The twelve hallmarks of aging

A scaffold for understanding why interventions work. Twelve interconnected biological processes — DNA damage, telomere shortening, mitochondrial dysfunction, cellular senescence, and others — track with aging and are widely treated as its drivers, though which of them actually cause aging rather than accompany it is unsettled. Every credible longevity intervention claims to target at least one of them. The framework is the most useful single map for evaluating whether something "anti-aging" is operating on real biology or just marketing.

The hallmarks-of-aging framework, introduced by López-Otín and colleagues in Cell in 2013 and substantially expanded in 2023, is the dominant biological framework for understanding the molecular and cellular processes that drive aging.^[1]^[2] The current version proposes twelve interconnected hallmarks, organised into three categories: five primary (root causes — accumulated damage), three antagonistic (responses to that damage, initially protective but deleterious when sustained), and four integrative (systemic consequences that emerge once the damage can no longer be compensated).^[3] The framework is a descriptive organising scaffold rather than a validated causal theory, and parts of it are actively debated (see Criticisms and open questions).

Why this framework matters

The hallmarks framework is the theoretical basis for nearly every "anti-aging" intervention — supplements, drugs, lifestyle protocols. Understanding them lets you evaluate whether a proposed intervention actually targets aging biology, or just markets itself that way.

To qualify as a hallmark, a process must:

Manifest naturally during physiological aging
Experimentally accelerate aging when exaggerated
Slow aging and extend healthspan when targeted

How to read the evidence ratings below

Each hallmark carries two separate judgements, because they come apart badly. Descriptive is how well established it is that the process happens and tracks with aging. Intervention is how well established it is that doing something about it slows aging — the framework's own third criterion. For several hallmarks the first is Strong and the second is Weak or absent, and the gap between them is the single most important thing on this page.

Primary hallmarks: root causes

1. Genomic instability

Descriptive: Strong. Intervention: Weak — no human trial shows that reducing DNA damage slows aging.

Accumulation of DNA damage from environmental toxins, ultraviolet (UV) radiation, and endogenous oxidative stress. As the cell's DNA-repair systems themselves become error-prone with age, somatic mutations accumulate, compromising cellular function and driving oncogenesis.^[4] The cells with the highest mutation burden are also the cells most likely to become senescent or transformed. Crucially, the damage doesn't stay local: DNA that leaks into the cytoplasm — from ruptured micronuclei or damaged mitochondria — is picked up by an innate-immune sensor as if it were a virus, driving the chronic inflammation (inflammaging, #11) that turns a per-cell problem into a systemic one. See Genomic instability for the full picture.

Targeted by: an antioxidant-rich Mediterranean-pattern diet (Dietary patterns), reduced exposure to documented environmental toxins (Environmental toxins), sufficient sleep (both DNA repair and the brain's overnight waste-clearance system peak during sleep), and exercise, where the cohort signal on biological-aging clocks is consistent in direction though the optimal dose is not established.

2. Telomere attrition

Descriptive: Strong. Intervention: Caution — the relationship is U-shaped, and genetically longer telomeres causally raise cancer risk.

Telomeres are protective nucleoprotein caps on chromosome ends. They progressively shorten with each cell division — the so-called "end-replication problem" — and, when critically short, trigger replicative senescence or apoptosis.^[5] Shorter telomeres correlate with cardiovascular disease, dementia, central obesity, and all-cause mortality in pooled cohort data. But the relationship is U-shaped, not linear: genetically longer telomeres causally raise cancer risk, so "longer is better" is false and consumer telomere tests are too noisy to act on — the genetic evidence is set out under Telomere attrition. See Telomere attrition for the full picture.

Targeted by: regular cardiovascular exercise (associated with longer telomeres across cohorts, though a pooled meta-analysis found the overall effect non-significant), chronic-stress reduction, adequate sleep, Mediterranean-pattern dietary intake. Pharmacological telomerase reactivation is a speculative target with substantial cancer-risk concerns.

3. Epigenetic alterations

Descriptive: Strong. Intervention: Moderate — caloric restriction slowed a methylation clock in a randomised trial, but a clock is a surrogate, not an outcome.

Progressive degradation of the cellular "software" — DNA methylation patterns, histone modifications, chromatin organisation. Cells lose their precise molecular identities; tissues drift from their original transcriptional programs. The marks scramble in two directions at once — a global loss of silencing alongside over-silencing of specific protective genes — and, because they are reversible, this is the most intervenable primary hallmark. See Epigenetic alterations for the full picture.

Epigenetic clocks — Horvath, GrimAge, DunedinPACE, DamAge / AdaptAge — read these methylation patterns to estimate biological age, and they are the most developed human biomarkers of aging — though not unchallenged: the same blood sample measured twice can return noticeably different ages, and a composite of seven inflammatory markers out-predicted thirteen of these clocks for four-year mortality.^[6]^[7]^[8]^[9]

Targeted by: structured exercise, time-restricted and intermittent fasting protocols, sustained caloric restriction — which slowed the DunedinPACE pace-of-aging clock in the randomised CALERIE trial, though participants achieved roughly 12% restriction against a 25% target, the slowing was a few percent, and it did not appear consistently on the other clocks, methyl-donor-adequate diet (folate, B12, choline), reductions in added-sugar intake (a recent NIMHD analysis tied added-sugar intake to dose-response acceleration of two of these clocks in midlife women).^[10] Partial cellular reprogramming — briefly switching on the four genes that can reset a cell to an embryonic-like state — is an experimental frontier with major safety concerns.

4. Loss of proteostasis

Descriptive: Moderate — the human data are largely cultured cells, the decline looks biphasic rather than steady, and one tissue survey found no universal decline at all. Intervention: none proven in mammals — the widest mechanism-to-lever gap of the twelve.

Failure of protein quality control — the molecular chaperone systems, autophagy, and the ubiquitin-proteasome pathway. Misfolded protein aggregates accumulate, and they are the defining lesion of Alzheimer's disease, Parkinson's disease, and the systemic amyloidoses. Aggregate burden and clinical disease are dissociable, though — by age 90 a large minority of cognitively normal people carry amyloid, and the antibodies that clear it move the plaque far more than they move the patient. See Loss of proteostasis for the full picture — including why the human evidence is thinner than the mechanism's elegance suggests.

Targeted by: nothing proven, and this is the hallmark with the widest gap between mechanism and available lever. Boosting protein quality control extends lifespan in worms and flies, but no mammalian experiment has done so. The commonly cited levers — heat-shock activation via sauna, spermidine, overnight brain waste clearance in deep sleep — rest on mechanistic inference rather than human outcome data, and the supplement trials aimed here have missed. The one drug class that genuinely corrects a folding defect — tafamidis and its two successors, for transthyretin amyloidosis — works because it targets a single known protein in one defined disease, which is the opposite of a general proteostasis intervention. See Loss of proteostasis.

5. Disabled macroautophagy (added 2023)

Descriptive: Strong. Intervention: Moderate in animals, Weak in humans.

Autophagy is the cell's protein-and-organelle recycling program. It slows with age, leaving cells burdened by damaged mitochondria, oxidised proteins, and aggregated cargo that no longer get cleared.^[11] The 2023 expansion added disabled autophagy as a hallmark in its own right — classed as a primary one, because declining autophagy is the loss of a maintenance function (a form of damage, specifically to organelle quality control) rather than a compensatory response. It links nutrient sensing, mitochondrial quality, and proteostasis, and its decline feeds neurodegeneration, heart failure, sarcopenia, and — via failed mitochondrial clearance and innate-immune activation — inflammaging. See Disabled macroautophagy for the full picture.

Targeted by: caloric restriction and time-restricted eating (Fasting), endurance exercise, spermidine (a polyamine that induces autophagy and extends lifespan in animal models — though the trial the field wanted, 100 older adults over 12 months, improved neither memory nor the biomarkers), and direct inhibition of the mTOR growth pathway with rapamycin.

Antagonistic hallmarks: responses to damage that turn deleterious

6. Deregulated nutrient sensing

Descriptive: Strong. Intervention: Strong in mice (rapamycin extends lifespan reproducibly); Weak in humans, where no trial has tested a hard endpoint — and the target is oscillation, not maximum suppression.

The most heavily targeted hallmark across the entire field. The growth-signalling network built around mTOR (mechanistic target of rapamycin) and the opposing energy-sensing network built around AMP-activated protein kinase and the sirtuins become miscalibrated; constant nutrient surplus keeps the growth side switched on, driving growth while suppressing repair. The mechanism unifies metabolic flexibility, cellular senescence, autophagy, and mitochondrial biogenesis. The goal is oscillation between growth and repair signalling, not permanent suppression of growth: too little anabolic signalling brings frailty, muscle loss, and impaired immunity, and in 960 genetically diverse mice the deepest restriction extended lifespan most but cost lean mass and shifted the immune repertoire — with genetics influencing lifespan more than diet did. The four fuel gauges — insulin and insulin-like growth factor 1, mTOR, AMP-activated protein kinase, and the sirtuins — and the growth-versus-repair logic are laid out under Deregulated nutrient sensing; see also Protein and Metabolic flexibility for the dietary detail.

Targeted by:

Caloric restriction, fasting, time-restricted eating
Exercise (the most reliable AMPK activator)
Pharmacology: rapamycin (a direct inhibitor of the mTOR growth pathway)^[12]; metformin (activates the same energy sensor, with a consistent observational signal in type 2 diabetes); GLP-1 receptor agonists, which slowed several DNA-methylation clocks (see Ozempic-class drugs) — though that finding is a post-hoc, exploratory analysis of a 32-week phase 2b trial in a narrow special population (adults with HIV-associated fat redistribution; semaglutide n=45, placebo n=39), and it remains a non-peer-reviewed preprint with epigenetic profiling performed by a commercial clock vendor^[13]. It is also not the first intervention to move a clock: caloric restriction slowed DunedinPACE in the general-population CALERIE trial two years earlier^[14]
Reduced chronic insulin and growth-factor stimulation through diet quality and body-composition optimisation

7. Mitochondrial dysfunction

Descriptive: Strong. Intervention: Moderate — exercise reliably improves mitochondrial function; no human lifespan data.

Mitochondria accumulate mitochondrial-DNA mutations and structural damage. ATP production falls; free radicals and leaked mitochondrial fragments are read by the cell's innate-immune sensors as danger signals, fuelling inflammaging.^[15] Aged tissue accumulates enlarged, dysfunctional mitochondria that the cell can no longer split off and recycle. The key practical principle is mitohormesis: brief, controlled mitochondrial stress (exercise, fasting, heat, cold) triggers adaptive renewal — which is why those levers work and high-dose antioxidants don't. See Mitochondrial dysfunction for the full picture.

Targeted by: aerobic exercise — particularly zone 2 training, the most potent stimulus for building new mitochondria, acting through the cell's energy sensor and the master switch for mitochondrial biogenesis — plus high-intensity intervals (which preferentially drive mitochondrial quality rather than quantity). Pharmacological adjuncts: urolithin A (a gut-bacterial metabolite that induces mitophagy; in 66 older adults it missed both primary endpoints, and the muscle-endurance signal at two months had gone by four) and the NAD⁺ precursors nicotinamide mononucleotide (NMN) and nicotinamide riboside (NR), where human longevity evidence is preliminary.

8. Cellular senescence

Descriptive: Strong. Intervention: Strong in mice (senolytics); Caution in humans — the flagship randomised knee-osteoarthritis trial missed, and six months of dasatinib + quercetin in healthy adults accelerated epigenetic clocks and shortened telomeres.

"Zombie cells" — irreversibly growth-arrested but metabolically active, secreting a continuous mix of inflammatory cytokines, chemokines, and matrix-degrading proteases known as the senescence-associated secretory phenotype (SASP). Useful in youth as a tumour-suppressive and wound-healing mechanism; deleterious when senescent cells accumulate in aged tissues.^[16] See Cellular senescence for the full picture.

Targeted by: exercise and dietary patterns (modest effect via reduced inflammatory drive), fasting (some senolytic signal in animal models). Senolytics — drugs that selectively kill senescent cells, most notably dasatinib plus quercetin and fisetin — show striking efficacy in mice. The human record is much weaker and includes a harm signal: the flagship randomised knee-osteoarthritis trial missed its goals, the first-in-human study needed a corrigendum, and in a 19-person study in healthy adults six months of dasatinib plus quercetin accelerated several epigenetic-aging clocks and shortened a methylation estimate of telomere length. See Cellular senescence — there is a concrete reason healthy, non-frail adults should not self-administer these.

Integrative hallmarks: systemic consequences

9. Stem-cell exhaustion

Descriptive: Strong. Intervention: Weak — exercise preserves satellite-cell pools; stem-cell therapies remain experimental.

Progressive depletion or functional decline of tissue-regenerating stem-cell pools — including muscle satellite cells, intestinal crypt stem cells, and blood-forming stem cells. Whether adult brain neurogenesis meaningfully persists at all is genuinely contested, with post-mortem studies flatly disagreeing, so treat that particular pool as unsettled. Tissues progressively lose the capacity to regenerate or to replace damaged cells. See Stem-cell exhaustion for the full picture.

Targeted by: exercise (preserves muscle satellite-cell pools and improves their responsiveness), hormonal sufficiency where clinically indicated, and nutrient adequacy. Caution on stem-cell clinics: unapproved stem-cell injections are sold by hundreds of clinics and carry a documented harm record — hundreds of people injured, deaths, and several cases of blindness from intraocular injections. See Stem-cell exhaustion.

10. Altered intercellular communication

Descriptive: Moderate. Intervention: Weak — largely correlative, and the parabiosis work is animal.

The hormonal, immune, neuroendocrine, and paracrine signalling networks that coordinate tissues break down with age. Cells stop hearing each other accurately — senescent-cell secretions add inflammatory noise, the physical channels between cells (gap junctions, the extracellular matrix) degrade, and the blood itself accumulates pro-aging factors that experiments joining the circulations of young and old animals show actively drive decline. See Altered intercellular communication for the full picture.

Targeted by: hormonal optimisation when indicated — menopausal hormone therapy initiated within the critical window, testosterone therapy in symptomatic hypogonadism, thyroid management; reduced systemic inflammation; and — perhaps surprisingly — social engagement, since social signals modulate the immune-endocrine system in measurable ways (Purpose).

11. Chronic inflammation ("inflammaging") (added 2023)

Descriptive: Strong. Intervention: Moderate — lifestyle levers lower inflammatory markers; the step to slowed aging is inference.

A sterile, low-grade, persistent inflammatory state pervades aging tissues — fuelled by senescent cells, leaked mitochondrial DNA, gut-barrier breakdown, and persistent low-level antigen exposure. Inflammaging directly drives atherosclerosis, osteoarthritis, insulin resistance, and neurodegeneration.^[17] See Chronic inflammation for the full picture.

Targeted by: Mediterranean-pattern diet (Dietary patterns), regular exercise, adequate and consistent sleep, reduced visceral adiposity, and long-chain omega-3 intake. Fasting belongs on this list only indirectly — a 12-month trial found neither time-restricted eating nor standard calorie restriction significantly changed inflammatory cytokines, and the anti-inflammatory effect appears to ride on fat loss rather than on the schedule.

12. Dysbiosis (added 2023)

Descriptive: Moderate. Intervention: Weak — among the least causally established of the twelve.

An age-related shift in the gut microbiome — loss of diversity, overgrowth of pathogenic taxa, and depletion of short-chain-fatty-acid producers such as Faecalibacterium and Roseburia, alongside Akkermansia muciniphila, which supports the gut barrier rather than producing butyrate. The resulting increased intestinal permeability allows endotoxin leakage, exacerbating inflammaging through the gut-immune axis. Centenarian guts suggest the goal is functional resilience — sustained short-chain-fatty-acid production and suppression of troublesome species rather than a fixed "young" species list. See Dysbiosis for the full picture.

Targeted by: high-fibre diet, fermented foods (the food-matrix effect is more important than live colony counts), prebiotic vegetables, judicious use of antibiotics, and possibly targeted probiotic strains for specific conditions.

How interventions map to hallmarks

Intervention	Hallmarks targeted
Mediterranean / MIND diet	Inflammation, dysbiosis, epigenetic alterations, genomic instability
Exercise (aerobic)	Mitochondrial dysfunction, telomere attrition, inflammation, autophagy
Resistance training	Stem cell exhaustion (muscle), intercellular communication
Fasting / time-restricted eating	Nutrient sensing, autophagy, epigenetic alterations, mitochondrial — mechanism robust, but no human lifespan trial, and the CALERIE clock result belongs to sustained caloric restriction, not to fasting
Sleep adequacy	Genomic instability (DNA repair), proteostasis (overnight brain waste clearance), inflammation
Sauna	Proteostasis (heat-shock response), inflammation — mechanistic
Cold exposure	Mitochondrial (modest) — mechanistic; no human longevity outcome data
Rapamycin	Nutrient sensing, autophagy
Metformin	Nutrient sensing, inflammation
GLP-1 drugs (semaglutide, tirzepatide)	Nutrient sensing, inflammaging — slowed DNA-methylation clocks, but a post-hoc result in a special population and still an unpublished preprint; caloric restriction did so first in CALERIE. Wider hallmark claims for this class are not supported
Senolytics	Cellular senescence
NAD⁺ precursors	Mitochondrial — mechanistic; limited human evidence

Causality-enriched epigenetic clocks: DamAge and AdaptAge

A 2024 Nature Aging paper introduced two new metrics^[18]:

DamAge — tracks methylation changes causally linked to accelerating aging (irreversible damage)
AdaptAge — tracks protective methylation changes that retard aging (compensatory adaptation)

These are causality-enriched clocks: the methylation sites they read were selected using genetic-instrument methods that favour changes appearing to drive rather than merely accompany aging^[19]. That construction invites a reframing of aging not as passive decay but as a dynamic struggle between damage and adaptive resilience — and the further suggestion that interventions be judged not just on halting damage but on upregulating adaptive resilience. That inference is a hypothesis, not a demonstrated result: it has not been shown that raising AdaptAge-type marks extends healthspan.

It offers one candidate explanation for why some interventions (exercise, dietary patterns) may work despite modest individual effects on traditional biomarkers — if they upregulate adaptation — but this remains speculative.

A practical biological-aging proxy: heart rate variability

Epigenetic clocks are powerful but require lab work and bring their own measurement caveats. A more accessible — and continuously trackable — proxy is heart rate variability (HRV), the millisecond fluctuation between heartbeats. HRV correlates with biological aging via the autonomic nervous system, the vagal anti-inflammatory pathway, and downstream cardiometabolic resilience. A 2025 machine-learning analysis of about 1,000 adults derived an "Autonomic Age" from 29 cardiovascular features and reported that high-risk cardiovascular profiles ran roughly a decade older than calendar age, while optimal profiles ran a couple of years younger^[20]. Treat that as preliminary: it is a non-peer-reviewed preprint derived in a single modest sample, not an established clock. See Heart rate variability for the full picture, including what consumer wearables actually measure accurately.

What the framework does and doesn't tell you

What it tells you:

The biology of aging is mechanistically complex, not random decay.
Multiple interventions target multiple hallmarks simultaneously — explaining why pattern-based approaches (Mediterranean diet, exercise) outperform single-molecule interventions.
Pharmacological interventions are real but not ready for healthy adults absent specific indications.

What it doesn't tell you:

That any specific supplement or drug currently extends human life.
That the hormetic lifestyle levers do more than engage the biology. The mechanisms are real, but fasting and cold exposure have no human longevity-outcome trials — fasting also carries safety caveats (an observational cardiovascular-mortality signal for very short eating windows; a refeeding-proliferation concern), and cold's popular metabolic and anti-inflammatory claims largely don't survive scrutiny.
That there's a single "master switch" for aging.
That the field has solved aging.

The honest message: the framework is sophisticated; the interventions matched to it (in healthy adults, for hard outcomes) remain modest at best.

Criticisms and open questions

The hallmarks are the field's dominant organising framework, but they are a descriptive scaffold, not a settled causal theory — and that distinction is genuinely contested in the primary literature.

The hallmarks describe, they don't explain. Gems and de Magalhães argue the scheme borrows its authority from the hallmarks of cancer without earning it, calling the aging version arbitrary and incomplete and warning that it quietly smuggles in the contested molecular-damage theory of aging.^[21]
Damage versus program. A vocal minority holds that aging is not damage accumulation at all but the harmful continuation of developmental and growth programs (chiefly mTOR-driven) — the "hyperfunction" view, under which some hallmarks are consequences rather than causes.^[22] A deeper unresolved problem is telling drivers of aging from passengers: many hallmarks are strongly age-associated without being demonstrated causes.^[23]
No expert consensus. A survey of dozens of aging biologists found no majority view on what aging even is, what causes it, when it begins, or what counts as rejuvenation — a useful reminder that the framework's tidiness overstates the field's agreement.^[24]
The evidence is not evenly strong across the twelve. The third criterion — that targeting a hallmark slows or reverses aging — is met robustly for some (rapamycin on nutrient sensing, senolytics on senescence, both in mice) and weakly or not at all for others (dysbiosis, altered intercellular communication), where the evidence is largely correlative. Treat the list as a gradient of confidence, not twelve equally proven levers.

The twelve hallmarks are not the only map. The seven pillars of geroscience cover much of the same ground but foreground the interconnection between processes rather than cataloguing them.^[25] The inverse framing, the Hallmarks of Health, describes the features that actively maintain physiology rather than the ways it fails.^[26] Organ-specific hallmark sets (cardiovascular, brain) also exist, and the hallmarks do not appear to switch on synchronously across tissues — a caution against treating "biological age" as a single number.

A note on the framework authors' interests

Commercial interests are worth naming on all sides of this field, including the framework's own authors. Per the 2023 paper's declaration of interests, several hold direct equity in companies built around specific hallmarks: Maria Blasco founded a telomere-measurement company, Manuel Serrano is a shareholder and adviser to senescence- and reprogramming-focused firms, and Guido Kroemer co-founded several biotech companies and holds patents on targeting aging. Carlos López-Otín and Linda Partridge declared none. This is context, not grounds to discount the framework.