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, and the safe route is indirect. People who inherit longer telomeres get fewer degenerative diseases but more cancers,[5] with no clear net effect on lifespan,[6] so "longer is better" is false. Lifestyle seems to protect telomeres mainly by lowering chronic inflammation, not by acting on them directly (observational evidence only).
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.[7] Shorter telomeres correlate with cardiovascular disease, dementia, central obesity, and all-cause mortality in pooled cohort data. But the genetics cut both ways: genetically longer telomeres protect against some degenerative diseases but 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 — a few randomised trials have now slowed methylation clocks slightly, but a clock is a stand-in measure, not a health 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.[8][9][10][11] They are not unchallenged, though. Measuring the same blood sample twice can give ages several years apart on the older clocks (Horvath, Hannum, PhenoAge), while GrimAge gives much more consistent repeat readings in the lab,[12] as does DunedinPACE.[13] And in one US cohort of adults over 50, a score built from seven inflammation blood measures predicted four-year mortality slightly better than the six clocks it was compared with.[14]
Targeted by: sustained caloric restriction. In the CALERIE trial, two years of restriction slowed the DunedinPACE pace-of-aging clock by a few percent, but people managed only about 12% restriction against a 25% target, the clock analysis was done after the fact, and two other clocks (PhenoAge and GrimAge) showed no significant change.[15] Later analyses of other trials found similarly small slowing: 1 g of omega-3 a day over three years in Swiss adults over 70, where vitamin D and a simple home exercise programme moved no clock on their own,[16] and a daily multivitamin, which slowed two of five clocks over two years.[17] Every other lever has only weak direct evidence: time-restricted and intermittent fasting; structured exercise, which is linked to younger clock readings in observational studies but has not been shown to slow them on its own in a trial; a methyl-donor-adequate diet (folate, B12, choline); and cutting added sugar (one cross-sectional study of 342 midlife women tied higher added-sugar intake to a small acceleration of the GrimAge2 clock, an association its authors call for longitudinal studies to confirm).[18] 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 mouse study 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. In mammals only one small mouse study, never repeated and with short-lived control animals, claims longer life,[19] and a protein-folding drug did not extend life in a large US government mouse-testing programme.[20] 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 real success is in transthyretin amyloidosis: two drugs (tafamidis and acoramidis) hold a single misfolding blood protein in its proper shape, and a third (vutrisiran) stops the body making it. Tafamidis and vutrisiran both improved outcomes in company-funded trials.[21][22] A second silencer, eplontersen, missed its main goal in the largest trial yet.[23] The drugs that work do so because there is one known culprit protein in one defined disease, which is the opposite of a general fix for aging. See Loss of proteostasis.
5. Disabled macroautophagy (added 2023)
Descriptive: Strong. Intervention: Moderate for exercise, the one lever with human muscle-biopsy data; Weak for fasting and for every drug and supplement; Caution on chronic rapamycin.
Autophagy is the cell's protein-and-organelle recycling program. It is thought to slow with age (shown in animals; the few direct human measurements are mixed), leaving cells burdened by damaged mitochondria, oxidised proteins, and aggregated cargo that no longer get cleared.[24] 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 rapamycin, which directly inhibits the mTOR growth pathway and switches autophagy on in animals, but it has not yet shown a benefit in healthy adults (see #6). Autophagy can so far be measured only in cells taken out of the body, mostly blood cells, so no test can show whether it is working better in a person's tissues.
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 — no trial has measured lifespan or deaths, a 1,000-person trial of another drug that blocks the same pathway did not reduce respiratory illness,[25] and both placebo-controlled trials of weekly rapamycin missed their main goal (details below). 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 female 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)[26], which has not yet delivered in people: 48 weeks of a compounded low-dose version, which reached only about a third of the blood levels of standard pharmacy rapamycin (sirolimus), did not change visceral fat,[27] and 13 weeks of standard rapamycin in 40 adults aged 65–85 did not add to the benefit of home exercise, may have slightly reduced it, and caused more side-effect events than placebo;[28] metformin (activates the same energy sensor; the early hint that people with diabetes taking metformin outlived people without diabetes did not hold up when it was re-tested in Danish national registers);[29] 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 two of its authors work for TruDiagnostic, a commercial epigenetic-testing company, whose scientific adviser is the first author; it has since been peer-reviewed and published, which retires the preprint caveat but none of the design ones[30]. It is also not the first intervention to move a clock: caloric restriction slowed DunedinPACE, though not two other clocks, in the CALERIE trial of healthy, non-obese adults aged 21–50 (see #3).[31]
- 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.[32] 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 — zone 2 training and high-intensity intervals alike, acting through the cell's energy sensor and the master switch for mitochondrial biogenesis. Across hundreds of human training studies both raised mitochondrial content by about the same amount, with total training load (how much times how hard) as the predictor.[33] 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). When the randomised trials in older adults were pooled, these did not improve muscle mass, grip strength, walking speed or chair-stand time, and in people with mild cognitive impairment NR was linked to slightly worse physical performance.[34]
8. Cellular senescence
Descriptive: Strong. Intervention: Strong in mice (senolytics); Weak in humans, with a reason for Caution — the flagship randomised knee-osteoarthritis trial missed, and in an uncontrolled study of 19 healthy adults, six months of dasatinib + quercetin pushed several DNA-methylation clocks the wrong way and lowered a methylation estimate of telomere length (the clocks best linked to health outcomes did not move).[35]
"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.[36] 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, and the first small study to show senescent cells being cleared in people later needed a correction.[37] A 20-week randomised trial of dasatinib plus quercetin in 60 postmenopausal women did not slow bone breakdown; a bone-building marker rose only briefly, and a gain in bone density appeared only in an exploratory look at the third of women with the most senescent cells.[38] In a 19-person study in healthy adults with no comparison group, six months of the same combination sped up several epigenetic-aging clocks and lowered a methylation estimate of telomere length.[39] 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: resistance training, which keeps muscle stem cells working; regular, unbroken sleep, which in mice protects the bone-marrow stem cells that make blood;[40] and moderate calorie restriction, which in animals helps keep stem cells in a protective resting state. Stem-cell therapy is still experimental. The best trial so far gave 148 frail older adults one infusion of donor bone-marrow stem cells; nine months later, people on the highest dose could walk about 63 metres further in six minutes than the placebo group, but the main six-month comparison missed and only one lower dose also helped. That is a single, company-funded result on a walking test, not proof that the therapy slows aging.[41] Caution on stem-cell clinics: unapproved stem-cell injections are sold by thousands 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; the young-blood experiments are in animals.
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.
The closest human test is a small trial of plasma exchange, which dilutes old-blood factors rather than adding young ones. It enrolled 44 healthy adults over 50, assigned treatments by order of sign-up rather than true randomisation, and was run by authors linked to a plasma-exchange company. Clock readings improved against a sham procedure at the mid-point but not by the end, and no health outcomes were measured.[42] Plasma exchange sold for "rejuvenation" is running ahead of this evidence.
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 is closely linked to atherosclerosis, osteoarthritis, insulin resistance and neurodegeneration, and the biology of how it could drive them is well worked out.[43] It is still a mechanistic model, though: no trial has followed the chain from start to finish, and most of the intervention evidence shows lifestyle changes lowering inflammation markers, not lower markers preventing these diseases. 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, though the details vary from study to study: more potentially harmful species and fewer of the bacteria that make short-chain fatty acids, such as Faecalibacterium and the Lachnospiraceae family.[44] Not every helpful species declines: in one Italian study, the barrier-supporting Akkermansia was more common in people aged 105 to 109 than in younger adults.[45] In old mice these shifts make the gut lining leakier, letting bacterial products into the blood and exacerbating inflammaging;[46] in people the leak-and-inflame link shows up mainly in the frail or ill. 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 — extends lifespan in mice; both placebo-controlled trials of weekly rapamycin so far missed their main goal |
| 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, profiled by the clock's vendor; caloric restriction did so first in CALERIE. Wider hallmark claims for this class are not supported |
| Senolytics | Cellular senescence |
| NAD⁺ precursors | Mitochondrial — mechanistic; pooled trials in older adults show no gain in muscle or physical function |
Causality-enriched epigenetic clocks: DamAge and AdaptAge
A 2024 Nature Aging paper introduced two new metrics[47]:
- 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[48]. 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. Choosing DNA sites with genetic "cause-finding" methods also does not by itself make a clock reliable from one measurement to the next. In a 2026 study of 18 clocks, the sister clock from the same paper (CausAge) was among the less stable: when the analysis was re-run on repeat measurements of the same samples, its estimated effect of a diet change sometimes reversed direction. The study's authors develop a competing clock.[49] DamAge and AdaptAge themselves have not been tested this way; for now they are research tools, not personal tests.
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 require lab work and come with their own measurement caveats. In large cohorts the newer, outcome-trained clocks predict future disease and death better than the original clocks trained only on calendar age.[50] No clock has yet been shown to work as a stand-in for real outcomes, though: nobody has shown that lowering your clock score lowers your risk, so a "younger" result after a supplement or protocol is not evidence that it helps.[51][52] 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 analysis of about 1,060 healthy adults built an "autonomic age" from 29 heart and blood-pressure measures and found that people with a high-risk cardiovascular profile scored about 11 years older than their calendar age, and low-risk people about a year younger.[53] Treat that as preliminary: it is a single cross-sectional study, not an established clock, and its measures were taken in a lab, not by a wearable. 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 somewhat arbitrary and incomplete and warning that it quietly smuggles in the contested molecular-damage theory of aging.[54]
- 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.[55] A deeper unresolved problem is telling drivers of aging from passengers: many hallmarks are strongly age-associated without being demonstrated causes.[56]
- 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.[57]
- The evidence is not evenly strong across the twelve. The third test, that treating a hallmark slows aging, comes closest to being met in mice, for example rapamycin for nutrient sensing and senolytic drugs for senescence. Even there, a critical review of the original nine hallmarks points out that most supporting studies use lifespan as a stand-in for aging and rarely treat young animals as well as old, so they cannot tell "slowed aging" apart from "fixed a problem that would improve at any age"; in one large mouse study, only some of rapamycin's benefits looked like slowed aging.[58] For others (dysbiosis, altered intercellular communication), the evidence is still largely correlative. Treat the list as a gradient of confidence, not twelve equally proven levers.
Related and competing frameworks
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.[59] The inverse framing, the Hallmarks of Health, describes the features that actively maintain physiology rather than the ways it fails.[60] Organ-specific hallmark sets (cardiovascular, brain) also exist,[61][62] and in mice the age-related changes in gene activity do not run in step across organs.[63] Aging is uneven inside one human body too: research models that estimate the age of individual organs from blood proteins found that nearly one adult in five had a single organ aging much faster than the rest, and that people with more "old" organs were more likely to die during follow-up.[64][65] That is a good reason to distrust any test that sums you up as one "biological age". These are observational findings, and several of the researchers hold equity in a company developing the method.
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.
Further reading
- López-Otín C et al. The Hallmarks of Aging. Cell 2013.[66]
- López-Otín C et al. Hallmarks of aging: An expanding universe. Cell 2023.[67]
- Horvath S. DNA methylation age of human tissues and cell types. Genome Biol 2013.[68]
- Lu AT et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging 2019;11:303–327.[69]
- Belsky DW et al. DunedinPACE — a DNA methylation biomarker of the pace of aging. eLife 2022.[70]
- Ying K et al. Causality-enriched epigenetic age uncouples damage and adaptation (DamAge / AdaptAge). Nature Aging 2024.[71]
- Mannick JB, Lamming DW. Targeting the biology of aging with mTOR inhibitors. Nature Aging 2023.[72]
- Chiu DT, et al. Essential Nutrients, Added Sugar Intake, and Epigenetic Age in Midlife Black and White Women: NIMHD Social Epigenomics Program. JAMA Netw Open 2024.[73]
- Waziry R, Belsky DW et al. Effect of long-term caloric restriction on DNA methylation measures of biological aging (CALERIE trial). Nature Aging 2023.[74]
- Gems D, de Magalhães JP. The hoverfly and the wasp: a critique of the hallmarks of aging as a paradigm. Ageing Research Reviews 2021.[75]
- de Magalhães JP. Distinguishing between driver and passenger mechanisms of aging. Nature Genetics 2024.[76]
- Gladyshev VN et al. Disagreement on foundational principles of biological aging. PNAS Nexus 2024.[77]
- Kennedy BK, Berger SL, Brunet A et al. Geroscience: linking aging to chronic disease (the seven pillars). Cell 2014.[78]
- López-Otín C, Kroemer G. Hallmarks of Health. Cell 2021.[79]