This article was written and published on Rogue Health by PD Mangan, and has been reproduced here in its entirety.
When a person fuels themselves with much more fat than glucose they enter into the metabolic state known as ketosis. As an energy source, fat combusts with oxygen to make ATP in mitochondria (via oxidative phosphorylation) and glucose ferments in the absence of oxygen to make ATP in the cytoplasm (via glycolysis).
Ketosis is a normal metabolic state for humans to be in. You can be in ketosis in a variety of scenarios; when pregnant, as a breast-fed infant, or as an adult when fasting intermittently (and especially if on a low-carb high-fat diet). There’s even post-exercise ketosis that happens when you exercised hard enough to extensively deplete glycogen. And of course ketosis happens when fasting or starving. It’s physiological (normal) in the former and pathological (abnormal) in the latter.
So being in ketosis is normal. A weight-stable person eating a standard ketogenic diet consisting of something like salmon, eggs, coffee and a kale salad can pump out about 185 g of ketones a day from their liver (specifically the ketones beta-hydroxybutyrate and acetoacetate) .
But being out of ketosis is normal too.
Humans regularly drop out of ketosis from short bursts of explosive activity which momentarily raising their blood sugar and glucocorticoids. Most commonly though, people leave ketosis after eating a couple of baked potatoes or some fruit. The reason is that insulin rises enough in the absence of sufficient glucagon, thus sequestering fat in fat cells and rendering it unavailable for being turned into ketones.
So if both ketosis and not-ketosis are normal metabolic states to be in, which one should an athlete choose when trying to perform at their best? Should it be constant ketosis? No ketosis ever? Or ‘targeted’ ketosis, a la ‘train low (carb), compete high (carb)’?
Let’s look at data on endurance, high-intensity efforts and body composition.
Let’s start with mice put on an ad lib 8-week ketogenic diet (76.1% fat, 8.9% protein and 3.5% carbs) or a control diet (7% fat, 17.8% protein and 64.3% carbohydrate) . Note that the control mice are getting nearly twice as much protein than the keto mice are. On the ‘time to exhaustion’ task, eyeballing the graph suggests that the keto mice lasted a just under 300 minutes whilst the control mice ~240 min. That’s a 20% difference, not bad!
We don’t know about the quality of their diet although it was probably terrible (e.g. seed oils, grain flour). But we do know that, at rest, the keto mice had 2.4 mmol/L of blood BhB and the controls had 0.29 mmol/L. However, the control mice showed post-exercise ketosis with blood BhB levels climbing to 2.8 whilst levels dropped to 0.72 mmol/L in the keto mice. Interestingly, the authors noted that the ketogenic mice appeared somewhat protected from acute liver and kidney injuries, as measured by BUN, ALT and AST right after they stopped exercising.
I can already hear the Calories In-Calories Out choir singing about how the performance difference came down to body weight. It didn’t.
Let’s now jump to a human keystone study from 2015 by Volek et al. . The design is simple: get 20 male endurance freaks who habitually eat a standard high-carb diet or a ketogenic one. Just before they hop on the treadmill to keep a pace of about 65% of their VO2 max, give the keto folk a keto shake (81% fat, 14% protein and 5% carbs) and the controls a standard shake (50% carbs, 14% protein and 36% fat). The first result is this beautifully crisp difference in maximal fat oxidation rates (g/min). The most fat burning carb burner can’t burn as much fat as the least fat burning fat burner.
By the way, this data required exercise physiology textbooks to update the higher maximal fat oxidation rates. It also shows that you can still burn a fair bit of fat at high intensities. Another exercise physiology dogma that falls.
What we also see is that the keto guys are burning proportionally less glycogen and less absolute amounts of glycogen overall. So is it that these keto guys don’t need all that glycogen or is it that they actually can’t use it like their counterparts can? The answer to that question is key to discerning whether or not low-carb diets are appropriate for high-intensity efforts. We’ll get to that in a bit.
Funnily enough, muscle glycogen content between the controls and keto guys was the same, both before and after exercising. I like the individual data points because they reveal the counterintuitive fact that some keto folk can have more replete glycogen stores than people eating lots of carbs.
Even back in 2004 it was known that post-exercise ‘carb-loading’ to recover faster by replenishing glycogen probably didn’t work . Funny how ‘industry trends’ can give researchers amnesia…But Fournier et al. reminds us that
“during recovery from exercise, it is possible for skeletal muscles to replenish their glycogen stores under conditions expected to be highly unfavourable to glycogen synthesis such as fasting or active recovery. The rates of muscle glycogen synthesis can be very high under these conditions […] This capacity of skeletal muscles to replenish their glycogen stores under extreme conditions is clearly advantageous as it allows muscles to maintain adequate levels of glycogen stores for fight or flight responses”
If you use an evolutionary lens (or common sense), this ability to replenish glycogen without carbs isn’t surprising given the following: we evolved as apex predators subsisting on lots of animal foods without access to sugary sport gels at all times. That being said, maybe carb-loading does have a performance benefit. It’s probably not because of any supposed glycogen loading repletion effects from carb-loading but rather from the fact that carbs are ergogenic aids. Evidence this fact comes from experiments showing performance benefits in athletes whether who just tasted the sugary drink or drank it [5, 6]. This effect isn’t always reproducible however [7, 8].
We have talked about athletes, but what about non-athletes exercising on a keto diet? This 8-week study advised severe calorie restriction of about 30% to the 60 obese men and women that were asked to either follow a keto diet unusually high in protein (61% fat, 35% protein and 4% carbs) or a high-carb control diet (46% carbs, 30% fat and 24% protein) . We don’t know what either group actually ate unfortunately. The keto group were only mildly ketotic by week 2 with 0.49 mmol/L blood BhB. Curiously though, the keto group’s average RER stayed above 1. This signifies that, at the moment of measure, they weren’t in ketosis. The closer to 1 you are, the more carbs you are burning, and the closest to 0.6 the more fat you’re burning. The blood BhB and RER values don’t make sense.
Both groups had their grip strength, maximal knee extensor strength, time to exhaustion on a graded treadmill, weight loss and metabolic markers taken before and after the 8 weeks. Both groups basically fared the same but for a few metrics. The keto group lost more weight, 8.4 kg vs the 6.7 kg of the high-carb control. No surprise there. Both groups gained leg strength but lost grip strength.
As mentioned beforehand, the RER didn’t lower much in either group which you would expect if they were losing significant amounts of weight – as this entails increased burning of fat (especially if much of that weight loss comes from fat rather than muscle). The takeaways here are that:
- if you’re obese and wanting to lose weight, a 30% restriction of calories is unsustainable and excessive given how it worsens grip strength. Grip strength is a decent mortality predictor of all-cause mortality, with hazard ratios of 1.96 (1.30-1.52) with a 95% confidence interval 
- if you’re obese and wanting to lose weight, a keto diet on the higher protein side of things is a good idea
- if you’re obese and wanting to lose weight, maybe try some resistance training rather than ‘low and slow’ modalities like the treadmill. Keep the low and slow stuff for walking and not being sedentary
So it’s clear that humans can perform very well in endurance events on keto. Better levels of fat adaptation that cannot be reached on high-carb diets may even turn out to be advantageous. But what about shorter-duration higher-intensity efforts?
15 male and female Crossfitters with very little experience in the sport (3 months) were fed a keto or higher-carb control diet for 6 weeks . There were no notable differences between both groups except for mean power output; it increased beyond significance for the controls from 8.24 to 8.7 W/kg but not for the keto folk. It’s hard to interpret this difference since both groups did equally well on all other ‘power’ tests like the 500m row, Wingate Anaerobic Test and 3-repetition maximum (3RM) deadlift). The authors, however, conclude the following:
“A 6-week ketogenic diet did not affect the performance of short-duration high-intensity exercise. Our data does not support the hypothesis that ketogenic diets induce detriments in the performance of activity that is anaerobic in nature. The current study took place over a 6 week period, allowing for keto-adaptation to occur; results may be different if a shorter time period were utilized”
Full-text wasn’t published as of 15/12/2018 so it doesn’t count for anything yet. However, it would be consistent with the majority of anecdotal reports around keto and Crossfit.
The nocebo effect (your health worses from negative expectations) is the opposite of the placebo effect (your health improves from positive expectations). I think the nocebo effect from no longer having carbs you believe you need is significant and underappreciated. Additionally, the supposedly positive effects of glycogen loading are likely exaggerated.
My friend Alessandro Ferretti and his colleagues studied the effect of ketogenic diets versus high-carb control diets on HIIT performance (high-intensity interval training) over 4 weeks . The diets were considered ketogenic if subjects reported eating < 50 g of carbs a day. The HIIT sessions was a 2:1 work-to-rest ratio; 5 high-intensity repetitions consisting of 3 min at 100% vVO2max separated by 1.5 min of passive recovery.
The authors found that the ketogenic diet
- improved fat oxidation during the graded treadmill test and HIIT session
- didn’t impair performance or cardiorespiratory fitness during the HIIT session
These results are promising but there are limitations. My analysis arises from discussions I’ve had with Dr.Tommy Wood.
- Blood BhB levels in the keto group increased to 0.7 mmol/L after two weeks but decreased to 0.4 mmol/L after four weeks, suggesting the athletes weren’tincreasingly keto-adapting as intended. Carb-creep is the likeliest explanation.
- The mean total energy intake was higher in the high-carb control group during the 4 weeks which could give them a potential advantage over the keto group. Indeed, the keto group lost a little weight unlike the control group
- Time to exhaustion tests (TTEs) aren’t sufficiently representative of real world racing. This is a fair criticism. But you can make it for any other such exercise physiology studies and it casts shade on all results (positive or negative)
- The reporting of the effect sizes is confusing. It’s unclear if the control group had a statistically significant relative increase in TTE. If it is the case though, it would suggest an advantage from carbs and/or an increased caloric intake relative to the keto group.
A bunch of Polish researchers found 8 off-road cyclists to put on a ketogenic (70% fat, 15% protein and 15% carbs) or high-carb diet (50% carbohydrates, 30% fats and 20% protein) . They used a cross-over study design, where subjects serve as their own controls by being on both diets with a washout period in between (to avoid carry over effects). Bonus points to the authors for including a few meagre details about the composition of the diets.
The authors explain that blood ketone levels in the keto group elevated 4-fold compared to the high-carb controls, thus confirming dietary adherence. Problem is, 4 times 0.04 gives you 0.16 mmol/L (they measured 0.15 mmol/L). That’s not nutritional ketosis so the study can’t answer the question it’s asking. It can still be useful to learn things about a low-carb diet though.
On average, these subjects ate 3,865 kcals a day, weighed 80.3 kg and had 14.5% body fat. These off-road cyclists were expending a lot of energy! Their fitness is probably pretty good. The pseudo-keto group ended up weighing 2 kg less with 0.6% less body fat, while the control group stayed around the same weight and body fat percentage.
The authors found that “when maximal intensity was introduced, FFA metabolism was inhibited by glycolysis, which was evidenced by significant increases in LA concentration”. This occured in both groups but more so in the pseudo-keto one. It means that the diet was indeed low-carb enough to increase the cyclist’s reliance on fat for fuel and thus down-regulate carbohydrate metabolism. This is further supported by the high-carb control group’s comparably higher RER and lactate concentrations at lower intensity levels. In comparison, the pseudo-keto group had a higher heart rate and improvements in VO2max and lactate threshold VO2 relative to their new (lower) weight.
As expected in a 4-week intervention that’s too short for full keto adaptation, “the power output during work at maximal intensity was compromised on the ketogenic diet[maybe because of] reduced activity of glycolytic enzymes due to the four-week diet intervention”. Keep in mind that there are different levels of adaptation. Some happen over a couple of days (e.g. basal fat burning), weeks (e.g. appetite) and months and months (notably, top end intensity efforts).
The title of this last study by Caryn Zinn and colleagues from 2017 says it all .
Ketogenic diet benefits body composition and well-being but not performance in a pilot case study of New Zealand endurance athletes
It studied 4 females and 1 male for 10 weeks, checking their dietary adherence by taking blood ketones levels. They always stayed above 0.5 mmol/ L from week 2 onwards
- Blood ketones ranged from 0.5 – 4.2 mmol/L
- Females ranged 0.5 – 1.9 mmol/L and never exceeded 1.9 mmol/L
- The male athlete ranged between 1.0 – 3.5 mmol/L, went below 1.0 mmol/L on 2 occasions (0.8 and 0.6 mmol/L) and measured at 4.2 mmol/L on 1 occasion
Some ran, others cycled. Kudos to Zinn and colleagues who gave examples of the suggested meals. For example, the male participant’s breakfast was
- ½ cup of granola (*nuts, seeds, coconut threads and coconut oil), 150 mL of coconut cream and 100 g of mixed frozen berries and 30 mL. The daily diet had 2,450 kcals, with 24 g of net carbs, 103 g of protein (1.4 g/kg) and 215 g of fat.
His lunch and dinner had smallish amounts of salmon and other fish, respectively, which is great but still leaves his breakfast without any animal sourced food. This is not good for maintaining a nutrient dense diet, optimising exercise performance and especially body composition.
So what happened?
- Mean max oxidation improved 43% !
- The average level of exercise intensity at which VO2max occured increased 31.2% !
- Their average work rate increased 21.5% from 39.5 (± 11.9%) to 48.0 (± 8.9%) !
- 2 out of the 5 athletes showed an increase in Fatmax relative to WR max, and the remaining 3 showed no change
- All the while average fat loss was 4.0 (± 3.1%) kg !
- All participants lowered their time to exhaustion by 2 min (± 0.7 min) !
- Maximal aerobic performance was reduced !
What does this mean? Looking at the individual responses we can see
(a) big confidence limits, a statistical term representing the the upper and lower bounds of possible results you’re willing to consider. Big confidence limits represent inaccurate knowledge of risk.
(b) relatively small effect sizes in a small sample size. No surprise there and the authors are upfront about it.
That’s why I don’t quite see the need to go into the detailed justifications for the apparent decreases in time to exhaustion. They invoked glycogen depletion, pyruvate dehydrogenase downregulation and counter-regulatory glycolytic downregulation due to increased fat burning.
I think it’s rather simpler.
The participants were in a possibly involuntary but significant caloric deficit given their weight loss. After looking over their meals, their animal food intake could have been quite a bit higher. This is all the more true given they’re consistently exercising seriously and thus placing increased demands for animal foods.
And what did they report?
General well-being improved overall. Several important subjective measures changed for the best, from improved skin to better recovery and prostate issues that resolved. Nevertheless, some felt a loss of power on the top end of their efforts, others constipation and others yet were bored by the diet. That last thing? I cannot explain for the life of me.
The takeaway is even when the diet is likely hypercaloric you can still exercise well and feel generally better, but the top end of your performance won’t be at its best. I think that all makes sense but I don’t think that being a better fat burner (i.e. more metabolically flexible) makes you a worse carb burner.
There’s the physiology of increased fat burning lowering glycolysis, which is correct. Then there’s the inference from that, that better fat burners are worse carb burners. That’s incorrect. This then leads to the assumption that low-carb or ketogenic diets must specifically disadvantage adherents when it comes to performing at high-intensity.
If it turns out that better fat burners are indeed worse carb burners, it won’t simply be because as your burn more fat you burn less glucose and vice-versa.
When it comes to body composition changes on a ketogenic diet, results are more clearly are more clearly positive overall.
For 3 months 12 male and female Crossfitters were either put on a high-carb control diet or a ketogenic one and changes in their performance and body composition were measured. Both groups had similar changes in their resting energy expenditure (REE), one-repetition maximum (1-RM) on the back squat movement, 400 m run times and peak VO2 values. Knowing their sample size was small, the authors conclude that “these preliminary data suggest that adopting a ketogenic diet causes marked reductions in whole-body adiposity while not impacting performance measures in recreationally-trained CrossFit trainees” .
I’ll take it.
The keto group seemed to actually be ketotic given the recorded blood BhB levels. As in many studies with self-reported dietary intake, we see here a decline in blood BhB levels after the first couple of weeks.
Volek’s recent study is entitled “Keto-adaptation enhances exercise performance and body composition responses to training in endurance athletes” . This study had 20 male endurance athletes aged 33 and weighed 80 kg. It had an interestingly intense training phase up to the last test day.
- 7+ hours a week of endurance training (moderate intensity 56 – 68% VO2max)
- 2 strength sessions per week; 6 sets of 8-10 reps on a leg press, or free squat (70–80% of participants 1RM)
- 2 HIIT sessions every week (10 sets of 1 minute bouts at 70% peak power with 1 minute recovery)
The authors note that during endurance training the keto group kept carbs to a minimum before training and limited food consumption during exercise. But before being tested again the subject’s breakfast was either high-carb for controls (60% carbs, 26% fat and 14% protein) or keto for the others (80% fat, 15% protein and 5% carbs). As shown in the picture above, they performed a six second (SS) sprint, then a 100 km time trial (TT) and finished on a critical power test (CPT). During the testing the controls took in 30 – 60 grams of carbs per hour (glucose, maltodextrin, sucrose and fructose) whilst the keto group only drank water and 0 calorie electrolyte drinks.
The keto group lost 5.9 kg (of which 78% was from fat and water) compared to 0.8 kg in the high-carb control group (of which 62.5 % was from fat and water). That’s a net win for keto in terms of body composition.
But it tanked their performance, right? No.
Despite the high-carb controls consuming 5.3g/kg of carbohydrate for 12 weeks – so 424 g of carbs a day for an 80 kg man – they didn’t equal the keto group’s mean time trial performance done right after the training period. Seems this study by Volek and friends really gave the high-carb group all the carbs they wanted and ensured the keto group wasn’t underfuelling either.
The keto group had very favorable body composition changes and showed that “endurance performance can be maintained [on a ketogenic diet], and in some cases improved compared to a [high-carb] diet”. Of course, this is less surprising given that the advice these athletes were given was better than in most keto-sports studies; the diet was 81% fat, protein was set at nearly a third higher than in other studies at 1.9 g/kg of lean body mass (LBM) and carbs were kept low at 41 g per day. The keto group had ketones of 0.1 mmol/L before adapting to the diet and increased them to 0.5 by the end of it. They were mildly ketotic but with a good RER (i.e. lower), which is entirely appropriate for testing a ketogenic diet.
In 2017, D’Agostino, Lowery and Wilson took 25 college students to see how the ketogenic diet might affect resistance training and body composition, strength and power . The keto did better for losing fat mass, with 2.2 kg (± 1.2) instead of 1.5 kg (± 1.6). The high-carb group put on more lean body mass in 10 weeks, 4.4% versus 2.4% (p < 0.01). When measured again a week later, the keto group shot up higher than the high-carb group at 4.8% ! Why? Because the keto group carb-loaded with 265 g of carbs. This will increase water retention which gives a false positive for increased for lean body mass, which also happens to equalize the comparison with high-carb diets. When equalized in this way we can see that the high-carb diet doesn’t fare better.
This German study from 2017 without a control group lasted 6 weeks, had 42 subjects that were 37 years old and were asked to monitor ketosis via urine strips . Their results were positive but unimpressive and their tools to measure body composition (bioimpedance analysis) were inaccurate. They reported moderate weight loss from 70.3 kg (± 11.5) to 68.4 kg (± 10.3).
But 2 interesting things happened.
First, folic acid improved from 7.81 (± 3.32) ng/mL to 10.04 (± 3.92). Dietitians are worried low-carb diets are micronutrient deficient in folic acid but they’re wrong about that.
Second, their triglyceride-to-HDL cholesterol ratio improved from 0.89 (0.43 – 4.24) to 0.76 (0.48 – 3.37). Dietitians and doctors are also worried keto diets worsen cardiovascular disease factors, which they’re also wrong about. And of course LDL cholesterol increased a little bit (12 mg/dL) which is totally uninteresting in terms of risk. Why? Because it’s not an independent risk factor for. This is clear on mechanistic grounds but unclear on epidemiological ones.
Studies 12 and 13
This 7-day study shows that carbohydrate restriction is a potentially useful tool for athletes needing to quickly cut weight for weight class competitions without the side-effects of big or small drops in performance, depending on the individual .
This study shows how glycogen concentrations don’t suppress the anabolic response to resistance exercise . So even if your glycogen stores are sometimes lower than those of high-carb athletes, it won’t stop you from putting on muscle mass.
Here’s how I’d summarize the ‘keto for sports’ evidence so far:
- The longer the study…or the longer its keto-adaptation phase…or the more keto-adapted the subjects are…the more likely the study is to find favorable performance results
- Keto is worth trying for anyone in any sport (but start in the off-season!)
- It’s highly unlikely keto is better for high-intensity
- It’s unlikely that keto is bad for high-intensity
- It’s likely that keto is neutral for high-intensity
- It’s likely that keto diets are better for endurance
- It’s very likely keto diets are better for body composition
- It’s very likely keto diets are generally healthier than standard high-carb diets for athletes
This article was written and published on Rogue Health by PD Mangan, and has been reproduced here in its entirety.