and other clocks (especially hematologic clocks) allow clinicians to determine exactly how much younger a smoker's phenotype would be if they had quit smoking at a certain point in the past. The DNAm changes associated with smoking have been linked to previously recorded smokingassociated alterations in gene expression that include the activation of senescence genes (Li et al., 2015;Walters et al., 2014). Smoking also affects DNAm levels in other tissues such as blood (Tsaprouni et al., 2014).
Giving up smoking is one of the most common medical recommendations for all age groups (Stead et al., 2013). Several studies have shown the adverse effects of smoking to be partly mediated by epigenetic mechanisms. More specifically, smoking increases the epigenetic age of airway epithelial and lung tissues by four to five years (Wu et al., 2019). Smoking cessation at an early enough stage has the potential to enable the restoration of the normal epithelial aging rate in the airways, yet the lungs of ex-smokers maintain the footprints of aging acceleration.
Some studies indicate that smoking in youth results in irreversible damage that can be detected in later life (Klopack et al., 2022). A time-to-death clock, GrimAge, is accelerated by both smoking in youth, pack-years, and parents' smoking. In the same study, two other clocks (DunedinePoAm, PhenoAge), however, were only affected by adult pack-years. Knowledge from such biogerontological studies can be used in clinical practice to estimate whether smoking cessation could restore a patient's normal pace of aging. Another common health recommendation is to reduce alcohol consumption due to its detrimental effects on the cardiovascular and neural systems (Mende, 2019). Several recent studies examined epigenetic age acceleration in regular drinkers (Bøstrand et al., 2022). One of these finds that people with alcohol use disorder are on average 2.2 years biologically older than healthy controls (Luo et al., 2020, p. 2). The type of alcohol consumed appears to play a significant role in alcohol-induced epigenetic aging. While liquor, beer, and total alcohol amount are associated with higher biological age, wine consumption displays no such association (Nannini et al., 2023). The authors suppose that this observation might be linked to the high polyphenol content in wine, while liquor contains alcohol in high concentrations and is depleted in polyphenols.
Interestingly, the severity of aging acceleration has been associated with polymorphism in the gene APOL2 and, possibly, its expression level in the hippocampus. APOL2 also has known links to addiction and schizophrenia, which implies that accelerated epigenetic aging might also be a symptom of some mental and substance abuse disorders (Lehrmann et al., 2006).
It should be noted, however, that the measured effects vary among aging clocks and some may return statistically insignificant results (Kresovich et al., 2021b).

Epigenetic diet
Giving up smoking and alcohol are effective in decreasing accelerated biological aging. However, those lifestyle interventions are by far insufficient for longevity medicine practice. Studies on the dietary factors affecting epigenetic aging have a wider reach and can be used to discover potential geroprotectors.
The most well-known dietary intervention with a proven effect on epigenetics is caloric restriction (CR). A prolonged up-to-40% reduction in calorie intake has been shown to prevent age-related methylome changes in model organisms (Gensous et al., 2019). Even a short-term (four weeks) CR remodels the DNAm profiles of genes involved in diabetes, inflammation, and cardiovascular health-related pathways in old rats (Kim et al., 2016). Similar results have been obtained with rhesus monkeys; long-term CR delayed the onset of age-associated pathologies and greatly improved the survival rate (Colman et al., 2009). In the CR cohort, 20% of the monkeys died from age-related causes by the age of 30, while 50% died in the control cohort. After adjusting for the lifespan difference between species, a similar drop in mortality for humans would be observed at the age of 91 years (Tacutu et al., 2018).
Until recently, the efficiency of CR in humans had not been measured, although some attempts were made to extrapolate the results obtained in primate studies (Maegawa et al., 2017). A2021 human study of CR involved 43 older adult males who underwent an eight-week program combining CR with exercise, dietary supplements, and guidance (Fitzgerald et al., 2021). The regimen resulted in an epigenetic age decrease of 2-3 years, as measured by the 2013 Horvath's clock. In the 2023 CALERIE study, 128 people had their caloric intake reduced by 25% and observed for two years (Waziry et al., 2023). A slight yet statistically reliable change in the pace of aging, as measured with Dun-edinPACE, was observed in a dose-dependent manner in response to CR. The other two aging clocks used in the study (GrimAge, PhenoAge) were unable to detect any aging-related changes in the CR group. The 2-3% deceleration of aging detected in CALERIE was interpreted by the authors as a 10-15% reduction in mortality rate.
Some other non-CR diets have also been assessed for their ability to decelerate epigenetic aging. In an exploration of epigenetic age in 2694 adult women, their pace of aging was put in the context of adherence to four dietary indices: DASH, HEI, aHEI, aMED (Kresovich et al., 2022). These indices represent the healthiness of a diet by assigning points based on diet composition with higher scores obtained by people who consume more vegetables, fruits, protein, polyunsaturated and omega-3 fatty acids, and those who consume less alcohol, sugars, and sweetened foods (Chiuve et al., 2012;Fung et al., 2009Fung et al., , 2008Krebs-Smith et al., 2018). While four epigenetic clocks were tested in the study, only two (PhenoAge and GrimAge) showed significant associations with diet index adherence. Interestingly, even in these two clocks the beneficial effect of healthy eating was higher in people with low levels of physical activity. For example, women with dietary scores in the upper quartile are 0.8-1.5 years younger compared to the lower quartile, based on PhenoAge estimates. In women with more than 2.5 h of weekly activity, however, this aging decelerating is not observed at all. The authors hypothesize that diet and exercise act on the same epigenetic pathways reflected in PhenoAge and thus their effects do not stack.
Although in (Kresovich et al., 2022) Horvath's (2013) clock did not detect a significant effect of diet quality on the aging rate, a more recent study has displayed its utility in a longitudinal setting. In (Fitzgerald et al., 2023), six people underwent an eight-week "methylation-supportive diet and lifestyle program" which involved breathing exercises, physical training, dietary and supplement prescriptions, and intermittent fasting. At the end of observation, the epigenetic age of the participants was reduced by 4.60 years on average (0.0-11.0 years range). In another recent study featuring a custom clock using only 70 CpG sites from six genes, one year of epigenetic diet resulted in a 0.58 year decrease in epigenetic age ). Compared to cross-sectional or retrospective studies, such designs allow us to assess the efficiency of anti-aging diets and other interventions in realistic settings and measure adherence.
A study of eating habits conducted on 407 subjects showed that the consumption of fish and poultry significantly decreased the pace of aging (Quach et al., 2017). Similarly, a diet rich in fruits and vegetables, as indicated by high blood carotenoids, has also been associated with aging deceleration (Quach et al., 2017). However, due to the methodology of the study, the effect of these factors cannot be translated into a specific number of years. Similar results were obtained in another recent study conducted on 219 women who followed a two-year plant-based diet plan (Fiorito et al., 2021). The participants in the intervention arm were, on average, 0.41 years younger than their chronological age at the end of observation, although this effect may have been confounded by the increased physical activity in which some were required to engage.
Some individual compounds (polyphenols in particular) have recently gained a lot of attention from the anti-aging community due to their effects on the epigenome mediated by their interactions with DNMTs, histone modifiers, and miRNAs (Abdul et al., 2017). There are also strong indications that curcumin, a major component of curry spice, may promote favorable epigenetic changes (Benameur et al., 2021;Reuter et al., 2011). Other plant polyphenols, such as quercetin and pterostilbene, have also been associated with epigenetic remodeling (Arora et al., 2020;Busch et al., 2015). However, the effect of these compounds on DNMTs has not been quantified in the context of aging. Thus, the concept of an epigenetic rejuvenation diet is still rather vaguely understood.
The positive effect of polyphenols is considered to be partially caused by their interactions with sirtuins, a group of proteins involved in DNA repair, circadian rhythms, and stress response. In particular, SIRT1 is involved in cellular senescence and prevents telomere attrition (Arora et al., 2020). The green tea polyphenol epigallocatechin-3-gallate (EGCG) has been shown to attenuate systemic inflammation and extend rat lifespan by 8-12 weeks (Niu et al., 2013). Another common polyphenol, resveratrol-shown to inhibit senescence and boost SIRT1 expression-is found in berries, grapes, and peanuts. Finally, genistein, which naturally occurs in soybeans, is known to directly inhibit DNMT1, DNMT3A, and DNMT3A, and consequently affect the global methylation pattern.
Although some reports suggest that other foods, such as garlic, Brazil nuts, parsley, and coffee, can affect epigenetic aging, the epigenetics of nutrition at large remains an underexplored field (Lea et al., 2001;Xiang et al., 2008). In most cases, the rejuvenative potential of specific ingredients or diets has not been properly measured due to a large number of confounders. Hopefully, the widespread use of consumer epigenetic screening will bring more clarity regarding this matter.

Physical fitness and epigenetic aging
Physical fitness is one of the key factors of human longevity. Largescale studies have shown that people who perform well during endurance training have a mortality rate that is 3-5 times lower than that of the least fit individuals of the same age (Blair et al., 1989).
The self-evident benefits of physical training may also be interpreted as aging deceleration, which can be registered with aging clocks. As such, numerous studies have explored whether the epigenetic signatures of exercise resemble those associated with a slowed pace of aging. On a genomic scale, people with a lifelong history of physical activity display lower DNAm levels on gene promoters in muscle tissue (Sailani et al., 2019). Differential methylation occurs mostly in genes involved in the electron transport chain, insulin signaling, and oxidative stress resistance. In adipose tissue, however, physical activity has been reported to increase DNAm levels on gene promoters (Rönn et al., 2013). This discrepancy may be attributed to tissue differences, but it may also reflect different experimental designs. While the muscle study was cross-sectional and involved generally active or inactive subjects, the adipose study measured DNAm levels in generally inactive subjects who had undergone a six-month training program.
This example is illustrative of some of the issues encountered in specialized aging clock studies. First, the effect of exercise on the epigenome is not uniform across the body; thus, aging clocks may not be able to detect aging deceleration in all tissues. Second, the persistence of epigenetic changes in response to physical activity is greatly underexplored. The DNAm profile acquired after a six-month training regimen may dissipate within months of cessation. These research gaps have not been addressed and greatly limit our ability to predict the long-term impact of physical activity on the pace of aging.
The effects of exercise and weight-management via diet are welldocumented to reduce various measures of aging in longitudinal settings (Ho et al., 2022). For epigenetic clocks, however, few longitudinal observations are available. Although multiple studies highlight the potentially beneficial DNAm changes caused by exercise, epigenetic aging clocks commonly fail to register them as aging deceleration (Marioni et al., 2015;Sillanpää et al., 2019). In studies that detect a significant effect of exercise on aging, its magnitude greatly depends on the definition of epigenetic age (McCrory et al., 2020;Quach et al., 2017).
While exercise itself is arguably linked to the pace of aging, epigenetic aging shows a significant and reproducible association with another metric of physical fitness-body mass index (BMI). According to twin studies and studies in obese people, 10 units of BMI correspond to 1-3 years of epigenetic age across blood, liver, and adipose tissues (de Toro-Martín et al., 2019;Horvath et al., 2014;Lundgren et al., 2021). Interestingly, bariatric surgery is known to reduce epigenetic aging proportional to the drop in BMI (Fraszczyk et al., 2020). While BMI and physical activity both affect epigenetic age, the associations between adiposity and epigenetic aging are not affected by physical activity in most cases (Kresovich et al., 2021a). Thus, BMI and physical activity cannot be used interchangeably as indicators of fitness in the context of epigenetic aging.

Psychological stress and epigenetic aging
Epigenetic aging is not determined only by physical factors, such as dietary compounds or exercise intensity. Current and historical psychological states also significantly contribute to one's pace of aging.
Psychological stress has long been known to affect human longevity on a molecular level through the promotion of oxidative processes and telomere attrition (Epel et al., 2004). More recently, accumulated lifetime stress has also been shown to be imprinted on the epigenome (Zannas et al., 2015). Aberrant glucocorticoid signaling in people with high cumulative stress is hypothesized to alter the DNAm profiles of glucocorticoid response elements that coincide with the location of epigenetic aging signatures. As a result, lifetime stress may be responsible for up to 3.6 years of biological age difference between individuals. Although this study does not report any significant aging acceleration in association with childhood stress, other studies suggest that stress accumulated at any stage of life can have long-lasting effects on the epigenome. For example, traumatic stress experienced in adulthood has been associated with 2.0 additional years of epigenetic age among Dutch soldiers deployed in Afghanistan (Boks et al., 2015). Similarly, psychological stress during childhood is reported to influence the aging rate in later life (Brody et al., 2016). More specifically, children whose primary caregivers frequently display depressive symptoms are 1.8 years epigenetically older when they reach adulthood than children of non-depressed parents. In addition, directly being the victim of violence is also associated with higher epigenetic age in children (Jovanovic et al., 2017). Additionally, research has shown that interventions aimed at reducing stress such as mindfulness-based stress reduction can have a positive impact on epigenetic aging markers. For example, meditation has been shown to decrease the epigenetic age in a cumulative manner, with each year of practicing meditation being equivalent to a 0.24 year decrease in biological age (Chaix et al., 2017).
Studies in baboons hint that less chronic stress, as opposed to trauma, may also lead to accelerated aging (Anderson et al., 2021). Alpha-male baboons are, on average, one year older epigenetically than their chronological age, and upward movement along the male dominance hierarchy results in aging acceleration. Interestingly, dominance rank was not significantly associated with epigenetic age in female baboons. The authors suggest that high competitiveness at the top of the hierarchy and the associated changes in glucocorticoid levels may be the drivers of aging acceleration in alpha-male baboons (Anderson et al., 2021;Gesquiere et al., 2011).
According to studies featuring hematological and psychological aging clocks, low mental well-being may have an even more harmful effect on one's pace of aging than smoking (Galkin et al., 2022a(Galkin et al., , 2022b. The insights gained from applying aging clocks to the field of psychology highlight the importance of maintaining mental health to achieve healthy longevity.

Supplements and epigenetics
Most recent surveys show that people are generally skeptical about experimenting with extreme rejuvenation technologies (Barnett and Helphrey, 2021). In a cohort of 911 adult US citizens, only one-third would take a hypothetical pill that "enabled them to live forever at their current age." However, the desire to live a longer life in good health is more popular. Even if only every third person took a hypothetical longevity pill, longevity pharmaceuticals would remain a multi-trillion dollar market (Scott et al., 2021). For example, metformin is a potential geroprotector that has been calculated to extend the lifespan by 3-4 years when the course is started at 75 years of age (Scott et al., 2021;Wang et al., 2017). In turn, the monetary equivalent of this gain in lifespan is at least $700,000 based on the willingness-to-pay estimations.
Potential antiaging compounds such as metformin are thus an exciting opportunity for pharmaceutical companies. However, the emerging and established longevity supplements most commonly lack the evidence base for their geroprotective efficacy. In some cases, existing studies present conflicting and inconclusive statements about emerging geroprotective supplements. For example, in a recent preprint, metformin has been reported to decrease the epigenetic age of diabetic patients by 2.77 years (Man Li et al., 2021). However, an earlier study conducted in a cohort of healthy individuals did not show any signs of age deceleration due to metformin usage (Quach et al., 2017).
Other popular supplements, such as NAD+ boosters (NMN, NR, niacin, and nicotinamide) and polyphenol formulations, are also greatly underexamined in clinical trials on their effects on biological age, despite their apparent involvement in aging and epigenetic mechanisms (Nadeeshani et al., 2021;Soma and Lalam, 2022). Official clinical trials featuring aging clocks can be applied to monitor the changes in patients' aging rate and thus boost consumer confidence and the rate of adoption of antiaging therapies. In addition, the integration of aging clocks into the pharmaceutical industry practices may lead to the discovery of new classes of geroprotectors.

Challenges of epigenetic rejuvenation
While epigenetic aging clocks are a popular instrument of scientific research, they have yet to attain ubiquitary levels of use in healthcare settings. The main obstacle to their widespread use remains the Cross-sectional Better diet quality can reduce the epigenetic age by 0.8-1.5 years. The effect is observed with PhenoAge and GrimAge, but not clocks described in (Horvath, 2013) and (Hannum et al., 2013). Age deceleration is less significant in women who exercise more than 2.5 h per week. (Kresovich et al., 2022) " F. Galkin et al. prohibitively high cost of analysis. Most epigenetic clocks were trained on datasets obtained with Illumina Human Methylation 27 K and 450 K arrays (Galkin et al., 2020, p.). These platforms measure DNAm levels on approximately 27,000 and 450,000 CpG sites, respectively. A newer EPIC methylation platform was released in 2015, which has superior coverage of 850,000 CpG sites. Over time, the high-end EPIC array has become the industry standard, and the older arrays have been discontinued. Although this scale of analysis is desirable in research settings, consumer applications require just a tiny fraction of all available CpGs of any array. For example, Horvath's clock uses just 353 CpGs, while the available Illumina platform offers almost 2500 times more CpGs. Information from any extra probes used by a platform might be discarded with no impact on an aging clock.
Another major obstacle to the adoption of aging clocks is their variety. Each aging clock defines biological age in its own way using a distinct set of CpG sites. Consequently, the associations between biological age, health conditions, and dietary and other interventions may not align with one another (Bell et al., 2019). This uncertainty is a source of consumer and institutional skepticism toward epigenetic clocks. As is the case with some studies cited in this review, only one of the several tested aging clocks shows a significant association with an investigated factor (Klopack et al., 2022;Kresovich et al., 2021a;McCrory et al., 2020). All clocks are different in how they are trained: the characteristics of their primary domain (tissue, health conditions, age range) and algorithm (target definition, machine learning model) can limit their applications and the scope of problems they can be used to solve. Epigenetic clocks trained on blood samples exclusively may be applicable to other tissues via tissue-specific adjustments and pan-tissue clocks, such as Horvath's, are expected to perform well in multiple tissue types (Galkin et al., 2021;Hannum et al., 2013;Horvath, 2013). Nonetheless, they might struggle with picking up the aging signal in certain tissues and it is advised to use a solution that has been developed to work with a tissue in question . The exact definition of biological age used by an aging clock and defined via its target variable (chronological age, phenotypic age, time-to-death, pace of aging) also carries great influence on its ability to detect the effects of various factors on one's biological age . Thus, choosing an aging clock to fit a research question is an important step in the practical applications of epigenetic aging clocks.
On a more fundamental level, epigenetic clocks suffer from low interpretability. The biological significance of a certain methylation level at a subset of CpG sites is difficult to establish due to our inability to manipulate such sites in controlled settings. While the role of specific genes in aging pathways can be dissected using knockout organisms or by adjusting their activity with selective compounds, that of DNAm marks is much more difficult to define. As a result, using DNAm clocks to generate therapeutic antiaging targets is a challenging task.
Moreover, differentially methylated sites identified in epigenomewide studies of potentially geroprotective treatments may not colocalize with the few sites captured by an aging clock. In such cases, a promising compound with proven antiaging effects may have zero impact on epigenetic age. To avoid such "false negative" results, a new kind of epigenetic clock may be necessary. In standard machine learning approaches, important features are selected blindly without relying on a priori knowledge of their biological function. Perhaps a better alternative would be to train an aging clock based on the DNAm levels of the aging-associated genes.
These fundamental issues may be relieved by the development of multi-omic clocks that unite data sources of different biological origins: epigenetics, transcriptomics, and proteomics. Such clocks could highlight the causative links connecting these branches of molecular regulation and help us reach a new level of understanding regarding the aging process.

Conclusion
Different processes are involved in the maintenance of epigenetic states. In the last 10 years, multiple models have been created based on DNA methylation data. This review summarizes the general epigenetic recommendations that are available for individuals to decrease the pace of aging (Table 1). Furthermore, some of them demonstrate antiinflammatory, antioxidant, antiangiogenic, and anticancer properties that can potentially prolong the human lifespan.
Although the exact magnitude of the effects that different lifestyle components exert on epigenetic aging remains undetermined in most cases, the widespread adoption of aging clocks can bridge this knowledge gap and, ultimately, enable a new mode of healthcare decisionmaking to fight the problem of global population aging.

Declaration of Competing Interest
AZ and FG are employed at Deep Longevity, a wholly-owned subsidiary of Endurance Longevity (SEHK:0575. HK), a publicly traded company. DK was a Deep Longevity employee at the time the first draft was written.

Data Availability
No data was used for the research described in the article.