Fasting

72-Hour Fasting, Autophagy, and Apoptosis: What the Science Actually Shows

July 12, 2026 · by Brendon Kennedy

I keep coming back to the 72-hour fast because it sits right at the edge of where the interesting cell biology is supposed to happen — and right at the edge of what anyone has actually measured in a human being. This is my running catalog of what I’m reading, and I want to be honest about where the evidence is solid, where it’s animal-and-cell-only, and where it’s basically hope wearing a lab coat.

The setup: the metabolic switch

Nothing exotic happens in the first day. You burn through liver glycogen and flip what Rafael de Cabo and Mark Mattson, in their 2019 New England Journal of Medicine review, call the “metabolic switch” — the shift from glucose to fat-derived ketones, typically 12 to 36 hours in, depending on your glycogen stores and activity. By hour 72 you’re deep in that fat-burning, low-insulin, low-IGF-1 state. That’s the physiological substrate everything else is layered on top of. The switch itself is well established. The claims stacked on top of it are where I slow down.

Autophagy: a Nobel-winning mechanism, mostly proven outside humans

Autophagy is the cell’s recycling system — it wraps up damaged proteins and worn-out organelles and digests them. Yoshinori Ohsumi won the 2016 Nobel Prize in Physiology or Medicine for working out its machinery, and he did it in yeast, in the 1990s. The detail that matters for fasters: autophagy is switched on by nutrient scarcity. The nutrient sensor mTOR normally suppresses it; starve the cell and mTOR backs off, releasing the ULK1 complex to start building autophagosomes.

That mechanism is rock-solid in yeast and mice. In humans it gets thin fast. The most-cited human data point is a small study of eight men fasted for 72 hours: mTOR signaling dropped about 50% and the autophagy marker LC3B-II rose about 30%. Sounds like a clean win — except the same paper found that p62 (which should fall if autophagy is genuinely clearing cargo) rose slightly too, and the authors explicitly could not rule out that they were seeing autophagosomes piling up rather than flux moving through them. Measuring true autophagic flux in a living person is genuinely hard; a snapshot of protein markers in one muscle biopsy is not the same thing as proving the recycling ran to completion. So when someone tells you “72 hours equals autophagy,” the honest version is narrower: 72 hours lowers mTOR and nudges autophagy markers in the expected direction in a handful of people, and almost everything else is extrapolated from animals.

Apoptosis and the cancer question

Apoptosis is programmed cell death — the tidy self-destruct sequence a damaged cell is supposed to run instead of turning malignant. The compelling fasting story here comes largely from Valter Longo’s lab: fasting and the fasting-mimicking diet (FMD) appear to create “differential stress resistance,” where normal cells hunker down while stressed or cancerous cells — cut off from glucose and growth signals — get pushed toward apoptosis. Almost all of that causal mechanism is preclinical. The human trials so far test safety and biomarkers, not cures. A 2022 Cancer Discovery trial of 101 patients found that cyclic five-day FMD was safe alongside standard therapy and reproducibly lowered blood glucose, insulin, and IGF-1 while increasing T-cell infiltration into tumors.

That’s meaningful — but it is a safety-and-biomarker result, not evidence that fasting treats or prevents cancer in humans. And the real picture is messier than the wellness version: in some models the FMD actually inhibits autophagy to sensitize cancer cells, so “more autophagy is always good” is simply not true. Autophagy can protect a tumor as readily as it protects you. The honest frame I keep landing on: fasting and FMD are associated with pro-apoptotic, anti-growth signaling in cancer models and are feasible and safe in early human trials — not proven to prevent cancer in people.

The part the enthusiasts skip: prolonged fasting isn’t free

A 2025 Molecular Metabolism study of a roughly 10-day water-only fast found that prolonged fasting acutely raised inflammation (CRP up 129%) and platelet-activation markers. And refeeding is the genuinely dangerous end of this: coming off a long fast can trigger refeeding syndrome — dangerous drops in phosphate, potassium, and magnesium, plus thiamine depletion — which is why clinical protocols reintroduce calories slowly, give thiamine first, and monitor electrolytes. A 72-hour fast is milder than a 10-day one, but the direction of these risks is identical, and they scale with duration and with how depleted you already are going in.

What I’m actually doing

I treat a prolonged fast as an experiment on one subject, logged — not a lifestyle I recommend to anyone.

• Frequency: occasional and unscheduled — something I try now and then when the week allows for it, not a routine I run on a calendar.
• What I track: whatever I’m actually measuring that week — usually some mix of bodyweight, energy, and sleep notes, plus a glucose or ketone reading if I have a meter on hand.
• Refeeding: I break it with something small and low-carb, not a feast — usually a little protein and some vegetables, then I wait and see how it settles before going back to normal meals.
• Hard stops: anything that stops feeling like ordinary discomfort and starts reading like a warning — lightheadedness that won’t pass, a resting heart rate that won’t settle, or a genuinely bad night of sleep. If I’m arguing with myself about whether to stop, I take that as the answer and stop.
• Medication timing: anything I take on a daily schedule gets planned around rather than improvised — some medications simply don’t sit well on an empty stomach, and any change to how or when I take one is a conversation with the prescriber first, never something I wing on my own.

I don’t attempt this near a hard training week, and I wouldn’t start a multi-day fast without talking to a doctor first — the platelet and refeeding data alone are enough reason to keep this supervised rather than heroic.

Not medical advice. This is a personal catalog of research I’m reading and habits I’m testing on myself. Nothing here diagnoses, treats, or prevents any disease, and it isn’t a substitute for a qualified clinician. Talk to your doctor before changing diet, fasting, exercise, or medication — especially with ADHD medication, alcohol, or a personal or family cancer history.

← All writing