The biology won a Nobel Prize; the timeline attached to it was never measured#
Autophagy — cells dismantling and recycling their own damaged components — is settled science. It won the 2016 Nobel Prize in Physiology or Medicine, awarded to Yoshinori Ohsumi "for his discoveries of mechanisms for autophagy"1. What has never been established is the number everyone quotes. There is no human study behind "autophagy kicks in at 16 hours," or 18, or 24, or 72. Those thresholds were not measured in people; they were inferred from yeast and rodent work and then hardened into a schedule by repetition.
So the position this article defends is narrow and specific: the mechanism is genuine and important, the human dosing is unknown, and the three human studies that have looked hardest at it disagree in instructive ways. That is a different claim from "autophagy is a myth." It is closer to knowing that a drug works in cell culture while nobody has run the dose-finding trial. The intermittent fasting pillar is the place for what a window does to your weight; this page is about the one claimed benefit that the scale can't see and, so far, neither can anyone else.
The trigger is a fuel gauge, not a stopwatch#
The reason an hours-based threshold is the wrong shape of claim becomes clear once you look at what actually starts the process. Autophagy is governed by two opposing nutrient-sensing kinases. AMPK, which activates when cellular energy runs low, switches on the initiating kinase ULK1 by phosphorylating it at Ser 317 and Ser 777. mTOR, which is active when nutrients are plentiful, does the reverse — it phosphorylates ULK1 at Ser 757 and physically breaks up the interaction between ULK1 and AMPK2.
Read that as a control system and the popular framing collapses. Neither kinase can read a clock. They read the energy status of the cell, which depends on how much glycogen you had when you stopped eating, how much you have moved since, how much muscle you carry, and what your last meal contained. Two people at hour sixteen of the same window can sit on opposite sides of that switch. An elapsed-time threshold is a proxy for a proxy — and a poor one, because the variance between people at any given hour is exactly the thing the number pretends away.
It is also worth noticing that fasting is not the only input. Exercise activates AMPK too, which is one reason the fasting-versus-fed training question has its own separate answer and why an "autophagy fast" and a hard training session are not cleanly separable interventions.
The entire human record fits in one table#
Animal and cell work on autophagy runs to tens of thousands of papers. Human measurement, in the context of fasting or time-restricted eating, is close to three studies — and each measured something different.
| Study | What was measured | Duration & n | Result |
|---|---|---|---|
| Dethlefsen et al., 2018 | LC3 and p62 protein in muscle biopsies at 2, 12, 24, 36 h | 36-hour fast; trained vs untrained | Fasting reduced LC3I, LC3II and p62 — in untrained subjects only |
| Jamshed et al., 2019 | Gene expression of LC3A in blood | 4 days early time-restricted eating, n = 11 | LC3A expression higher in the morning (p < 0.04) |
| Bensalem et al., 2025 | Autophagic flux of LC3B-II in blood mononuclear cells | 6 months, n = 121 | Differed from standard care at 6 months (P = 0.04); no significant rise from baseline within the fasting group |
Three things follow from reading down that middle column. The first study measured static protein levels; the second measured messenger RNA; only the third measured flux — the rate at which material is actually moving through the pathway, which is the only one of the three that answers "is autophagy happening faster?" A rise in LC3 protein can mean autophagosomes are being built faster or that they are being cleared more slowly, which is why Dethlefsen's result — markers going down after a day and a half without food — is not the paradox it first looks like, and is also why it cannot be scored as a win for either side.
Second, look at the durations against the claims. The one human time-course anyone has run went to 36 hours, sampling four times along the way, and found the response modest and dependent on training status. That is the empirical basis of the entire "hours to autophagy" genre — a single study, in muscle, in which the direction of the marker change was down.
Third, the strongest study is the most recent and the most deflating. Bensalem's team measured autophagic flux in 121 people over six months, the largest human autophagy measurement in a nutritional trial to date. The fasting-plus-time-restricted-eating group differed from standard care at six months, but did not significantly increase from its own baseline — meaning part of the between-group signal comes from where the control group drifted. The authors' own framing is that further studies are required to confirm whether nutrient restriction modifies human autophagy at all. When the best available evidence is exploratory and its own authors label it that way, "unlock autophagy at hour sixteen" is not a cautious summary of it.
A mouse's 24-hour fast is not your 24-hour fast#
The rodent literature is where the dramatic numbers live, and it does not convert. A review of fasting procedures in mice makes the mismatch explicit: mice are nocturnal, their metabolic rate is far higher than a human's, gastrointestinal transit takes roughly 6 hours against 40 to 70 in people, and they are grazers rather than meal-eaters — so overnight fasting of a mouse is not comparable to overnight fasting of a person, and fasting duration in mice has substantial effects on hormone balance, metabolism, body temperature and body weight6.
Run the consequence forward. A 24-hour fast in an animal that empties its gut in a quarter of the time and burns through its stores far faster is, in physiological terms, a much deeper deprivation than 24 hours in you. Which means the translation error in the popular timelines does not merely add uncertainty — it runs in a known direction. Hours borrowed from mice systematically overstate what the same clock time does in a human body. The metabolic-switch numbers people cite for ketones have the same problem in milder form, and the pillar handles that one.
What to do with a mechanism nobody has dosed#
The practical answer is unexciting, which is usually the sign it is the right one. Do not pick your eating schedule for autophagy. There is no threshold to aim at, no way to tell from the outside whether you crossed one, and no human trial linking a particular fasting duration to a clinical outcome by way of autophagy. Anything you might do to chase it — a longer fast, a tighter window, a full day off food — has costs that have been measured: more lean-mass risk as the window tightens, harder protein targets, and worse adherence.
Pick your schedule for the reason the trials actually support — whether it helps you eat less without misery — and treat any cellular housekeeping as a possible bonus you cannot observe. If the weight-independent effects are what interest you, the better-evidenced ones are metabolic rather than cellular: blood pressure, fasting glucose and insulin all move a little under time-restricted eating, with the effects concentrated in people who start out metabolically worst, and that literature is worth reading properly because it has real trials behind it. The autophagy story may eventually join it. Right now it is a mechanism with a Nobel Prize and no dose.
FAQ#
How many hours of fasting does it take to trigger autophagy?#
Nobody knows, and no human study has established a threshold. The switch is controlled by AMPK and mTOR reading the cell's energy status, not elapsed time2, so the same hour means different things in different people. The only human fasting time-course, sampling muscle at 2, 12, 24 and 36 hours, found a modest response that depended on training status3. Every specific number circulating online — 16, 18, 24, 72 — is extrapolated from animals.
Has autophagy ever been measured in a fasting human?#
Yes, but rarely and indirectly. The largest attempt measured autophagic flux in blood mononuclear cells across 121 people over six months and found the fasting group differed from standard care at six months (P = 0.04) without significantly rising from its own baseline5. The authors describe the analysis as exploratory and say further work is needed to confirm that nutrient restriction modifies human autophagy.
If a study found more LC3, does that mean autophagy increased?#
Not on its own. LC3 is a component of the autophagosome, so a higher level can mean the cell is building them faster or clearing them more slowly — the static amount doesn't distinguish the two. Measuring flux, the rate of movement through the pathway, is what settles it, and only one human fasting study has done that. It is the main reason marker-based results in this field point in different directions.
Sources#
- Rubinsztein DC, Frake RA. Yoshinori Ohsumi's Nobel Prize for mechanisms of autophagy: from basic yeast biology to therapeutic potential. J R Coll Physicians Edinb. 2016;46(4):228-233.
- Kim J, Kundu M, Viollet B, Guan KL. AMPK and mTOR regulate autophagy through direct phosphorylation of Ulk1. Nat Cell Biol. 2011;13(2):132-141.
- Dethlefsen MM, Bertholdt L, Gudiksen A, et al. Training state and skeletal muscle autophagy in response to 36 h of fasting. J Appl Physiol (1985). 2018;125(5):1609-1619.
- Jamshed H, Beyl RA, Della Manna DL, Yang ES, Ravussin E, Peterson CM. Early Time-Restricted Feeding Improves 24-Hour Glucose Levels and Affects Markers of the Circadian Clock, Aging, and Autophagy in Humans. Nutrients. 2019;11(6):1234.
- Bensalem J, Teong XT, Hattersley KJ, et al. Intermittent time-restricted eating may increase autophagic flux in humans: an exploratory analysis. J Physiol. 2025;603(10):3019-3032.
- Jensen TL, Kiersgaard MK, Sørensen DB, Mikkelsen LF. Fasting of mice: a review. Lab Anim. 2013;47(4):225-240.
