Autophagy: What Humans Actually Have Evidence For

Autophagy is your cells' built-in recycling program. Damaged proteins, worn-out organelles, and surplus bits of cytoplasm get wrapped in a double-membrane vesicle (called an autophagosome), shuttled to the lysosome (the cell's acid bath), and broken down into amino acids, fatty acids, and nucleotides the cell reuses.

It is housekeeping. Like all housekeeping, it ticks along quietly at a low rate most of the time. It ramps up under stress: starvation, low oxygen, infection, or a build-up of misfolded protein junk. And when it breaks, things go wrong in predictable ways.

We know this thanks to one person. Yoshinori Ohsumi ran a yeast-genetic screen in 1993 that pulled out the first autophagy genes, and the Nobel committee gave him the 2016 Nobel Prize in Physiology or Medicine for that single line of work. The Karolinska press release notes that although autophagosomes had been seen under electron microscopes back in the 1960s, "its fundamental importance in physiology and medicine was only recognized after Yoshinori Ohsumi's paradigm-shifting research in the 1990's." The founding paper is Tsukada M, Ohsumi Y. 1993 FEBS Letters 333(1-2):169-174. They starved yeast of nitrogen and screened for mutants that could not accumulate autophagic bodies in their vacuoles. Out came 75 mutants in 15 complementation groups: "At least 15 APG genes are involved in autophagy in yeast." That single screen turned autophagy from a microscopy curiosity into a tractable genetic pathway.

The conjugation machinery. Five years later, Mizushima N, Noda T, Yoshimori T, Tanida I, Ishii T, George MD, Klionsky DJ, Ohsumi M, Ohsumi Y. 1998 Nature 395:395-398 showed that "a unique covalent-modification system is essential for autophagy to occur." Apg12 gets covalently tacked onto Apg5 via an enzyme called Apg7. It looked exactly like the ubiquitin tagging system the cell uses to mark proteins for the trash, except this was the first non-ubiquitin protein doing the same trick. The same machinery (renamed ATG12-ATG5-ATG16L1) runs in human cells.

The marker scientists actually measure. Kabeya Y, Mizushima N, et al. 2000 EMBO Journal identified LC3 as the mammalian version of yeast Apg8p. LC3-I floats around in the cytoplasm; LC3-II gets a fatty tail attached and sticks to the autophagosome membrane. "The amount of LC3-II is correlated with the extent of autophagosome formation." Twenty-five years on, LC3-II is still the workhorse readout in human muscle-biopsy autophagy studies.

The cargo-selection adaptor. Bjørkøy G, Lamark T, et al. 2005 Journal of Cell Biology described p62/SQSTM1: it grabs ubiquitin-tagged junk with one hand and LC3 with the other, dragging the cargo to the autophagosome. Because p62 itself gets eaten by autophagy, tissue p62 levels work as a flux readout (with caveats). High p62 plus high LC3-II usually means autophagy is jammed, not running fast.

The methodology bible. Klionsky DJ et al. 2021 Autophagy is the fourth-edition consensus on how to measure autophagy properly. Over 2,800 contributing authors, 382 journal pages. The single line every lay reader should remember: "a single measurement (such as autophagic protein 1 or LC3 lipidation) is not adequate." You cannot read autophagy flux off one LC3-II snapshot. You need pulse-chase methods, lysosomal-inhibitor controls, or paired tissue samples.

Critically, Klionsky 2021 also states that LC3 and p62 in blood are unreliable proxies for what is happening in tissue. So when a private DACH clinic offers you an "Autophagie-Panel" from a vial of blood, the methodology bible disagrees with the sales pitch.

One last reframe that matters: autophagy is renovation, not demolition. Mizushima N, Komatsu M. 2011 Cell 147:728-741: autophagy "produces new building blocks and energy for cellular renovation and homeostasis." The popular "detox" framing badly misses the point. This is not garbage collection. It is the cell's primary recycling and rebuilding pathway.

The single strongest citation for the claim that autophagy is real aging biology (not supplement-marketing wishful thinking) is López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. 2023 Cell. "Hallmarks of aging: An expanding universe." The 2023 update promoted autophagy from a sub-component of proteostasis to its own hallmark, "disabled macroautophagy."

The full twelve hallmarks: genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, disabled macroautophagy, deregulated nutrient-sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, altered intercellular communication, chronic inflammation, and dysbiosis. To make the list, a hallmark has to (i) show up with age, (ii) accelerate aging when you experimentally enhance it, and (iii) be at least partly reversible. Autophagy clears all three bars.

The model-organism case is overwhelming. Hansen M, Rubinsztein DC, Walker DW. 2018 Nature Reviews Molecular Cell Biology. The headline finding is striking: every major life-extending intervention tested in yeast, worms, flies, and mice (caloric restriction, reduced TOR signalling, reduced insulin/IGF-1, mitochondrial perturbation, spermidine) needs functional autophagy genes to deliver its lifespan benefit. Knock out the core ATG genes and the lifespan extension disappears. Autophagy is the common denominator of every validated lifespan-extending intervention in model organisms.

The disease-side companion is Levine B, Kroemer G. 2019 Cell 176(1-2):11-42, a decade-on follow-up to their 2008 landmark (Cell 132(1):27-42). The 2019 review adds an honest framing: ATG genes have "diverse physiologically important roles in other membrane-trafficking and signaling pathways" beyond degradation. So "more autophagy = better" is wrong as a slogan. Autophagy is essential, context-dependent, and pleiotropic (it does many things at once).

Why the mouse-to-human jump is partial. Three problems converge:

1. Measurement. In mice you sacrifice the animal, fix the tissue, and read LC3-II/p62 directly from liver, brain, heart, and muscle. In humans you get skin punch biopsy, occasional skeletal muscle biopsy in research settings, and serum markers that the Klionsky guidelines explicitly warn against trusting.
2. Lifespan endpoints. A mouse lifespan study costs three years. A human lifespan study costs a researcher's entire career. Surrogate endpoints (DunedinPACE, biological age, biomarker panels) are still being validated.
3. The cancer paradox. Autophagy suppresses tumour initiation: Beclin-1 monoallelic deletion shows up in 40-75 % of sporadic breast and ovarian cancers (Liang XH et al. 1999 Nature 402:672-676). But once a tumour is established, autophagy helps it survive chemotherapy. So "boost autophagy with spermidine and resveratrol" is not safe blanket advice if you have active or recently treated cancer.

The honest 2026 framing: autophagy is real aging biology with overwhelming model-organism evidence. The human translation is partial, the measurement is hard, and the simplistic "more is better" intuition is wrong.

Autophagy is controlled by two opposing kinase complexes that monitor whether the cell has plenty of food or is running on fumes. The cleanest mechanistic paper for the switch is Kim J, Kundu M, Viollet B, Guan KL. 2011 Nature Cell Biology. The money quote: "AMPK promotes autophagy by directly activating Ulk1 through phosphorylation of Ser 317 and Ser 777" during glucose starvation, while "mTOR activity prevents Ulk1 activation by phosphorylating Ulk1 Ser 757 and disrupting the interaction between Ulk1 and AMPK."

Three phospho-sites on one kinase (ULK1) decide whether autophagy runs. AMPK punches the gas at S317 and S777. mTORC1 hits the brake at S757.

mTORC1 (mechanistic target of rapamycin complex 1) is the "food is here, grow!" sensor. It is activated by amino acids (especially leucine), insulin/IGF-1, and high cellular energy. When mTORC1 is on, the cell builds protein and stores fat. Autophagy gets suppressed. Saxton RA, Sabatini DM. 2017 Cell is the master review: "deregulated mTOR signaling is implicated in the progression of cancer and diabetes, as well as the aging process." Rapamycin is the textbook mTORC1 inhibitor. Its lifespan effect in mice (covered later) runs through ULK1 derepression and autophagy induction (plus other downstream effects).

AMPK (AMP-activated protein kinase) is the cellular fuel-low sensor. When the AMP:ATP ratio climbs (energy charge falls), AMPK switches on. Glucose starvation and sustained muscle contraction are the canonical triggers. Hardie DG, Ross FA, Hawley SA. 2012 Nature Reviews Molecular Cell Biology is the master review. AMPK phosphorylates ULK1 to start autophagy, blocks lipogenesis at ACC, and shuts down mTORC1. Metformin's autophagy linkage runs through AMPK.

The downstream complex. Once ULK1 is active, it phosphorylates Atg13 and FIP200/RB1CC1. The nucleation complex assembles around Beclin-1 and the class III PI3K VPS34. The Beclin-1 paper is Liang XH, Jackson S, Seaman M, Brown K, Kempkes B, Hibshoosh H, Levine B. 1999 Nature, "Induction of autophagy and inhibition of tumorigenesis by beclin 1." Beclin-1 monoallelic deletion shows up in 40-75 % of sporadic breast and ovarian cancers, the historical anchor for the "autophagy as tumour suppressor" framing.

The acetyl-CoA / EP300 side route. Not everything that boosts autophagy hits mTOR. Mariño G, Pietrocola F, Eisenberg T, et al. 2014 Molecular Cell showed that low cytosolic acetyl-CoA plus reduced EP300 acetyltransferase activity induces autophagy independently of mTOR. The companion review (Mariño G, Pietrocola F, Madeo F, Kroemer G. 2014 Autophagy 10(11):1879-1882) coined the label "caloric restriction mimetics" for natural compounds that hit this axis instead of mTOR directly. This is the mechanism that connects spermidine and resveratrol to autophagy.

The practical version for a reader. Anything that lowers insulin/IGF-1 (fasting, low protein), raises the AMP:ATP ratio (sustained exercise, ketosis), or depletes cytosolic acetyl-CoA (spermidine via SAT1, resveratrol via SIRT1/EP300) will, in principle, induce autophagy. Whether it does so meaningfully in human tissue, at tolerable exposures, with measurable clinical benefit, is what the rest of this guide tries to grade honestly.

This is the single most important section in the guide, because here is where popular DACH wellness messaging diverges most sharply from what humans have actually been measured to do.

The popular claim. Almost every German lifestyle magazine, fasting clinic brochure, and longevity influencer states some version of "ab 16 Stunden Fasten setzt die Autophagie ein" or "Autophagie startet nach 16 Stunden." Apotheken Umschau, FOCUS Gesundheit, and almost every Buchinger-aligned book carries this line. It is usually attributed, implicitly or explicitly, to Ohsumi's Nobel research.

What humans have actually been measured doing. Vendelbo MH, Møller AB, Christensen B, Nellemann B, Clasen BF, Nair KS, Jørgensen JO, Jessen N, Møller N. 2014 PLOS One. This is the cleanest published human autophagy-marker biopsy study, and the popular-press framing almost never cites it.

Design: 8 healthy men, studied twice. Once after a 10-hour overnight fast, once after 72 hours of fasting. Vastus lateralis (thigh) muscle biopsies at both timepoints. Findings: at 72 h, LC3B-II rose roughly 30 % and p62 rose roughly 10 %, with decreased mTOR and ULK1 phosphorylation.

The original authors themselves flag two caveats:

1. LC3B-II and p62 both rising at the same time makes flux interpretation ambiguous. Without a lysosomal-inhibitor control (e.g. bafilomycin A1 in cell models) or a pulse-chase, you cannot cleanly separate "more autophagosomes forming" (high flux, what you want) from "autophagosomes piling up because the lysosome cannot clear them fast enough" (blocked flux, bad).
2. n = 8, healthy young men, vastus lateralis only. No women, no older adults, no other tissues. This is the cleanest data we have. It is still small.

Now the important bit: there is no comparable peer-reviewed muscle-biopsy dataset in healthy adults at 16 hours of fasting that shows clean autophagy activation. The Pinto et al. 2022 and Bagherniya et al. 2022 reviews of intermittent fasting and tissue autophagy markers reach a consistent conclusion. Mice activate liver autophagy easily on the rodent equivalent of a 16-hour fast. Human muscle, in the published biopsy series, does not. The response is inconsistent, tissue-dependent, and often opposite-direction between mouse and human.

Where did the 16-hour number come from? Three sources converged:

1. Mouse-to-human extrapolation. A mouse has a metabolic rate around 7x per kg compared to a human. A 12-16-hour mouse fast is the protein-restriction-driven autophagy peak in mouse liver. Naive species-time scaling gives "about 16 hours" for humans. Real human metabolic kinetics are not 1:1.
2. The Ohsumi Nobel halo. Ohsumi won the 2016 Nobel Prize and gave many interviews. He consistently talked about yeast and mouse biology. He has not, to our knowledge, endorsed any specific human fasting time threshold. The "Ohsumi said 16 hours" framing is misattribution.
3. Popular IF books and clinical influencers. Jason Fung's The Obesity Code (2016), Mark Mattson's lay communications, and Buchinger-Wilhelmi-style fasting literature generalised rodent thresholds into a fixed human number. The German wellness press repeated it.

What the 16:8 RCT evidence actually shows. Lowe DA, Wu N, Rohdin-Bibby L, et al. 2020 JAMA Internal Medicine. The TREAT randomized clinical trial. n = 116 adults with overweight or obesity. TRE group ate noon to 8 pm; control ate three structured meals daily. 12 weeks. Negative trial. Between-group weight difference -0.26 kg, P = 0.63, with no significant change in any secondary metabolic outcome. The verbatim conclusion: "Time-restricted eating, in the absence of other interventions, is not more effective in weight loss than eating throughout the day."

TREAT is the cleanest evidence that the 16:8 schedule alone, the exact schedule DACH lifestyle media most often markets as an "autophagy trigger," does not produce meaningful metabolic benefit beyond the calorie change it inadvertently causes. And TREAT showed no such calorie change either.

Where fasting does have stronger human evidence. Cienfuegos 2020 Cell Metabolism (4- and 6-hour eating windows in n=58 adults with obesity) and Wei 2017 Science Translational Medicine (three monthly 5-day fasting-mimicking diet cycles in n=100) both show real metabolic improvements, but the mechanism is reduced calorie intake, not a magic autophagy threshold.

Stekovic S et al. 2019 Cell Metabolism.e6 is the Madeo Graz lab's flagship alternate-day fasting RCT. 60 healthy non-obese adults, 4 weeks. Result: weight loss, improved fat-to-lean ratio, reduced sICAM-1, lower fT3, decreased trunk fat. This is the strongest DACH-led human signal connecting a fasting protocol to autophagy-relevant molecular markers. Direct LC3/p62 muscle biopsy data remain limited.

The defensible 2026 framing: the cleanest peer-reviewed human muscle autophagy-marker change occurs at roughly 72 hours of fasting in 8 healthy men, with measurement caveats. No peer-reviewed human RCT shows clean autophagy activation at 16 hours. The 16-hour number is mostly mouse extrapolation, repeated until it sounded canonical.

If fasting is the popular mental model of autophagy induction, protein restriction is the better-evidenced version of the same idea. mTORC1 is exquisitely sensitive to amino acids, especially leucine. Lower the leucine signal, and autophagy goes up.

The headline human cohort paper. Levine ME, Suarez JA, Brandhorst S, et al. 2014 Cell Metabolism. NHANES 1988-94 data, 18-year follow-up, n = 6,381. Adults aged 50-65 who ate high protein had a 75 % higher all-cause mortality and a 4-fold higher cancer mortality vs the low-protein group. The signal flipped after 65, where low protein became associated with higher mortality. Plant-vs-animal protein modulated the effect. Caveats: observational data, food-frequency questionnaire measurement of protein, healthy-user confounding likely.

The strongest rodent geometric-framework study. Solon-Biet SM, McMahon AC, Ballard JWO, et al. 2014 Cell Metabolism. 25 diets, 858 mice. Longevity peaked at low-protein, high-carbohydrate ratios. The proposed mechanism: hepatic mTOR, circulating branched-chain amino acids, and glucose. The author's verbatim line: "calorie restriction via high-protein diets or dietary dilution showed no beneficial effects on lifespan." This is the strongest rodent paper for the framing: protein quantity and quality move longevity more than absolute calories do.

Methionine restriction. Miller RA et al. 2005 Aging Cell. Methionine-restricted mice lived longer, with reduced IGF-1, T4, and insulin: the canonical autophagy-permissive hormonal milieu. Methionine is enriched in animal protein. The plant-skew in the Levine 2014 protective signal is consistent with this.

The metformin / methionine bridge. Cabreiro F, Au C, Leung KY, et al. 2013 Cell. In C. elegans, metformin's lifespan benefit depends on the drug perturbing microbial folate and methionine metabolism. This is the bridge citation that links metformin to AMPK to autophagy, and metformin to microbiome to methionine restriction.

CALERIE-2: the cleanest human calorie-restriction trial. Kraus WE, Bhapkar M, Huffman KM, et al. 2019 Lancet Diabetes & Endocrinology. n=218 enrolled (healthy non-obese adults aged 21-50), randomised 2:1 to 25 % CR vs ad libitum for 2 years. Achieved restriction was roughly 12 %, not 25 %. CR group lost 7.5 kg, mostly fat. Persistent improvement in LDL-C, blood pressure, CRP, and HOMA-IR. CALERIE-2 did not publish a formal human autophagy-marker time series; the autophagy linkage is mechanistic, not biopsy-proven in this trial.

The biological-aging readout came later. Waziry R, Ryan CP, Corcoran DL, Sugden K, Caspi A, Moffitt TE, Belsky DW. 2023 Nature Aging. "Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial." Waziry's analytic sample was n=220 randomised with n=197 having DNA methylation data (slightly different from Kraus's enrolment count). The CR group showed a statistically significant slowing of the DunedinPACE pace-of-aging clock vs controls. Effect size is small in absolute terms. The headline is direction (slowdown), not magnitude.

The practical translation. In your 50s and 60s, lower-protein eating (Levine 2014's range was below 0.8 g/kg/day, well below typical Western intake) with a plant skew is the lifestyle that most closely mimics the autophagy-permissive endocrine milieu. After 65, the equation flips (Levine 2014 showed protective inversion) and sarcopenia risk dominates. This is why the longevity community's consensus around "protein periodisation" exists: slightly lower in midlife, higher in older age.

Three molecules dominate the pharmacological-autophagy conversation. The evidence quality varies enormously between them.

Rapamycin is the strongest mammalian lifespan signal. Harrison DE, Strong R, Sharp ZD, et al. 2009 Nature. "Rapamycin fed late in life extends lifespan in genetically heterogeneous mice." NIA Interventions Testing Program. Rapamycin started at 600 days of age (mouse equivalent of late middle age) in UM-HET3 mice. Median lifespan extension: +14 % in females, +9 % in males, replicated across three independent ITP sites. This remains the only small molecule with replicated lifespan extension when started in middle age in mammals. The autophagy linkage runs through mTORC1 inhibition and ULK1 derepression.

The credible human signal is Mannick JB, Del Giudice G, Lattanzi M, et al. 2014 Science Translational Medicine 6(268):268ra179. RAD001 (everolimus, a rapalog) "enhanced the response to the influenza vaccine by about 20 % at doses that were relatively well tolerated" in elderly subjects. The follow-up: Mannick JB et al. 2018 Sci Transl Med 10(449):eaaq1564. n=264 elderly, RAD001 + BEZ235 (dactolisib) combination vs placebo. Significant reduction in self-reported infection rate (P = 0.001) at 1 year. This is the cleanest human signal that a TORC1 inhibitor improves an objective immunological readout in older adults. The caveat is that the regimen included a PI3K/mTOR co-inhibitor, not pure rapamycin.

Rapamycin in DACH practice. Sirolimus is licensed in Germany only for immunosuppression after organ transplantation. Off-label longevity prescribing happens on Privatrezept through a small number of private clinics. It is not GKV-reimbursed. Advertising is restricted under the HWG (Heilmittelwerbegesetz). Dosing protocols vary widely between providers. The longevity dose (typically 4-6 mg weekly) sits below the transplant dose but above zero, and trough monitoring is not standardised in this off-label population. See our rapamycin in Germany guide for the practical and regulatory detail.

Spermidine: the Madeo Graz axis. The sister guide Spermidine and Longevity covers this in full. Two points matter most for the autophagy reader:

1. Eisenberg T et al. 2016 Nature Medicine. Lifespan extension in mice, improved cardiac function in aged mice, and Bruneck cohort association of dietary spermidine with cardiovascular health. The lifespan extension was abolished when autophagy was genetically blocked. One of the cleanest mammalian demonstrations that an intervention's lifespan benefit is autophagy-dependent.
2. Schwarz C, Benson GS, Horn N, et al. 2022 JAMA Network Open. SmartAge. n=100, 12 months, roughly 0.9 mg/day spermidine via wheat germ extract. Negative trial. "No significant changes were observed in mnemonic discrimination performance" between groups. The longest, largest spermidine cognition RCT to date is negative on its primary endpoint. Do not soft-pedal that. It is the single most important counterweight to the Madeo-lab pre-clinical case and to DACH wellness messaging.

The newest mechanistic synthesis: Hofer SJ, Daskalaki I, Bergmann M, et al. 2024 Nature Cell Biology. Fasting raises endogenous spermidine across species including human volunteers, and "perturbing the polyamine pathway in vivo abrogated the lifespan- and healthspan-extending effects" of fasting. Mechanism: autophagy plus eIF5A hypusination. The 2024 paper positions spermidine as a fasting-mimicking metabolite rather than a standalone intervention.

The Bruneck cohort: the strongest human spermidine signal. Kiechl S, Pechlaner R, Willeit P, et al. 2018 American Journal of Clinical Nutrition. n=829 adults aged 45-84 in South Tyrol, 20-year follow-up, 341 deaths. HR for all-cause mortality per 1-SD higher dietary spermidine: 0.74 (95 % CI 0.66-0.83, P<0.001) unadjusted; 0.76 (95 % CI 0.67-0.86, P<0.001) with covariates. Validated in a Salzburg cohort. Observational, food-frequency-questionnaire-based: the strongest human signal we have, but it cannot establish causation.

Resveratrol: the over-promised molecule. Morselli E et al. 2010 Cell Death & Disease. Caloric restriction and resveratrol promote longevity via SIRT1-dependent autophagy induction. The mechanistic case via cytosolic acetyl-CoA / EP300 (Mariño 2014 Mol Cell) is defensible in cell models. The honest counter-evidence: NIA ITP mouse lifespan studies (Strong et al. 2013 Aging Cell) showed resveratrol did not extend lifespan in genetically heterogeneous mice on a standard diet. Human RCTs for hard endpoints are absent. Resveratrol earns the same skepticism the CoQ10 guide applies to the ubiquinol bioavailability claim: defensible mechanism, weak clinical translation, retail framing that overstates both.

Metformin (TAME context). Barzilai N, Crandall JP, Kritchevsky SB, Espeland MA. 2016 Cell Metabolism. The original TAME framing paper. As of 2026, TAME has not yet read out hard-endpoint data. The AMPK to autophagy mechanism is defensible (Cabreiro 2013 worm data). The human longevity-endpoint case is not yet there.

Exercise is the under-marketed autophagy inducer. It does not need a supplement, a clinic, or a fasting schedule. And the mechanistic evidence for exercise-induced autophagy in mammals is unusually clean.

The cleanest mouse paper. He C, Bassik MC, Moresi V, Sun K, Wei Y, Zou Z, An Z, Loh J, Fisher J, Sun Q, Korsmeyer S, Packer M, May HI, Hill JA, Virgin HW, Gilpin C, Xiao G, Bassel-Duby R, Scherer PE, Levine B. 2012 Nature. "Exercise-induced BCL2-regulated autophagy is required for muscle glucose homeostasis." The authors engineered BCL2 AAA mice carrying a knock-in mutation that blocks exercise-stimulus disruption of the BCL2-Beclin-1 interaction. These mice have normal basal autophagy but cannot mount exercise-induced autophagy. The phenotype:

• Decreased exercise endurance.
• Altered glucose metabolism during acute exercise.
• Loss of exercise's protective effect against high-fat-diet-induced glucose intolerance.

This is one of the cleanest mammalian demonstrations that an intervention's metabolic benefit specifically requires functional autophagy in the relevant tissue. Exercise protects against insulin resistance partly through muscle autophagy.

Human exercise/autophagy data is sparser. Tissue biopsies in exercising humans are technically and ethically constrained. The Schwalm et al. 2015 FASEB Journal work on autophagy markers in human muscle after endurance vs resistance exercise gets cited often, but the body of clean head-to-head intensity-vs-volume autophagy human biopsy data is small.

The Vendelbo 2014 study (from the fasting section above) is the most-cited human muscle autophagy/mTOR readout in the field, and it tested fasting, not exercise.

Practical synthesis. Cardiorespiratory training (Zone 2, intervals, mixed) lowers mTOR signalling acutely, raises AMPK signalling acutely, and drives PGC-1α-mediated mitochondrial biogenesis chronically. All three converge on autophagy. The Zone 2 / VO₂max guide (Zone 2 Training & VO₂max) covers the cardiorespiratory-fitness mortality evidence in depth. Resistance training matters too (sarcopenia is itself a hallmark of aging), but its autophagy linkage is less clean.

The honest editorial point. Of all the things popular DACH wellness media calls an "autophagy trigger" (16-hour fasting, autophagy supplements, autophagy fasting clinics), exercise has the cleanest mechanistic mammalian evidence (He 2012 BCL2 AAA mice) and the strongest observational human survival evidence (the entire cardiorespiratory-fitness-mortality literature). It is also the one thing nobody can sell you in a bottle.

Read the autophagy-and-longevity literature of the last 15 years and Frank Madeo's lab at the University of Graz shows up in roughly 60 % of the high-impact papers. Autophagy as a longevity topic in DACH is not an American import. It is largely a DACH research domain, and the editorial frame of this guide reflects that.

The Graz network. Karl-Franzens-Universität Graz / BioTechMed-Graz is the central node. The 2009 Nature Cell Biology spermidine-induced-autophagy paper (Eisenberg T, Knauer H, Schauer A, et al.), the 2016 Nature Medicine cardioprotection paper (Eisenberg 2016), the 2018 Science spermidine-in-health-and-disease review (Madeo F, Eisenberg T, Pietrocola F, Kroemer G. 2018 Science 359:eaan2788, the SmartAge Charité-collaboration RCT, and the 2024 Nature Cell Biology fasting-mimicking spermidine paper (Hofer 2024) all run through this network.

The Bruneck cohort (Kiechl 2018 AJCN) is South Tyrol, the German-speaking northern Italian autonomous province, recruited and followed via Innsbruck/Bolzano collaboration. The validation cohort is Salzburg (Paracelsus Medical University). DACH-adjacent.

Charité Berlin and DZNE Berlin (Agnes Flöel's group, Claudia Schwarz, Miranka Wirth) ran the SmartAge cognition RCT (Schwarz 2022 JAMA Netw Open) and the Wirth 2018 Cortex pilot. Berlin plus Graz is the operational axis of human spermidine trials.

Fachhochschule Wiener Neustadt (Pekar group) ran the 2020 Wiener Klinische Wochenschrift trial in established dementia in Styrian nursing homes (n=85). Result: a small MMSE benefit in the higher-dose arm. The trial had no placebo arm.

Longevity Labs+ (Graz, AT) is the spin-off that holds the EU Novel Food authorisation for spermidine-rich wheat germ extract. The flagship product is spermidineLIFE. A non-trivial share of DACH human-trial funding flows through Longevity Labs+ industry partnerships. This is a disclosed, normal academic-industry arrangement, not a scandal, but it is context every reader should bring to the literature.

Why this matters editorially. When DACH wellness media talks about "Autophagie-Fasten," "Autophagie Kur," or "Autophagie-Lebensmittel," they are usually pulling from this network's mouse and observational data. The Madeo group itself has been honest about the limits. Frank Madeo and Tobias Eisenberg have publicly told German-language outlets that eating wheat germ is a reasonable alternative to buying a spermidine supplement. The science is real, the people doing it are credible, the research is European, and the gap between DACH wellness retail framing and the actual peer-reviewed evidence (especially the SmartAge null result) is exactly what this guide tries to close.

Where the autophagy literature meets clinical medicine, the picture is mixed. Powerful in some disease areas, context-dependent in others, and over-claimed in retail wellness messaging.

Neurodegeneration. Menzies FM, Fleming A, Caricasole A, et al. (Rubinsztein DC senior). 2017 Neuron. Autophagy is "protective against neurodegeneration linked to protein accumulation," which is relevant to Alzheimer's (amyloid-β, tau), Huntington's (mutant huntingtin), Parkinson's (α-synuclein), and motor neuron disease. The therapeutic hope is to enhance autophagic clearance of misfolded protein aggregates. The honest counterweight: no autophagy-enhancing therapy has yet read out a positive phase 3 trial in a major neurodegenerative indication.

Cancer: the double-edged sword. Levine B, Kroemer G. 2008 Cell. The original framing paper. Autophagy is context-dependent in cancer: tumour-suppressive at the initiation stage (Beclin-1 monoallelic deletion in 40-75 % of sporadic breast and ovarian cancers, Liang 1999 Nature), but pro-survival in established tumours under chemotherapy stress. "Do not start spermidine or strong autophagy inducers without oncology input if you have active or recently treated cancer." This is the same caveat the existing spermidine guide carries, and it is the single most important practical safety point in the whole autophagy literature.

Chaperone-mediated autophagy in aging. Cuervo AM, Wong E. 2014 Cell Research. CMA (the LAMP2A/HSC70-mediated selective degradation pathway) is distinct from macroautophagy and declines with age separately. CMA is implicated in lysosomal storage disorders, Parkinson's (α-synuclein clearance), and metabolic syndrome. Practical implication: "autophagy" in research usually means macroautophagy; CMA is a related but distinct pathway with its own decline.

Cardiovascular. Mostly covered by the Eisenberg 2016 Nature Medicine and Kiechl 2018 AJCN Bruneck cohort signal in the spermidine section. The Pietrocola / Bravo-San Pedro reviews on cardiac autophagy biology are useful for a deeper dive but not essential to a generalist guide.

Lysosomal storage diseases. Pompe disease (GAA deficiency), Danon disease (LAMP2 deficiency), and the COQ-related primary mitochondrial diseases are paradigmatic of "autophagy fails when lysosomal degradation fails." These are clinical genetics, not lifestyle wellness, but they are useful as the disease-mechanism counterpoint to retail claims that "more autophagy is always better."

Infection and immunity. Xenophagy (selective autophagy of intracellular pathogens) is the autophagy arm of host defence against M. tuberculosis, Salmonella, and several viral infections. This is more biology than longevity-relevant therapy, but it explains why autophagy genes are evolutionarily conserved and broadly essential.

The honest synthesis. Autophagy biology is essential and pleiotropic. Some clinical indications (lysosomal storage diseases, primary mitochondrial disease) are real treatment indications. Others (neurodegeneration prevention, cardioprotection) are mechanistically plausible but have not produced positive phase 3 trials in healthy adults. And in cancer, the simplistic "more autophagy = better" framing is wrong: stage and tumour type matter.

DACH wellness marketing has built an entire vertical around autophagy. Some of it has defensible physiology underneath. Most of it does not have the RCT evidence the retail claims imply.

Spermidine retail. Covered in detail in the spermidine guide. EU Novel Food authorisation under Implementing Regulation (EU) 2017/2470 (Union list), amended by Commission Implementing Regulation (EU) 2020/443 of 25 March 2020, covers "spermidine-rich wheat germ extract from Triticum aestivum" with a maximum daily intake equivalent to 6 mg spermidine for food supplements for the adult population, excluding pregnant and lactating women. Longevity Labs+ (Graz, AT) holds the original authorisation. Synthetic spermidine is not authorised. Representative DACH products: spermidineLIFE, Doppelherz Spermidin, Sunday Natural Spermidin, Sanct Bernhard Spermidin, Green Naturals Spermidin, spermidinPLUS. Monthly cost typically 30-70 €. The largest cognition RCT (SmartAge) was negative on its primary endpoint. Keep that in mind when reading retail copy.

"Autophagie Kur" and "Autophagie Fasten": the fasting-clinic vertical. Buchinger-Wilhelmi (Überlingen am Bodensee, Marbella) and other Buchinger-tradition fasting clinics use autophagy language in their educational material. The general framing is that therapeutic fasting (often 7-14+ day water-and-broth protocols) induces autophagy. Most clinical claims stay hedged ("supports autophagy," "fördert die zelluläre Erneuerung") rather than making explicit disease-prevention claims, which keeps them outside the Article 14 disease-claim regulation under Reg (EC) 1924/2006.

Here is an honest editorial read: Buchinger Wilhelmi's underlying physiology framing (that long fasts of 5+ days do something measurable to autophagy in humans) is more defensible than the popular 16:8 claim. The Vendelbo 2014 muscle biopsy data and the broader extended-fast clinical literature (Wilhelmi de Toledo F et al. 2019 PLOS One on over 1,400 patients in 4-21-day fasting protocols) do show measurable physiology. Direct flux-validated autophagy biopsy time series in Buchinger-style cohorts are still sparse in the peer-reviewed literature, but the broader physiological case is real. The 16-hour wellness slogan is mouse extrapolation. The 7-day Buchinger fast at least sits on top of a small clean human muscle-biopsy paper. That is not an endorsement of fasting clinics; it is honest grading of the underlying evidence.

Wellness apheresis ("Blutwäsche" or "Blutreinigung"). Some private DACH clinics (e.g. Apherese Frankfurt, Vital Path Germany) market therapeutic apheresis as a longevity therapy. Marketing copy sometimes invokes "Zellerneuerung" or autophagy-adjacent framing. There is no peer-reviewed evidence that therapeutic apheresis induces autophagy in any tissue, nor any RCT supporting apheresis as a longevity intervention. Apheresis has legitimate medical indications (familial hypercholesterolaemia, certain autoimmune disorders, ANCA vasculitis) at €1,500-3,000 per session. The "longevity / autophagy" framing is marketing, not evidence.

Rapamycin off-label. A small but growing private-clinic vertical in DE/AT/CH. The Rapamycin in Germany guide covers this in detail. Sirolimus is licensed only for transplant immunosuppression. Longevity prescribing happens on Privatrezept, is not GKV-reimbursed, and HWG advertising restrictions apply. Dose protocols vary widely between providers (typically 4-6 mg weekly, with or without trough monitoring). The evidence base is the Mannick STM trials and the mouse ITP data: both real, neither a phase 3 longevity RCT.

"Autophagie-Komplex" combination supplements. Multiple DACH supplement vendors market combination products (spermidine + resveratrol + green tea EGCG + niacinamide) as "Autophagie-Komplex." None of these claims are EFSA-authorised (see next section). The wording "supports cellular renewal" or "fördert die Zellerneuerung" is the regulatorily safer alternative to explicit autophagy-induction claims. Actual evidence for combinations beyond their individual components is essentially zero.

No GKV/ÖGK/LAMal reimbursement for any autophagy intervention. Not in the wellness baskets at TK, Barmer, DAK, AOK, ÖGK, or LAMal as of 2026.

Short answer: no. EFSA's Panel on Dietetic Products, Nutrition and Allergies (NDA) has not authorised any health claim that uses the word "autophagy" under Article 13(1) or Article 14 of Regulation (EC) No 1924/2006 on nutrition and health claims made on foods.

The closest EFSA opinion in this space is EFSA NDA Panel, 2011, EFSA Journal 9(12):2466. "Scientific Opinion on the substantiation of a health claim related to spermidine and prolongation of the growing phase (anagen) of the hair cycle pursuant to Article 13(5) of Regulation (EC) No 1924/2006." The regulatory record shows this claim was not authorised: no positive opinion appears in the Union list of authorised claims.

What this means in practice.

1. EU food-supplement labels and DACH marketing copy may not legally state "induces autophagy," "promotes autophagy," or "supports autophagy" as an Article 13(1) or Article 14 health claim. Where such language appears in DACH retail, it is either non-compliant or routed through softer "cellular renewal" / "Zellerneuerung" wording that does not trigger the strict claim regime.
2. Spermidine products fall under the Novel Food regime (Reg (EU) 2015/2283, listing via IR (EU) 2017/2470, amended by IR (EU) 2020/443) for ingredient authorisation. But no autophagy-specific health claim has been authorised at the substance level.
3. The EFSA NDA panel has historically required a measurable physiological outcome in humans for an Article 13(1) claim. Autophagy-marker biology is currently considered a mechanism, not an authorised endpoint. This is consistent with the Klionsky 2021 guidelines' position that blood LC3/p62 are not reliable proxies for tissue autophagy.
4. Switzerland sits outside the EU. Swiss food law (BLV / LMG / LIV) mirrors EU rules closely on health claims, and the practical effect on labelling is the same: no authorised autophagy claims, identical wheat germ extract products marketed under softer language.

Why this matters for the editorial frame. Throughout this guide we describe what trials measured and what cohorts were associated with. We do not phrase findings as label-style benefit assertions, because the regulatory reality is clear: food supplement, no authorised autophagy claim, no GKV / ÖGK / LAMal reimbursement, no medical-device-grade quality regime. The science is real. The human evidence is partial. And no EU-authorised health claim exists.

Is autophagy real biology? Yes, unambiguously. The 2016 Nobel Prize (Ohsumi), the López-Otín 2023 hallmarks framework (disabled macroautophagy listed as one of 12 hallmarks of aging), and the Hansen 2018 Nature Reviews (autophagy is the common denominator of every validated lifespan intervention in model organisms) are not the soft end of the literature. They are the consensus.

Does "16 hours of fasting trigger autophagy" in humans? The cleanest peer-reviewed muscle-biopsy data (Vendelbo 2014 PLOS One, n=8 healthy men) shows LC3B-II rises and mTOR signalling drops at 72 hours, not 16. No peer-reviewed human RCT shows clean autophagy activation at 16 hours. The 16-hour number is mostly mouse extrapolation. The TREAT RCT (Lowe 2020 JAMA Intern Med, n=116) showed 12 weeks of 16:8 produced no weight or metabolic benefit beyond the calorie change it inadvertently caused. The popular DACH wellness claim is wrong.

Does spermidine extend human lifespan or prevent dementia? Mechanism: yes. Mouse lifespan extension: yes (Eisenberg 2016, autophagy-dependent). Bruneck cohort observational mortality: HR 0.74 per 1-SD higher dietary spermidine. SmartAge RCT (Schwarz 2022 JAMA Netw Open, n=100, 12 months) for cognition: NEGATIVE. That is the largest, longest interventional human study and it did not find a benefit. Do not soft-pedal that. The strongest human signals are: (a) the Bruneck observational cardiovascular-mortality association, (b) the mechanistic 2024 Hofer paper positioning spermidine as a fasting-mimicking metabolite. Dietary spermidine from wheat germ, soy, mature cheese, mushrooms, and pulses is reasonable. Buying a 50 EUR/month supplement is not RCT-evidenced for hard endpoints.

Does rapamycin work? In mice yes (Harrison 2009, +14 % / +9 % median lifespan). In humans the Mannick STM trials (immunosenescence) are credible signals. A phase 3 longevity RCT does not exist. In DE the drug is prescription-only for organ-transplant indication; off-label longevity prescribing is rare, Privatrezept-only, and not standardised.

Does resveratrol work? Defensible cell-model mechanism (Mariño 2014 Mol Cell, acetyl-CoA/EP300 axis). Negative in NIA ITP mouse lifespan trials on standard diet. No positive human RCT for hard endpoints. Apply the same skepticism the CoQ10 guide applies to the ubiquinol bioavailability claim.

What about exercise? Cleanest mammalian causal evidence (He 2012 Nature, BCL2 AAA mice: exercise's metabolic protection specifically requires functional muscle autophagy). Strongest observational human mortality association (cardiorespiratory fitness, see Zone 2 guide). And the one thing nobody can sell you in a bottle.

Wellness apheresis, "Autophagie Kur" boutique packages, "Autophagie-Komplex" combo supplements? No EU-authorised autophagy health claim exists. None of these products has phase 3 RCT evidence for autophagy induction in tissue. Some are harmless and expensive. Some are not harmless. None are GKV/ÖGK/LAMal-reimbursed.

The defensible practical takeaway. The lifestyle that mimics autophagy-permissive physiology is the same lifestyle that mimics caloric restriction and exercise. Eat fewer calories overall. Eat less animal protein in your 50s and 60s, with a plant skew. Exercise (it has its own validated autophagy mechanism via BCL2-regulated autophagy). Get plenty of wheat germ, pulses, mature cheese, and mushrooms as dietary spermidine sources. Do not expect a magic 16-hour-fasting threshold or a 50 EUR supplement to do anything special. If you want to try a longer fast (3-5 days), do it with medical supervision if you have any chronic condition, and recognise the human evidence base at long fast durations is still based on small biopsy cohorts.

The biology is real. The mouse data is overwhelming. The human translation is partial. Don't let DACH wellness marketing tell you otherwise.