A disclosure before we go further: I develop soup cleanse products, which is a reason to read the evidence sections closely rather than take my word for it. Where the research is thin, I’ve said so.
What autophagy is
Autophagy means self-eating. It describes cells breaking down their own worn-out parts and reusing the pieces. It runs all the time at a low level, and stress, including a shortage of nutrients, can speed it up. Yoshinori Ohsumi won the 2016 Nobel Prize in Physiology or Medicine for working out how it operates, and that work was done in yeast. His experiments in baker’s yeast in the early 1990s identified genes the process depends on. The prize recognized the mechanism and says nothing about whether a few days of fasting improves anyone’s health.
How researchers measure it
Measuring autophagy is harder than it sounds. The field’s guidelines, published in 2021 in their fourth edition at 382 pages and written by more than 2,900 researchers, state that no single assay suits every situation and call for several techniques at once. The common markers are LC3-II, which rises when autophagosomes form, and p62, a protein consumed as autophagy proceeds.
A rise in LC3-II alone is ambiguous. It can mean autophagy has been switched on, or that the pathway is blocked further along. The guidelines separate induction from flux, the flow through the whole pathway, and flux is the measure that counts. That is the ambiguity in the 72-hour muscle study, where LC3B-II and p62 both rose. Settling it takes flux assays on tissue, and in healthy people that means muscle biopsies or blood cells.
Where the hour counts come from
Most claims about when autophagy starts trace back to animal work. In mice, studies report a strong increase in autophagy in skeletal muscle after 24 and 48 hours of fasting. Mice burn energy so fast that 48 hours without food is an extreme stress for them, so the same clock can’t be assumed for a person. When an article names a specific hour at which autophagy begins in humans, the number is a mouse figure or an estimate.
What human studies found
Liver tissue is hard to sample in healthy people, so human fasting studies mostly look at muscle or blood. A 2022 paper reported that intermittent fasting activated autophagy markers in mouse liver but not in muscle from mice or humans.
A 36-hour fast, studied in the Journal of Applied Physiology in 2018, affected muscle autophagy only modestly. Several markers fell in untrained participants, and the authors saw no sign that autophagy had been initiated in untrained muscle.
A 2014 study in PLOS One took muscle biopsies from eight men after a 72-hour fast. One marker of autophagy, LC3B-II, rose by about 30%. A second, p62, a protein that should fall when autophagy is active, rose by about 10%. The authors wrote that this made the results hard to interpret.
The most encouraging human result is longer and in a different setting. In an exploratory analysis in the Journal of Physiology in 2025, 121 adults with obesity were randomized to standard care, calorie restriction, or intermittent fasting with time-restricted eating for six months. In this exploratory analysis, autophagic flux in blood cells rose with intermittent fasting. That is six months of an eating pattern, measured in blood cells, and it says little about a three-day cleanse.
Why a soup cleanse is a different case
Nutrient signals hold autophagy back. Amino acids and insulin switch on a growth regulator called mTOR, which holds back the first step of the process. In the 72-hour fast study, mTOR signaling in muscle fell, which fits the idea that fasting lifts the brake. How far a soup cleanse differs from a water fast on this score depends on the pot. Soup with legumes or grains in it keeps protein and calories in the day, so those signals stay higher than they would in a water fast, while a thin plan of broth and vegetable purees sits closer to a fast. By that reasoning, a richer soup is a weaker push toward autophagy than a fast. That is an inference from mechanism, and I haven’t found a measurement that supports or refutes it.
What isn’t evidence
Several things get offered as proof that autophagy has started, and none of them measures it. Ketones show that the body is burning fat. Weight loss shows lost fluid and some fat. Feeling clearer or lighter on day two is a subjective report, and the day-two feeling has an ordinary explanation in sodium and caffeine, as the headache post lays out. A mouse study shows what happens in mice. The only measurement of autophagy is a flux assay on tissue or cells.
What would settle it
A clean answer needs a design nobody has run. Take a muscle sample or draw blood before a cleanse and again at 24, 48, and 72 hours, measure flux and not a single marker, and run the same schedule on a soup cleanse, a water fast, and ordinary eating in the same people. Until a study like that exists, the claim that a soup cleanse switches on autophagy is a prediction.
The mTOR reasoning above gives that prediction a direction: a richer pot should produce a weaker signal, and a thin one a stronger signal. A trial could test the direction as well as the claim.
How I treat the claim
The 3-day guide makes no autophagy claim, and this is the reason. A claim that a soup cleanse triggers autophagy would rest on mouse data, mixed human muscle data, and a mechanism argument that points the other way. What changes measurably in three days is more ordinary: what you eat, and a few pounds of mostly water, as the post on why the weight comes back lays out.
Common questions
- Does fasting trigger autophagy?
- In mice, yes, within a day or two. In humans, studies of short fasts in muscle have given mixed or modest results, and a six-month study in blood cells points to a rise with intermittent fasting.
- How long do you have to fast for autophagy to start?
- Nobody has established it for humans. A clinical trial was registered to find out how long fasting must last, which shows the question is open. The hour counts quoted online come from animal studies or are estimates.
- Does eating soup stop autophagy?
- Autophagy runs at a low level all the time, and nutrient signals such as amino acids and insulin turn it down. How much a given soup does that has not been measured.
- How is autophagy measured in people?
- With markers such as LC3-II and p62, ideally as flux assays on muscle or blood cells. A single marker can’t separate autophagy being switched on from the pathway being blocked.