Insulin is one of the body’s most important signals for determining whether you burn glucose or fat: when insulin rises, fat release and fat burning fall, while lower insulin allows stored fat to become available for fuel. This helps explain why reducing carbohydrates can dramatically increase fat oxidation even without changing the body’s basic ability to burn fat.

::: spoiler generated summary

Insulin and the glucose-fatty-acid cycle

• Fuel choice is not just about which substrate is available. Carbohydrates raise blood glucose and insulin; insulin suppresses fat-cell lipolysis and makes long-chain fatty-acid entry into mitochondria harder, so higher insulin shifts oxidation toward glucose and lower insulin releases stored fat for oxidation.
• The glucose fatty-acid cycle began with a 1963 paper: fat oxidation inhibits glucose oxidation at pyruvate dehydrogenase, intermediates then back up through the pathway, and glucose uptake falls. The original observations came from isolated rat muscle and heart. [1]
• Human clamp studies hold insulin and glucose at chosen levels and can use lipid emulsion to hold fatty acids constant, isolating glucose-fat competition.

Human evidence that fat oxidation suppresses glucose oxidation

• Healthy volunteers underwent two insulin clamps; one kept fatty acids elevated by lipid infusion, while the other let them fall as insulin normally makes them fall. [2]
• With insulin and fatty acids both high, leg glucose uptake fell by about one-third; glucose supplied a little over half of energy versus about three-quarters when fatty acids fell, and pyruvate dehydrogenase activity was lower. [2]
• Respiratory quotient (RQ) tracks fuel use from CO2 produced relative to O2 consumed: carbohydrate is near 1.0, fat near 0.7, and whole-body respiratory exchange ratio is the analogous whole-body measurement.
• High insulin plus high fatty acids is an artificial combination in a healthy person because insulin normally suppresses fatty-acid release; lipid infusion overrides that normal hormonal effect.

The reverse direction: glucose and insulin suppress fat oxidation

• Most people eat predominantly carbohydrate; based on a nutrient database, about 70% of calories globally come from carbohydrate.
• In healthy volunteers, a hyperinsulinemic, hyperglycemic clamp held fatty acids constant while glucose oxidation rose more than threefold and fat oxidation fell; cellular glucose availability determined fuel oxidation even without a fall in fat availability. [3]
• A companion experiment infused tagged oleate, a long-chain fat that requires the carnitine shuttle, and octanoate, a medium-chain fat that bypasses it. Under the glucose-insulin clamp with fatty acids held constant, oleate oxidation fell almost by half while octanoate oxidation did not change. [4]
• The control is at long-chain fatty-acid transport into mitochondria through the carnitine shuttle/CPT1: glucose and insulin can reduce long-chain fat oxidation while circulating fatty acids remain available. [4]
• After an ordinary carbohydrate-containing meal, glucose and insulin rise, muscle and adipose glucose uptake rises, adipose fatty-acid release falls, and long-chain fatty-acid transport into mitochondria is inhibited; the cell therefore burns more glucose. This follows the insulin signal whether total calories are in deficit or surplus.

Insulin acts at two control points

• Insulin controls fat oxidation upstream by suppressing adipose lipolysis and downstream by limiting long-chain fatty-acid entry into mitochondria.
• In a clamp controlling insulin, growth hormone, and glucagon, less than 2 µU/mL of insulin reduced fat release by half in healthy people; fasting insulin is commonly 5-10 µU/mL, so adipose lipolysis is "exquisitely sensitive to insulin." [5]
• In another study using four increasing insulin infusion rates, the insulin concentration needed to halve fatty-acid release was roughly one-third of that needed to halve fat oxidation. Insulin suppresses fat supply at lower concentrations than mitochondrial fat oxidation. [6]

Meals and exercise expose both control points

• Six active men cycled for an hour after an overnight fast, glucose, or fructose. Fructose does not increase insulin; in these trials, pre-exercise insulin averaged about 8 µU/mL fasted, 17 after fructose, and 38 after glucose. After glucose, fatty-acid release was about half the fasted value and fat oxidation fell with it. [7]
• A fourth glucose trial restored circulating fatty acids with lipid infusion: fat oxidation rose about 30% above the glucose-only trial but remained below the fasted trial, separating reduced fatty-acid supply from inhibited mitochondrial entry. [7]
• Exercise makes fuel use easier to measure, but the same mechanisms matter at rest; several carbohydrate-containing meals can keep insulin elevated through much of the waking day.
• In six endurance-trained subjects cycling for 40 minutes, a large glucose drink reduced total fat oxidation by 34%; roughly half of the reduction came from circulating fatty acids and half from intramuscular triglyceride. Because intramuscular fat was already inside muscle, its reduced oxidation points again to mitochondrial-level inhibition. [8]

Sustained carbohydrate restriction changes the fuel mix

• Five trained cyclists ate a eucaloric diet with about two-thirds of calories from carbohydrate for one week, then a calorie- and protein-matched ketogenic diet below 20 g carbohydrate/day for four weeks. RQ fell from 0.83 to 0.72, glucose oxidation fell about threefold, and muscle glycogen use fell about fourfold, while VO2max and time to exhaustion were unchanged. [9]
• Twenty matched elite ultramarathon/Ironman-distance athletes included 10 high-carbohydrate and 10 low-carbohydrate athletes; the low-carbohydrate group averaged about 10% carbohydrate and 70% fat for about 20 months. Peak fat oxidation was more than twice as high with no overlap between groups, and fat supplied 88% versus 56% of energy over a three-hour run. [10]
• Resting muscle glycogen, glycogen used during the run, and glycogen restored afterward were the same between those athlete groups, so chronic low-carbohydrate intake did not leave their muscle glycogen depleted. [10]

Insulin resistance and metabolic inflexibility

• In insulin-resistant muscle, fuel selection goes wrong in both directions: between meals and overnight it keeps burning more carbohydrate when it should shift toward fat, while a rise in insulin produces a weaker shift toward carbohydrate. This is metabolic inflexibility. [11]
• The Randle cycle is a valid physiological principle but may not explain insulin resistance in skeletal muscle; in this model, simply eating more fat is not what creates insulin resistance. [11]
• In obese subjects studied with arterial-venous leg sampling, fasting leg RQ remained near 0.85, and muscle biopsies had lower carnitine-palmitoyltransferase activity and lower activity of fat-oxidizing enzymes. [12]
• Chronic high insulin creates opposite tissue problems: insulin-resistant adipose tissue keeps releasing fatty acids despite high insulin, while muscle remains responsive to insulin's block on mitochondrial fatty-acid entry. Fatty acids are delivered but not burned and can accumulate in muscle, liver, pancreas, and other ectopic sites.

Fuel oxidation versus body-fat loss

• Which fuel is being burned at a moment is not automatically the same question as how much body fat is ultimately lost.
• In 28 overweight or obese adults, two calorie-restricted diet periods compared a ketogenic diet below 10% carbohydrate with a low-fat diet near 60% carbohydrate, allowing within-person comparison. The low-carbohydrate period had higher self-recorded intake, about 1,900 versus 1,500 kcal/day, yet produced more weight, total-fat, and trunk-fat loss. [13]
• The food was not provided and intake was self-recorded, so there is potential error; having each person act as their own control is a strength of the comparison. [13]
• A meta-analysis of 13 trials lasting at least 12 months found greater weight loss with very-low-carbohydrate ketogenic diets than low-fat diets, but the difference was modest at about 1 kg. [14]

Why calorie-restricted comparisons can wash out the insulin difference

• Every low-carbohydrate versus low-fat trial in this comparison also cut calories in both groups. The higher-carbohydrate group therefore ate fewer carbohydrates than before the study and lowered insulin too, so these trials compare lowered insulin with more-lowered insulin, not unchanged high insulin with low insulin.
• A cleaner test is caloric overfeeding in both groups, because nobody then reduces carbohydrate simply by eating less and any difference can be attributed to diet composition.

The overfeeding case and conclusion

• The only study in nutrition science that tested this overfeeding design is Sam Feltham's n-of-1 case: 5,800 kcal/day for 21 days on each of three diets, with about three months between periods to return toward baseline—low-carbohydrate at about 6% carbohydrate, low-fat, and very-low-fat plant-based vegan. [15]
• Weight gain was a little over 1 kg on low carbohydrate, over 7 kg on low fat, and about 5 kg on the very-low-fat vegan diet. It is one case in one person, but it is the only study that tested the overfeeding question this way. [15]
• Across these human data, carbohydrate restriction increases fat oxidation, produces at least as much fat loss as fat restriction and usually more, requires less of a caloric deficit, and loses relatively more visceral/abdominal fat.
• Fuel selection is not decided simply by which fuel is more abundant in blood: clamp studies held fatty acids constant and fat oxidation still fell when glucose and especially insulin rose. Insulin controls both fat release from adipose tissue and long-chain fatty-acid entry into mitochondria.
• Carbohydrate is the main dietary driver of insulin, and insulin is the main controller of which fuel is burned. Lowering carbohydrate lowers the insulin signal at both control points and lets the body burn more fat: "you burn what you eat."

References

1. [01:37] The glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus — https://doi.org/10.1016/S0140-6736(63)91500-9
2. [04:25] Interaction between glucose and free fatty acid metabolism in human skeletal muscle — https://doi.org/10.1172/JCI116603
3. [09:16] Glucose and insulin-induced inhibition of fatty acid oxidation: the glucose-fatty acid cycle reversed — https://doi.org/10.1152/ajpendo.1996.270.4.E733
4. [09:58] Glucose plus insulin regulate fat oxidation by controlling the rate of fatty acid entry into the mitochondria — https://doi.org/10.1172/JCI119034
5. [14:10] Insulin regulation of lipolysis in nondiabetic and IDDM subjects — https://doi.org/10.2337/diab.38.12.1595
6. [15:18] Regulation of free fatty acid metabolism by insulin in humans: role of lipolysis and reesterification — https://doi.org/10.1152/ajpendo.2006.263.6.E1063
7. [16:01] Lipolytic suppression following carbohydrate ingestion limits fat oxidation during exercise — https://doi.org/10.1152/ajpendo.1997.273.4.E768
8. [18:12] Fatty acid oxidation is directly regulated by carbohydrate metabolism during exercise — https://doi.org/10.1152/ajpendo.1997.273.2.E268
9. [19:28] The human metabolic response to chronic ketosis without caloric restriction: preservation of submaximal exercise capability with reduced carbohydrate oxidation — https://doi.org/10.1016/0026-0495(83)90106-3
10. [20:47] Metabolic characteristics of keto-adapted ultra-endurance runners — https://doi.org/10.1016/j.metabol.2015.10.028
11. [22:26] Fuel selection in human skeletal muscle in insulin resistance: a reexamination — https://doi.org/10.2337/diabetes.49.5.677
12. [23:28] Skeletal muscle fatty acid metabolism in association with insulin resistance, obesity, and weight loss — https://doi.org/10.1152/ajpendo.1999.277.6.E1130
13. [26:02] Comparison of energy-restricted very low-carbohydrate and low-fat diets on weight loss and body composition in overweight men and women — https://doi.org/10.1186/1743-7075-1-13
14. [27:16] Very-low-carbohydrate ketogenic diet v. low-fat diet for long-term weight loss: a meta-analysis of randomised controlled trials — https://doi.org/10.1017/S0007114513000548
15. [29:30] A case study of overfeeding 3 different diets — https://doi.org/10.1097/MED.0000000000000668 :::