Deregulated nutrient sensing
Your cells run four molecular "fuel gauges" that read how much food is around and decide whether to grow and store or to repair and recycle — and with age that decision gets stuck on "grow." The honest lesson from recent human work is balance rather than maximal suppression: the goal is to restore the youthful oscillation between feeding and fasting, not to starve the growth machinery into the ground.
Deregulated nutrient sensing is the sixth of the twelve hallmarks of aging and arguably the most central — it is the hub where diet meets the rate of biological aging. The cell constantly measures the availability of energy and nutrients and uses that reading to balance anabolism (building and storing) against catabolism (breaking down and repairing).[1] The governing principle of geroscience here is blunt: the evidence strongly supports the idea that growth signalling accelerates aging and that turning nutrient signalling down extends life.[2] The 2023 update of that framework adds a nuance: in youth this growth signalling drives beneficial building processes, and it takes on pro-aging properties in adulthood.[3] Most of the interventions on this site work, in part, by tilting that balance back toward repair. This is the most heavily targeted hallmark in all of longevity science.
How confident should you be
Descriptive — Strong. Intervention — Strong in animals, Weak-to-Moderate in humans, and on biomarkers rather than survival.
The mechanism is as well established as anything in aging biology: the four pathways exist, they respond to food, and turning the growth side down lengthens life in worms, flies and mice with a reproducibility almost nothing else in the field matches. The gap opens at the species boundary. No trial has tested any of these levers against human mortality, and the few that tested a clinical endpoint have given mixed results. A six-week course of mTOR inhibitors cut infections over the following year in one trial of 264 older adults, but a phase 3 trial of one of those drugs in 1,024 adults aged 65 and over found no fewer respiratory illnesses, despite a positive phase 2b result, and weekly rapamycin did not improve the gains from an exercise programme. Everything else is a biomarker: a two-year calorie-restriction trial that moved one biological-age clock by a few percent, a fasting-mimicking result from a post-hoc analysis, and a rapamycin trial that missed what it set out to measure. The drugs section is the weakest part of this page for a reason.
The four fuel gauges
Strong. The four-sensor architecture and its wiring are settled cell biology, mapped in detail across species.
Nutrient sensing runs on four interconnected molecular networks. Two read abundance (insulin/IGF-1 signalling, which tracks glucose, and mTOR, which tracks amino acids), and two read scarcity: AMPK and the sirtuins both detect low energy.[4] That gives a convenient accelerator pair and brake pair, though the split is a simplification, because the four are tightly cross-linked by feedback loops.[5]
The accelerators (pro-growth, pro-aging):
- Insulin and insulin-like growth factor 1 (IGF-1) signalling, usually shortened to IIS, is the most evolutionarily conserved aging-control pathway. Insulin is the signal that tells cells glucose is around, and IGF-1 runs through the same internal pathway, which drives growth and storage. Among its many targets are the forkhead box O (FOXO) transcription factors, the downstream switch that matters most for longevity in worms and flies.[6] Active growth signalling keeps FOXO out of the cell nucleus; when growth factors or nutrients run short, FOXO moves back in.[7]
- mTOR (mechanistic target of rapamycin) is the master growth hub, integrating signals from amino acids and growth factors to ramp up protein and fat synthesis while shutting down recycling.[8]
The brakes (pro-repair, pro-longevity):
- AMPK (AMP-activated protein kinase) is the low-fuel sensor: when energy runs low it activates, switches on catabolic energy-producing pathways, inhibits mTOR, and triggers autophagy and the building of new mitochondria.[9]
- Sirtuins, a family of seven repair enzymes, need NAD⁺ (nicotinamide adenine dinucleotide) to work, which ties them directly to the cell's metabolic state and to mitochondrial health.[10] NAD⁺ is widely said to fall with age, and it does in mice,[11] but in people the picture depends on where you measure. A 2025 review found a consistent age-related decline in only a limited number of human studies.[12] Muscle NAD⁺ is lower in older than in younger adults, most of all in the physically impaired,[13] while a study pooling seven independent human cohorts found whole-blood NAD⁺ remarkably stable with age.[14]
Metabolic overdrive: when the oscillation flattens
Strong for the cellular mechanism; Moderate for the claim that this is what drives human aging. The four sensors and their feed–fast responses have been mapped in cells and animals; the oscillator picture built on them, and the idea that its flattening is what ages people, is inference, not measurement.
Overeating switches on the abundance sensors such as mTOR, inhibits the scarcity sensors (AMPK and the sirtuins SIRT1 and SIRT3), and suppresses the cell's adaptive stress responses, including autophagy, antioxidant defence and DNA repair. Fasting and dietary restriction do the reverse.[15] The idea that health depends on a regular swing between those two states, the youthful oscillator this page keeps returning to, is a reasonable reading of that wiring rather than something that has been measured directly.
With age — and with constant grazing on a modern diet — that oscillation flattens into a state of chronic metabolic overdrive: IIS and mTOR idle high, AMPK and sirtuin activity fall, and the cell is locked in grow-and-store mode with the repair program switched off.[16] Two things accelerate the decline of the brake side. NAD⁺, the fuel sirtuins need, is used up largely by two enzymes: PARP1, a DNA-damage sensor that may burn more of it as damage builds up with age, and CD38, an NAD⁺-destroying enzyme whose levels rise in the tissues of old mice and humans, possibly driven by chronic low-grade inflammation. So sirtuin activity can be starved just when it is needed.[17][18] And persistent mTOR activity suppresses autophagy, letting damaged components and leaked mitochondrial DNA accumulate, which inflames tissue and further deafens cells to insulin — a self-reinforcing loop that ties nutrient sensing directly to inflammaging.[19]
Insulin: the part of this you can actually measure
Moderate. That chronic high insulin marks the problem is well supported; the thresholds used to interpret it come from cohort data rather than trials, and live under midlife labs.
Most of this hallmark is invisible from the outside. One arm of it is not. As insulin clearance falls with age, the body compensates by running chronically high insulin, which keeps the IIS accelerator pressed and suppresses the FOXO stress-response program — the whole-body version of what the cell-level picture describes.[20] That drift shows up in insulin sensitivity well before blood sugar reaches the diabetic range. Among 6,538 British civil servants followed for a median of 9.7 years, the 505 who developed type 2 diabetes saw their insulin sensitivity fall steeply over the 5 years before diagnosis, while their fasting glucose crept up slowly and climbed steeply only in the last 3.[21] That is why fasting insulin, and the HOMA-IR score calculated from fasting insulin and glucose together, is the most direct window a routine blood test gives on this hallmark. It is a window, not a verdict: across 7 cohort studies of 26,976 adults without diabetes, those in the highest HOMA-IR group had about 34% higher all-cause mortality than the lowest (relative risk 1.34), while fasting insulin on its own fell just short of a clear link.[22] See metabolic flexibility for the physiology of losing and regaining the feed–fast swing.
The growth-signalling paradox: less is more, but not zero
Strong in animals; Weak for the human U-shape — inferred from animal work and from observational associations between IGF-1 levels and mortality, not from any trial.
The most counterintuitive finding in this field is that dialling down the body's main growth signal extends lifespan. Mutations that reduce insulin/IGF-1 or mTOR signalling are among the most reproducible longevity levers in biology, lengthening life across worms, flies, and mice.[23]
But the relationship is a U-shape, not a straight line, and this is the crucial calibration. Shutting the pathway off entirely is incompatible with life: mice lacking either of its two core kinases die as embryos.[24] In people, both ends of the range carry more risk. Pooling 19 cohort studies of 30,876 adults, those with low IGF-1 had about 33% higher all-cause mortality (HR 1.33, a hazard ratio comparing death rates between groups) and those with high IGF-1 about 23% higher (HR 1.23) than those in the middle.[25] Among 380,997 UK Biobank adults with no history of cancer, cardiovascular disease or diabetes, the lowest tenth of IGF-1 carried about 39% higher all-cause mortality and the highest tenth about 17% higher than the middle.[26] These are associations, not proof that moving IGF-1 changes risk. Aging muscle shows what the low end looks like: it develops anabolic resistance, needing more amino acids to switch on growth signalling, a major cause of age-related muscle loss.[27] The goal is to restore the youthful swing between growth and repair, not to abolish growth, a theme that recurs every time these two forces are in tension.
mTOR: the central hub and the protein question
Strong for the mechanism; Weak for any single amino acid as a practical lever.
Of the four, mTOR is the most directly actionable. It operates as two complexes — mTOR complex 1 (mTORC1), the well-understood nutrient-sensing one, and complex 2 (mTORC2). mTORC1 reads amino acids, oxygen, and growth factors and responds by ramping up building while switching off autophagy; chronic mTORC1 activation from constant nutrient surplus is a key driver of cellular senescence and age-related disease.[28]
The dietary lever here is protein, and the textbook explanation is that the amino acid leucine is a potent direct activator of mTORC1, tying protein intake straight to this pro-aging pathway.[29] That explanation turns out to be too simple — as the amino-acid section below shows, restricting leucine specifically does not reproduce the benefit, and the total load matters more than the single most potent trigger. What survives is the tension covered in depth under protein: lower protein keeps mTOR quiet (good for longevity signalling), but adequate protein is needed to preserve muscle and prevent frailty (good for healthspan), and the right balance shifts with age — the same growth-versus-repair trade-off as the IGF-1 U-shape, played out on the dinner plate.
Diet, fasting, and the resilience paradox
Caloric restriction — Strong in animals, Moderate in humans (biomarker endpoints only). Reducing calorie intake without malnutrition is the most reproducible lifespan-extending intervention across species, working precisely by hitting all four hubs — turning the IIS and mTOR accelerators down and the AMPK and sirtuin brakes up.[30]
The human test is CALERIE, which randomised 220 healthy adults without obesity to a 25% calorie-restriction target or normal eating for two years. It improved cardiometabolic markers, but not all of the weight that came off was fat. After two years the restricted group had lost about 7.6 kg, including about 2.0 kg of fat-free (lean) mass, roughly in proportion to the weight lost.[31] Absolute leg strength fell, even though strength per kilogram of body weight rose,[32] and bone mineral density dropped modestly at the spine and hip, which the investigators flagged as a potential limitation of prolonged restriction.[33] The trial also slowed the DunedinPACE methylation clock by roughly 2–3%, meaning the restricted group accumulated biological age about 2–3% more slowly than controls, a real but small shift. Three deflators belong with that result every time it is quoted: the methylation analysis was post hoc rather than the trial's designed test; participants achieved only about 12% restriction against the 25% target; and the effect appeared on that one pace-of-aging clock, with no significant change on the other methylation clocks tested, including PhenoAge and GrimAge.[34] It remains the cleanest human evidence that this lever does something — and it is a biomarker result, not a survival result. Two later analyses of the same trial point the same way. Organ-by-organ biological-age scores built from routine clinical biomarkers, in 185 of the participants, rose more slowly under restriction for metabolic, cardiovascular, immune and whole-body age, though not kidney age; whole-body age ended about 1.3 years lower than in the normal-eating group after two years.[35] And the gains depended on keeping the weight off: the 20 participants who regained more than 5% of their starting weight in the second year lost the improvements in insulin and IGF-1 signalling.[36] Both are surrogate measures, and the second grouped people by what their weight did rather than by randomised arm. The clocks themselves are covered under epigenetic alterations.
Fasting-mimicking diets — Moderate. Because sustained restriction is hard and can carry costs, periodic fasting-mimicking diets (FMD) — short, structured low-calorie, low-protein cycles every few months — aim to trigger the same cellular cleanup without chronic deprivation. A secondary analysis of two randomized trials found that three monthly FMD cycles were associated with less insulin resistance and other pre-diabetes markers, lower liver fat, a younger immune-cell profile, and a roughly 2.5-year drop in a biological-age measure, independent of weight loss.[37] Three deflators belong with that figure: the measure is not a methylation clock but a composite of seven routine blood and blood-pressure readings, and the improvement tracked changes in four of its own seven inputs; between 21% and 31% of participants on the diet came out biologically older, depending on the trial; and the trial diet was supplied by a company in which two of the authors hold equity. That is one small trial set measuring a biological-age estimate rather than survival. Two later randomised pilot trials in healthy adults confirmed the short-term effects on the fuel gauges themselves. In one, 46 healthy adults were randomised to one of two 7-day fasting-mimicking diets or a normal-calorie diet, and both fasting-mimicking groups saw fasting glucose fall by about 10% and IGF-1 by about 35% compared with the control group;[38] in the other, 30 healthy adults were randomised to one of two 6-day versions or a control diet, and the fasting-mimicking groups showed lower fasting glucose and insulin resistance, higher ketones and a shift in an autophagy marker in blood cells, though not at every time point.[39] Neither ran beyond a single cycle or measured biological age, and both had company involvement, so the biological-age claim still rests on the one analysis above. The science of intermittent fasting and time-restricted eating is the everyday version of the same idea.
The resilience paradox is the essential caveat — Moderate, and the most important result on this page. A large 2024 study put 960 genetically diverse female mice on graded caloric restriction (20% or 40% fewer calories than free feeding) or intermittent fasting (one or two fasting days a week). Eating less did extend lifespan in proportion to the degree of restriction — but genetics influenced lifespan more than the diet did, and, strikingly, the traits that best predicted a long life were holding onto body weight through stressful handling, keeping a high lymphocyte proportion, and carrying more fat in late life. The deepest restriction extended lifespan most but cost lean mass and shifted the immune repertoire in ways that could raise susceptibility to infection, and the improvements in fasting glucose and adiposity that restriction produced were not the things associated with living longer.[40] The lesson: better metabolic numbers don't guarantee a longer life if they come at the expense of physiological reserve. A robust body needs enough muscle, lean mass and immune capacity to survive infection and stress, so the deepest, most continuous restriction is not automatically the best choice. Intermittent fasting is no free pass either: in the same study it did not extend life in mice that started out heavy, and two fasting days a week went with disruption of the cell populations that make red blood cells.[41] Note the single-sex cohort: given how sharply the amino-acid results below split by sex, a female-only study is a real limit on how far this generalises. A companion study shows why. It tested the same two-days-a-week fast in 800 mice of both sexes from 10 genetically distinct inbred strains and found a modest lifespan gain in males (median about 1.7 months) and none in the inbred females, although the fast had lengthened life in females of the original mixed-ancestry population, and the effect varied from strain to strain. Again, bigger metabolic improvements did not mean longer lives, and keeping more fat and more lymphocytes late in life predicted living longer.[42]
It's not just calories: amino-acid restriction
Strong in animals; Weak in humans. The animal work is consistent and replicated across labs, but the few human trials are short, limited to metabolic markers, and none has measured an aging outcome.
A calorie is not merely a calorie. In rodents, restricting the single amino acid methionine extends lifespan by 30–40% regardless of total calorie intake, so the composition of dietary protein tunes these pathways in its own right.[43]
- Branched-chain amino acids (BCAAs) — leucine, isoleucine, valine — are potent mTORC1 activators, and high blood and dietary levels have been linked to insulin resistance and obesity in both humans and rodents. In standard lab mice, a lifelong low-BCAA diet reduced frailty and raised median lifespan by about 32% in males, but female lifespan was no different from that of normally fed females; started in midlife, the diet lowered frailty in both sexes.[44] The effect turns out to be specific rather than generic. The metabolic harm of a high-BCAA diet is carried by isoleucine and valine, and restricting leucine alone — the most potent mTORC1 activator of the three — does almost nothing, which is the clearest sign that "activates mTORC1 hardest" is not the same as "matters most."[45] Restricting isoleucine alone extends lifespan in genetically diverse mice of both sexes, more so in males.[46] Restricting valine alone reduces frailty, cancer and senescent-cell burden in both sexes, but extends lifespan only in males, whose median lifespan ran about 23% longer than unrestricted males — one of the sharpest sex splits in the whole dietary-restriction literature, and a warning against assuming any of these results transfer evenly.[47]
- Methionine restriction is one of the best-validated single-nutrient interventions in animals. Methionine is the precursor to the cell's universal methyl donor (S-adenosylmethionine, or SAM). When SAM is plentiful, a dedicated sensor called SAMTOR lets mTORC1 switch on. Methionine restriction is thought to work partly by lowering SAM, which turns mTORC1 activity down and frees up autophagy; in rats it also raises total energy expenditure, and it improves insulin sensitivity, possibly through a rise in the hormone fibroblast growth factor 21 (FGF21).[48] In people the picture is thinner. In a blinded trial, 59 adults with overweight or obesity were given all their food for 8 weeks on a plant-based diet either low (about 2 g a day) or high (about 5.6 g a day) in the sulfur amino acids methionine and cysteine. The low group lost about 20% more weight (1.14 kg more), but fasting insulin and glucose did not differ, and the rise in FGF21 and fall in IGF-1 seen in animals did not appear, although an earlier 4-week feeding study had reported both.[49]
In practice this is much of the science under dietary patterns: plant-forward eating is naturally lower in methionine and BCAAs than a meat-heavy diet.
The drugs: caloric-restriction mimetics
Weak in humans, across the board. Every drug below has a strong mechanism and no trial showing it slows human aging as such, and the animal record is uneven: a 2025 meta-analysis of lifespan studies across eight vertebrate species found that rapamycin, but not metformin, significantly extended lifespan.[50] The GLP-1 drugs are the partial exception — they have large hard-outcome trials — but those trials tested disease, not aging. The site covers this class in depth under geroprotectors.[51]
- Rapamycin directly inhibits mTORC1 and is the most reproducible pharmacological lifespan-extender in animal models. The catch is dosing: daily high doses (as used in transplant medicine) eventually also hit mTORC2, causing insulin resistance and immune suppression, so longevity research uses low, intermittent — typically weekly — dosing to pulse autophagy while sparing mTORC2, a schedule worked out in mice before it was tried in people.[52][53] The human record is thinner than the enthusiasm. PEARL, which its authors describe as the largest and longest trial to date of low-dose weekly rapamycin for healthy aging, ran 48 weeks in generally healthy adults aged 50–85 and missed its primary endpoint: no reduction in visceral fat. It was well tolerated, and women on the higher dose gained lean mass and reported less pain, but those were secondary, subgroup-confined findings, and the trial used a compounded drug later found to deliver only about a third as much drug into the blood as generic sirolimus, rapamycin's standard pharmacy form.[54] Another small placebo-controlled trial, in 40 sedentary adults aged 65–85, used standard sirolimus at 6 mg once a week for 13 weeks on top of a home exercise programme: rapamycin did not add to the exercise gains, the per-protocol analysis favoured placebo, and although 85% of each group reported at least one side effect, the rapamycin group logged more in total (99 vs 63 events).[55] Across 19 human studies of rapamycin and its relatives, a 2024 systematic review found no serious drug-attributed side effects in healthy people, but more infections and higher cholesterol and triglycerides in people who already had age-related disease.[56] The strongest clinical human result for damping this pathway is older and comes from rapamycin relatives: a randomised trial gave 264 older adults six weeks of two low-dose mTOR inhibitors and found significantly fewer infections reported over the following year — the trial's central question — along with a better antibody response to influenza vaccination.[57] But a phase 3 trial in 1,024 adults of one of those drugs then found no reduction in respiratory illness (26% of people on the drug vs 25% on placebo),[58] so even the one place where turning the accelerator down produced a measurable clinical benefit in people has not held up at scale — and it was a six-week course judged on one year of infections, not a longevity result.
- Metformin activates AMPK, mimicking part of the calorie-restriction signal. In people with type 2 diabetes the observational record is mixed. A Danish registry study that included twin pairs found that people starting metformin alone had higher mortality than people without diabetes, contradicting the earlier finding that metformin users outlive non-diabetics,[59] while among 438 women aged 60 and over with new diabetes, starting metformin rather than a sulfonylurea went with about 30% lower risk of dying before 90 (HR 0.70), a comparison its authors say cannot show cause and effect.[60] In healthy adults aged 65 and over, metformin blunted the muscle gains from 14 weeks of resistance training.[61] The much-discussed TAME trial was designed to test whether metformin delays age-related disease in older adults without diabetes, as a randomised, placebo-controlled trial whose endpoint is a composite of major age-related diseases;[62] if it succeeded, it could set a precedent for recognising aging as a treatable condition. It has never enrolled anyone: about $5 million set aside by the US National Institute on Aging, plus private pledges, falls far short of its estimated $45–70 million cost,[63] so it remains a proposal rather than a study in progress (geroprotectors).
- NAD⁺ precursors — nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN) — aim to raise NAD⁺ and so keep the sirtuins supplied. The first human NR trial confirmed they safely and dose-dependently raise blood NAD⁺,[64] and a 60-day randomized trial of NMN in 80 healthy middle-aged adults, fully funded by the companies that make the NMN tested, reported raised blood NAD⁺ and longer six-minute walking distance, with benefits peaking at the middle dose (600 mg a day). Its "biological age" score, calculated from routine lab markers rather than a methylation clock, held steady on NMN, but the contrast rests mostly on the placebo group's score jumping from 39.8 to 45.4 years within 60 days, which points to a noisy measure.[65] Pooled trials since then are less encouraging. In placebo-controlled trials in adults averaging over 60, NMN did not improve muscle mass, grip strength, walking speed or chair-rise time, and NR was linked to worse physical-performance scores in people with mild cognitive impairment.[66] A 2026 pooling of 15 NMN trials found it well tolerated but without effect on body weight, fasting glucose, long-term average blood sugar (HbA1c), blood lipids or systolic blood pressure.[67] The repletion is real; a health payoff in healthy people, let alone a longevity one, has not been shown.
A newer addition acts on the same axis from a different angle: GLP-1 receptor agonists — glucagon-like peptide-1 drugs such as semaglutide and tirzepatide — reduce nutrient load and have produced a notable pharmacological signal on DNA-methylation aging clocks. Three caveats travel with that claim and should never be separated from it: the result is a post-hoc, exploratory analysis of a 32-week trial in a narrow special population (semaglutide n=45, placebo n=39), it has since passed peer review but the design limits remain, and two of the authors work for TruDiagnostic, a commercial epigenetic-testing company, whose scientific adviser is the first author.[68] Dietary caloric restriction moved one of these clocks (DunedinPACE) first, in a general healthy population — see GLP-1 drugs.
A practical, age-aware reading
Moderate for the direction of travel; Weak for the specifics — the age threshold below comes from one observational survey analysis that a re-analysis of the same survey and a larger cohort did not support.
The synthesis this page draws from all of the above (our reading, not any single study's conclusion) is oscillation, not suppression: protect metabolic flexibility and the youthful feed–fast swing while preserving functional reserve. The reserve half rests mainly on the mouse finding that better metabolic numbers did not track a longer life, even though in that study the deepest continuous restriction still extended lifespan most.[69] Two practical points stand out, both well covered elsewhere on the site:
- Protein needs may shift with age, but the mortality evidence for a sharp switch is thin and conflicting. Through midlife, a moderate, plant-forward protein intake keeps mTOR and IIS from idling high; in later life the priority tilts toward enough protein to defend against muscle loss and frailty. The well-known age "flip" comes from one US national-survey analysis of 6,381 adults aged 50 and over, built on one day of recalled diet, in which high protein went with higher death rates between 50 and 65 and lower ones after 65.[70] A 2025 re-analysis of the same survey, using estimates of people's usual intake and looking at protein gram by gram rather than high versus low, found no link between protein and death from any cause in either the under-65 or the over-65 group; its authors consult for the food industry.[71] And among about 19,400 UK Biobank adults, eating at least 1.8 g of protein per kilogram of body weight a day went with about 36% more major cardiovascular events after age 55 (HR 1.36) but no significant excess before 55, the opposite direction.[72] Treat the age threshold as unsettled: the case for more protein in older age rests on muscle and frailty, not on mortality data. The dose tables live under protein, and the training side under resistance training.
- Exercise has the best combined evidence-and-safety profile of any lever here. Exercise switches on AMPK in working muscle, and harder effort switches it on more: in eight sedentary men, a cycling bout at 80% of aerobic capacity raised AMPK activation 2.8-fold, while an easy bout at 40% that burned the same energy did not raise it measurably.[73] Training on an empty stomach adds little, if anything. In a crossover study of eight young trained men, AMPK activation was about 30% higher after fasted than after fed cycling, with no extra fat burning,[74] and a later, more tightly controlled crossover study in eight active men found no difference in AMPK activation itself.[75] On the NAD⁺ side, exercise training raises muscle levels of NAMPT (nicotinamide phosphoribosyltransferase), the enzyme that rebuilds NAD⁺, by about 1.46-fold on average across five human studies,[76] and older adults who train carry muscle NAD⁺ levels closer to young adults' than less active older adults do, though that comparison is a single snapshot rather than a trial.[77] All of this works the brake pathways with none of the dosing and tolerability problems that constrain the drugs. Zone 2 training is one good way to build the volume, but it is not the strongest AMPK stimulus.
What this does and doesn't tell you
What it tells you: nutrient sensing is the master integrative hallmark — four pathways (insulin/IGF-1, mTOR, AMPK, sirtuins) are the best candidate mechanism linking what you eat to the rate at which you age, a link established in cells and animals and inferred in humans. The accelerator-versus-brake logic explains why the best-evidenced longevity levers — eating with restraint, fasting, and exercise — plausibly work: they push the balance from growth toward repair. It also explains where the drug candidates are aimed.
What it doesn't tell you: that maximally suppressing growth signalling is the goal. The IGF-1 U-shape, the leucine result, and the resilience paradox all say otherwise — too little growth signalling, or too much weight and muscle loss, brings frailty and immune failure, and the longest-lived animals under restriction are the ones that keep their reserve. Nor does it mean any supplement or drug reliably extends human life: rapamycin's small randomised trials in generally healthy older adults have shown no clear benefit (the largest missed its primary endpoint), metformin's flagship aging trial has never raised enough money to start, and the NAD⁺ case rests mostly on raising a molecule whose downstream longevity payoff isn't yet demonstrated. The durable message is the familiar one — eat with restraint, keep the feed–fast rhythm, move, and don't keep the growth machinery switched on around the clock.
Further reading
- López-Otín C, et al. The hallmarks of aging. Cell 2013.[78]
- López-Otín C, et al. Hallmarks of aging: an expanding universe. Cell 2023.[79]
- Fernandes SA, Demetriades C. The multifaceted role of nutrient sensing and mTORC1 signaling in physiology and aging. Front Aging 2021.[80]
- Papadopoli D, et al. mTOR as a central regulator of lifespan and aging. F1000Res 2019.[81]
- Sadria M, et al. Interactions among mTORC, AMPK and SIRT: a computational model for cell energy balance and metabolism. Cell Commun Signal 2021.[82]
- Barcena ML, et al. Role of AMPK and sirtuins in aging heart: basic and translational aspects. Aging Dis 2024.[83]
- Ching TT, Hsu AL. The impacts of different dietary restriction regimens on aging and longevity: from yeast to humans. J Biomed Sci 2025.[84]
- Waziry R, Ryan CP, Belsky DW, et al. Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. Nat Aging 2023 — post hoc analysis within a randomised trial.[85]
- Das SK, et al. Body-composition changes in the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy (CALERIE)-2 study: a 2-y randomized controlled trial of calorie restriction in nonobese humans. Am J Clin Nutr 2017.[86]
- Di Francesco A, et al. Dietary restriction impacts health and lifespan of genetically diverse mice. Nature 2024 — the resilience-paradox study.[87]
- Brandhorst S, et al. Fasting-mimicking diet causes hepatic and blood markers changes indicating reduced biological age and disease risk. Nat Commun 2024.[88]
- Yu D, Richardson NE, Green CL, et al. The adverse metabolic effects of branched-chain amino acids are mediated by isoleucine and valine. Cell Metab 2021.[89]
- Calubag MF, et al. Lifelong restriction of dietary valine has sex-specific benefits for health and lifespan in mice. Nat Aging 2026.[90]
- Zhang N, et al. Methionine restriction — association with redox homeostasis and implications on aging and diseases. Redox Biol 2022.[91]
- Rahmani J, et al. Association between IGF-1 levels ranges and all-cause mortality: A meta-analysis. Aging Cell 2022.[92]
- Tabák AG, et al. Trajectories of glycaemia, insulin sensitivity, and insulin secretion before diagnosis of type 2 diabetes: an analysis from the Whitehall II study. Lancet 2009.[93]
- Papanikolaou Y, et al. Animal and plant protein usual intakes are not adversely associated with all-cause, cardiovascular disease-, or cancer-related mortality risk: an NHANES III analysis. Appl Physiol Nutr Metab 2025.[94]
- Moel M, et al. Influence of rapamycin on safety and healthspan metrics after one year: PEARL trial results. Aging 2025.[95]
- Stanfield B, et al. Exercise and Weekly Sirolimus (Rapamycin) in Older Adults: RAPA-EX-01 Randomised, Double-Blind, Placebo-Controlled Trial. J Cachexia Sarcopenia Muscle 2026.[96]
- Mannick JB, et al. TORC1 inhibition enhances immune function and reduces infections in the elderly. Sci Transl Med 2018.[97]
- Ivimey-Cook ER, et al. Rapamycin, Not Metformin, Mirrors Dietary Restriction-Driven Lifespan Extension in Vertebrates: A Meta-Analysis. Aging Cell 2025.[98]
- Trętowicz MM, et al. Human whole-blood NAD(+) levels do not vary with age or lifestyle interventions. Nat Metab 2026.[99]
- Corley MJ, Dwaraka V, et al. Semaglutide slows epigenetic aging in a randomized trial of HIV-associated lipohypertrophy. Nat Commun 2026.[100]
- Delrue C, Speeckaert MM. Rewinding the clock: emerging pharmacological strategies for human anti-aging therapy. Int J Mol Sci 2025.[101]