Glycogen depletion and gluconeogenesis

Name: Glycogen depletion and gluconeogenesis: what has been measured in people
Creator: Aaron Cuha

Also called: Glycogenolysis, Making new glucose

Written by Aaron CuhaReviewed Sep 2026

The body does not burn through its stored sugar and then switch to making new sugar. It is already making most of its sugar from scratch within the first day. So the "metabolic switch" everyone talks about is not really a switch.

What it is

Glycogen is sugar stored in long branching chains, packed with water. That is why the first pounds lost on any low-carb plan come off fast. They are mostly water, not fat.

There are two separate stores, and you cannot swap one for the other. Sugar stored in the liver can leave the liver and go into the blood. Sugar stored in muscle can never leave the muscle. That is not a matter of degree. It comes down to one missing enzyme, explained below.

Gluconeogenesis means making new sugar from things that are not sugar. There are four main raw materials. Glycerol is the backbone left over when fat is broken down. Alanine is a protein building block that comes out of muscle. Glutamine is another one, and the kidney uses a lot of it. Lactate comes back from working muscle. Only some of those make truly new sugar.

Rounded figures for how many grams each store holds are quoted everywhere. The research behind this page could not trace those gram figures to an original source, so none of them are printed here.

How it works

When any tissue breaks down its glycogen, the sugar comes out with a phosphate tag on it. That tag gives it a charge, and the sugar doorways in a cell wall only let through plain, untagged sugar. To send sugar out, a cell must first strip off the tag. That takes an enzyme called glucose-6-phosphatase.

The liver, kidney and gut have this enzyme. Muscle does not. So a person can carry several hundred grams of glycogen in their legs and still get low blood sugar, because none of it can reach the blood. Muscle glycogen feeds only the muscle that stores it. That one fact is why liver and muscle stores should never be added together as sugar the body can use.

Muscle glycogen does help in a roundabout way, through the Cori cycle. Muscle breaks sugar down to lactate. The lactate travels to the liver. The liver turns it back into sugar, and the sugar goes back to the muscle. The energy maths is poor. Breaking down sugar in muscle nets 2 energy units, while rebuilding it in the liver costs 6. So the cycle burns energy overall and moves the work from muscle to liver. It is not there for efficiency. And importantly, it makes no new sugar. The carbon that arrives as lactate came from sugar in the first place. It recycles sugar. It does not create any.

For truly new sugar, four sources matter. Glycerol comes off each fat molecule as it is broken down, and the liver feeds it straight into sugar making. Alanine carries carbon and nitrogen out of muscle to the liver. Glutamine matters most for sugar making in the kidney. It also carries the nitrogen the kidney uses to make ammonia, which is how the kidney handles the extra acid from ketosis. Odd-chain fats can make a little new sugar, but they are a tiny share of stored fat.

The glycerol point answers a question people ask all the time. The three fatty acid tails of a fat molecule cannot become new sugar in a person. The chemistry only runs one way, and their carbons are lost as carbon dioxide. The glycerol backbone can become sugar, and it does, in large amounts. In a 21 day fast, glycerol supplied as much of the new sugar as all the protein building blocks put together.

What switches it on

Falling insulin and rising glucagon act on the liver. They switch on the enzyme that breaks down glycogen, switch off the one that builds it, and turn up the enzymes that make new sugar. Cortisol plays a support role. It keeps those enzymes being made and supplies protein building blocks from muscle.

On the timing
This is the process that knocks down the popular clock, and it does so with unusually good data. Liver glycogen was measured again and again in living people with a special MRI scan across a 68 hour fast. Total sugar output was measured at the same time with a tracer, so new sugar making could be worked out by subtraction. In the first 22 hours, new sugar making already supplied 64% of all sugar output. It was 82% from 22 to 36 hours, and 96% after that. A completely different method gave about 47% at 14 hours, 67% at 22 hours and 93% at 42 hours. That method used heavy water that people drank, which leaves a trace in newly made sugar. Two unrelated methods closely agree. There is no hour when the body switches from glycogen to making sugar. Both run from the first hours, and the balance shifts steadily.

What has been measured

In people

This is the most human of all the processes on the site. Nearly every number here comes from a person. The methods include repeated MRI scans of the living liver, heavy water tracing, needle samples, tubes in blood vessels, and 21 day hospital ward studies that tracked every gram of protein lost.

New sugar making, deep into starvation, in 11 people with obesity. At 5 to 6 weeks, total sugar output was down to about 86 g a day, roughly half from the liver and half from the kidney. Fat in the blood and ketones did not level off until after day 17. Protein building blocks in the blood did not fall until after day 17 either (Journal of Clinical Investigation, 1969).
The alanine finding, from the same series. Of the 20 protein building blocks, alanine fell fastest and furthest during long starvation. Keeping alanine low is what finally slows the liver's sugar making and saves protein (Journal of Clinical Investigation, 1969).
The cost in protein, and this is the most careful count there is. Five people with obesity fasted 21 days on a hospital ward, with every gram of protein tracked, plus blood tests, body make-up measures and tubes in blood vessels. They lost body fat and lean mass side by side. Lean mass loss did not stop once ketosis set in. Protein loss in the urine fell compared with early starvation, which is the well known protein-saving change, but it never reached zero. Late in long starvation, burning protein supplied about 7% of energy needs. The lowest amount of protein the body had to burn was 0.27 g per kg of body weight per day. On days 19 to 21, a drug called phenylacetate was given to pull glutamine out of the body. It raised protein loss in the urine but did not cut sugar making in the liver and kidney (American Journal of Clinical Nutrition, 1998).
From the same body of work: over a 21 day fast, glycerol supplied as much of the new sugar as all the protein building blocks put together. That is the exact sense in which fat does make new sugar in a person.
Liver glycogen measured with repeated MRI scans, every 3 to 12 hours, across a 68 hour fast, with total sugar output measured by tracer. New sugar making supplied 64% of all sugar output in the first 22 hours, 82% over the next 14 hours, and 96% over the next 18 (Science, 1991).
A separate check using heavy water. Healthy people drank heavy water, and the trace it left in blood sugar was used to split new sugar from stored sugar. New sugar making supplied about 47% at 14 hours, 67% at 22 hours and 93% at 42 hours. The paper's own summary is that after an overnight fast, new sugar making covers about half of sugar output, and after 42 hours, almost all of it. The spread figures in the paper's online summary do not add up and look like a typing error, so only the averages are given here (Journal of Clinical Investigation, 1996).
The two direct measurements of stored glycogen in people. Liver glycogen was measured by liver sample during total starvation and then eating carbs again (Scandinavian Journal of Clinical and Laboratory Investigation, 1973). Muscle glycogen was measured by needle sample, in relation to diet and physical performance (Acta Physiologica Scandinavica, 1967). Neither record was read in full for this page, so the gram figures usually credited to them are not repeated here.

In animals and in cells in a dish

A lot of the detail here comes from rat liver, mouse models and liver cells in a dish. That includes where the tag-stripping enzyme is found, how the Cori and alanine cycles work, and how the liver and kidney control sugar making. This page deliberately does not quote any rodent timelines for glycogen running out.

In rat liver, rat livers kept working outside the body, mouse models and liver cells in a dish: the enzyme chemistry and the control of sugar making. A current review sums this up with both rodent and human evidence (Nature Reviews Endocrinology, 2017).
In mouse liver cells: how the liver's energy level controls sugar making, studied in the context of metformin (Journal of Clinical Investigation, 2010).
In mice, glycogen runs out far faster than in people, because a mouse burns energy much faster for its size. An hour of fasting for a mouse is not an hour for a person in any metabolic sense. Rodent timelines are often wrongly carried over into fasting advice for people. That is why none are quoted here.

Why it matters

Almost every fasting timeline out there is built on one idea: glycogen runs out at some hour, and that is the switch that turns on fat burning, ketosis and everything after it. The measurements in people say new sugar making is already supplying about two thirds of sugar output within the first day. A second, unrelated method found it at about half by 14 hours. There is no cut-off and no switch. There is a steady shift in balance that starts right away.

That matters beyond nitpicking, because the switch idea is what all the clock-hour claims elsewhere on this site are built on. If the real change is gradual and already well along by hour 14, then a set hour when some later process begins has nothing to stand on.

The second thing to carry away is the protein cost. Protein loss during a fast is real and steady. The body keeps it as low as it can, but never to zero. Ketosis cuts the rate a lot compared with early fasting. It does not stop it. The 21 day study exists to explain why death in starvation follows the loss of lean body mass, while body fat is still left.

A claim you will see repeated

Two claims get repeated a lot here, and neither is printed on this page. The first is the set gram figures for how much glycogen the liver and muscle hold, and the claim of 3 grams of water per gram of glycogen. They are quoted everywhere, but the research behind this page could not trace them to an original source, so they are left out, not repeated. The second is that the body switches to burning fat for sugar once glycogen is gone. Fatty acids cannot become new sugar in a person. The glycerol backbone of fat can. Over a long fast, it supplies as much new sugar as all the protein building blocks combined. That is the accurate, and more interesting, version of the same idea.

Sources

Study in people1991
Quantitation of hepatic glycogenolysis and gluconeogenesis in fasting humans with 13C NMR
Science
Researchers used a special scanner to track liver sugar stores in people across a 68 hour fast. In the first 22 hours, 64% of the body's new blood sugar was made from scratch rather than taken from stores. That rose to 82% over the next 14 hours and 96% over the next 18 hours.
Hepatic glycogen measured serially at 3 to 12 hour intervals across a 68 hour fast by carbon-13 magnetic resonance spectroscopy, with whole-body glucose production by tritiated glucose. Gluconeogenesis accounted for 64 plus or minus 5% of total glucose production in the first 22 hours, 82 plus or minus 5% over the next 14 hours, and 96 plus or minus 1% over the next 18 hours.
Study in people1996
Contributions of gluconeogenesis to glucose production in the fasted state
Journal of Clinical Investigation
Healthy people drank specially labelled water so researchers could tell where blood sugar came from. Sugar made from scratch supplied about 47% at 14 hours of fasting, 67% at 22 hours and 93% at 42 hours. The rest came from stored sugar.
Healthy subjects ingested deuterated water, and deuterium enrichment at specific carbons of blood glucose was used to partition gluconeogenesis from glycogenolysis. Gluconeogenesis contributed about 47% at 14 hours, 67% at 22 hours and 93% at 42 hours. The dispersion values printed in the indexed abstract are internally inconsistent and are not reproduced.
Study in people1998
Protein, fat, and carbohydrate requirements during starvation: anaplerosis and cataplerosis
American Journal of Clinical Nutrition
Five obese people fasted 21 days while researchers tracked their bodies closely. They lost fat and lean tissue together, and lean tissue loss did not stop once ketosis set in. Late in the fast, protein supplied about 7% of energy. Glycerol, from fat, made as much new blood sugar as all the protein building blocks combined. A compound given on days 19 to 21 raised protein loss but did not cut sugar making.
Five obese subjects fasted 21 days with nitrogen balance, blood metabolites and hormones, indirect calorimetry, body composition and catheterisation. Fat and fat-free mass were lost in parallel and fat-free mass loss did not stop once ketosis was established. Late in starvation amino acid oxidation supplied about 7% of energy requirements, with minimum obligatory oxidation of 0.27 plus or minus 0.08 g amino acid and 1.53 plus or minus 0.21 g fat per kg body weight per day. Glycerol's contribution to gluconeogenesis equalled that of all amino acids combined. Phenylacetate from day 19 to 21 raised urinary nitrogen loss without curtailing hepatic and renal gluconeogenesis.
Study in people1969
Liver and kidney metabolism during prolonged starvation
Journal of Clinical Investigation
11 obese people went without food for weeks. Blood sugar and insulin fell quickly over the first 3 days. Protein building blocks in the blood did not fall until after day 17, and fats and ketones did not level off until then either. At 5 to 6 weeks, the body made about 86 g of sugar a day, roughly half from the liver and half from the kidneys.
11 obese subjects. Blood glucose and insulin fell acutely over the first 3 days. Alpha amino nitrogen did not fall until after day 17, and free fatty acids, beta-hydroxybutyrate and acetoacetate did not plateau until after day 17. At 5 to 6 weeks total glucose production was approximately 86 g per 24 hours, roughly half hepatic and half renal, with total urinary nitrogen 4.66 plus or minus 0.62 g per 24 hours.
Study in people1969
Amino acid metabolism during prolonged starvation
Journal of Clinical Investigation
People on long fasts. Of the 20 protein building blocks, alanine fell fastest and furthest. Keeping alanine low is what finally slows the liver's sugar making and spares protein.
Of the 20 amino acids, alanine fell fastest and furthest during prolonged starvation, and maintenance of that low alanine is what ultimately throttles hepatic gluconeogenesis and spares protein. This is the human demonstration of the glucose-alanine cycle as the rate-limiting arm.
Study in people1973
Liver glycogen in man, the effect of total starvation or a carbohydrate-poor diet followed by carbohydrate refeeding
Scandinavian Journal of Clinical and Laboratory Investigation
Liver sugar stores were measured directly with a liver sample, during total fasting and while eating carbs again. The full paper was not read for this page, so no amounts are given.
Hepatic glycogen measured directly by liver biopsy during total starvation and during carbohydrate refeeding. The PubMed record carries no abstract and the full text was not read for this page, so no gram figures are quoted from it.
Study in people1967
Diet, muscle glycogen and physical performance
Acta Physiologica Scandinavica
Muscle sugar stores were measured with a needle sample and compared with diet and exercise performance. The full paper was not read for this page, so the amounts often quoted from it are not repeated here.
Muscle glycogen measured by needle biopsy in relation to diet and physical performance. The full text was not read for this page, so the gram figures commonly attributed to it are not quoted.
Review of studies2017
Regulation of hepatic glucose metabolism in health and disease
Nature Reviews Endocrinology
A review of rat, mouse and human research on how the liver controls its sugar output. It covers where in the body the enzyme that releases sugar is found, and control from the liver itself versus from other organs.
Synthesis of hepatic glucose flux regulation across rodent and human evidence, including the tissue distribution of glucose-6-phosphatase and the distinction between direct hepatic and indirect extrahepatic control of gluconeogenesis.
Animal study2010
Metformin inhibits hepatic gluconeogenesis in mice independently of the LKB1/AMPK pathway via a decrease in hepatic energy state
Journal of Clinical Investigation
In mice and in mouse liver cells. The liver's sugar making is set by the liver's own energy level. It does not need LKB1 or AMPK.
Mouse hepatocytes and mice. Hepatic gluconeogenesis is controlled by hepatic energy state directly, without requiring LKB1 or AMPK, which is the clearest demonstration that the liver's glucose output is set by its own energetic condition.