miniphim · longevity · markers · built 2026-10-12 by Morphyx
Aging clocks and markers: what links them to lifespan
The main page counts only lifespan. This page catalogues what that page leaves out: the clocks and markers that stand in for lifespan, because nobody can wait 80 years for a human result. For each one it gives the link to lifespan it's supposed to carry, and how far that link has been tested.
Every claim below quotes its source's archived abstract, and the build stops if a quote isn't there. Abstracts are what the authors chose to put first. A full-text reading could add to a row, but it shouldn't contradict one.
The ladder
- tracks age. It changes with chronological age, or was trained to.
- predicts death. In people followed forward, it predicts death or disease beyond what age alone does.
- causal test. Genetics (MR) or a direct manipulation asks whether the marker itself drives the outcome.
- movable in a trial. A randomized trial has moved it.
- moving it moved outcomes. A randomized trial moved it AND changed hard outcomes. Only this rung makes a surrogate.
Predicting is not causing, and causing is not the same as being safe to push. A marker can predict death because it is a gauge (CRP: genetics says it doesn't cause heart disease), or because it is a lever pulled the wrong way (HbA1c: pushed too low, more people died). Only rung 4 tells you that changing the marker changes the outcome.
The table
Green yes, red no, yellow mixed, grey open. Hover over a cell for the finding, or read the details further down. An empty cell means not shown in the sources archived here, which is not the same as untested.
| marker | 0
tracks age | 1
predicts death | 2
causal test | 3
movable in a trial | 4
moving it moved outcomes |
|---|
Horvath pan-tissue clock
DNAm clock (1st generation) | yes | yes | | | |
|---|
Hannum blood clock
DNAm clock (1st generation) | yes | yes | | | |
|---|
PhenoAge (DNAm)
DNAm clock (2nd generation) | | yes | | no | |
|---|
GrimAge
DNAm clock (2nd generation) | | yes | | no | |
|---|
DunedinPACE
DNAm pace-of-aging (3rd generation) | | yes | | yes | open |
|---|
Proteomic aging clock (204 proteins)
plasma protein clock | | yes | | | |
|---|
Organ-specific proteomic ages
plasma protein clock, per organ | | yes | | | |
|---|
Telomere length
chromosome-end length | | | mixed | | |
|---|
Senescent cells (p16Ink4a+)
cell-state marker | | | yes | | |
|---|
Clonal hematopoiesis (CHIP)
somatic mutation in blood | yes | yes | | | |
|---|
C-reactive protein (CRP)
inflammation marker | | yes | no | | |
|---|
Frailty index
count of deficits | yes | yes | | | |
|---|
Gait speed
physical function | | yes | | | |
|---|
Grip strength
physical function | | yes | | open | |
|---|
Cardiorespiratory fitness (VO2max, METs)
physical function | | yes | | | |
|---|
LDL cholesterol
blood lipid (comparison row) | | | | | yes |
|---|
Systolic blood pressure
vital sign (comparison row) | | | | | yes |
|---|
HbA1c
blood sugar average (comparison row) | | | | | no |
|---|
What the table says
Of the 15 aging markers, 0 reach rung 4. Counting each by the highest rung it has been tested at: 3 at rung 3, 3 at rung 2, 9 at rung 1, and 0 only at rung 0. Rung 1 is where most of the field stands: many markers predict death well, and none has been shown to be a target. The highest rung any clock reaches is 3: DunedinPACE was slowed by calorie restriction in a randomized trial, and the effect was small. In that same trial, PhenoAge and GrimAge, the two clocks built to predict death, didn't move. So which clock you use decides whether a trial "worked".
The strongest causal result belongs to a marker, not a clock: removing p16+ senescent cells extended lifespan in mice. The cleanest warning comes from outside aging: HbA1c is an FDA-accepted surrogate, and pushing it too low raised deaths.
The three comparison rows (LDL, blood pressure, HbA1c) show what rung 4 takes: tens of thousands of people randomized, and years of counting deaths. That is the bill no aging clock has paid. TAME was designed to be a step toward paying it and has not been funded.
Who decides what counts
In the US, the FDA decides which markers can stand in for outcomes in drug approval. It keeps a public table of surrogate endpoints. LDL, blood pressure and HbA1c are on it, each tied to a disease. No aging clock, frailty index or telomere measure is, and aging itself isn't a disease the table lists. (Read 2026-10-12. The page says "Content current as of 04/29/2026". FDA's server refused our archiver, so this one is read, not archived.) No clock can be the endpoint of an approval trial until it is on that table, so a company with a clock sells it to consumers or to researchers, not to a regulator. The researchers' own response is a consensus effort to set standards (the Biomarkers of Aging Consortium; Cell 2023), which opens with the current lack of standards and consensus on the properties of a reliable aging biomarker
.
One more caution on measurement: technical noise produces deviations up to 9 years between replicates for six prominent epigenetic clocks
(Nature Aging 2022). The same paper offers principal-component versions that agree within 1.5 years. A single clock reading from a single sample can be off by years before any biology enters.
Every marker, in detail
Horvath pan-tissue clock DNAm clock (1st generation)
The proposed link: DNA methylation age measures the work of an "epigenetic maintenance system"; a clock running ahead of the calendar would mean that system is wearing out faster.
- Rung 0 (tracks age): yes. Built to predict calendar age across 51 tissues; near zero in stem cells.
multi-tissue predictor of age
(Genome biology 2013) - Rung 0 (tracks age): yes. Running ahead of the calendar is heritable, so it is not just noise.
highly heritable measure of age acceleration
(Genome biology 2013) - Rung 1 (predicts death): yes. Four cohorts of older people; 16% after adjusting for health, IQ, class and APOE. The abstract says "DNA methylation age"; the paper computed it with the Horvath and Hannum clocks (full text, not archived).
A 5-year higher Δage is associated with a 21% higher mortality risk
(Genome biology 2015)
Hannum blood clock DNAm clock (1st generation)
The proposed link: Each person's methylome ages at a rate, and a faster rate is a faster-aging body.
- Rung 0 (tracks age): yes. 656 people, aged 19 to 101; the rate differs by sex and gene variants.
measures the rate at which an individual's methylome ages
(Molecular cell 2013) - Rung 1 (predicts death): yes. Four cohorts of older people; 16% after adjusting for health, IQ, class and APOE. The abstract says "DNA methylation age"; the paper computed it with the Horvath and Hannum clocks (full text, not archived).
A 5-year higher Δage is associated with a 21% higher mortality risk
(Genome biology 2015)
PhenoAge (DNAm) DNAm clock (2nd generation)
The proposed link: Trained on a clinical score that predicts death, not on calendar age, so the methylation it picks up should be the part that matters for lifespan.
- Rung 1 (predicts death): yes. Predicts all-cause mortality, cancers, physical function, Alzheimer's.
including all-cause mortality
(Aging 2018) - Rung 3 (movable in a trial): no. Two years of 25% calorie restriction (CALERIE, randomized, n = 220) did not move it.
did not lead to significant changes in biological age estimates measured by various DNAm clocks including PhenoAge and GrimAge
(Nature aging 2023)
GrimAge DNAm clock (2nd generation)
The proposed link: Methylation stand-ins for seven plasma proteins and for smoking pack-years, combined into a "predictor of lifespan".
- Rung 1 (predicts death): yes. The strongest death predictor among the clocks of its time.
time-to-death (Cox regression P=2.0E-75)
(Aging 2019) - Rung 3 (movable in a trial): no. CALERIE did not move it either.
did not lead to significant changes in biological age estimates measured by various DNAm clocks including PhenoAge and GrimAge
(Nature aging 2023)
DunedinPACE DNAm pace-of-aging (3rd generation)
The proposed link: Not "how old" but "how fast": trained on 20 years of decline in 19 organ measures in one 1972–73 birth cohort.
- Rung 1 (predicts death): yes. Also adds prediction beyond GrimAge; high test-retest reliability.
was associated with morbidity, disability, and mortality
(eLife 2022) - Rung 3 (movable in a trial): yes. The only clock here moved in a randomized trial, and the paper says "Treatment effect sizes were small".
CALERIE intervention slowed the pace of aging, as measured by the DunedinPACE DNAm algorithm
(Nature aging 2023) - Rung 4 (moving it moved outcomes): open. No trial has shown that slowing it changes disease or death. The CALERIE authors say so themselves.
will require trials with long-term follow-up to establish effects of intervention on primary healthy-aging endpoints
(Nature aging 2023)
Proteomic aging clock (204 proteins) plasma protein clock
The proposed link: The blood's protein mix shifts with age across many organ systems, so a protein age ahead of the calendar flags a body ahead of the calendar.
- Rung 1 (predicts death): yes. UK Biobank n = 45,441; 18 chronic diseases; age prediction held up in China and Finland.
as well as with multimorbidity and all-cause mortality risk
(Nature medicine 2024)
Organ-specific proteomic ages plasma protein clock, per organ
The proposed link: Organs age at different speeds in one person; proteins shed by one organ date that organ.
- Rung 1 (predicts death): yes. About 20% of people have one organ aging fast; accelerated heart age raises heart failure risk by 250%.
Accelerated organ aging confers 20-50% higher mortality risk
(Nature 2023)
Telomere length chromosome-end length
The proposed link: Telomeres shorten with each division, so short telomeres were read as a worn-out body and long ones as a young one.
- Rung 2 (causal test): mixed. MR (420,081 cases): genetically longer telomeres raise glioma, lung adenocarcinoma and melanoma risk and lower heart disease risk. Longer is not simply better.
longer telomeres increase risk for several cancers but reduce risk for some non-neoplastic diseases
(JAMA oncology 2017)
Senescent cells (p16Ink4a+) cell-state marker
The proposed link: Cells that stop dividing but keep secreting inflammatory signals pile up with age and damage the tissue around them.
- Rung 2 (causal test): yes. Killing p16+ cells with a transgene, from one year of age: the strongest causal result in this table. It is in mice, with a genetic switch no drug has.
extended median lifespan in both male and female mice of two distinct genetic backgrounds
(Nature 2016)
Clonal hematopoiesis (CHIP) somatic mutation in blood
The proposed link: Blood stem cells with cancer-linked mutations take over a share of the blood, and those cells may drive vessel disease as well as cancer.
C-reactive protein (CRP) inflammation marker
The proposed link: "Inflammaging": low-grade inflammation rises with age and drives disease.
- Rung 1 (predicts death): yes. In 160,309 people, each SD of log CRP carried an RR of about 1.5 for vascular and for non-vascular death.
non-vascular mortality
(Lancet (London, England) 2010) - Rung 2 (causal test): no. MR in 194,418 people: CRP marks heart risk, it doesn't cause it. A gauge, not a lever.
unlikely to be even a modest causal factor in coronary heart disease
(BMJ (Clinical research ed.) 2011)
Frailty index count of deficits
The proposed link: Aging is the accumulation of many small failures, so the share of deficits a person has reflects how close they are to death, whichever deficits they are.
- Rung 0 (tracks age): yes. In elderly Canadians.
deficits accumulated at 3% per year
(TheScientificWorldJournal 2001) - Rung 1 (predicts death): yes. The founding paper; widely re-tested since (not archived here).
serves as an important prognostic factor for life expectancy
(TheScientificWorldJournal 2001)
Gait speed physical function
The proposed link: Walking speed integrates heart, lungs, muscle, nerves and balance, so it reads the whole system at once.
- Rung 1 (predicts death): yes. 9 cohorts, 34,485 people aged 65+.
pooled hazard ratio per 0.1 m/s, 0.88
(JAMA 2011)
Grip strength physical function
The proposed link: Muscle strength stands for the body's reserve.
- Rung 1 (predicts death): yes. PURE, 139,691 people in 17 countries.
hazard ratio per 5 kg reduction in grip strength 1.16
(Lancet (London, England) 2015) - Rung 3 (movable in a trial): open. The authors' own open question: nobody has shown that getting stronger moves the death rate.
to test whether improvement in strength reduces mortality
(Lancet (London, England) 2015)
Cardiorespiratory fitness (VO2max, METs) physical function
The proposed link: Peak oxygen use sets how much the whole body can do.
- Rung 1 (predicts death): yes. Elite vs low fitness, 122,007 patients referred for treadmill tests; no upper limit seen. Retrospective, and people who were already sick were sent for testing.
adjusted hazard ratio [HR], 0.20
(JAMA network open 2018)
LDL cholesterol blood lipid (comparison row)
The proposed link: Not an aging marker. It is here as the template: what a validated surrogate looks like.
- Rung 4 (moving it moved outcomes): yes. 26 randomized trials, about 170,000 people. On FDA's surrogate table.
each 1·0 mmol/L reduction reducing the annual rate of these major vascular events by just over a fifth
(Lancet (London, England) 2010)
Systolic blood pressure vital sign (comparison row)
The proposed link: Not an aging marker; a second validated surrogate.
- Rung 4 (moving it moved outcomes): yes. SPRINT: aiming for <120 instead of <140 mm Hg cut major events and deaths from all causes. On FDA's surrogate table.
death from any cause
(The New England journal of medicine 2015)
HbA1c blood sugar average (comparison row)
The proposed link: Not an aging marker. It is here as the warning: an accepted surrogate, moved hard, and more people died.
- Rung 4 (moving it moved outcomes): no. ACCORD: aiming for HbA1c <6.0% in high-risk type 2 diabetes raised deaths (HR 1.22). HbA1c is still on FDA's surrogate table for diabetes drugs.
increased mortality and did not significantly reduce major cardiovascular events
(The New England journal of medicine 2008)
Abbreviations
- ITP
- Interventions Testing Program. The NIA's mouse lifespan program: each compound is tested at three sites (Jackson Laboratory, University of Michigan, UT Health San Antonio) on genetically mixed mice, and the failures are published too.
- NIA
- National Institute on Aging. The US federal institute (part of NIH) that funds the ITP.
- NIH
- National Institutes of Health. The US federal medical research agency.
- MPD
- Mouse Phenome Database. The Jackson Laboratory's public database. Its ITP center page is where the cohort table here comes from.
- UM-HET3
- University of Michigan heterogeneous stock, 4-way cross. The ITP's mouse: each animal is a genetically different grandchild of four inbred strains, so a result can't ride on one strain's quirks.
- NR
- nicotinamide riboside. A vitamin B3 form, sold as an NAD+ booster.
- NAD+
- nicotinamide adenine dinucleotide. A coenzyme whose levels fall with age in some tissues.
- AKG
- alpha-ketoglutarate. A metabolite of the citric acid cycle.
- NDGA
- nordihydroguaiaretic acid. A plant-derived antioxidant; in the ITP it extended lifespan in males only, in both cohorts that tested it.
- CI
- confidence interval. Here, 95% Clopper–Pearson: the range of true rates that the count can't rule out.
- HR
- hazard ratio. How much faster the event (here usually death) happens in one group than another. 1.4 means 40% faster at any moment.
- RR
- risk ratio. Risk in one group divided by risk in another.
- SD
- standard deviation
- RCT
- randomized controlled trial. Chance, not the participant or doctor, decides who gets the intervention. The only design here that can show moving a marker causes the outcome.
- MR
- Mendelian randomization. Using gene variants that raise or lower a marker from birth as a natural experiment: if people born with higher levels don't get more disease, the marker probably isn't the cause.
- DNAm
- DNA methylation. Methyl groups on DNA, read at CpG sites. Most epigenetic clocks are weighted sums of these.
- CpG
- cytosine–phosphate–guanine site. A place in DNA where a C is followed by a G; the usual site of methylation.
- CHIP
- clonal hematopoiesis of indeterminate potential. A blood stem cell carrying a cancer-linked mutation that has expanded into a visible share of the blood, without cancer.
- CRP
- C-reactive protein. A liver protein that rises with inflammation.
- LDL
- low-density lipoprotein (cholesterol)
- HbA1c
- glycated hemoglobin. The share of hemoglobin with glucose stuck to it; a 2–3 month average of blood sugar.
- CHD
- coronary heart disease
- CALERIE
- Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy. A 2-year randomized trial of 25% calorie restriction in adults without obesity.
- PURE
- Prospective Urban Rural Epidemiology study. A cohort of about 140,000 people in 17 countries.
- SPRINT
- Systolic Blood Pressure Intervention Trial
- ACCORD
- Action to Control Cardiovascular Risk in Diabetes
- TAME
- Targeting Aging with Metformin. A proposed human trial of about 3,000 people; not fully funded.
- FDA
- US Food and Drug Administration. It decides which markers count as surrogate endpoints for drug approval.
- doi
- digital object identifier. A permanent link to a paper: https://doi.org/ followed by the number.
Sources (archived, sha256)
- Biomarkers of aging for the identification and evaluation of longevity interventions. (Cell 2023) doi 10.1016/j.cell.2023.08.003 ·
epmc-10_1016_j_cell_2023_08_003.json 2fd480c0ef2f9382… - A computational solution for bolstering reliability of epigenetic clocks: Implications for clinical trials and longitudinal tracking. (Nature aging 2022) doi 10.1038/s43587-022-00248-2 ·
epmc-10_1038_s43587-022-00248-2.json 497d0de57610b7f1… - DNA methylation age of human tissues and cell types. (Genome biology 2013) doi 10.1186/gb-2013-14-10-r115 ·
epmc-10_1186_gb-2013-14-10-r115.json d7c249b4b5bcc554… - DNA methylation age of blood predicts all-cause mortality in later life. (Genome biology 2015) doi 10.1186/s13059-015-0584-6 ·
epmc-10_1186_s13059-015-0584-6.json 271f6ff8a2bef180… - Genome-wide methylation profiles reveal quantitative views of human aging rates. (Molecular cell 2013) doi 10.1016/j.molcel.2012.10.016 ·
epmc-10_1016_j_molcel_2012_10_016.json 35cca58a92992f43… - An epigenetic biomarker of aging for lifespan and healthspan. (Aging 2018) doi 10.18632/aging.101414 ·
epmc-10_18632_aging_101414.json 455d663d29b9da07… - Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. (Nature aging 2023) doi 10.1038/s43587-022-00357-y ·
epmc-10_1038_s43587-022-00357-y.json 02e1cb1161965a28… - DNA methylation GrimAge strongly predicts lifespan and healthspan. (Aging 2019) doi 10.18632/aging.101684 ·
epmc-10_18632_aging_101684.json aa61a89661512a87… - DunedinPACE, a DNA methylation biomarker of the pace of aging. (eLife 2022) doi 10.7554/elife.73420 ·
epmc-10_7554_eLife_73420.json b8a406310ed453ec… - Proteomic aging clock predicts mortality and risk of common age-related diseases in diverse populations. (Nature medicine 2024) doi 10.1038/s41591-024-03164-7 ·
epmc-10_1038_s41591-024-03164-7.json b2547f27a7567e76… - Organ aging signatures in the plasma proteome track health and disease. (Nature 2023) doi 10.1038/s41586-023-06802-1 ·
epmc-10_1038_s41586-023-06802-1.json bc74b768647ded29… - Association Between Telomere Length and Risk of Cancer and Non-Neoplastic Diseases: A Mendelian Randomization Study. (JAMA oncology 2017) doi 10.1001/jamaoncol.2016.5945 ·
epmc-10_1001_jamaoncol_2016_5945.json 7f4ee6bc780cf7b1… - Naturally occurring p16(Ink4a)-positive cells shorten healthy lifespan. (Nature 2016) doi 10.1038/nature16932 ·
epmc-10_1038_nature16932.json 0a1a330a32f26348… - Age-related clonal hematopoiesis associated with adverse outcomes. (The New England journal of medicine 2014) doi 10.1056/nejmoa1408617 ·
epmc-10_1056_NEJMoa1408617.json ac20b68603f05085… - C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. (Lancet (London, England) 2010) doi 10.1016/s0140-6736(09)61717-7 ·
epmc-pmid-20031199.json 80e2cd5a5b515bb6… - Association between C reactive protein and coronary heart disease: mendelian randomisation analysis based on individual participant data. (BMJ (Clinical research ed.) 2011) doi 10.1136/bmj.d548 ·
epmc-10_1136_bmj_d548.json 19056eed79ccaad1… - Accumulation of deficits as a proxy measure of aging. (TheScientificWorldJournal 2001) doi 10.1100/tsw.2001.58 ·
epmc-10_1100_tsw_2001_58.json 404db11284ef7333… - Gait speed and survival in older adults. (JAMA 2011) doi 10.1001/jama.2010.1923 ·
epmc-10_1001_jama_2010_1923.json dd1da76523a517b7… - Prognostic value of grip strength: findings from the Prospective Urban Rural Epidemiology (PURE) study. (Lancet (London, England) 2015) doi 10.1016/s0140-6736(14)62000-6 ·
epmc-pmid-25982160.json 76fff05ef48465ba… - Association of Cardiorespiratory Fitness With Long-term Mortality Among Adults Undergoing Exercise Treadmill Testing. (JAMA network open 2018) doi 10.1001/jamanetworkopen.2018.3605 ·
epmc-10_1001_jamanetworkopen_2018_3605.json 031a33f8fa2e516f… - Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. (Lancet (London, England) 2010) doi 10.1016/s0140-6736(10)61350-5 ·
epmc-pmid-21067804.json fcbec0c00f82fb1c… - A Randomized Trial of Intensive versus Standard Blood-Pressure Control. (The New England journal of medicine 2015) doi 10.1056/nejmoa1511939 ·
epmc-10_1056_NEJMoa1511939.json a9c35f2c7b5fd8ac… - Effects of intensive glucose lowering in type 2 diabetes. (The New England journal of medicine 2008) doi 10.1056/nejmoa0802743 ·
epmc-10_1056_NEJMoa0802743.json 26dc875ab2f6ca69…
Built by research/longevity/markers.mjs, which checks every hash and every quote. Abstracts are from Europe PMC.