The highest fat oxidation rates ever recorded came with a higher oxygen cost and no faster race

What exercise economy is

Economy is how much oxygen an athlete needs to move at a given speed, and two people with the same aerobic capacity can be minutes apart over a race if one needs less oxygen per kilometre. It matters here because burning fat instead of carbohydrate yields less energy per unit of oxygen, so a change in fuel is also a change in economy.

Drawn from the paper's discussion, which sets out the stoichiometry.

Why they ran it

Adaptation to a ketogenic diet had been proposed in lay, social and peer reviewed media as a universal benefit to endurance sport, on the strength of the increase in fat oxidation it produces. What had not been done was to test it in world class athletes across the intensities at which they actually compete, with the diet controlled rather than self reported, and real performance on the end of it.

Drawn from the paper's introduction.

Twenty nine elite race walkers completed three weeks of intensified training with a mild energy deficit on one of three isoenergetic diets. High carbohydrate availability throughout, nine athletes. The same macronutrients periodised within or between days to alternate low and high availability, ten athletes. A ketogenic low carbohydrate high fat diet at under 50 grams of carbohydrate a day and 78 percent of energy as fat, ten athletes.

Peak aerobic capacity rose in every group, at P less than 0.001, with a 90 percent confidence interval on the increase of 2.55 to 5.20 percent. The training worked regardless of what anyone ate.

The ketogenic arm reached peak fat oxidation of 1.57 plus or minus 0.32 grams per minute during two hours of walking at about 80 percent of peak oxygen uptake. The authors state these are the highest rates of fat oxidation reported, and note that some individuals exceeded 1.9 grams per minute.

The oxygen cost of walking at a speed approximating 20 kilometre race pace fell in the high carbohydrate and periodised groups, with 90 percent confidence intervals of minus 7.05 to minus 2.55 and minus 5.18 to minus 0.86 percent respectively. In the ketogenic group it stayed at pre intervention levels.

Ten kilometre race walk times improved by 6.6 percent in the high carbohydrate group, 90 percent confidence interval 4.1 to 9.1, and by 5.3 percent in the periodised group, interval 3.4 to 7.2. The ketogenic group changed by minus 1.6 percent, with an interval from minus 8.5 to 5.3 that spans both a slower and a faster race, which is to say no improvement was detected.

The numbers

Intervention length	3 weeks
High carbohydrate arm	8.6 g/kg/day carbohydrate, n equals 9
Periodised arm	same macronutrients, timing alternated, n equals 10
Ketogenic arm	under 50 g/day carbohydrate, 78 percent of energy as fat, n equals 10
Peak fat oxidation, ketogenic arm	1.57 plus or minus 0.32 g/min at about 80 percent of peak oxygen uptake
Peak aerobic capacity, all groups	increased, P less than 0.001, 90 percent CI 2.55 to 5.20 percent
10 km time change, high carbohydrate	6.6 percent faster, 90 percent CI 4.1 to 9.1
10 km time change, periodised	5.3 percent faster, 90 percent CI 3.4 to 7.2
10 km time change, ketogenic	minus 1.6 percent, 90 percent CI minus 8.5 to 5.3
Oxygen cost at 20 km race pace, ketogenic	maintained at pre intervention levels while the other two arms fell

Why this might happen

Proposed by the authors This is the explanation the authors offer in their discussion. This study did not test it.

The authors are explicit that they did not investigate the mechanism behind the rise in fat use, pointing instead at earlier biopsy work reporting more intramuscular triglyceride, higher hormone sensitive lipase activity, more fatty acid transporter protein and more carnitine palmitoyltransferase. For the fall in carbohydrate use they cite their own earlier finding of lower glycogen breakdown and a lower active form of pyruvate dehydrogenase, which reduces the capacity to oxidise carbohydrate even when the supply is adequate.

For the oxygen cost they give the stoichiometry rather than a hypothesis. Carbohydrate metabolism produces a higher ratio of NADH to FADH2 than beta oxidation does, and oxidative phosphorylation yields more ATP per unit of oxygen from NADH, so carbohydrate delivers more ATP per litre of oxygen even though fat delivers more per gram of substrate. They note this has been known empirically for a century, and add that it appears to have been largely ignored in the recent argument.

Drawn from Discussion, PMC5407976.

What this does not show

It does not show that fat adaptation makes everyone slower. The ketogenic arm's performance interval spans zero in both directions. What the study demonstrates is the absence of the improvement the other two arms got, in an event contested at a high fraction of aerobic capacity, not a measured slowing.
Three weeks is the tested exposure. The authors argue that three weeks was long enough because the arm reached the fat oxidation rates seen in athletes adapted for years. That is an argument about the metabolic marker. Whether other adaptations take longer is not settled by it.
No muscle was sampled. Everything is whole body respiratory measurement and performance. The study says what happened to fuel use and to the race, and it does not say what changed inside the muscle to produce either.
Elite male race walkers in one sport. Ten athletes in the ketogenic arm, in an Olympic event contested at a very high fraction of aerobic capacity. An ultra event run at a much lower fraction is a different question and this design does not answer it.

Where this leaves us

The trade now has both halves measured in the same athletes at the same time. Adaptation delivered the largest increase in fat oxidation in the literature and simultaneously removed the economy gain that the other two arms got from the same training block. This contradicts the claim that the metabolic adaptation is a free upgrade, and it does so in the population the claim was aimed at.

Elite male race walkers over three weeks of intensified training, competing over 10 kilometres at a high fraction of their aerobic capacity.

The same controlled design in an event contested at a much lower fraction of aerobic capacity, where the authors' own argument predicts there is aerobic reserve to absorb the economy penalty.

Caveats worth holding

Parallel groups of nine and ten, not a crossover, so between athlete differences are carried into the performance comparison.
The performance result for the ketogenic arm is a wide interval spanning zero, and should be read as no detected improvement rather than as a measured decline.
Confidence intervals are reported at 90 percent rather than 95 percent throughout.
The training camp involved a mild energy deficit for all arms, which is a co intervention.
No muscle biopsies were taken, so the mechanism is inferred from other studies rather than measured here.
Male athletes only, one sport.