Cellular senescence
Senescent cells stop dividing but refuse to die, leaking inflammatory signals into the tissue around them. In mice, clearing them out makes old animals healthier, which set off a race for "senolytic" drugs and supplements. In people, every senolytic trial so far has been small or has missed its goal, and in healthy adults the best-known combination pushed several biological-age markers the wrong way.
Don't take senolytic drugs or supplements outside a trial. If you want to act on this biology, the levers with human data are eating less (two randomised trials) and treating insomnia (one), both judged only on senescence-linked blood markers. Exercise is well supported in animals and worth doing for many other reasons.
What a senescent cell is, and why the body makes them
Cellular senescence is the eighth of the twelve hallmarks of aging. A senescent cell has permanently stopped dividing in response to stress — DNA damage, the erosion of telomeres (the protective caps on chromosome ends), an activated cancer gene, or mitochondrial breakdown — but has not died. It resists the normal signal to self-destruct and keeps broadcasting inflammatory molecules.[1]
This is settled cell biology, and it starts as a safety mechanism. A damaged cell that might turn cancerous is pushed into permanent arrest, because a cell that cannot divide cannot become a tumour. Two molecular brakes hold the arrest in place, one reacting to acute damage and the p16 pathway locking it in for the long term. Senescence also helps sculpt tissues in the embryo, and in wound healing a short-lived wave of senescent cells limits scarring before immune cells clear them away.[2]
Healthy senescence is usually temporary: the cell does its job and is removed. Not always, though. In mice, some long-lived senescent cells do lasting work: many line the liver's small blood vessels, and wiping them out broke down the barrier between blood and tissue, caused scarring and worsened overall health.[3] That is one reason clearing as many senescent cells as possible may not be the right goal.
The problem with age is that they pile up, for two reasons: the body makes more of them as damage accumulates, and it gets much slower at clearing them. In mice, a senescent cell is removed within days in youth but lingers for weeks in old age, partly because the senescent cells themselves slow their own removal.[4] Weakening immune surveillance, covered under immunosenescence, is one reason for that slowdown.[5]
How a few bad cells can age a whole body
What makes a lingering senescent cell harmful is what it secretes: a mixture of inflammatory signalling proteins, tissue-dissolving enzymes and growth factors called the senescence-associated secretory phenotype (SASP). It degrades the surrounding tissue, recruits inflammatory immune cells, and can turn nearby healthy cells senescent too, spreading the state from cell to cell.[6] Across millions of cells and decades, this is one of the main engines of inflammaging, the chronic low-grade inflammation of aging tissue.
In mice, transplant experiments show that senescent cells can cause decline, not merely accompany it. Injecting about half a million senescent cells into young mice weakened their grip and slowed their walking, and the slowdown lasted up to six months. The same dose given to older mice (17 months old) also shortened their lives: over the following year they had about five times the risk of dying of mice given normal cells.[7]
In people the link is still indirect. In 1,678 adults aged 70 to 79 followed for about 11 years, higher blood levels of the proteins senescent cells release went with a higher risk of death, loss of mobility and heart failure — but other cells release the same proteins too.[8]
Where the damage shows up
The evidence here is mechanistic and from animals or human tissue in the lab; none of it rests on a human outcome trial.
- Muscle. Senescent cells build up in the muscle's repair niche, including the stem cells that rebuild fibres, and impair regeneration — one contributor to sarcopenia, the loss of muscle mass and strength with age.[9] It is one more reason resistance training matters.
- Joints. Stress pushes cartilage cells (chondrocytes) into senescence, and their secretions degrade the cartilage and spread the state to neighbouring cells, driving osteoarthritis. In lab tests on human cartilage, one senolytic (navitoclax) made the cells worse at building cartilage.[10]
- Blood vessels. Cells lining the arteries are highly prone to senescence; as they accumulate they reduce the nitric oxide vessels use to relax and stiffen the arteries, covered under genomic instability.[11]
- Brain. In the brain's immune cells (microglia), aging mitochondria leak DNA that a viral-DNA sensor mistakes for an infection, switching on a lasting inflammatory program linked to the loss of nerve cells. In old mice, a drug blocking this sensor calmed brain inflammation and improved memory in maze and fear-learning tests — animal evidence only.[12]
Two drug strategies: kill the cells or quiet them
Clearing senescent cells reverses frailty, bone loss and cardiovascular dysfunction in mice, and that set off the most active drug race in aging research.[13]
Senolytics kill senescent cells by briefly switching off the anti-death defences they rely on. The best-known is dasatinib plus quercetin (D+Q): dasatinib, a leukaemia drug, paired with quercetin, a plant flavonoid, because each works best on different kinds of senescent cells.[14] Other senolytics are held back by toxicity. Navitoclax blocks a family of survival proteins that platelets and some white blood cells also rely on, so continuous treatment can cause a sharp drop in platelets, with a risk of internal bleeding, and in infection-fighting white cells.[15]
Senomorphics quiet the harmful secretions without killing the cells. Rapamycin and metformin fall here.[16] The site covers both, with the senolytics, under geroprotectors.
Researchers are now working on precision versions — antibody-guided drugs and engineered immune cells aimed at markers unique to senescent cells.[17] The most advanced attempt to sidestep toxicity, foselutoclax, has stalled. It blocks the same survival proteins as navitoclax but is injected straight into the eye. In 65 people with diabetic swelling of the retina, vision after 48 weeks was about 5.6 letters better on an eye chart than after a sham injection, a result also compatible with no benefit at all.[18] A follow-up trial in 52 people, designed to show it was no worse than a standard eye injection, missed that main goal at weeks 20 to 24,[19] and the company, the field's best-known senolytic developer, dissolved in September 2025.[20]
In people, the record so far is mostly misses
The gap between mouse and human results is the most important thing to understand about this field. Most human studies enrolled fewer than twenty people and several had no control group; the two randomised trials large enough to test a benefit both missed their main goal. No senolytic has yet shown a clinical benefit in people; the one small randomised trial that met its main goal, in fatty-liver disease, measured liver scarring on biopsy and has not been repeated.
- The knee-osteoarthritis drug failed outright. UBX0101, a senolytic that works through a different route from navitoclax (it disrupts the p53/MDM2 interaction), cleared senescent cells and regrew cartilage in rodents. In a randomised, placebo-controlled trial of a single knee injection in 183 people with knee osteoarthritis, knee pain after 12 weeks had improved no more on any of three doses than on placebo, and the long-term follow-up study was stopped early.[21] Some reviews still describe this trial as having reduced pain; the results posted on the trial registry show it did not.[22]
- The randomised D+Q trial for bone missed too. In 60 postmenopausal women, 20 weeks of intermittent D+Q did not slow bone breakdown, the main goal. There were no serious side effects; a bone-density gain appeared only in an exploratory look at the women with the highest levels of a senescence marker.[23]
- A small randomised liver trial met its goal, for now. In 31 adults with scarring fatty-liver disease (fibrotic MASH, metabolic dysfunction-associated steatohepatitis), about half of those given intermittent D+Q for 21 weeks showed at least one stage less liver scarring on biopsy, against one in fourteen on placebo. Side effects were more common on D+Q, though all passed on their own. The result rested on few responders and lost significance under the most pessimistic handling of missing data, so the authors treat it as preliminary until a larger trial repeats it; two of them have a financial interest in Mayo Clinic senolytic patents.[24]
- The early D+Q successes were small and uncontrolled. The first study to show senescent cells being cleared in people had nine patients, no control group, and needed a corrigendum that revised some of its conclusions.[25][26] A twelve-person memory pilot in older adults at risk of Alzheimer's disease had no control group, and its changes in thinking scores and inflammation were compatible with chance.[27] In five people with mild Alzheimer's disease, dasatinib barely reached the fluid around the brain and quercetin was not detected.[28]
- When a placebo group was added, the benefit did not show up and side effects did. Better walking after D+Q in an uncontrolled lung-scarring (idiopathic pulmonary fibrosis) study did not show up in a 12-person placebo-controlled repeat, though that trial was too small to be sure; sleep problems and anxiety affected four of the six people on D+Q and none of the six on placebo.[29][30]
The warning for healthy people. In 19 healthy adults aged 43 to 87, with no control group, a six-month course of D+Q, at the doses people self-administer, had accelerated all three first-generation DNA-methylation "clocks" (lab estimates of biological age) by three months, moved a second-generation one too, and lowered a methylation estimate of telomere length.[31] The clocks best validated against health outcomes, GrimAge and DunedinPACE, did not move, and two of the accelerated clocks were back to baseline by six months — but two others (Hannum and PhenoAge) were not, and the telomere estimate had not recovered by six months. This is not proof that senolytics "age you". It is a concrete reason healthy adults should not self-administer them: a tool designed to clear sick cells can stress healthy ones, and nineteen people cannot tell you how much.
Laboratory work points the same way. Healthy young blood-vessel muscle cells exposed to D+Q briefly took on the features of senescent cells, recovering only after the drugs were withdrawn,[32] and in rodents with acute kidney injury caused by folic acid, D+Q worsened the damage.[33] Clearing senescent cells is context-dependent, and more is not safely better.
The supplement versions don't rescue the idea
Quercetin and fisetin (plant flavonoids) and curcumin (the turmeric polyphenol) are sold as supplements on the strength of this story, and the evidence for them is weak.
Fisetin is the most talked-about. One lab found it extended median and maximum lifespan in aged mice,[34] but the US National Institute on Aging's multi-site lifespan testing programme found no lifespan extension in either sex.[35] The only placebo-controlled trial with results so far, in 74 people with knee osteoarthritis, found high-dose fisetin no better than placebo on pain, function or inflammation.[36] And in the small clock study described above, a follow-on group that added fisetin to D+Q saw one key biological-age clock rise significantly, not fall.[37]
Curcumin shows why "natural" is not the same as "safe at any dose". In cultured vascular cells, high concentrations flip it from anti-inflammatory to pro-oxidant, pushing the cells into senescence. Whether any dose people swallow reaches that range is unknown, so this is a reason for caution, not a demonstrated harm.[38][39]
All three are poorly absorbed: they are water-insoluble, unstable in the gut and rapidly broken down by the liver, so unformulated powder produces negligible blood levels.[40] Black-pepper extract (piperine) is often added to curcumin on the strength of a roughly 20-fold increase in absorption, but that figure comes from one small 1998 study co-authored by an ingredient supplier,[41] and a 2025 independent crossover study in 9 healthy men found piperine added nothing.[42]
Rapamycin and the immune system: an early signal that didn't hold
Inhibiting mTOR, the cell's master nutrient-sensing switch, slows the formation of senescent cells and quiets their secretions; chronically active mTOR suppresses the cell's self-cleaning autophagy.[43] Rapamycin, which does this, remains the most consistently life-extending drug in animal studies. In healthy people it is tried at low weekly doses, to avoid the immune suppression of the daily doses given to transplant patients.[44]
The human results are thin. A randomised trial gave 264 older adults six weeks of low-dose mTOR inhibitors, including the rapamycin relative everolimus, and found fewer infections over the following year and a better response to flu vaccination.[45] It did not hold up at scale: a 1,000-person phase 3 trial of one of those drugs found no reduction in respiratory illness (26% of people on the drug vs 25% on placebo).[46] Reviewers have suggested a dose threshold above which mTOR inhibition stops boosting immunity and starts suppressing it, but no trial in older adults has shown that switch, and the trials do not point to a clear best dose.[47] The year-long placebo-controlled PEARL trial of weekly rapamycin in healthy adults found it reasonably safe but missed its main goal.[48] The details are on the rapamycin page.
What actually helps
None of these has been tested against a hard health outcome in this context; the human evidence is on blood markers. The idea behind all of them is to limit damage and restore the body's own clearance rather than do the immune system's job with a drug.
Eating less — the best human evidence, from two randomised trials. Chronic overeating keeps growth signalling (insulin, insulin-like growth factor 1 and mTOR) switched on and favours senescence.[49] In 58 middle-aged and older adults with obesity and prediabetes, 18 weeks of moderate calorie restriction (about 11% weight loss) left senescence markers in immune cells unchanged but significantly altered a composite of circulating SASP proteins, not uniformly in one direction, and only loosely in step with weight lost.[50] In the longer CALERIE 2 trial, healthy young and middle-aged adults of normal weight or moderately overweight who cut calories for two years had lower levels of nine senescence-linked proteins after one year, four of them still lower at two years — on only about 12% fewer calories, half the 25% target.[51] None of these markers is specific to senescent cells, so this shows the secretions calming down, not the cells being cleared. See fasting and time-restricted eating.
Sleep — thinner, but one randomised trial. One night of short sleep in 29 adults aged 61 to 86 switched on SASP genes in their blood immune cells; an early rise in the senescence marker p16 did not hold up on a second test.[52] In a randomised trial of 231 adults over 60 with insomnia, p16 in blood immune cells rose over two years with sleep education alone but stayed flat with cognitive behavioural therapy for insomnia (CBT-I).[53] The repeated oxygen dips of obstructive sleep apnea also drive oxidative stress and premature senescence.
Exercise — consistent in animals, unproven in people. In animals, regular exercise reduces senescent cells in the heart, kidney, liver, muscle and fat and sharpens the immune cells that hunt them.[54][55] In people, the one positive study, 12 weeks of strength and endurance training that lowered p16 in older adults' T-cells, had no comparison group,[56] and a 2026 review found only three randomised trials in people over 60; none measured p16, and in the two that could be pooled, blood inflammatory proteins did not differ from controls.[57] Extreme training may backfire: in rats, swimming to exhaustion pushed brain cells into senescence.[58]
Stress — mechanistic only. Sustained stress hormones generate oxidative damage that attacks the vulnerable DNA at chromosome ends, speeding telomere erosion and pushing cells toward senescence.[59] The stress and recovery page covers what lowers that load.
What to actually do
- Leave senolytics to trials. That includes D+Q, fisetin and quercetin sold for senescence; if one interests you, look for a trial to join.
- Eat a little less. CALERIE 2's effects came from cutting about 12% of calories.
- Treat insomnia with CBT-I, and get suspected sleep apnea tested.
- Train consistently across zone 2 and high-intensity intervals, and recover properly — for its many proven benefits, with senescence a possible bonus.
Further reading
- Li W, et al. Cellular senescence: from homeostasis to pathological implications and therapeutic strategies. Front Immunol 2025.[60]
- Fu TE, Zhou Z. Senescent cells as a target for anti-aging interventions: from senolytics to immune therapies. J Transl Int Med 2025.[61]
- Alum EU, et al. Targeting cellular senescence for healthy aging: advances in senolytics and senomorphics. Drug Des Devel Ther 2025.[62]
- Lelarge V, et al. Senolytics: from pharmacological inhibitors to immunotherapies, a promising future for patients' treatment. NPJ Aging 2024.[63]
- Martel J, et al. Lifestyle interventions to delay senescence. Biomed J 2024.[64]
- Justice JN, Leng XI, LeBrasseur NK, et al. Caloric restriction intervention alters specific circulating biomarkers of the senescence-associated secretome in middle-aged and older adults with obesity and prediabetes in an 18-week randomised controlled trial. J Gerontol A Biol Sci Med Sci 2024.[65]
- Lee E, Carreras-Gallo N, Lopez L, et al. Exploring the effects of dasatinib, quercetin, and fisetin on DNA methylation clocks: a longitudinal study on senolytic interventions. Aging (Albany NY) 2024 — n=19 Phase I; read the tables, not the abstract.[66]
- Millar CL, et al. A pilot study of senolytics to improve cognition and mobility in older adults at risk for Alzheimer's disease. EBioMedicine 2025 — STAMINA; single-arm, open-label, n=12.[67]
- Yousefzadeh MJ, et al. Fisetin is a senotherapeutic that extends health and lifespan. EBioMedicine 2018.[68]
- Picos A, et al. Vascular senescence and aging: mechanisms, clinical implications, and therapeutic prospects. Biogerontology 2025.[69]
- Hickson LJ, et al. Senolytics decrease senescent cells in humans: preliminary report from a clinical trial of dasatinib plus quercetin in individuals with diabetic kidney disease. EBioMedicine 2019 — open-label Phase 1, n=9.[70]
- Mannick JB, et al. TORC1 inhibition enhances immune function and reduces infections in the elderly. Sci Transl Med 2018.[71]
- Hands JM, et al. What is the clinical evidence to support off-label rapamycin therapy in healthy adults? Aging (Albany NY) 2025.[72]