Issue 08 · 27 August 2026
A special issue on metabolic plasticity. Eight papers from 1983 to this year on how fast skeletal muscle rebuilds itself around whatever fuel it is given, why building up one pathway means turning the other one down, and what that trade costs when the two directions run on different clocks.
What the muscle does when the diet changes
Five studies, five different exposures, one shape. Each row carries its own outcome as a direct label, because the measurements are not comparable to each other.
The rebuild is measured in days, not months
Every study here found its change at the end of the exposure shown. The bars mark how long the diet ran, not how long the change took, and the outcomes are different measurements in different units.
Main 2026 sits off this axis in the other direction. No adaptation at all, just one day of eating, and the cost was already measurable in watts.
The fat side goes up a long way
Three separate studies, three different protocols. Each card compares a group against its own control, and no card should be read against another.
1.43
grams of fat per minute after six days more than a 200 percent increase in world class race walkers, reached in under a week rather than over months. Burke 2020
1.57
grams of fat per minute, sustained in ketogenic race walkers across two hours at about 80 percent of peak oxygen uptake. The highest rates the authors could find published. Burke 2017
0.72
respiratory quotient after four weeks down from 0.83, which is the difference between a mixed fuel and almost entirely fat. Phinney 1983
Carbohydrate oxidation after a full day of carbohydrate
One study, one protocol, one set of units. World class race walkers adapted to a ketogenic diet, then given 24 hours of high carbohydrate and a pre race meal, measured at the same two treadmill speeds as their own baseline.
Stellingwerff 2006 found the enzyme behind this. After five days of fat adaptation and one carbohydrate day, muscle glycogen came back identical, 873 against 868, and pyruvate dehydrogenase activity did not, 1.69 against 2.39.
Same training block, three diets, one arm that did not improve
Elite race walkers, three weeks of intensified training, 10 kilometre race before and after. Peak aerobic capacity rose in all three arms.
+6.6%
high carbohydrate arm faster over 10 km. 90 percent confidence interval 4.1 to 9.1 percent. Burke 2017
+5.3%
periodised carbohydrate arm faster over 10 km. 90 percent confidence interval 3.4 to 7.2 percent. Burke 2017
none
ketogenic arm no improvement detected. Minus 1.6 percent, interval minus 8.5 to 5.3, which spans both a slower and a faster race. Burke 2017
six of the seven fat adapted race walkers in the 2020 study were slower than their own baseline, after a full day of carbohydrate restoration
In this issue
Fourteen trained cyclists ate five days of high fat and five days of high carbohydrate in a crossover. Resting muscle taken after the fat diet carried more of the message for the main fatty acid transporter, and more of the transporter protein itself.
The American Journal of · Cameron-Smith et al. · Deakin University · 1 February 2003 · PMID 12540388
Seven men emptied their muscle glycogen with exhausting exercise, then ate either almost no carbohydrate or a great deal of it for two days. The high carbohydrate arm raised the glucose transporter message and lowered the message for the brake on burning carbohydrate. The low carbohydrate arm did the reverse on the fat side.
American Journal of Physiology, · Arkinstall et al. · RMIT University · 3 February 2004 · PMID 14761878
Nine well trained female cyclists were emptied of glycogen and then fed either a great deal of carbohydrate or almost none for the next day. Power at the first ventilatory threshold fell by nineteen watts, and gross cycling efficiency fell with it. The all out sprint at the end was unaffected.
European Journal of Applied · Main et al. · the Auckland University of Technology · 2 June 2026 · PMID 42228164
Seven cyclists trained for five days on either a high fat or a high carbohydrate diet, then ate a full carbohydrate day. Muscle glycogen ended up identical. Pyruvate dehydrogenase activity did not. After the fat week it was lower at rest and throughout exercise, and glycogen was broken down more slowly.
American Journal of Physiology, · Stellingwerff et al. · the University of Guelph · 27 September 2005 · PMID 16188909
Five well trained cyclists ate four weeks of a ketogenic diet with calories held constant. Glucose oxidation fell threefold, muscle glycogen use fell fourfold, and endurance time at a moderate pace was unchanged. This is the study the whole low carbohydrate argument in endurance sport is built on.
Metabolism 32(8):769 to 776 · Phinney et al. · 1 August 1983 · PMID 6865776
Twenty nine elite race walkers trained hard for three weeks on one of three controlled diets. The ketogenic arm reached fat oxidation rates higher than any previously published. It also needed more oxygen to walk at race pace, and it was the only arm that did not improve its ten kilometre time.
The Journal of Physiology · Burke et al. · the Australian Institute of Sport · 14 February 2017 · PMID 28012184
Thirteen world class race walkers spent five to six days on a ketogenic or a high carbohydrate diet, then everyone ate a full carbohydrate day and raced. The retooling arrived in under a week. It did not leave in a day. Carbohydrate oxidation came back to 61 and 78 percent of where it started.
The Journal of Physiology · Burke et al. · the Australian Catholic University · 19 August 2020 · PMID 32697366
Seven cyclists ate the same food in two different orders around an evening interval session. Going to bed with the carbohydrate already spent raised the fat side of the programme by morning, including a shift in DNA methylation. It did not raise the markers of mitochondrial building the practice is usually sold on.
Journal of Applied Physiology · Lane et al. · RMIT University · 25 June 2015 · PMID 26112242
Flexible is not the same as free
Opinion Our reading of what these papers add up to. Everything above this line came from the papers. This part did not.
The first thing worth saying is that the plasticity itself is not controversial and has not been for a long time. Two days of eating differently moved the transcripts. Five days moved the transporter protein. Six days produced fat oxidation rates that people had assumed took a year to reach. Nobody in this literature disputes that muscle rebuilds itself around its fuel supply, and the disagreement that consumes the field is entirely about what the rebuild is worth.
The second thing is the part we think gets skipped. The machinery is reciprocal, and reciprocal is not a poetic description of it. Arkinstall watched the carbohydrate genes and the fat genes move in opposite directions in the same tissue in the same two days. Stellingwerff then showed what that means in practice, which is that you can refill a muscle with glycogen completely and still find the enzyme that spends it turned down. Those two results together are why we would resist the word flexibility as it is usually used. Flexibility suggests a muscle that can do both. What is actually on the record is a muscle that commits.
The third thing is the clock, and it is the finding we would put in front of anyone deciding what to eat before a race. The tooling goes in fast and comes out slowly. Six days to adapt, five to six days to wash out, and a single day of carbohydrate at the end of it recovered somewhere between two thirds and four fifths of the carbohydrate oxidation that had been lost. The strategy of adapting to fat and topping up at the last minute is, on this evidence, a bet that the two directions run at the same speed. They do not.
Now the other side, because it deserves its own paragraph and because we think it is stronger than the tidy version of this argument allows. Every performance result here comes from events contested at a very high fraction of aerobic capacity. Race walking over ten kilometres is close to the worst case for a fuel that costs more oxygen. Phinney measured at 62 to 64 percent of maximal oxygen uptake and found nothing broken. Burke's own explanation for why the economy penalty bites says so directly: when exercise is undertaken at a modest fraction of aerobic capacity there is room to compensate, and at threshold there is not. An athlete spending twenty hours at a genuinely easy pace is in the first case, not the second, and this issue does not contain the study that settles what happens to them.
The newest paper here does something the older ones cannot. Main and colleagues did not adapt anyone to anything. They emptied the muscle, gave one group a day of carbohydrate and the other almost none, and found the ceiling on sustainable pace had already moved by nineteen watts, with efficiency down alongside it and the all out sprint untouched. We read that as the same trade at its shortest possible timescale. It is also the only study in this issue conducted in women, which is a sentence we would rather not have to write about a body of work this influential.
What we take from the Lane study is narrower and slightly awkward for the practice it belongs to. Sleeping on low glycogen turned the fat handling programme on overnight, which is a striking demonstration that the switch responds to timing and not only to totals. But the outcome the practice is usually justified by, the mitochondrial building, is the one the authors report as not augmented. We think that gap between what a method demonstrably does and what it is sold as doing is the more useful thing to notice.
Where our confidence drops is on the human evidence base itself. Five, seven, seven, nine, thirteen. Those are the sample sizes carrying an argument that has shaped how a generation of endurance athletes eat, and until this year almost none of those people were women. The effects are large enough to survive small samples in a way a performance effect never would, but the studies that would settle the applied question, at the intensities and durations where the answer might differ, mostly have not been run.
The reading we would defend is this. Metabolic plasticity is real, fast and reciprocal, and treating it as a free upgrade is the error. Choosing a fuel is choosing what to be good at, the choice takes days to make and longer to unmake, and the honest framing of any diet in this space is not what it adds but what it trades. That is our view. The papers above are the evidence, and everything in this section is opinion.
Substrate Digest
Newsletter
Each issue by email, when the newsletter launches. Leaving your address puts you on the list, nothing is sent yet.