Longevity science
The hallmarks of aging, and the three tests that decide the list
The hallmarks of aging grew from nine entries to twelve over 10 years, between the 2013 paper and its 2023 update. Almost nobody reproduces the criteria behind that list, and the criteria are the part that lets you judge whether a longevity claim is serious.
What do the hallmarks of aging actually claim?
That aging can be organized into a short list of processes, which stood at nine entries for 10 years, from the 2013 paper to its 2023 update. That first paper proposed the bewildering biology could be sorted this way.[1] The original entries were:
- genomic instability, telomere attrition and epigenetic alterations
- loss of proteostasis and deregulated nutrient sensing
- mitochondrial dysfunction and the buildup of senescent cells
- stem cell exhaustion and altered intercellular communication
The paper has since become one of the most cited in the biology of aging, which tells you how badly the field wanted an organizing structure.
What is often lost is that the framework is a proposal about how to think, not an inventory of facts. It was built to be argued with, and it has been.
What does a process need to count as a hallmark?
Strong human evidence It has to pass three tests, and those criteria are the intellectual core of the whole thing. A candidate process must pass all three:
- It must manifest with age.
- Experimentally aggravating it must accelerate aging.
- Experimentally ameliorating it must slow aging or extend healthy lifespan.[2]
Notice how demanding the third condition is. Plenty of things change with age, and rather fewer accelerate aging when you make them worse.
Very few have ever been improved in a way that measurably extended healthy lifespan in an animal. That is why the list is twelve items rather than two hundred, and it is a ready-made filter for any supplement claim you encounter.
Why did the list grow from nine to twelve?
Because the 2023 update added disabled macroautophagy, chronic inflammation and dysbiosis, bringing the count to twelve.[2] None of the three was new as a concept. Each had simply accumulated enough evidence over a decade to satisfy the criteria that had excluded it earlier.
- Disabled macroautophagy was separated from proteostasis because its decline turned out to deserve its own line.
- Chronic inflammation, sometimes called inflammaging, earned inclusion on the strength of a large intervention literature.
- Dysbiosis arrived with the microbiome field.
A framework that revises itself on evidence is behaving correctly.
Show the numbers as a table
| Measure | Value |
|---|---|
| 2013 framework | 9 hallmarks |
| 2023 update | 12 hallmarks |
What are the 14 hallmarks of aging?
There is no such published list. The two peer-reviewed frameworks are the nine of 2013 and the twelve of 2023. Meeting reports have floated further candidates, including mechanical properties of tissue and splicing dysregulation, and none of these has become canonical.
Where the number fourteen usually appears is on commercial pages that need a slightly bigger list to hang more products on. If a page presents fourteen as settled without naming the paper, that is the tell. The closest thing to an official waiting list is the discussion in the 2023 update, which names candidates without adding them.
How do the hallmarks connect to each other?
In a loop rather than a line: mitochondrial dysfunction feeds inflammation, inflammation accelerates senescence, and senescent cells release signals that damage neighbors and deplete stem cell pools.[4] The framework groups them into primary damage, antagonistic responses to that damage, and integrative consequences that produce the phenotype.
The consequence for anyone reading intervention claims is direct. Because everything is connected to everything, almost any compound can be described as targeting a hallmark. The description is nearly free; the demonstration is not.
What are the 12 hallmarks of aging?
They are twelve processes that the 2023 update sorts into three tiers.[2] Primary hallmarks are sources of damage, antagonistic hallmarks are responses that help at first and harm when chronic, and integrative hallmarks are where the accumulated damage becomes the aging you can see.
- Primary (damage): genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis and disabled macroautophagy.
- Antagonistic (responses): deregulated nutrient sensing, mitochondrial dysfunction and cellular senescence.
- Integrative (phenotype): stem cell exhaustion, altered intercellular communication, chronic inflammation and dysbiosis.
The grouping is a reading aid rather than a law. The authors describe it as a way to order causes and consequences, and they are explicit that the tiers feed back into one another. The sections below take each hallmark in turn, asking the same three questions every time.
- What is the process, in plain terms?
- What experiment connects it to aging rather than just to old age?
- Is there anything, drug or behavior, that has improved it in a living animal?
The numbering below follows the order of the 2023 paper and is not a ranking. The authors present the hallmarks as a connected system, so a lower number does not mean a more important target.[2]
Where does the damage of aging start?
In the five primary hallmarks: genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis and disabled macroautophagy. Each is unambiguously harmful, with no phase in which it helps the organism, and each accumulates over time.[2] They are also the group with the fewest practical interventions.
1. Genomic instability
Every cell faces a steady stream of DNA damage from outside, such as radiation and chemicals, and from inside, such as replication errors and reactive oxygen species. Repair systems fix most of it, and what they miss accumulates as mutations, chromosome losses and structural damage to the nucleus.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Somatic mutations accumulate in human and animal tissues, and blood cells carrying the same mutation expand into clones as people get older. |
| Aggravating it | Inherited defects in DNA repair produce human progeroid syndromes, and mice engineered with similar defects age prematurely. |
| Ameliorating it | The evidence is thinner, coming mainly from genetically modified mice with strengthened chromosome safeguards that stayed healthier for longer. |
Damage to the nuclear lamina, the scaffold lining the nucleus, belongs here too. A faulty lamin protein causes Hutchinson-Gilford progeria, the best known accelerated aging disease in children, and the original paper used it as a model of what architectural damage does.[1]
No supplement or drug has been shown to reduce genomic instability in a way that extends healthy life. Products that claim to "protect your DNA" are describing an antioxidant effect in a dish, which is the first criterion at most.
2. Telomere attrition
Telomeres are protective caps at chromosome ends, and in most adult cells they shorten slightly every time a cell divides because the enzyme that rebuilds them, telomerase, is largely switched off. When a telomere becomes critically short, the cell treats it as broken DNA and stops dividing.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Telomere shortening with age is seen in humans and mice. |
| Aggravating it | Mice lacking telomerase show premature tissue decline, and human diseases of telomerase deficiency cause early failure of regenerative tissues such as bone marrow and lung. |
| Ameliorating it | The original paper cites telomerase reactivation in adult mice delaying aging without an increase in cancer, a result not yet turned into a human therapy. |
The catch is that short telomeres also act as a brake on runaway cell division, so the same process that ages tissue protects against tumors. That tension is why telomerase activation is treated with caution rather than enthusiasm in the review literature.
For an individual, a telomere test gives a number with no established action attached to it. The hallmark is real, but the consumer product built on it answers a question nobody can yet use.
3. Epigenetic alterations
Epigenetic marks are the chemical tags on DNA and on the histone proteins it wraps around, and they decide which genes a cell reads. With age the pattern drifts: methylation shifts at specific sites, histone marks change and chromatin loosens or tightens in the wrong places.[2]
| Test | What the evidence shows |
|---|---|
| Manifests with age | The drift is regular enough that Horvath built a clock from methylation at a set of sites that estimates age across many tissues.[6] |
| Aggravating it | Disrupting chromatin regulators in model organisms shortens lifespan and produces aging features. |
| Ameliorating it | The 2023 update discusses partial reprogramming, briefly switching on cell identity factors, which reversed aging features in mice. |
This is the hallmark with the most commercial activity right now, mostly through biological age tests. Those clocks track the drift well across populations, but individual readings are noisy, and technical variation alone can move a result by years.[8]
Reprogramming is the most striking animal result in the whole framework and also the furthest from use. Pushing cells too far toward an embryonic state causes tumors in mice, which is the practical reason no human therapy exists.
| Hallmark | Best evidence for the third test | Status in people |
|---|---|---|
| Genomic instability | Mice with strengthened chromosome safeguards | No intervention shown |
| Telomere attrition | Telomerase reactivation in adult mice | No therapy; tests not actionable |
| Epigenetic alterations | Partial reprogramming in mice | Experimental; tumor risk |
| Loss of proteostasis | Boosted chaperones in worms and flies | No intervention shown |
| Disabled macroautophagy | Enhanced autophagy genes in mice | Inducers unproven in people |
4. Loss of proteostasis
Proteostasis is the machinery that keeps proteins correctly folded and clears the ones that are not. Chaperone proteins guide folding and the proteasome breaks down damaged proteins, and both systems lose capacity with age.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Misfolded and aggregated proteins accumulate, the pattern behind Alzheimer disease, Parkinson disease and cataracts. |
| Aggravating it | Impairing chaperones or the proteasome in worms, flies and mice accelerates age-related decline. |
| Ameliorating it | Boosting chaperone activity or proteasome function extends lifespan in worms and flies, with more limited mouse data. |
The lesson from this hallmark is that the aggregate is the visible end of a process that started much earlier. Clearing the visible deposit is not the same as restoring the machinery, which is the gap between criterion one and criterion three in practice.
5. Disabled macroautophagy
Macroautophagy is the cell's bulk recycling system: it wraps worn-out components, including whole damaged mitochondria, in a membrane and delivers them for digestion.[2] In 2013 it sat inside proteostasis, and the 2023 update gave it a line of its own because it handles far more than proteins.[2]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Autophagy activity declines in aging tissues across species. |
| Aggravating it | Deleting autophagy genes in mice produces neurodegeneration and other aging features. |
| Ameliorating it | Genetically enhancing autophagy extends lifespan in mice, and spermidine, which induces autophagy, extended mouse lifespan and improved heart function.[7] |
Rapamycin also belongs in this section, because blocking mTORC1 releases a brake on autophagy.[1] One drug touching two hallmarks at once is exactly the interconnection the framework warns about.
The spermidine mouse work shows how an autophagy claim should be built. Spermidine extended lifespan and protected the aging heart, and the cardiac benefit did not appear in mice whose heart muscle cells could not carry out autophagy.[7] That kind of dependency test is what separates a mechanism from a correlation.
Which hallmarks start out protective and turn harmful?
The three antagonistic hallmarks: deregulated nutrient sensing, mitochondrial dysfunction and cellular senescence. At low levels each is protective, a way of limiting harm or saving resources, and each becomes damaging when it runs too long or too hard.[2] That dual nature is why blunt attempts to switch them off tend to disappoint.
6. Deregulated nutrient sensing
Cells track how much fuel is available through a small set of pathways: insulin and IGF-1 signaling, mTOR, AMPK and the sirtuins. When nutrients are plentiful these pathways push growth, and when nutrients are scarce they shift the cell toward maintenance and repair.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Signaling through these pathways becomes poorly matched to the body's needs. |
| Aggravating it | Constantly high growth signaling is associated with shorter life in model organisms. |
| Ameliorating it | Dampening insulin and IGF-1 signaling, dietary restriction and mTOR inhibition extend lifespan in yeast, worms, flies and mice. |
This is the strongest hallmark by the third criterion.
- Rapamycin extended lifespan in genetically diverse mice across three independent sites even when started late in life.[3]
There is a paradox the original paper names directly. Growth signaling naturally falls with age, yet lowering it further extends life, which the authors read as the body's own defensive response that helps up to a point.[1]
Metformin is usually filed here too, through AMPK. The case for testing it rests on observational data and the TAME trial design, and TAME has published no efficacy results.[9]
7. Mitochondrial dysfunction
Mitochondria produce most of the cell's usable energy, and with age they become less efficient, accumulate damage to their own DNA and leak more reactive molecules. The older free radical theory held that this leakage was the core of aging.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Respiratory efficiency falls and mitochondrial DNA mutations accumulate in aging tissues. |
| Aggravating it | Mice engineered to make errors copying mitochondrial DNA age prematurely. |
| Ameliorating it | Mild mitochondrial stress can extend lifespan in worms, a phenomenon called mitohormesis, while broad antioxidant supplements have not extended life. |
That last point is the reason this hallmark is classed as antagonistic. Reactive oxygen species turned out to be signals as well as damage, so a supplement that mops them up can blunt a useful stress response rather than slow aging.
NAD, a molecule central to energy metabolism, falls with age, and NAD precursors are sold on that basis. Precursors reliably raise NAD levels in blood, yet human trials have not shown that raising it delivers the benefits the biology implies.[5]
| Hallmark | Best evidence for the third test | Status in people |
|---|---|---|
| Deregulated nutrient sensing | mTOR inhibition and dietary restriction across species | Small trials, no lifespan data |
| Mitochondrial dysfunction | Mild mitochondrial stress in worms | Precursors raise NAD, benefit unshown |
| Cellular senescence | Clearing senescent cells in old mice | Early pilot in kidney disease |
8. Cellular senescence
A senescent cell has stopped dividing permanently, usually in response to damage, but does not die. It then secretes a mix of inflammatory signals, the senescence-associated secretory phenotype or SASP, that can damage surrounding tissue.[4]
Much of that picture of the SASP traces to the lab of Judith Campisi at the Buck Institute, and she is a co-author of the 2019 consensus review that set out how senescent cells should be identified.[4]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Senescent cells accumulate in many tissues with age. |
| Aggravating it | Transplanting senescent cells into young mice caused physical dysfunction. |
| Ameliorating it | Clearing senescent cells with dasatinib plus quercetin improved physical function and extended remaining lifespan in old mice.[10] |
Detecting senescent cells is harder than the marketing implies. A consensus review recommended combining several markers, such as p16INK4a and senescence-associated beta-galactosidase, because no single marker identifies a senescent cell reliably.[4]
- In old mice, intermittent dasatinib plus quercetin increased remaining median lifespan by 36% (per Xu and colleagues, 2018).[10]
- The same study transplanted small numbers of senescent cells into young mice and found they were enough to cause measurable physical dysfunction.[10]
Senescence is protective in the short run: it stops damaged cells from becoming cancer and helps wounds heal. The harm comes from accumulation, when clearance by the immune system falls behind production.[4]
Human evidence is early. Fisetin, sold widely as a senolytic, did not extend lifespan in the NIA Interventions Testing Program.[11]
- That split between a promising mechanism and a disappointing product is the pattern of the whole antagonistic group.
- Our guide to cellular senescence goes deeper into the SASP and the trials.
Which hallmarks make aging visible?
The four integrative hallmarks: stem cell exhaustion, altered intercellular communication, chronic inflammation and dysbiosis. Here damage and failing responses add up to what a doctor would notice: tissues stop regenerating and inflammation settles in.[2] They are the hallmarks most tied to frailty.
9. Stem cell exhaustion
Adult stem cells replace the cells that tissues lose to wear, from blood to gut lining to muscle. With age their numbers and, more importantly, their function decline, so repair slows and some tissues simply thin out.[1]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Blood-forming stem cells lose output and shift toward producing certain immune cells over others, and muscle stem cells respond poorly to injury. |
| Aggravating it | Forcing stem cells to divide too often exhausts the pool early, the other side of too little division. |
| Ameliorating it | The original paper cites rapamycin improving the function of blood stem cells in old mice, and exposure to young blood factors rejuvenating old stem cells in parabiosis experiments. |
The balance matters here. Too little stem cell activity leaves tissue unrepaired, and too much drains the reserve and raises cancer risk, which is why the review treats this as an integrative outcome of the earlier hallmarks rather than a standalone target.[1]
Commercial stem cell clinics that offer injections for aging are not applying this research. Nothing in the framework supports infusing cells as an anti-aging treatment, and the animal results that exist concern restoring the function of cells already in place.
10. Altered intercellular communication
Cells coordinate through hormones, nerve signals and a constant exchange of molecules in blood and tissue fluid. With age that signaling degrades, and damage in one tissue can spread its effects to others through the circulation.[2]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Hormonal and neuroendocrine signals change, and blood from old animals carries factors that impair young tissues. |
| Aggravating it | The SASP from senescent cells can push nearby healthy cells into senescence, a spreading effect sometimes called bystander senescence. |
| Ameliorating it | In parabiosis experiments, sharing circulation with a young animal improved several tissues in an old one. |
This hallmark explains why aging looks systemic rather than organ by organ. It also explains the commercial interest in young plasma, an idea that has run well ahead of any evidence that infusions help people.
| Hallmark | Best evidence for the third test | Status in people |
|---|---|---|
| Stem cell exhaustion | Young blood factors in parabiosis | No intervention shown |
| Altered intercellular communication | Shared young circulation in parabiosis | Plasma infusions unproven |
| Chronic inflammation | Blocking inflammatory pathways in mice | No healthspan benefit shown |
| Dysbiosis | Microbe transfer in fish and progeroid mice | No probiotic proven |
11. Chronic inflammation
Inflammation is the immune system's short-term response to injury and infection. With age a low-grade, persistent version develops without an obvious trigger, a state often called inflammaging, and the 2023 update gave it its own place on the list.[2]
| Test | What the evidence shows |
|---|---|
| Manifests with age | Circulating inflammatory markers rise and the immune system becomes less precise, responding poorly to new threats while staying chronically activated. |
| Aggravating it | Sustained inflammatory signaling accelerates aging features in mice. |
| Ameliorating it | Blocking specific inflammatory pathways in mice has improved healthspan, and the 2023 update cites this intervention literature as the reason for inclusion. |
Senescent cells are a major source of that chronic signal, which is one of the clearest examples of the hallmarks feeding each other.[4] Mitochondrial damage and a changed gut microbiome add to it.
Anti-inflammatory supplements are sold heavily on this hallmark. The honest status is that general anti-inflammatory effects in a cell or a blood marker have not been shown to extend healthy life, and suppressing inflammation broadly carries infection risk.
12. Dysbiosis
Dysbiosis means a disturbed community of gut microbes. The composition and diversity of the gut microbiome change with age, and the 2023 update added it as the newest hallmark on the strength of transfer experiments.[2]
| Test | What the evidence shows |
|---|---|
| Manifests with age | The microbiome of older adults differs from that of younger adults, and very long-lived people show distinct profiles. |
| Aggravating it | Disturbed microbial communities are linked to inflammation and to a leakier gut barrier. |
| Ameliorating it | The 2023 update cites transplanting microbes from young donors, which extended lifespan in short-lived fish and in mice with premature aging. |
This is the hallmark where the gap between the animal result and the shelf is widest. Probiotic products routinely claim longevity effects, and no probiotic has been shown to meet the third criterion in people.
Dysbiosis also illustrates why the list is revisable. A decade earlier the microbiome field did not have the intervention data to qualify, and it earned a place the same way the other two additions did.
Which hallmarks do current interventions actually target?
Mainly four: nutrient sensing, senescence, mitochondrial function and autophagy, and the first has the drug with the most consistent animal lifespan record.[3]
Nutrient sensing
Deregulated nutrient sensing is the one with a real drug attached: rapamycin inhibits mTORC1 directly, and it remains the compound with the most consistent lifespan extension record in animals, including mice started in late life, which lived 9% to 14% longer (per Harrison et al. 2009).[3] Our explainer on how mTOR inhibition works covers that pathway.
Senescence, mitochondria and autophagy
- Cellular senescence has senolytics, an area with strong mouse data and thin human data, which we cover in the senolytic evidence review.
- Mitochondrial dysfunction and altered communication are the pitch for NAD precursors, where NAD levels do fall with age and the supplement case is far weaker than the biology implies.[5]
- Disabled macroautophagy is where spermidine sits, with real animal work and a serious human absorption problem.[7]
Show the numbers as a table
| When | Event | Detail |
|---|---|---|
| 2009 | Rapamycin in mice | 9% to 14% longer life |
| 2013 | Nine hallmarks | Original framework |
| 2016 | Spermidine study | Autophagy, mouse data |
| 2019 | Senescence review | Consensus on SASP |
| 2023 | Twelve hallmarks | Three added |
What touches the most hallmarks at once?
Exercise. Reviews have argued that it attenuates most of the hallmarks simultaneously, through separate mechanisms rather than a single pathway: mitochondrial biogenesis, autophagy induction, lower chronic inflammation, better proteostasis, preserved stem cell function.
Why no compound compares
No compound comes close, and it is worth being blunt about why that comparison is uneven. Exercise has been studied for decades in enormous populations with hard outcomes, while the supplements have mechanistic plausibility and mouse data.
Those are different tiers of evidence, and the hallmarks framework, used properly, makes the difference visible rather than hiding it.
How do the hallmarks map onto diseases of aging?
Through shared drivers: the framework's quiet ambition is to explain why so many diseases rise together with age. If a handful of processes drive aging, then heart disease, dementia and cancer should share them, and the reviews argue that they do.[2]
- Neurodegeneration: loss of proteostasis and disabled autophagy sit behind the protein aggregates of Alzheimer and Parkinson disease.
- Heart disease: spermidine's mouse benefits included better heart function, tying autophagy to cardiac aging.[7]
- Tissue fibrosis and frailty: senescent cells and their secretions damage the surrounding tissue.[4]
- Loss of immune function: chronic inflammation and stem cell exhaustion leave older adults with weaker responses to vaccines and infection.
Cancer is the awkward case
Cancer shares several hallmarks with aging, including genomic instability and epigenetic change, yet some aging hallmarks act against it. Senescence and telomere shortening both stop damaged cells from dividing, so switching them off could raise cancer risk.[1]
That is why the reviews describe aging and cancer as partly shared and partly opposed processes. It is also why any intervention that promises to reverse a protective hallmark deserves a question about tumors.
Why this matters for prevention
The geroscience argument follows from the shared mechanisms. Treating one disease at a time leaves the common drivers untouched, while slowing a hallmark could in principle delay several diseases at once, which is the premise the TAME trial was designed to test.[9]
That premise has not yet been demonstrated in people. No drug has been shown to delay age-related diseases as a group in a human trial, and TAME itself has not reported.[9]
What was added, what was split, and what is still waiting
Reading the 2013 and 2023 papers side by side shows the framework is conservative. Every one of the original nine survived, and the only structural change was pulling autophagy out of proteostasis.[1,2]
- Kept: all nine 2013 hallmarks, with updated evidence under each.
- Split out: disabled macroautophagy, formerly part of loss of proteostasis.
- Added: chronic inflammation, which in 2013 sat inside altered intercellular communication, and dysbiosis, which was new.
- Discussed but not added: candidates such as altered mechanical properties of tissues and dysregulated RNA splicing.
What changed inside the original nine
The update did more than add three names. Each surviving hallmark was rewritten around a decade of new work, and the emphasis shifted in ways that matter for anyone reading intervention claims.[2]
| Hallmark | Where the 2023 update puts the weight |
|---|---|
| Genomic instability | Somatic mutations and clonal expansions in aging tissues, plus damage to nuclear architecture |
| Telomere attrition | Telomere dysfunction in human disease, with telomerase activation still experimental |
| Epigenetic alterations | Epigenetic clocks as measures of aging and partial reprogramming as an experimental reversal |
| Loss of proteostasis | Chaperones and the proteasome, after autophagy moved to its own line |
| Deregulated nutrient sensing | Insulin signaling, mTOR, AMPK and sirtuins treated as one interacting network |
| Mitochondrial dysfunction | Mitophagy and mitohormesis, with reactive oxygen species treated as signals as well as damage |
| Cellular senescence | Drugs that kill senescent cells and drugs that quiet their secretions |
| Stem cell exhaustion | Restoring stem cell function through circulating factors and reprogramming |
| Altered intercellular communication | Inflammation separated out, leaving hormonal, neural and circulating signals |
The 2023 paper carries the subtitle "an expanding universe," a signal that its authors expect more revisions. Candidates on the waiting list have met the first criterion and parts of the second, and what they lack is the intervention evidence the third one demands.[2]
Where is the framework contested?
On overlap, uneven criteria and cause versus consequence. The framework is widely cited and not beyond criticism, and the most common objections come from inside the biology of aging rather than from outside it.
Hallmarks overlap
Several hallmarks describe the same events from different angles. Senescent cells drive chronic inflammation, and damaged mitochondria feed both, so the tidy count of twelve hides a single tangled network.[4] The 2023 authors acknowledge this interdependence directly.
The criteria are uneven in practice
The third criterion is often met in worms, flies or engineered mice rather than in normal mammals or people. A hallmark can therefore qualify on evidence that says little about what would work in a human, which is why a hallmark's presence on the list is not proof that targeting it will help anyone.
Cause versus consequence
The framework does not claim to identify the root cause of aging, and some researchers argue that no single root cause exists. That humility is a feature, but it means the list cannot tell you which hallmark to target first or whether fixing one will move the others.
Reading the original papers
Both papers appeared in the journal Cell, and both are long, dense reviews rather than experiments. The 2013 paper is the cleaner read for the original structure, and the 2023 update is the one to use for anything current.[1,2]
- Read the introduction of each paper for the three criteria in the authors' own words.
- Use the figures, which lay out the tiers, before tackling the text.
- Treat each section's intervention discussion as a list of animal results, not a list of treatments.
Summaries built on these papers, including commercial ones, tend to keep the list and drop the criteria. Going to the source is the fastest way to see how cautious the authors were about what the evidence supports.
Three worked examples of the third test
The criteria are easiest to understand by running real compounds through them. One compound below passes the third test in animals and stalls in people, another stalls before it reaches people, and the third moves a marker without showing an outcome.
Rapamycin: passes in mice, unproven in people
| Test | What the evidence shows |
|---|---|
| Criterion one | The mTOR pathway is part of the nutrient sensing that the framework says becomes deregulated with age.[1] |
| Criterion two | Persistently high growth signaling is associated with shorter life in model organisms.[1] |
| Criterion three | The Interventions Testing Program began feeding rapamycin at 600 days of age, and lifespan at the 90th percentile rose 14% in females and 9% in males (per Harrison et al. 2009).[3] |
The 600-day start is roughly equivalent to a 60-year-old human, and the effect held across three independent test sites.[3] In people the picture changes.
- PEARL, a 48-week placebo-controlled trial of weekly rapamycin in healthy adults, missed its primary outcome of visceral fat (per Moel et al. 2025).[12]
- Participants took 5 or 10 mg of compounded rapamycin weekly, and lean mass and pain improved only in women on the higher dose (per Moel and colleagues, 2025).[12]
- The PEARL authors were employed by a telehealth company that prescribes the drug, a conflict the paper discloses.
So rapamycin sits exactly where the framework predicts a serious candidate should: a strong animal case for the third criterion and no human lifespan data. The human trials are small, short and mostly aimed at safety and markers.
Fisetin: a single-lab result that did not replicate
Fisetin, a plant flavonoid, entered the senolytic market on the strength of earlier single-lab mouse studies reporting that it extended health and lifespan.[11] On paper that looked like the third criterion met for cellular senescence.
| Test | What the evidence shows |
|---|---|
| Criterion one | Senescent cells accumulate with age.[4] |
| Criterion two | Transplanting or accumulating senescent cells causes dysfunction in mice.[10] |
| Criterion three | The NIA Interventions Testing Program fed fisetin to genetically diverse mice at three sites and found no significant lifespan extension in either sex.[11] |
The replication failure does not mean senescence is the wrong target. It means one compound, at one dose and route, did not do what an earlier study suggested, which is precisely the kind of error the third criterion exists to catch.
The contrast with rapamycin is the practical lesson. Both compounds have a hallmark and a mechanism attached, and only one has survived a multi-site lifespan test.
NAD precursors: a marker that moves, an outcome that does not
NAD is a coenzyme at the center of energy metabolism, and its levels fall with age. Because NAD itself is a large molecule, most products sell precursors that the body converts into it.[5]
| Test | What the evidence shows |
|---|---|
| Criterion one | NAD levels decline with age, which ties the story to mitochondrial dysfunction.[5] |
| Criterion two | Mitochondrial dysfunction is on the list because disrupting mitochondria accelerates aging features in animals.[1] |
| Criterion three | No human trial has shown that raising NAD with a precursor extends healthy life or delivers the benefits the biology implies.[5] |
The precursor case is the most common shape of a longevity claim. A real hallmark, a real decline and a real change in a blood level are all true at once, and none of them is the outcome a buyer is paying for.
Can you measure a hallmark in yourself?
Mostly not in any way that changes a decision. The hallmarks are defined in tissues and model organisms, and the tests sold to consumers measure a proxy for one of them at best.
- Epigenetic clocks track epigenetic alterations well across groups, but replicate measurements of the same sample differed by up to 9 years on some clocks (per Higgins-Chen et al. 2022).[8]
- The first multi-tissue clock estimated age from DNA methylation at 353 sites across many tissues and cell types (per Horvath 2013).[6]
- Retrained clock versions brought most replicate measurements within 1.5 years, showing the noise is a design problem that can be reduced (per Higgins-Chen and colleagues, 2022, a study from the Yale lab of Morgan Levine).[8]
- Telomere length tests report a real hallmark with no established action attached to the result.
- NAD blood tests confirm that a precursor raised NAD, which is not the same as showing a benefit.[5]
- Senescent cell counts need several markers measured together in tissue, which no consumer blood test provides.[4]
- Inflammatory markers are ordinary clinical tests, useful for diagnosing disease rather than for grading aging.
The noise matters most for before-and-after claims. If a single clock reading can move by years without any change in the person, a test taken before and after a supplement cannot show whether the supplement worked.
How can you use the framework to judge a claim?
Apply the third criterion the next time a product says it targets a hallmark.
- Has improving this process, by this means, been shown to extend healthy lifespan in any organism?
- For rapamycin the answer is yes in mice, but for most of the supplement shelf, the honest answer is that the compound changes a marker in a cell.
That is the practical value of the hallmarks of aging: not as a list to memorize, but as a standard of evidence someone already wrote down. Our review of what the longevity supplement evidence actually shows applies exactly that standard, compound by compound.
Frequently asked questions
What are the 12 hallmarks of aging?
Genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, disabled macroautophagy, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, altered intercellular communication, chronic inflammation and dysbiosis. The 2023 update added the last three plus disabled macroautophagy to the original nine.
What are the 14 hallmarks of aging?
There is no canonical list of fourteen. The peer-reviewed frameworks are the nine proposed in 2013 and the twelve proposed in 2023. Various meeting reports and commercial articles have floated additional candidates, and a page presenting fourteen as established is usually selling something adjacent to it.
Why did the list change from nine to twelve?
Because a decade of research strengthened the case for candidates that had not met the criteria in 2013. Chronic inflammation, dysbiosis and disabled macroautophagy accumulated enough evidence to qualify. The framework was designed to be revisable, and that revision is a sign of it working.
What makes something a hallmark rather than just a feature of aging?
Three tests. It should appear with age, experimentally worsening it should accelerate aging, and experimentally improving it should slow aging or extend healthy lifespan. That third test is the hard one and is why many plausible candidates never qualified.
Which hallmark do longevity drugs target?
Mostly deregulated nutrient sensing. Rapamycin inhibits mTORC1 directly, which is the clearest example of a compound aimed squarely at one hallmark. Senolytics target cellular senescence, and NAD precursors are aimed at mitochondrial dysfunction, with much weaker evidence in both cases.
Are the hallmarks causes of aging or symptoms?
The framework deliberately does not settle that. The three criteria require that intervening on a hallmark changes the trajectory, which implies causal contribution, but the hallmarks interact so heavily that separating cause from consequence for any single one is largely unresolved.
What is cellular senescence?
A state in which a damaged cell stops dividing permanently but does not die, and instead secretes inflammatory signals that disturb the tissue around it. It is both protective, as a brake on cancer, and harmful when senescent cells accumulate, which is why targeting it is more delicate than it sounds.
What habits age you faster?
Smoking, chronic sleep deprivation, sustained inactivity, excess alcohol, chronic psychological stress and diets that keep insulin and inflammation elevated. Each maps onto several hallmarks at once, which is part of why lifestyle effects are larger than any single supplement.
Can any intervention address all twelve?
Exercise comes closest. Reviews have argued it attenuates most hallmarks simultaneously through separate mechanisms. No drug or supplement does anything comparable, and any product claiming to cover all twelve is describing marketing rather than pharmacology.
Is telomere length worth testing?
Not really, for an individual. Telomere attrition is a real hallmark and consumer telomere tests have poor reproducibility and no established action attached to the result. Epigenetic clocks are more informative and still not clinically actionable.
How do the hallmarks connect to each other?
Densely. Mitochondrial dysfunction drives inflammation, inflammation accelerates senescence, senescence depletes stem cell pools, and nutrient sensing modulates almost everything else. That interconnection is why isolating one hallmark for treatment has been so much harder than describing it.
Does the framework predict how long someone will live?
No. It is a map of mechanisms, not a clinical tool. Nothing in it produces an individual prognosis, and no combination of consumer tests currently converts hallmark status into a reliable personal number.
Why does every supplement claim to target a hallmark?
Because the framework is respected, freely citable and written at the level of cell biology, which makes it easy to attach a mechanistic story to almost any compound. Meeting the first criterion is trivial. Meeting the third one, that improving it extends healthy lifespan, is where nearly everything fails.
What is the difference between the 2013 and 2023 hallmarks papers?
The 2013 paper proposed nine hallmarks and the three criteria. The 2023 update kept all nine, split disabled macroautophagy out of proteostasis, gave chronic inflammation its own line and added dysbiosis, and rewrote each section around a decade of new evidence.
What does the phrase an expanding universe mean in the 2023 title?
It signals that the authors expect the list to keep growing. Candidates such as altered mechanical properties of tissues and dysregulated RNA splicing were discussed, and they lack the intervention evidence the third criterion requires.
Where should someone start reading?
The 2013 paper for the original architecture and the 2023 update for the current version. Both are open enough to read directly, and both are considerably more careful than the summaries built on top of them.
References
12 sources, all link-checked; oldest check
- The Hallmarks of Aging. López-Otín C et al. Cell 2013;153(6):1194–1217.Publisher blocks automated checks · last tried Sep 15, 2026
- Hallmarks of aging: an expanding universe. López-Otín C et al. Cell 2023;186(2):243–278.Updated framework: twelve hallmarks.Publisher blocks automated checks · last tried Sep 15, 2026
- Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Harrison DE et al. Nature 2009;460:392–395.9–14% mouse lifespan extension started mid-life.Verified Sep 15, 2026
- Cellular senescence: defining a path forward. Gorgoulis V et al. Cell 2019 (cited 3,603).Consensus definition: a cell state triggered by stressful insults and certain physiological processes, with prolonged and generally irreversible cell-cycle arrest, secretory features, macromolecular damage and altered metabolism.Publisher blocks automated checks · last tried Sep 15, 2026
- NAD+ supplements: can they really slow down aging?. Cleveland Clinic Health Essentials, 26 February 2026.Mainstream clinical overview; notes NAD+ itself is a large molecule and that most products supply precursors instead.Verified Sep 15, 2026
- DNA methylation age of human tissues and cell types. Horvath S. Genome Biology 2013;14:R115.Verified Sep 15, 2026
- Cardioprotection and lifespan extension by the natural polyamine spermidine. Eisenberg T et al. Nature Medicine 2016;22:1428–1438.Verified Sep 15, 2026
- A computational solution for bolstering reliability of epigenetic clocks. Higgins-Chen AT et al. Nature Aging 2022;2:644–661.Documents test–retest reliability limits of methylation clocks.Verified Sep 15, 2026
- TAME: Targeting Aging with Metformin. American Federation for Aging Research.The proposed multi-center trial designed to test whether metformin delays age-related disease; as of 2026 it has published no efficacy results.Publisher blocks automated checks · last tried Sep 15, 2026
- Senolytics improve physical function and increase lifespan in old age. Xu M et al. Nature Medicine 2018;24:1246–1256.Verified Sep 15, 2026
- Astaxanthin and meclizine extend lifespan in UM-HET3 male mice; fisetin, SG1002, dimethyl fumarate and mycophenolic acid do not. Harrison DE et al. GeroScience 2023 (NIA Interventions Testing Program; cited 64).The most rigorous multi-site mouse lifespan program tested fisetin and found no lifespan extension.Verified Sep 15, 2026
- Influence of rapamycin on safety and healthspan metrics after one year: PEARL trial results. Moel M et al. Aging (Albany NY) 2025;17(4):908–936.48-week RCT, 5/10 mg weekly; safe, no visceral-fat change; authors employed by AgelessRx; note conflict.Verified Sep 15, 2026
Every link above is re-requested on a schedule by an automated checker; the date shown is when it last answered. Publishers that block automated requests are marked as such rather than reported broken. Our editorial standards explain how a source gets cited here in the first place.
Filed by the Rapamycin.store evidence desk. Each source is re-checked when the page is reviewed.
Own a longevity or health brand?
Rapamycin.store publishes sponsored and guest posts on an aged, topically-relevant domain, and offers display advertising. Tell us what you want to promote and we'll send our current options.