Autophagy: what the human evidence actually shows

Autophagy is your cells' built-in recycling program. When a protein gets damaged or an organelle wears out, the cell wraps it in a double-membrane bubble called an autophagosome. That bubble carries it to the lysosome, the cell's acid-filled recycling bin. There it is broken down into amino acids, fatty acids and other building blocks the cell can use again.^[6] Most of the time this ticks along quietly in the background. It speeds up under stress, for example when nutrients run short, oxygen runs low or misfolded proteins pile up.^[6]

That reuse is the point, which is why researchers describe autophagy as renovation rather than demolition. A widely cited review frames it as a supply of fresh building blocks and energy that lets cells and tissues rebuild themselves.^[6] The popular "detox" image misses this, because the cell is not simply throwing things away; it is stripping old parts for material.

Most of what we know about the machinery goes back to Yoshinori Ohsumi, who found the first autophagy genes in yeast.^[2] His work turned a structure seen under the microscope since the 1960s into a genetic pathway researchers could study, which earned him the 2016 Nobel Prize in Physiology or Medicine.^[1] Later work found the same machinery in mammals, including a protein-tagging system the process cannot run without.^[3]

That shared machinery gave researchers something to measure. A protein called LC3, the mammalian version of one of Ohsumi's yeast proteins, comes in two forms.^[4] One floats free, while the other carries a fatty tail that sticks to the autophagosome membrane.^[4] More of the membrane-bound form (LC3-II) means more autophagosomes are present, which can mean that more are being built or that fewer are being cleared.^[7] A second protein, p62, ferries tagged cargo into the autophagosome and is broken down along with it.^[5] So when p62 and LC3-II are both high, autophagy may be jammed rather than running fast.^[7]

Because a snapshot cannot tell those two stories apart, the field's reference manual insists on measuring turnover, which researchers call flux. Those consensus guidelines, written in 2021 by more than 2,800 authors, say a single reading of LC3 or p62 is not enough.^[7] Researchers have to block the lysosome in a parallel sample and see how much LC3-II backs up, or track the cargo over time. Two snapshots, such as a biopsy before and one after, are still two snapshots.^[7]

The same guidelines warn that blood cells alone cannot tell you how autophagy is doing in the rest of the body.^[7] Studies in people should therefore include cells from the tissue that matters as well. Even within blood, the markers can disagree: in freshly isolated T cells, p62 did not move together with LC3-II.^[7] So if a clinic offers you an "Autophagie-Panel" from one blood draw, with no flux measurement, that is not a validated test. Any claim that a diet, a pill or a cure "boosts autophagy" in people stands or falls on how it was measured.

Hard as it is to measure in people, autophagy has a firm place in aging biology. The 2023 update of the "Hallmarks of aging" framework promoted disabled autophagy from a sub-point under protein upkeep to one of twelve standalone hallmarks.^[9] To make that list, a change has to appear with age, speed up aging when made worse in lab animals, and slow aging when treated.^[9]

The strongest evidence for that place comes from lab animals. A 2018 review of yeast, worms, flies and mice found that many of the best-studied ways to extend lifespan need working autophagy genes.^[8] Those ways include eating less, dialling down growth and insulin signals, and giving spermidine.^[8] In worms and flies above all, knocking the genes out wipes out the extra lifespan.^[8] So the case for autophagy as a common thread in animal longevity is strong, and researchers had argued for it for years before that review.^[44]

Strong in animals does not mean "more is better", though. The genes that run autophagy also do many other jobs around the cell. A 2019 review of those genes in disease describes their effects as depending heavily on context.^[10] That is one reason the step from mouse to human is shorter on paper than in practice.

Three things make that step hard, and the first is measurement. In a mouse, researchers can read the markers straight from liver, brain and muscle. A person offers only the occasional muscle biopsy, plus blood markers that cannot stand in for tissue on their own.^[7] Time is the second obstacle. A mouse lifespan study fits into a few years,^[29] but a human one would take decades, so trials in people measure stand-ins rather than lifespan.

Cancer is the third, because autophagy is thought to help stop some tumours from starting but can help established tumours survive treatment.^[11] The section on diseases covers what that means for anyone with a cancer history. Together, the three obstacles mean a human claim about autophagy is only as good as its evidence on which switch was pulled and which tissue was measured.

Autophagy is controlled by two opposing sensors, mTOR and AMPK, and both answer the same question: does the cell have plenty of fuel, or is it running low? A 2011 study showed that both act on ULK1, the protein that starts autophagy.^[12] When glucose runs low, AMPK tags ULK1 at two sites and switches autophagy on. When nutrients are plentiful, mTOR tags it at a different site, which keeps autophagy off and stops AMPK from binding.^[12] AMPK is the accelerator and mTOR the brake, and both work through one protein.

mTOR is the "food is here, time to grow" sensor. Amino acids, especially leucine, switch it on, and so do insulin and a full energy supply.^[13] While it is on, the cell builds protein and stores fuel instead of recycling, and when its signalling goes wrong, it feeds into cancer, diabetes and aging.^[13] Rapamycin is the textbook mTOR blocker, and its lifespan effect in mice is thought to come partly from releasing that brake.^{[13, 29]}

AMPK works the other way round. It is a low-fuel sensor that switches on when the cell's energy charge drops, classically when glucose is short or muscles work hard.^[14] Once active, it starts autophagy, stops fat production and turns mTOR down.^[14] The diabetes drug metformin is thought to act partly through AMPK,^[14] a link the section on drugs looks at more closely.

Both sensors read the cell's fuel, but autophagy also has a side route that works even without directly blocking mTOR. In a 2014 study, lowering a cellular fuel molecule called acetyl-CoA, or turning down EP300, the enzyme that uses it to tag proteins, was enough to trigger autophagy.^[16] A companion review grouped natural compounds that act on this tagging as caloric restriction mimetics, a name for substances that copy some effects of eating less.^[17] In cell and rodent studies, spermidine and the green-tea compound EGCG block the enzymes that add the tags, while resveratrol boosts the enzymes that remove them.^[17] Spermidine blocked the purified EP300 enzyme directly and switched on autophagy in cultured cells, where the mTOR brake also eased.^[69]

In principle, then, three kinds of lever should push autophagy up.^{[12, 14, 16]} One lowers insulin and amino-acid signals (fasting, eating less), one drains the cell's energy charge (hard, sustained exercise), and one reduces that protein tagging. Whether a given lever does that in human tissue, at a dose people tolerate, is a separate question. For fasting, the lever people ask about most, that question takes a precise form: after how many hours?

Nobody has shown when autophagy starts during a fast in humans. No human study has found it switching on at 16 hours of fasting, and no human study has pinned down a start time at all.^{[49, 66]} That matters because "ab 16 Stunden setzt die Autophagie ein" is one of the most repeated lines in German-language fasting advice, often tied, openly or by hint, to Ohsumi's Nobel Prize.

What the muscle biopsies show

The study most often quoted on the question compared an overnight fast with a 72-hour fast in eight healthy men, with a thigh-muscle biopsy after each.^[49] After 72 hours, membrane-bound LC3B-II was up by about 30% and the mTOR signal had dropped, but p62 had also risen by about 10%.^[49] The authors themselves say that makes the flux hard to read, because both markers rising can mean more recycling or a jam at the lysosome.^[49] The study also did not test 16 hours, and its group was small: eight men, one muscle, no women and no older adults.

Shorter fasts give no clearer signal. Another study biopsied muscle at 2, 12, 24 and 36 hours of fasting in trained and untrained volunteers and found only modest changes.^[66] In the untrained group LC3-II and p62 actually fell, and nothing in the data looks like a switch at 16 hours.^[66] A third study, in women on an intermittent-fasting diet, found muscle autophagy markers lower rather than higher after 12- and 24-hour fasts, probably in response to weight loss.^[77] Across these small studies, human muscle shows modest or ambiguous change rather than a start time.

Blood gives a different picture, and an equally tentative one. An Australian trial measured autophagy flux, the actual turnover rather than a snapshot, in blood immune cells of 121 adults with obesity.^[52] After six months, people who fasted for about 20 hours on three days a week had higher flux than the standard-care group.^[52] Flux did not rise significantly within the fasting group itself, though, and the authors call the analysis exploratory.^[52] Because it tested 20-hour fasts in blood cells, it says nothing about a 16-hour mark in muscle or liver.

Where the 16 hours come from

If no human study shows it, where does the 16-hour figure come from? We found no source that traces it. It resembles the hours over which autophagy rises in starved rodents. A mouse, however, burns through its energy reserves far faster than a person, so scaling those hours to humans would be guesswork. We also found no statement in which Ohsumi gave a human fasting time; his prize was for yeast genetics.^[1]

What the 16:8 trials measured

The 16:8 trials that do exist measured weight, not autophagy. In the TREAT trial, 116 adults with overweight or obesity either ate only between noon and 8 pm or ate three structured meals a day for 12 weeks.^[24] Weight loss barely differed between the groups, and neither did the metabolic measures.^[24] In a subgroup measured in person, the 16:8 group had a lower lean-mass index for arms and legs, which is not the direction anyone fasting for health wants.^[24]

A longer trial saw more weight loss, though for a familiar reason. Adults with obesity ate within a noon-to-8-pm window for six months and then within a wider 10-am-to-8-pm window for another six.^[53] They were compared with a calorie-restriction group and a control group.^[53] The time-restricted group lost about 4.6 kg more than controls and reported eating about 425 kcal a day less, but it did no better than plain calorie restriction.^[53] Eating less is the likely engine, though the trial did not test that directly. Neither trial measured autophagy, so neither result supports calling 16:8 an "autophagy trigger".

Where fasting does show measurable benefits in people, the engine looks like fewer calories rather than an autophagy threshold. Trials of 4- and 6-hour eating windows.^[22] and of five-day fasting-mimicking cycles^[23] reported metabolic improvements. In the Madeo lab's alternate-day fasting trial in Graz, healthy, non-obese, middle-aged adults ate about a third fewer calories^[20] After four weeks they had less belly fat.^[20] Lower levels of sICAM-1, a marker linked to inflammation, showed up only in people who had already fasted this way for more than six months.^[20]

Two years of moderate calorie restriction in the CALERIE-2 trial improved weight, blood pressure, LDL cholesterol and insulin sensitivity.^[18] A later analysis of that trial found a slightly slower pace of aging on a DNA-based aging clock.^[19] None of these trials measured autophagy, so any link to it is inferred from the mechanism.

The timeline at a glance

The table puts the human fasting studies side by side, from the shortest fast to the longest, with what each one actually measured.

Sources: the studies cited in each row.

The hour on the clock is therefore the wrong thing to watch, since the benefits that fasting trials do show travel with eating less. If 16:8 helps you eat less without losing muscle, that is a fine reason to keep it. Just do not expect a cellular clean-out at hour sixteen.

Cutting protein rather than calories eases the autophagy brake in animals, but in people the one direct test found no effect. The idea is sound on paper, because the mTOR brake responds strongly to amino acids such as leucine.^[13] Yet when 63 healthy young adults spent four weeks on a reduced-protein diet without eating fewer calories, autophagy flux in their blood immune cells did not change.^[67]

The 2026 trial was randomised and used a crossover design, so each volunteer ate both diets in turn and served as their own comparison.^[67] It measured blood only, in young adults, and two of its authors hold a patent on the flux test.^[67] So it does not settle the question for muscle or for older people.

The reason many longevity fans eat less protein in midlife is an observational study, one that watched what people ate rather than changing it. A 2014 analysis followed 6,381 American adults from the NHANES survey for 18 years.^[25] People aged 50 to 65 who ate a lot of protein had about 75% higher all-cause mortality than the low-protein group, and about four times the cancer mortality.^[25]

That link was weaker or gone when the protein came from plants, and after 65 the pattern flipped, so low protein went with higher mortality.^[25] The catch is how protein was measured: a single 24-hour dietary recall at the start stood in for 18 years of eating.^[25] And the study counted deaths, not autophagy.

Mouse studies lean the same way, but they stay in mice. In a mouse study that tested 25 diets, lifespan peaked when protein was low and carbohydrates high, and cutting calories through high-protein diets brought no lifespan benefit.^[26] Another mouse study narrowed this to one amino acid: a diet poor in methionine extended lifespan and lowered the growth signal IGF-1, along with insulin and thyroid hormone.^[27] That study did not measure autophagy, so the link to autophagy is a hypothesis.

What follows for a reader is modest. Eating a bit less protein in midlife, with more of it from plants, rests on observational and animal data.^{[25, 26]} That makes it an option to discuss with your doctor rather than proven advice. If you cut back, do not go below the German reference intake.^[68] It is 0.8 g of protein per kg of body weight a day, or 1.0 g from age 65.^[68]

Pair any change with strength training, and if you have a low appetite, an illness or low muscle mass, do not cut protein at all. From about 65, the same survey data point towards more protein rather than less, and losing muscle (sarcopenia) becomes the bigger risk.^[25]

Four molecules lead the research on drugs and supplements for autophagy, each with real animal or cell data and an open question in people: their human evidence ranges from thin to absent, so it pays to know what each one has shown before you spend money on it. Rapamycin has the strongest animal record and the most serious risks, and spermidine has the most research from German-speaking countries. Resveratrol and metformin show the widest gap between mechanism and outcome.

Rapamycin: the strongest animal signal, and real risks

Rapamycin (sirolimus) has the strongest lifespan record of the four in mice. The US National Institute on Aging's Interventions Testing Program runs each test at three independent sites.^[29] There, rapamycin started at 600 days of age (late middle age for a mouse) extended lifespan by about 9–14% in both sexes.^[29] Its link to autophagy runs through blocking mTOR, which takes the brake off ULK1.^{[12, 29]}

The human evidence is about immune aging, not lifespan. In a 2014 trial, the related drug everolimus improved older adults' response to a flu vaccine by about 20%, at doses they tolerated reasonably well.^[30] A later trial combined a low dose of everolimus with a second mTOR blocker, dactolisib.^[31] Older adults on the combination reported fewer infections over the following year than those on placebo.^[31] That signal did not hold up in a larger phase 3 trial, the final stage of testing, where dactolisib alone did not reduce respiratory illness.^[56]

There is no established longevity dose. PEARL, a 48-week placebo-controlled trial in healthy adults, used 5 or 10 mg of compounded rapamycin once a week.^[65] It found no change in its main outcome, visceral fat (the fat packed around the organs), and side effects were about as common as on placebo.^[65] A lean-mass benefit appeared only in women on 10 mg, and the trial was run by a company that prescribes rapamycin.^[65]

Two smaller placebo-controlled trials in older adults found no benefit either. Daily 1 mg for eight weeks changed none of the measured functions, though some red-blood-cell values fell slightly.^[78] In an exercise trial, 6 mg a week for 13 weeks added nothing to the gains from a home exercise programme.^[79] Minor side effects were more common, and one serious infection may have been drug-related.^[79]

Sirolimus is an immunosuppressant, a drug that dampens the immune system. Its label lists infections, slow wound healing, low blood counts, raised blood lipids and blood sugar, high blood pressure, mouth sores and protein in the urine.^[64] These figures come from transplant patients on higher doses.^[64] Drugs that strongly block or speed up the liver enzyme CYP3A4 and the transporter P-gp (for example clarithromycin, ketoconazole, rifampicin) are not recommended with it.^[64] Live vaccines should be avoided as well.^[64]

Women need effective contraception during treatment and for 12 weeks after, and it should not be used in pregnancy.^[64] Anyone considering it needs a physician who checks blood values. With an active infection, a planned operation, pregnancy plans or other immunosuppressants, it is not a do-it-yourself option.

In Germany, sirolimus is prescription-only and licensed to prevent rejection after a kidney transplant and for a rare lung disease, LAM.^[64] A sirolimus gel is approved for a skin condition in tuberous sclerosis, and there is no longevity indication. Off-label longevity prescriptions are written on a private prescription (Privatrezept) by a small number of clinics, and statutory health insurance (GKV) does not pay. Our rapamycin in Germany guide covers the practical detail. The German law on advertising for medicines (HWG) also limits how the drug may be promoted (as of October 2026).

Spermidine: the Graz molecule

Spermidine, a compound found in wheat germ, soybeans and other foods,^[73] is the molecule most tied to research from German-speaking countries. Our spermidine guide covers it in full. In mice it extended lifespan and improved heart function in old animals, and the heart protection vanished in mice whose heart cells lacked a key autophagy gene (Atg5).^[33] That makes it one of the clearer mammal demonstrations that a benefit depends on autophagy, though the lifespan gain itself was not tested that way.^[33]

In people, the main trial did not find the hoped-for benefit. The SmartAge trial gave 100 older adults with self-reported memory decline about 0.9 mg of spermidine a day for 12 months.^[38] Their memory did not improve compared with placebo.^[38] At the full authorised 6 mg a day, a 2026 pilot trial in 40 adults over 65 found the supplement well tolerated over 13 weeks.^[75] People who had responded poorly to vaccines before also made more antibodies.^[75] The trial is small and exploratory, and two of its authors consult for The Longevity Labs, which makes spermidine supplements.^[75]

The best human signal is observational. The Bruneck study in South Tyrol followed 829 adults aged 45 to 84 for 20 years.^[34] Each standard step up in dietary spermidine went with about a quarter lower risk of dying from any cause.^[34] A second group followed in Salzburg showed the same, but both rest on food questionnaires and show a link, not a cause.^[34]

A 2024 study then recast spermidine as part of how fasting works.^[39] Fasting raised the body's own spermidine in yeast, flies, mice and people on supervised fasts lasting several days.^[39] Blocking spermidine production removed fasting's benefits, but those cause-and-effect tests were done in model organisms and cultured cells,^[39] so the study does not show that a supplement helps people.

Resveratrol: the mechanism holds, the mouse result does not

Resveratrol has a neat mechanism, and its file in people is still thin. In human cells and in the worm C. elegans, resveratrol needed autophagy and the enzyme SIRT1 to extend the worms' lifespan, just as calorie restriction did.^[43] The review on caloric restriction mimetics also lists it as a booster of the enzymes that remove the protein tags.^[17] In genetically diverse mice on a standard diet, however, the same testing program that confirmed rapamycin found that resveratrol did not extend lifespan.^[58] We found no human trial showing a benefit on hard outcomes such as disease or death. Like the ubiquinol absorption story in our CoQ10 guide, resveratrol has a defensible mechanism and retail copy that runs ahead of it, so the useful check is the human data, not the label.

Metformin: waiting for TAME

Metformin, the diabetes drug, is linked to aging mainly through the planned TAME trial, whose design was laid out in 2016.^[45] We found no published TAME results as of October 2026. In worms, its lifespan effect depends on gut microbes and on AMPK, though the authors read part of the AMPK result as protection from the drug's toxicity.^[28] An autophagy step was not measured there, and in the mouse testing program metformin alone did not extend lifespan.^[55]

Of the four, only rapamycin has a human signal worth weighing, and it comes with the risks of a transplant drug. That makes the lever that needs no prescription worth a closer look.

Hard exercise shows signs of faster autophagy in human muscle.^[60] Among the popular levers, it is the one with both a causal mouse study and human biopsy data behind it.^{[46, 60]} The human data come from one small study of trained athletes, though.^[60]

The causal case comes from mice. A 2012 study used mice with a mutation that leaves everyday autophagy intact but blocks the extra autophagy that exercise or starvation normally switches on.^[46] These mice had less endurance, handled blood sugar worse during exercise and lost exercise's usual protection against the blood-sugar problems a high-fat diet causes.^[46] In mice, then, exercise-triggered autophagy appears to contribute to exercise's metabolic benefit.

Human data are thinner, mostly because repeated muscle biopsies during exercise are hard to get. In a small study of trained cyclists, a hard two-hour ride showed signs of faster autophagic turnover in thigh muscle, which an easy ride did not.^[60] The cargo protein p62 fell, and autophagy genes became more active.^[60] Intensity mattered more than whether they had eaten, which fits with hard exercise switching on the low-fuel sensor AMPK.^{[14, 60]} Whether exercise's benefit in people runs through autophagy, though, no study has shown.

The practical upshot is simple. If autophagy is your reason to move, intensity seems to matter, so adding some hard sessions to easier endurance work is a reasonable choice. Our Zone 2 guide covers the evidence that links fitness to a longer life. Strength training counts for healthy aging too, though its autophagy link is less clear.

Much of the spermidine and autophagy research you read about comes from one network centred on Frank Madeo's lab at the University of Graz. Its papers include the study that tied spermidine-induced autophagy to longer life in model organisms.^[32] and the mouse heart study^[33] An overview of spermidine in health and disease.^[35] and the 2024 fasting study^[39] came from the same network. Autophagy as a longevity topic in Germany, Austria and Switzerland is therefore not an American import, and knowing who produced a finding helps you weigh it.

The human data come from a wider circle. The observational numbers stem from the Bruneck study in German-speaking South Tyrol, checked against a second group in Salzburg^[34] Researchers at Charité Berlin and in Greifswald ran SmartAge, the main human memory trial.^[38] It followed an earlier pilot in older adults at risk of dementia.^[36] and a safety study that found spermidine well tolerated^[37] A smaller Austrian study shows why that trial's placebo group matters. In people with dementia in Styrian nursing homes, a higher spermidine dose went with a small gain on a standard cognition test over three months.^[42] Without a placebo group, though, that result is hard to interpret.^[42]

Industry is part of the network too. The Graz spin-off The Longevity Labs notified its spermidine-rich wheat germ extract for the EU Novel Food list in 2017.^[51] The company sells the extract under the spermidineLIFE brand, and some trial authors disclose consultancies with it.^[75] That is a normal, disclosed link between academia and industry, not a scandal. It is still worth knowing when a press release sounds more certain than the trial behind it.

Read that way, the network's own record is balanced: strong mouse data, an observational link and a null result in the main human trial, all published.^{[33, 34, 38]} When wellness copy promises an "Autophagie-Kur" or "Autophagie-Lebensmittel", it usually draws on the mouse and cohort work from this network. The null result belongs in every fair summary of that work.

Outside the longevity debate, autophagy is linked to brain degeneration, cancer and a few rare inherited diseases. In none of the common ones has boosting it yet proven itself as a treatment. In brain diseases driven by protein build-up, such as Alzheimer's, Parkinson's, Huntington's and motor neuron disease, autophagy helps clear misfolded protein clumps.^[47] That is why researchers hope to speed it up. A second, more selective cleanup route, chaperone-mediated autophagy, also declines with age and is tied to Parkinson's and metabolic disease.^[48] Even so, we found no autophagy-boosting therapy for Alzheimer's, Parkinson's or Huntington's disease that has succeeded in a phase 3 trial, the large final stage of testing (as of October 2026).

Cancer is where "more is better" fails most clearly. Autophagy is thought to help stop some tumours from starting, but once a tumour is established, it can use autophagy to survive stress and chemotherapy.^[11] Even the evidence for the protective half is weaker than it once looked. Deletions of Beclin-1, a hub protein the autophagy machinery assembles around, turn up in many sporadic breast and ovarian cancers and were long cited as proof.^[15] Later sequencing showed that they mostly travel with the neighbouring BRCA1 gene, so they say little about Beclin-1 itself.^[57]

The practical rule therefore follows from the other half. If you have active or recently treated cancer, do not start spermidine, long fasts or any strong autophagy inducer without your oncologist on board.^[11]

In a few inherited diseases, such as the lysosomal storage diseases Pompe and Danon and some mitochondrial diseases, the lysosome or the recycling machinery itself is broken. These diseases show what failure looks like, and they are treated with disease-specific therapies, not with autophagy boosters. Across the whole list, a link to autophagy marks a research target, not an established treatment.

Wellness businesses in Germany, Austria and Switzerland (DACH) sell autophagy as supplements, fasting cures, blood treatments and off-label prescriptions. None of these has been shown to raise autophagy in human tissue. Some rest on real physiology, most promise more than their evidence carries, and the parts below show what each one can and cannot deliver.

Spermidine on the shelf

Spermidine supplements are the biggest category. The EU authorises a spermidine-rich wheat germ extract as a Novel Food for adult supplements, up to the equivalent of 6 mg spermidine a day.^[61] Pregnant and breastfeeding women are excluded.^[61] The Longevity Labs notified the extract in 2017, and Implementing Regulation (EU) 2020/443 later changed only its purity specification.^[51] That 6 mg is the authorised maximum, not a dose shown to work.

The best-known brand lists a monthly pack from about €62 (as of October 2026). What that money buys is a standardised dose within the authorised limit; a memory benefit has not been shown, because the main cognition trial, SmartAge, found none.^[38] Food is the cheaper source: adults in Europe already take in an estimated 10 to 14 mg of spermidine a day from their normal diet.^[73]

Fasting cures

Fasting clinics sell the "Autophagie-Kur". Buchinger fasting is a medically supervised very-low-calorie fast, often 7 days or longer, with vegetable broth, juice and a little honey, around 200–250 kcal a day.^[59] Autophagy has not been measured during Buchinger fasting itself. Two small trials of six- and seven-day fasting-mimicking diets reported higher autophagy readouts in blood, though one was sponsored by the diet's maker.^{[80, 81]} Blood cells cannot show what happens in other tissues,^[7] and the muscle-biopsy studies stop at 72 hours of a water-only fast.^[49]

The Buchinger Wilhelmi clinic's own study of 1,422 guests found that weight, blood pressure and blood values improved and serious problems were rare.^[59] It measured no autophagy, though, and clinic staff were among the authors.^[59] So "fasting cures switch on autophagy" is plausible but unmeasured. The same study also excluded several groups for whom a long fast is not safe, and the final section lists them.^[59]

Apheresis as "Blutwäsche"

Some private clinics sell therapeutic apheresis ("Blutwäsche", a blood-cleaning procedure) as a longevity treatment, sometimes with "Zellerneuerung" or autophagy language. There is no evidence that apheresis switches on autophagy. One small placebo-controlled trial in adults over 50 reported shifts in epigenetic aging clocks, DNA-based estimates of biological age, after plasma exchange.^[74] Its lead clinician co-founded a plasma-exchange company, and it measured lab markers, not health outcomes.^[74] That is far from proof of a longer or healthier life, and the medical uses of apheresis for specific diagnoses are a separate matter.

Rapamycin and combination products

Off-label rapamycin is a small private-clinic category. Its approval status, the missing longevity dose and its risks are set out in the section on drugs and supplements and in our rapamycin in Germany guide. The evidence rests on immune trials of rapamycin-like drugs.^{[30, 31]} and the mouse lifespan data^[29] A few small placebo-controlled trials in healthy adults add to it, and none of them showed a benefit on its main outcome.^{[65, 78, 79]} Dosing and blood-level monitoring are not standardised between providers.

"Autophagie-Komplex" products combine spermidine, resveratrol, green-tea EGCG, niacinamide and similar ingredients. No health claim using the word autophagy is authorised in the EU,^[63] and we found no evidence for these combinations beyond what each ingredient does alone. Our longevity supplements guide applies the same grading to other products. One ingredient needs particular care, because concentrated green-tea extract can harm the liver. Since 2022, EU law has banned supplement portions with 800 mg EGCG a day or more.^[71] The label must also warn against use on an empty stomach, by pregnant or breastfeeding women and under-18s, and alongside other green-tea products on the same day.^[71]

What insurers pay

Statutory and private insurers do not pay for spermidine supplements, autophagy blood panels or wellness fasting packages (as of October 2026). Medically indicated treatments, such as apheresis for specific diagnoses, are a separate matter. Buyers therefore pay out of pocket for products that are legal to sell but have not been shown to do what the word "autophagy" on the pack suggests.

No health claim using the word "autophagy" has been authorised in the EU. EFSA, the EU food-safety agency, assesses the science behind a claim, and the European Commission decides whether it goes on the list of permitted claims.^[63] Without that authorisation, a health claim on a food or supplement is prohibited.^[63]

The closest the system came was spermidine and hair. In 2011, EFSA assessed a claim that spermidine prolongs the growing phase (anagen) of the hair cycle.^[50] It did not reject the claim for weak evidence: the panel said the claimed effect was about treating a disease, which falls outside the health-claim rules.^[50] Commission Regulation (EU) No 1017/2013 then formally refused it.^[62]

That leaves marketers with softer words, and those do not escape the rules either. Any wording that implies a food is good for your health counts as a health claim, so "fördert die Zellerneuerung" (supports cell renewal) is covered too.^[63] A general health promise is only allowed alongside a specific authorised claim.^[63] Even the spermidine extract's Novel Food authorisation is a safety approval for a dose, not a statement about autophagy.^[61]

That makes the label a useful filter: a pack that promises to "activate autophagy" is making a claim no EU authority has accepted. Switzerland sits outside the EU, but its food law follows the EU claim rules closely, so Swiss labels carry no authorised autophagy claim either.

Clinics follow different rules, because the EU health-claims regulation covers foods, not clinic treatments. In Germany, advertising for treatments falls under the Heilmittelwerbegesetz, the law on advertising for medicines and therapies, which bans claiming effects a treatment does not have.^[70]

Autophagy is real aging biology, and many of the best-studied lifespan-extending interventions in animals need it.^{[8, 9]} In people, though, no study has found a fasting hour, food, supplement or clinic package that reliably switches it on in tissue.^{[49, 66, 67]} That makes the everyday choices the useful ones, and most of them cost little.

A 16:8 eating window is fine if it helps you eat less, but it is not an autophagy switch. The trials found weight loss only alongside eating less, and none of them measured autophagy.^{[24, 53]}

Exercise hard some of the time. In mice, exercise-triggered autophagy contributes to the metabolic benefit of exercise.^[46] In one small human study, a hard ride showed stronger signs of autophagic turnover in muscle than an easy one.^[60]

Do not overeat, but keep enough protein. If you eat less protein in midlife, stay at or above 0.8 g per kg of body weight a day, or 1.0 g from 65.^[68] Pair the change with strength training.

Spermidine is the easiest of these to try: food is the simple way in, and capsules are an optional extra. Wheat germ has about 35 mg per 100 g and soybeans about 20 mg, and other legumes, mushrooms and some mature cheeses also contain it.^[73] The human signal in its favour comes from food questionnaires and shows a link, not a cause,^[34] while the main supplement trial found no memory benefit.^[38] If you prefer capsules for a measured dose, they are a reasonable experiment for a healthy adult within the authorised limit; give it a fixed trial period, knowing that around €60 a month buys a standardised dose, not a proven outcome.

Rapamycin belongs with prescription drugs, not supplements. There is no established longevity dose, and the placebo-controlled trials in healthy adults have shown no benefit on their main outcomes.^{[65, 78, 79]} The label lists infections, slow wound healing and drug interactions, and it rules out use in pregnancy.^[64]

Who should not fast for days or self-experiment

Multi-day fasting is not for everyone. The Buchinger clinic's own study excluded people with severe wasting (cachexia), anorexia nervosa, advanced kidney, liver or cerebrovascular disease, dementia, and anyone pregnant or breastfeeding.^[59] If you have diabetes, take regular medication or have any chronic condition, plan a fast of several days with a physician, ideally one trained in fasting therapy.^[76] Anyone who is underweight or losing weight without trying should not use eating less as a lever. With active or recently treated cancer, talk to your oncologist before spermidine, long fasts or any strong autophagy inducer.^[11]

For everyone else, a reasonable starting point is regular exercise with some harder sessions, enough protein and no excess calories, with a spoon of wheat germ as an easy extra.