Four weeks of ketosis cut muscle glycogen use fourfold and the time to exhaustion did not move

What the respiratory quotient is

The ratio of carbon dioxide breathed out to oxygen taken in tells you what fuel a person is burning. Around 1.0 is almost entirely carbohydrate and around 0.70 is almost entirely fat, which is why a single number moving from 0.83 to 0.72 is a large statement rather than a small one.

Drawn from background physiology, not from this paper.

Why they ran it

The question was whether chronic ketosis affects exercise performance in endurance trained people, separated from the thing ketogenic diets are usually studied alongside, which is losing weight. So both diets were made eucaloric and isonitrogenous, matched for calories and for protein.

Drawn from the paper's stated objective and methods.

Five well trained cyclists ate one week of a eucaloric balanced diet, two thirds of the non protein calories as carbohydrate, then four weeks of a eucaloric ketogenic diet providing less than 20 grams of carbohydrate a day, matched for calories and protein.

Maximal oxygen uptake did not change between the control week and week 3 of the ketogenic diet. Mean time to exhaustion at 62 to 64 percent of maximal oxygen uptake was 147 minutes on the balanced diet and 151 minutes at week 4 of the ketogenic diet.

The fuel mix changed a great deal. The steady state respiratory quotient during that ride fell from 0.83 to 0.72, at P less than 0.01. Glucose oxidation fell threefold, from 15.1 to 5.1 milligrams per kilogram per minute, at P less than 0.05. Muscle glycogen use fell fourfold, from 0.61 to 0.13 millimoles per kilogram per minute, at P less than 0.01.

No clinical or biochemical evidence of low blood sugar appeared during the ride at week 4.

The numbers

Ketogenic diet duration	4 weeks
Carbohydrate on the ketogenic diet	less than 20 g per day
Time to exhaustion, balanced diet	147 minutes
Time to exhaustion, week 4 of ketosis	151 minutes
Steady state respiratory quotient	0.83 to 0.72, P less than 0.01
Glucose oxidation	15.1 to 5.1 mg/kg/min, P less than 0.05
Muscle glycogen use	0.61 to 0.13 mmol/kg/min, P less than 0.01
Maximal oxygen uptake	unchanged between control week and week 3

Why this might happen

Shown in this study The study measured this step directly.

They tracked glucose oxidation with a tracer and glycogen use by biopsy, and concluded that the preserved exercise capacity came from an adaptation that conserved the body's limited carbohydrate stores and made fat the predominant muscle substrate at that power level.

They also wrote the sentence the next forty years of this argument kept returning to. The price paid for such extreme conservation of carbohydrate during exercise, they said, appears to be a limitation on the intensity of exercise that can be performed.

Drawn from the paper's stated conclusions, with the closing sentence as quoted in Burke and colleagues 2017, PMID 28012184.

What this does not show

It does not show that endurance was preserved. Five cyclists, no control group, and a single mean of 147 against 151 minutes. A trial this size cannot detect anything smaller than a very large difference, in either direction, so unchanged here means undetected rather than equal.
The order was fixed, not randomised. Everyone ate the balanced diet first and the ketogenic diet second. Five weeks of training, familiarity with the test and seasonal effects all run in the same direction as the diet, and the design cannot separate them.
62 to 64 percent of maximal oxygen uptake is not race pace. That is a training intensity. Competitive endurance events are contested at a far higher fraction of aerobic capacity, and this protocol never goes there.
Sparing fuel is not the same as being able to spend it. The fourfold drop in glycogen use is reported here as conservation. Whether a muscle that has stopped spending glycogen can still spend it on demand is exactly the question this design does not put.

Where this leaves us

The founding observation stands and so does its own caveat. Four weeks of ketosis with calories held constant produced a very large shift in fuel use and no detectable loss of moderate intensity endurance in five people. What has been carried forward selectively is the authors' own warning about intensity, which was in the paper from the start.

Five well trained male cyclists riding to exhaustion at 62 to 64 percent of maximal oxygen uptake after four weeks of a eucaloric ketogenic diet.

A randomised crossover with enough participants to detect a small performance difference, run at an intensity a competitive endurance athlete would recognise.

Caveats worth holding

Five participants, no parallel control group, and a fixed diet order rather than a randomised crossover.
The endurance result is a comparison of two means with no dispersion reported in the abstract, so the precision of the 147 against 151 minute comparison cannot be judged here.
Calories and protein were matched between diets, which removes weight loss as an explanation. That is a genuine strength of the design and worth holding onto when reading the rest.
The exercise intensity tested is moderate. Nothing in this study speaks to high intensity or race pace work.
The full discussion was not openly reachable, so the mechanism section rests on the paper's stated conclusions, one of which reaches us through its quotation in a later paper.
PubMed lists no institutional affiliation for this paper, so none is stated here.