Epigenetic Aging: Biological Clocks, Cellular Senescence, and Age Reversal

Every cell in your body carries the same DNA sequence — the same three billion base pairs arranged in the same order. What determines whether a cell becomes a neuron, a liver hepatocyte, or a cardiomyocyte is not the sequence itself but the epigenome: the chemical modifications layered on top of DNA that control which genes are switched on or off. Chief among these modifications is DNA methylation, in which a methyl group is attached to cytosine nucleotides at specific locations — predominantly at sites where cytosine is followed by guanine, called CpG dinucleotides.

Epigenetic ageing is the gradual, systematic drift of these methylation patterns over time. Certain CpG sites become hypermethylated (more heavily tagged) with age, silencing genes involved in immune surveillance and tumour suppression. Others become hypomethylated, activating genes that drive inflammation and genomic instability. This is not random noise — the changes are highly reproducible across individuals of the same age, across tissues, and even across species. That reproducibility is what makes epigenetic methylation one of the most reliable molecular clocks in biology. For a broader view of how these changes fit into the full picture of cellular deterioration, see our guide on the twelve hallmarks of ageing.

Critically, epigenetic changes are not immutable. Unlike germline mutations, methylation marks are reversible in principle — enzymes called DNA methyltransferases (DNMTs) add them and TET enzymes remove them. This reversibility is what makes the epigenome such a compelling therapeutic target: if ageing-associated methylation drift can be halted or reset, a genuine biological rejuvenation becomes conceptually possible rather than merely a thought experiment.

An epigenetic clock is a machine-learning model trained on genome-wide methylation data from large cohorts. The algorithm selects a subset of CpG sites — typically hundreds out of the roughly 28 million in the human genome — whose methylation levels together produce the best prediction of a target outcome. The exact subset and their mathematical weights define the clock. The resulting score is expressed as an age in years, directly comparable to chronological age. For a detailed breakdown of this methodology, our article on how epigenetic clocks work covers the mechanics step by step.

Steve Horvath's 2013 clock — trained on 353 CpG sites across 51 tissue types — was a landmark achievement. It demonstrated that a single methylation signature could estimate age across radically different cell types with a mean absolute error of 3.6 years. However, it was trained purely to minimise chronological age prediction error, which means it measured the tick of a molecular clock without necessarily capturing health-relevant ageing.

Second-generation clocks corrected this limitation. Morgan Levine's PhenoAge (2018) was trained not on chronological age but on phenotypic age — a composite score derived from nine blood biomarkers (albumin, creatinine, glucose, CRP, lymphocyte percentage, mean corpuscular volume, red cell distribution width, alkaline phosphatase, and white blood cell count) that are themselves predictors of mortality. By training on this health-relevant composite, PhenoAge captures the dimension of ageing that matters clinically: how fast you are deteriorating relative to peers of the same birth year. Individuals whose PhenoAge runs ahead of their chronological age — so-called PhenoAge accelerators — show substantially elevated risk of chronic disease and death. GrimAge (2019) extended this further by training directly on time-to-death and time-to-first-disease-diagnosis data from the Framingham Heart Study, making it the most mortality-predictive clock currently available.

The most recent development is DunedinPACE (2022), which measures not static biological age but the current pace of ageing — how many biological years are accumulating per calendar year. A DunedinPACE of 0.8 means you are ageing at 80% of the population average; a score of 1.2 means you are ageing 20% faster. This longitudinal framing makes DunedinPACE particularly useful for clinical trials where researchers need to detect changes in the rate of ageing within months rather than years.

Cellular senescence is a state in which a cell permanently exits the cell cycle and ceases to divide. It is triggered by a range of insults — critically short telomeres, double-strand DNA breaks, oxidative stress, oncogene activation, and chemotherapy agents. In young organisms, senescence serves an important purpose: it prevents damaged cells from proliferating uncontrollably and becoming cancerous, and transiently senescent cells release signals that recruit immune cells to clear them. This is the beneficial, acute face of senescence.

The problem emerges when immune clearance fails, as happens progressively with age. Senescent cells that are not cleared accumulate in tissues, where they continue to secrete a complex cocktail of inflammatory cytokines, proteases, and growth factors known as the senescence-associated secretory phenotype, or SASP. The SASP drives chronic low-grade inflammation, degrades the extracellular matrix that provides structural support to tissues, and — in a particularly damaging feedback loop — induces neighbouring healthy cells to enter senescence themselves. A 2023 analysis estimated that senescent cells constitute roughly 10–15% of cells in the tissues of 70-year-old humans, up from less than 1% in young adults. The accumulation is not uniform: adipose tissue, liver, kidney, and lung are especially susceptible. To understand how telomere shortening feeds into this process, our guide on telomeres, telomerase, and ageing covers the relevant biology.

Senolytics are drugs that selectively kill senescent cells by targeting the pro-survival pathways these cells depend on to avoid apoptosis (programmed cell death). The most studied combination is dasatinib (a BCR-ABL tyrosine kinase inhibitor repurposed from oncology) plus quercetin (a flavonoid). In mouse models, intermittent senolytic dosing — a few days on, weeks off — has extended median lifespan by 17–36% and improved physical function, organ health, and cognitive performance. Early human data from the Mayo Clinic (2019, EBioMedicine) showed that three weeks of dasatinib plus quercetin reduced senescent cell markers in adipose tissue and skin of patients with diabetic kidney disease. More recent trials involving navitoclax (a BCL-2 family inhibitor) have shown promising senescent cell clearance results in pulmonary fibrosis, though thrombocytopenia limits its tolerability.

Nicotinamide adenine dinucleotide (NAD+) is a coenzyme present in every living cell, functioning as an electron carrier in energy metabolism and as a substrate for a class of enzymes — sirtuins, PARPs, and CD38 — that are central to DNA repair, gene expression regulation, and mitochondrial biogenesis. NAD+ levels decline by approximately 50% between young adulthood and age 60 in most tissues, and this decline is causally implicated in several hallmarks of ageing: impaired mitochondrial function, reduced DNA repair capacity, accumulation of damaged proteins, and dysregulated gene expression.

The two major NAD+ precursor supplements under clinical investigation are nicotinamide mononucleotide (NMN) and nicotinamide riboside (NR). Both are converted to NAD+ via distinct salvage pathways. A landmark 2023 placebo-controlled trial published in Nature Aging found that NMN supplementation (300 mg/day) significantly elevated blood NAD+ levels and improved muscle insulin sensitivity in prediabetic men aged 50–70 — replicating earlier findings in mouse models. A large 2022 NR trial (n=80) showed NAD+ restoration in peripheral blood mononuclear cells at 1,000 mg/day but modest functional effects, suggesting tissue-specific barriers to precursor uptake that oral supplementation alone may not fully overcome. The longevity evidence base for NAD+ precursors is detailed extensively in our guide on NAD+ and ageing science.

Sirtuin 1 (SIRT1) deserves particular attention. It is the best-studied of the seven mammalian sirtuins and functions as an epigenetic regulator — deacetylating histones and transcription factors to influence gene expression at hundreds of loci. Critically, SIRT1 activity is NAD+-dependent: as NAD+ declines with age, SIRT1 becomes less active, and the epigenetic regulation it maintains drifts. This is one proposed mechanism linking NAD+ depletion to epigenetic ageing. David Sinclair's information theory of ageing frames epigenetic drift as the central cause of biological ageing, driven in part by SIRT1 being recruited to repair DNA breaks and abandoning its epigenetic maintenance role — a model still contested but highly influential in the longevity research community.

Rapamycin — a macrolide originally derived from a soil bacterium on Easter Island — is the most robust pharmacological longevity intervention identified in mammals to date. It functions by allosterically inhibiting mTORC1 (mechanistic target of rapamycin complex 1), a master kinase that integrates nutrient availability, growth factor signalling, and energy status to coordinate anabolic processes including protein synthesis and cell growth. Inhibiting mTORC1 mimics the molecular response to caloric restriction: autophagy is upregulated, protein translation is reduced, and cells shift from a growth mode to a maintenance and repair mode.

The longevity evidence is unusually strong for a single drug. Rapamycin extended median lifespan by 14% in female mice and 9% in male mice even when treatment began at the equivalent of age 60 in humans — the first drug shown to extend lifespan in normally-aged mammals rather than disease models. In subsequent studies using intermittent dosing protocols (once weekly rather than daily), lifespan extension reached 23–26% with substantially fewer immunosuppressive side effects. Epigenetic clock analysis of rapamycin-treated mice showed reductions in biological age of 7–12 weeks after 3 months of treatment, a striking finding given the drug's primary mechanism is mTOR inhibition rather than direct methylation targeting. Our dedicated guide on rapamycin as a longevity drug covers the human evidence, dosing data, and the risk-benefit considerations in clinical practice.

Human data remain limited — rapamycin carries real immunosuppressive risk at transplant doses — but the Dog Aging Project has demonstrated epigenetic age reduction in companion dogs treated with intermittent low-dose rapamycin (0.05 mg/kg once weekly), providing mammalian translational data directly applicable to longevity medicine. Several physician-guided longevity clinics now offer off-label intermittent rapamycin protocols for healthy adults, though this practice remains controversial and sits outside any approved indication.

The most ambitious approach to epigenetic age reversal is partial cellular reprogramming — the use of Yamanaka factors to transiently reset a cell's epigenetic state toward a more youthful configuration. Shinya Yamanaka's 2006 Nobel Prize-winning discovery showed that four transcription factors (Oct4, Sox2, Klf4, and c-Myc, collectively OSKM) could convert any adult somatic cell into an induced pluripotent stem cell (iPSC) — in effect erasing its differentiated identity entirely. Full reprogramming is obviously dangerous in a living organism: it would erase tissue identity and cause teratomas.

The key insight driving the reprogramming longevity field is that the epigenetic reset precedes the loss of cell identity during OSKM expression. By using transient, intermittent, or partial factor expression — typically OSK without c-Myc to reduce oncogenic risk — researchers can achieve epigenetic rejuvenation without inducing full dedifferentiation. A 2020 Nature paper from David Sinclair's lab at Harvard showed that AAV-delivered OSK expression in retinal ganglion cells of mice with optic nerve injuries restored youthful methylation patterns and regenerated functional nerve fibres — a result previously considered biologically impossible. The mice regained significant visual function. This work, and the broader reprogramming landscape, is the scientific foundation of Altos Labs, which has assembled an unprecedented team of reprogramming researchers including Morgan Levine, Shinya Yamanaka, and Juan Carlos Izpisua Belmonte to pursue systemic reprogramming in humans. The current state of this work is covered in our article on Morgan Levine, PhenoAge, and Altos Labs in 2026.

Safety remains the central challenge. Any intervention that activates Oct4 in non-laboratory settings carries carcinogenic risk, and delivery mechanisms — primarily adeno-associated viruses — have their own immunogenicity and cargo-size limitations. The field is actively exploring mRNA-based transient delivery and small-molecule proxies that activate reprogramming pathways without viral vectors. Timeline estimates for clinical applications range from 10 to 25 years depending on the optimism of the forecaster, but the field has progressed faster than most observers expected in 2020.

A separate but intersecting line of longevity research concerns the circulating factors in blood that change with age and may themselves drive or slow epigenetic ageing. Heterochronic parabiosis experiments — in which the circulatory systems of young and old mice are surgically joined — showed that old mice paired with young partners exhibit measurable rejuvenation of muscle, liver, and brain tissue, while young mice paired with old partners show accelerated deterioration. This indicated that blood carries pro-ageing and pro-youth signals that can override local tissue programming.

GDF11 (growth differentiation factor 11) was initially identified as a candidate pro-youth circulating factor before subsequent replication failures complicated the picture. TGF-beta superfamily members, factors in the complement system, and various inflammatory cytokines collectively create a pro-ageing systemic environment as organisms age. Plasma dilution — replacing a portion of aged plasma with saline plus albumin — has been shown to reduce inflammatory markers and improve cognitive function in aged mice (Villeda lab, Nature Medicine, 2022), and the Alkahest AMBAR trial showed cognitive stabilisation in Alzheimer's patients receiving plasma exchange protocols. Clinical trial data from 2026 plasma exchange and senolytic studies are covered in our dedicated article on plasma exchange and senolytics: 2026 trial results.

For individuals seeking to monitor their epigenetic age longitudinally, the current gold standard is a commercial DNA methylation assay. TruDiagnostic (TruAge Complete) uses the Illumina EPIC array to measure over 850,000 CpG sites and returns results scored against multiple clocks including GrimAge, PhenoAge, and DunedinPACE — the pace of ageing metric. Elysium Health's Index test employs a validated composite of three clocks. Both require a blood spot sample sent by mail and cost $299–$499. Given assay-to-assay measurement variability of approximately 1–2 years, retesting at 6–12 month intervals provides the most meaningful longitudinal data.

Standard blood biomarker panels provide cheaper intermediate tracking points. The nine biomarkers underpinning PhenoAge — albumin, creatinine, glucose, hs-CRP, lymphocyte percentage, mean corpuscular volume, red cell distribution width, alkaline phosphatase, and white blood cell count — can be extracted from a comprehensive metabolic panel plus a complete blood count, which most laboratories offer for under $50. Computing a PhenoAge score from these values is straightforward using the published formula (Levine et al., 2018, EBioMedicine).

Wearable metrics — resting heart rate, heart rate variability, and estimated VO2 max from Apple Watch, Garmin, or Oura Ring — serve as daily proxies for epigenetic age trajectory. A resting heart rate decreasing over months signals improving cardiorespiratory fitness; a rising morning HRV trend reflects reduced sympathetic nervous system load. These readouts will not replace methylation assays, but they provide continuous feedback that annual or biannual assays cannot. Pairing wearable tracking with periodic methylation testing creates a layered monitoring approach with high temporal resolution.