Skip to content
The Longevity Archive
browse 192

The biology of fasting

FGF21

Also called: Fibroblast growth factor 21

Written by Reviewed Sep 2026

A liver hormone that in mice is the signal of fasting, and in humans turns out to respond mainly to how little protein you have eaten rather than how few calories, and not to a two-day fast at all.

What it is

FGF21 is an unusual member of the fibroblast growth factor family. Classical FGFs carry a domain that binds heparin and keeps them tethered near the cell that made them, so they act locally. FGF21 lacks it, diffuses away, and behaves as a circulating hormone. It is made chiefly by the liver, and also by adipose tissue, skeletal muscle and pancreas.

Its receptor requirement is what gives it tissue selectivity. FGF21 signals through FGF receptors 1c, 2c or 3c, but only when the co-receptor beta-Klotho is present. A tissue without beta-Klotho does not respond to FGF21 at any concentration. Beta-Klotho is expressed in adipose tissue, liver, pancreas and in the brain, including the hypothalamus and hindbrain, which is where its effects on what an animal chooses to eat originate.

The actions attributed to it across species are: more hepatic fat oxidation and ketogenesis, more glucose uptake and adiponectin release from fat, higher energy expenditure and browning of white fat, improved insulin sensitivity, suppressed preference for sweet things and for alcohol, and stimulated appetite for protein.

This is the biggest human-to-rodent divergence anywhere in fasting biology, and the divergence is the reason the page exists.

The machinery

In rodents, FGF21 was characterised as a target gene of PPAR alpha in the liver and a key mediator of hepatic lipid metabolism in ketotic states. PPAR alpha is the fatty-acid-activated nuclear receptor that runs the fasted liver programme, including HMGCS2, the rate controlling enzyme of ketogenesis. That places FGF21 squarely inside the ketogenic response, and it is why the rodent literature treats it as the hormone of fasting.

In humans the input that dominates is different. A two day fast did nothing to serum FGF21. A ketogenic diet did nothing. Seven days of fasting produced a 74% rise. Meanwhile a single meal experiment showed protein suppressing the FGF21 response outright: a sucrose drink produced a substantial rise, adding protein to the same sucrose load abolished it, and adding fat did not significantly change it.

So in a person, FGF21 tracks the balance of sugar to protein in what was eaten, more than it tracks energy deficit. The authors of the meal study interpret it as a signal that promotes macronutrient balance and adequate protein intake, which fits the behavioural effect of stimulating protein appetite. That is a different job from being the hormone of energy scarcity.

What switches it on

In humans, disproportionate sugar intake relative to protein, and prolonged fasting at seven days but not at two. In rodents, the fasted and ketogenic hepatic programme through PPAR alpha. Pharmacological PPAR alpha activation with fenofibrate raises it in people too.

On the timing
The human onset is later than almost anyone assumes, and it was measured directly. Ketosis induced by a two day fast did not influence FGF21 levels. Neither did feeding a ketogenic diet. A 74% increase occurred only after seven days of fasting. Any claim that a 16 or 24 hour fast raises FGF21 in humans is not supported by that study, which is the definitive human fasting measurement on this hormone.

What has been measured

In people

FGF21 is in blood, so human measurement is straightforward, and there are both fasting studies and randomised trials of engineered FGF21 analogs in patients. The picture that emerges is consistent: the lipid and liver effects reproduce in people, and the glucose lowering effect that makes FGF21 famous in rodents largely does not.

  • The definitive human fasting study, and its findings are frequently misreported. Serum FGF21 varied 250-fold among 76 healthy individuals, and did not relate to age, sex, BMI, serum lipids or plasma glucose. That degree of between-person spread means group means conceal enormous individual variation. FGF21 showed no diurnal variation. Ketosis induced by a two day fast did not influence FGF21 levels, and neither did feeding a ketogenic diet. A 74% increase occurred only after seven days of fasting. Hypertriglyceridaemic non-diabetic patients had two-fold elevated FGF21, further increased 28% by the PPAR alpha agonist fenofibrate. The authors' explicit conclusion is that the physiological role of FGF21 in humans may differ from that in mice (Cell Metabolism, 2008).
  • Protein abolishes the response within a single meal. Randomised, double-blinded crossover in healthy volunteers, comparing a sucrose drink, a sucrose plus protein drink, and a sucrose plus fat drink with sucrose content matched. Sucrose alone produced an incremental area under the curve of 484 plus or minus 127 pg/mL times hour. Adding protein gave minus 35 plus or minus 49, P < 0.001. Adding fat gave 319 plus or minus 102, not significantly different from sucrose alone. There was no evidence that glycaemic regulators explained the effect. Protein outweighs sugar in FGF21 regulation (American Journal of Physiology: Endocrinology and Metabolism, 2023).
  • 24 hour energy expenditure measured by whole-room calorimetry during 24 hour fasting in 20 volunteers. The decrease in energy expenditure varied between individuals and correlated with the thermogenic response to mild cold and with changes in FGF21 concentration. The exact percentage decrease is not given in the abstract and is therefore not quoted here (Diabetes, 2020).
  • First analog trial, 28 days, obese subjects with type 2 diabetes, randomised placebo controlled double blind, LY2405319 at 3, 10 or 20 mg daily. Significant improvements in dyslipidaemia: LDL cholesterol and triglycerides fell, HDL rose, and the apolipoprotein profile shifted toward less atherogenic. Favourable effects on body weight, fasting insulin and adiponectin. Only a trend toward glucose lowering. This is the key result and it is consistently under-reported: in rodents FGF21 is a potent glucose lowering agent, and in humans the lipid, weight and adiponectin effects reproduced while the glucose effect largely did not (Cell Metabolism, 2013).
  • Phase 2a in fatty liver disease. 75 patients with biopsy confirmed non-alcoholic steatohepatitis, fibrosis stage 1 to 3, and at least 10% hepatic fat fraction by MRI, randomised to placebo, pegbelfermin 10 mg daily or 20 mg weekly for 16 weeks. Co-primary outcomes were safety and absolute change in hepatic fat fraction (The Lancet, 2019).
  • Phase 2b on a histological endpoint, which is a high bar in liver disease. 128 patients with biopsy confirmed steatohepatitis and stage F2 or F3 fibrosis randomised to placebo or efruxifermin 28 mg or 50 mg weekly, with paired biopsies at baseline and week 24. Fibrosis improved by at least one stage with no worsening in 8 of 41 on placebo (20%), 15 of 38 on 28 mg (39%, risk ratio 2.3, 95% CI 1.1 to 4.8, P = 0.025) and 14 of 34 on 50 mg (41%). This is a genuine positive, in a diseased population, on a liver endpoint (The Lancet Gastroenterology and Hepatology, 2023).

In other species and in cell culture

The identity of FGF21 as the hormone of the fasted, ketotic liver was established in mice, and so was its role in muscle. Those findings are solid in the mouse. They are the source of the human claims that did not reproduce.

  • Mouse liver: FGF21 identified as a PPAR alpha regulated gene and a key mediator of hepatic lipid metabolism in ketotic states, including the fasted and ketogenic-diet-fed states (Cell Metabolism, 2007).
  • Mouse skeletal muscle: muscle-specific FGF21 knockout mice, studied fed and fasted, show that FGF21 controls mitophagy and muscle mass. FGF21 expression is very low in normal healthy mouse muscle but is induced and released by fasting, endoplasmic reticulum stress, mitochondrial myopathies and metabolic disorders. The same study overexpressed FGF21 in mouse skeletal muscle in vivo to test whether FGF21 alone induces atrophy (Journal of Cachexia, Sarcopenia and Muscle, 2019).
  • Non-human primate work on FGF21 and glucose control in diabetic monkeys is frequently cited as the bridge that motivated the human diabetes trials. The identifier surfaced in the research behind this page did not pass the dual-source verification procedure, so it is not cited here.

Why it matters

FGF21 is the clearest case on this site of a mechanism that is genuinely well established in one species and genuinely different in another, where the difference has been measured rather than assumed. In a mouse, FGF21 is the hormone of fasting, driven by the ketogenic liver programme and lowering glucose powerfully. In a person, a two day fast does not move it, a ketogenic diet does not move it, seven days moves it 74%, and a single protein-containing meal shuts down the response that sugar alone produces.

The analog trials complete the picture from the other direction. Giving people engineered FGF21 reliably improves lipids and hepatic fat and, in the strongest trial, liver histology. It does not reliably lower glucose in humans the way it does in rodents. The development programme has followed the human evidence into metabolic liver disease, which is where that evidence actually is.

Hold both halves. FGF21 does something real in people. What it does is not what the rodent fasting literature predicted.

A claim you will see repeated

The claim that intermittent fasting raises FGF21, or that a 16 to 24 hour fast does, is not supported by the human study that tested it. Two days of fasting produced no change. Neither did a ketogenic diet. Seven days produced a 74% rise, in a study where the same hormone varied 250-fold between individuals to begin with. The deeper misconception is the framing. FGF21 is described as a fasting hormone, which imports the rodent story wholesale. In humans the dominant variable at this node is protein relative to sugar, not energy deficit, which makes it a signal about the composition of what was eaten rather than about the absence of eating.

Citations

  1. Human2008
    The circulating metabolic regulator FGF21 is induced by prolonged fasting and PPARalpha activation in man

    Cell Metabolism

    Serum FGF21 varied 250-fold among 76 healthy individuals and did not relate to age, sex, BMI, serum lipids or plasma glucose, and showed no diurnal variation. Ketosis induced by a 2 day fast did not influence FGF21 levels, and neither did a ketogenic diet. A 74% increase occurred only after 7 days of fasting. Hypertriglyceridaemic non-diabetic patients had 2-fold elevated FGF21, further increased 28% by fenofibrate. The authors conclude the physiological role of FGF21 in humans may differ from that in mice.

  2. Human2023
    Meal sugar-protein balance determines postprandial FGF21 response in humans

    American Journal of Physiology: Endocrinology and Metabolism

    Randomised double-blinded crossover meal study in healthy volunteers with sucrose content matched across arms. Sucrose alone gave an incremental area under the curve of 484 plus or minus 127 pg/mL times hour; adding protein gave minus 35 plus or minus 49, P < 0.001; adding fat gave 319 plus or minus 102, not significant versus sucrose. No evidence that glycaemic regulators explained the effect. Protein outweighs sugar in FGF21 regulation.

  3. Human2020
    Metabolic Responses to 24-Hour Fasting and Mild Cold Exposure in Overweight Individuals Are Correlated and Accompanied by Changes in FGF21 Concentration

    Diabetes

    24 hour energy expenditure measured by whole-room calorimetry during 24 hour fasting in 20 volunteers, with a decrease whose magnitude varied between individuals and correlated with the thermogenic response to mild cold and with changes in FGF21. The exact percentage decrease is not stated in the abstract.

  4. Human2013
    The effects of LY2405319, an FGF21 analog, in obese human subjects with type 2 diabetes

    Cell Metabolism

    Randomised, placebo controlled, double blind trial of placebo or 3, 10 or 20 mg LY2405319 daily for 28 days in obese subjects with type 2 diabetes. Significant improvements in dyslipidaemia with decreased LDL cholesterol and triglycerides, increased HDL, and a less atherogenic apolipoprotein profile, plus favourable effects on body weight, fasting insulin and adiponectin. Only a trend toward glucose lowering.

  5. Human2019
    Pegbelfermin (BMS-986036), a PEGylated fibroblast growth factor 21 analogue, in patients with non-alcoholic steatohepatitis: a randomised, double-blind, placebo-controlled, phase 2a trial

    The Lancet

    75 patients with biopsy confirmed non-alcoholic steatohepatitis, fibrosis stage 1 to 3 and at least 10% hepatic fat fraction by MRI-PDFF, randomised to placebo, 10 mg daily or 20 mg weekly for 16 weeks. Co-primary outcomes were safety and absolute change in hepatic fat fraction.

  6. Human2023
    Safety and efficacy of once-weekly efruxifermin versus placebo in non-alcoholic steatohepatitis (HARMONY): a multicentre, randomised, double-blind, placebo-controlled, phase 2b trial

    The Lancet Gastroenterology and Hepatology

    128 patients with biopsy confirmed steatohepatitis and stage F2 or F3 fibrosis, randomised to placebo or efruxifermin 28 mg or 50 mg weekly, paired biopsies at baseline and week 24. Fibrosis improved by at least one stage with no worsening of steatohepatitis in 8 of 41 on placebo (20%), 15 of 38 on 28 mg (39%, risk ratio 2.3, 95% CI 1.1 to 4.8, P = 0.025), and 14 of 34 on 50 mg (41%).

  7. Animal2007
    Hepatic fibroblast growth factor 21 is regulated by PPARalpha and is a key mediator of hepatic lipid metabolism in ketotic states

    Cell Metabolism

    Mouse liver. FGF21 identified as a PPAR alpha regulated gene and a key mediator of hepatic lipid metabolism in ketotic states, including the fasted and ketogenic-diet-fed states. This is the origin of the description of FGF21 as the hormone of fasting.

  8. Animal2019
    Fibroblast growth factor 21 controls mitophagy and muscle mass

    Journal of Cachexia, Sarcopenia and Muscle

    Muscle-specific FGF21 knockout mice studied fed and fasted, plus in vivo overexpression of FGF21 in mouse skeletal muscle. FGF21 expression is very low in normal healthy mouse muscle but is induced and released by fasting, endoplasmic reticulum stress, mitochondrial myopathies and metabolic disorders, and it controls mitophagy and muscle mass.