The Science of Ketogenic Exercise

There is remarkably little scientific research available around exercise while on a Ketogenic Diet. For a long time, the only useful research was a single study by Phinney in 1983. While this is a fascinating study, it has few subjects, it has not been confirmed with other studies and the results of the study are somewhat ambiguous. A 2016 study on race walkers has provided more insight, and I've covered both studies in some detail. I've also included some other studies that are on Non-Ketogenic Low Carbohydrate Diets.

1 Burke's Study of Race Walkers

Unlike the Phinney study described below, the study by Burke is not performed by proponents of a low carbohydrate or ketogenic diet. The study seems to have a reasonable approach, though it's not without its flaws. The main problem with this study is that it looks at a relatively small number of subjects, and the subjects are all elite athletes. This means that the findings may not translate to recreational athletes, or even sub-elite athletes. However, I'd argue this is probably the best study we have so far on the ketogenic diet for athletes. The high-level finding of the study is that athletes performed worse on a low carbohydrate, ketogenic diet. The fundamental problem that the study highlights with a low carbohydrate, ketogenic diet is that it requires more oxygen to produce a given amount of energy from fat than it does from carbohydrate. This problem is highlighted not just in the somewhat artificial, laboratory tests, but also in race performances.

• The study subjects were 21 elite race walkers. Their event is the 20Km race walk, which takes an elite athlete 75-80 minutes and is completed at about 90% of V̇O₂max. That means the subjects are focused on a moderate lengths endurance event, roughly similar to runners in a half marathon. The high intensity of this event means that the results might not translate to ultramarathons in the same way.
• The study took place during a three-week intense training block, that included weight training, race walking, and endurance cross training using running, cycling, or swimming.
• The subjects were divided into 3 different groups; one group was given a standard high carbohydrate diet (HCHO), another group was given a Low CHO, high fat (LCHF), and the third group cycled between high carb and low carb (Periodised CHO or PCHO). All 3 diets were matched for overall energy intake.
  • The High Carb (HCHO) group got 60-65% of their calories from carbohydrate, 15-20% protein, 20% fat.
  • The Periodised CHO (PCHO) group had the same average nutrient balance as the high carb diet, but performed some of their training in a low carbohydrate or fasted state.
  • The Low CHO, high fat (LCHF) was 75-80% fat, 15-20% protein, and less than 50g of carbohydrate per day. The study doesn't give as many details on the ketone levels as I'd like, but they do measure BOHB (beta-hydroxybutyrate) and show levels well above 1.0 when measured at rest. This indicates the low carbohydrate athletes are clearly in ketosis during the testing phase. Given the overall protocol, I think it's reasonable to conclude that the athletes were keto adapted by the end of the study.
• There were a number of tests performed, including V̇O₂peak, 25Km long walk, and simulated 10Km race. The simulated race even had prize money as incentive, so one could perhaps think of this as more of a true race than a race simulation. In fact, four of the subjects set personal best times in the simulated race, and one even said a national record.
• The low carbohydrate athletes had problems during training, with higher rates of perceived exertion, and higher heart rates, and the inability to complete some of the training sessions. It's unclear if these problems were only at the beginning of the training, before they'd adapted to the low carbohydrate diet, or if they persisted through to the end.
• LCHF had problems completing the training, complaining of higher perceived effort and inability to complete the training sessions. (Adaptation period?)
• The oxygen cost of exercise was higher in the low carbohydrate group than a high carbohydrate group, while the periodised carbohydrate group had a slightly lower oxygen cost and the high carbohydrate group.
• The subjects performed the simulated race before and after the dietary intervention. The race times were improved in the high carbohydrate group by an average of 190 seconds (114-266) and in the periodised carbohydrate group by 124s (62-186). However, the low carbohydrate group was slower by an average of 23 seconds (162-208, a range that makes no sense and I've queried with the authors). It looks like the low carbohydrate athletes increased their V̇O₂peak at the end of the study, but this increase was negated by the impaired economy of higher oxygen cost of exercise.
• The low carbohydrate group burned more fat during the end of study tests by a rather dramatic amount.

2 Phinney's Study of Elite Cyclists

A classic study was performed in 1983 by Stephen Phinney on the effect of four weeks on a Ketogenic Diet on 5 elite cyclists, and found that on average their endurance was maintained. However, this average represents two subjects that dramatically improved their endurance, one that didn't change and two subjects that had dramatically reduced endurance. One of the subjects that had reduced performance was Overtrained, and if their results are excluded the average is a 13% improvement.

• The diet was 15% of calories from protein, 85% from fat, with less than 20g of carbohydrate.
• Subjects performed V̇O₂max and endurance tests before and after the Ketogenic Diet.
• The five subjects had an endurance time of 147 minutes on the normal diet and 151 minutes on the Ketogenic Diet. However, the individual responses are rather different.
  • Two subjects dramatically increased their endurance time on the Ketogenic Diet, one by 57% and another by 30%.
  • One subject had an almost identical endurance time.
  • Two subjects had a dramatic decrease in their endurance, one by 36% and another by 28%. The subject that decreased their performance by 36% had complained of overtraining and had reduced his training level the month before the trial.
• Of the five subjects (see table below for details):
  • JP: Endurance improved by 57% and had much lower final muscle Glycogen levels on the Ketogenic Diet than the normal diet (41.5 compared with 59.2).
  • IK: Endurance improved by 30% and had similar final Glycogen levels on normal and Ketogenic Diets.
  • MK: Endurance was the similar (2% improvement) and had higher final Glycogen levels on the Ketogenic Diets.
  • BK: Endurance reduced by 28% and had higher final glycogen levels after the ketogenic endurance test than the normal diet (38.1 compared with 58.4).
  • WB: Endurance reduced by 36%.
    • WB was Overtrained and reduced training volume from 300 miles/week to 100 miles/week.
    • WB had higher final glycogen levels after the ketogenic endurance test than the normal diet.
    • WB had higher VO₂ on the Ketogenic Diet.
• Unlike typical fat metabolism, the oxygen cost of calories was not different on the Ketogenic Diet. This is a huge deal, as the big problem with burning fat is that it requires more Oxygen to produce the same amount of energy. However Phinney provides no details of the power outputs.
• The RQ on the VO2max test dropped from 1.04 to 0.9, and on the endurance test from 0.83 to 0.72, indicating a shift in substrate metabolism. However, because Ketone metabolism can produce RQ values that vary significantly, it's not possible to estimate what fuel is being metabolized. (The conversion of fat to Ketones consumes Oxygen without producing Carbon dioxide, the metabolism of Ketones has an RQ of 1.0, with the overall RQ matching that of Fat.)
• The endurance test did not include any fuel, just water. It is likely that the endurance test would have had a better result in the control condition with carbohydrate supplementation.
• Blood glucose during the endurance test was similar after the Ketogenic Diet to before, but the rise and fall were somewhat reduced. At no point did blood glucose drop to the point of hypoglycemia. Blood glucose provided an estimated 28% of calories on the normal diet and 9% of calories on the Ketogenic Diet.
• Blood levels of the Ketone 3-hydroxybutyrate (BOHB), which indicates ketogenisis, were insignificant on the normal diet and elevated (1.28 mmol/L) on the Ketogenic Diet.
• Muscle Glycogen levels where higher before the endurance test on the normal diet than on the Ketogenic Diet (143 and 53 respectively). Both tests had similar muscle glycogen levels after the endurance tests. It is interesting to see that muscle glycogen levels were replenished somewhat on the Ketogenic Diet, even though the subjects continued normal training for the four weeks.

	Endurace-1	Endurance-2	Change	Glycogen Pre-1	Glycogen Post-1	Drop in Glycogen	Glycogen Pre-2	Glycogen Post-2	Drop in Glycogen	Change in final levels	Difference in drop	VO2-1	VO2-2	Change	RQ-2
JP	148	223	51%	144.8	59.2	-59%	81.7	41.5	-49%	-30%	-10%	3.30	3.20	-3%	0.65
IK	100	130	30%	179.7	62.9	-65%	67.5	63.5	-6%	1%	-59%	3.10	2.97	-4%	0.75
MK	178	181	2%	142.8	46.6	-67%	65.1	57.6	-12%	24%	-56%	3.09	3.01	-3%	0.72
BK	169	121	-28%	120.8	38.1	-68%	80.7	58.4	-28%	53%	-41%	3.79	3.97	5%	0.72
WB	140	89	-36%	124.9	57.8	-54%	82.6	60.1	-27%	4%	-26%	2.61	2.95	13%	0.74
Average			4%			-63%	75.5		-24%	10%	-38%			2%
Exclude WB			13%			-65%	73.8		-24%	12%	-41%			-1%

3 Other studies

There are a few other studies that are of interest, though they do not directly deal with exercise capacity and the Ketogenic Diet.

3.1 Metabolic characteristics of keto-adapted ultra-endurance runners (FASTER)

This study compares the characteristics of 20 elite endurance athletes, 10 of them having been on a long-term low carbohydrate diet. The study is sometimes referred to by the backronym FASTER (Fat Adapted Substrate use in Trained Elite Runners.) The study has some significant flaws and it's hard to find any useful conclusions. The athletes on the low carbohydrate diet had higher levels of fat burning than their high carbohydrate comparisons, but did not seem to have any of the benefits that high fat burning would normally convey.

• The subjects were 20 elite ultrarunners or triathletes, finishing in the top 10%, some of them with course records, or national/international records. The athletes had broadly similar average characteristics.
• While the title of the study suggests that the athletes are on a Ketogenic Diet, this does not appear to be the case. The low carbohydrate athletes had resting Ketone Levels of ~0.7 mmol/L, which I feel is borderline for Ketosis. Their diet indicates their Ketogenic Ratio was only 1:1, which is far too low for Ketosis and there is no information provided to indicate they are Keto-adapted.
• The study consists of two tests on consecutive days. The first day involved a V̇O₂max test and the second day a sub-maximal three-hour treadmill endurance run. The V̇O₂max test was after a 4 hour fast, and the endurance run was 90 minutes after a nutrient shake that was different for the high and low carbohydrate athletes. This difference in nutrient intake makes the results endurance run hard to interpret as it's impossible to know if the differences in results are due to the prior nutrition and training, or due to the different nutrient intake.
• The V̇O₂max test resulted in the low carbohydrate athletes had a 2.3x higher peak fat burning capability than the low carbohydrate athletes, both from an absolute and as a percentage of V̇O₂max where the peak fat burning occurred. This might suggest that athletes on a low carbohydrate diet adapt to being able to burn fat at a higher rate, or it might suggest that athletes with naturally higher fat burning capability are more likely to choose a low carb diet.
• The overall V̇O₂max for the two groups was broadly similar with an average of 64.7 and 64.3, though both groups had a wide range of 55-76. Therefore it's hard to reach any conclusion on the impact of a low carb diet on V̇O₂max.
• The endurance test was not to exhaustion and is more of a long training run. The intensity was 64% of V̇O₂max, which a pace of about 100 seconds/mile slower than marathon pace.
• The biggest problem with interpreting the endurance test is that not only are the two group of athletes different, but the groups were given a shake with different nutrition 90 minutes before the test. It's impossible to know what variation between the groups results is due to prior training and what is due to the shake.
• The low carb shake consisted of heavy cream, olive oil, whey protein, walnut oil, and strawberries, which is seems to be a good choice of ingredients. Constructing a shake that is low in carbs but palatable is tricky.
• The high carb shake is more problematic: It also consisted of the heavy cream, olive oil, whey protein, walnut oil, and strawberries from the low carb shake, plus bananas and agave syrup. The high carb shake is nothing like the nutrition a reasonable athlete would consume. The carbohydrate in agave syrup is 67-90% Fructose, and this level of Fructose makes this a remarkably poor choice for athletes (or anyone else) as it is extremely slow to digest. The inability to digest the Fructose can be seen in the way the high carb athletes' blood glucose does not rise. It's impossible to know if this was simply a poor choice or if it constitutes an intentional bias intended to promote low carb nutrition.
• During the endurance test the low carb athletes burned more fat and less carbohydrate than the high carb athletes. Normally an increased fat burning capability is seen as a good thing in athletes as it helps preserves Glycogen, which is typically a limiting factor in endurance events. However, both groups burned through similar amounts of Glycogen. The test was a fixed 3 hour run, so there's no way of knowing if either group would last longer before reaching voluntary exhaustion, but it seems likely that Glycogen levels would limit both groups equally. It also seems likely that a reasonable source of carbohydrate for the high carb group would have resulted in them using less Glycogen than the low carb group. (The study noted that the low carb athletes burned more Glycogen than the estimate for their total carbohydrate burn, which is rather strange.)

3.2 Non-Ketogenic Low Carbohydrate Diet (NKLCD) and Exercise

A 1994 study looked at the effect of a low carbohydrate diet on cycling performance. This study explicitly avoiding being ketogenic, providing calories from 67% fat, 7% carbohydrate and 25% protein, with a control condition of 70% carbohydrate. The exercise test protocol was rather unusual; a maximum strength test was followed by a Wingate test, then a high intensity (85% V̇O₂max) endurance test to exhaustion, a 20 minute rest, then a moderate intensity (50% V̇O₂max) endurance test to exhaustion. The high fat and high carbohydrate conditions were not statistically different except for the moderate intensity test where the low carbohydrate condition produced significantly better endurance (80 minutes rather than 43.) However, while the high intensity exhaustion test was not significantly different, the low carbohydrate condition was only 8.3 minutes rather than 12.5. This is not statistically significant, but it does suggest the high carbohydrate condition could have been far more tiring than the low carbohydrate condition. There have also been other studies showing that a rest period between exercise bouts can produce low blood glucose, so this study might be the result of the rather unusual structure.

3.3 Exercise capacity in obese subjects undergoing Protein Sparing Modified Fast

An earlier study by Phinney looked at untrained, obese subjects that were treated using a Protein Sparing Modified Fast (PSMF) that provided 500-750 Calories/Day. Because a PSMF requires medical supervision, the subjects were in admitted to a medical facility for the 8 week study. Using a treadmill walking test to exhaustion showed that endurance dropped to 80% of baseline after one week, but after Ketoadaptation their capacity increased to 155%. This is a fascinating study, but the nature of the subjects, their weight loss, and the difference between the PSMF and a Ketogenic Diet, all make it hard to draw general conclusions.

3.4 Post-exercise ketosis in post-prandial exercise- effect of glucose and alanine

A study of 33 trained endurance runners on a low carbohydrate diet looked at the effect of glucose and alanine (an amino acid). The diet was not ketogenic as their blood BOHB was only 0.23 and they were on the diet for only 48 hours before the tests (no Ketoadaptation). This study showed that 50g (200 Calories) of Glucose before or immediately after a 90 minute half-marathon distance run did not impact blood ketone levels. (Taking 50g 2 hours after exercise will cause a dramatic drop in ketone levels). However, the alanine reduced ketone levels when taken before or after the run.

3.5 The effects of a ketogenic diet on exercise metabolism and physical performance in off-road cyclists

This study looks promising, but sadly it does not appear to actually use the Ketogenic Diet. The diet the study used consisted of 70% fat, 15% protein and 15% carbohydrates, and provided an average of 2,865 Calories, which works out to ~145g of carbohydrate per day, a level generally considered far too high to be Ketogenic. In addition, while the Ketone Levels increased, the level of BOHB only rose to 0.15 mmol/L. The recommended minimum level to be considered Ketogenic is 0.2 mmol/L (BOHB+AcAc), though other researchers recommend a BOHB level of 0.5. While the study was not Ketogenic, it had the subjects on the low carbohydrate diet for four weeks to allow adaptation. The low carbohydrate diet reduced body fat, which explained some improved performance metrics. The low carbohydrate diet also resulted in a higher oxygen requirement.

4 See Also

• The classifications and types of Low Carbohydrate Diet.
• An introduction to the Ketogenic Diet.
• My experiences with ultrarunning on the Ketogenic Diet
  • How the Ketogenic Diet can be used for the treatment and management of disease.
    • Using the Ketogenic Diets to mitigate and treat Epilepsy
    • How Ketogenic Diets might help with Cancer
    • A look at the possible mechanisms of action of the Ketogenic Diet
  • Health Risks of the Ketogenic Diet
    • Health Checks for the Ketogenic Diet
    • Supplements for the Ketogenic Diet
  • The time frame and changes that occur with Ketoadaptation
  • What are Ketones
    • How to measure Ketones
    • What ketone levels to aim for
  • The pros and cons of the Ketogenic Diet for athletes
    • The Science of Ketogenic Exercise
  • The Types of Ketogenic Diet
    • The "traditional" Ketogenic Diet based on the Ketogenic Ratio
      • The ways the Ketogenic Ratio can be calculated
        • The different definitions of Net Carbohydrates
    • Low Glycemic Index Treatment
    • MCT Diet
      • MCT
    • Modified Atkins Diet (M.A.D.)
• My Ketogenic Recipes
• Non-Ketogenic Low Carbohydrate Diets