Low carbohydrate, high fat diet impairs exercise economy and negates the performance benefit from intensified training in elite race walkers

During continuous exercise lasting more than a few minutes duration, skeletal muscle is fuelled by both intra‐ and extramuscular carbohydrate (CHO) and lipid substrates, with only a minor contribution from amino acids (for review see Hawley et al. 2015). These observations were made over a century ago by Zuntz & Schumburg (1901) who manipulated the proportions of fat and CHO in the diet for several days and showed changes in the non‐protein respiratory exchange ratio (RER) during subsequent submaximal exercise. These experiments identified a potential benefit for CHO as a substrate for muscle metabolism by virtue of an ∼8% higher energy yield per litre of oxygen (O₂) consumed when CHO was the primary fuel oxidised. Krogh & Lindhard (1920) confirmed the findings of Zuntz & Schumburg (1901) and reported a 5.5% greater energy yield (per litre of O₂ consumption) during laboratory cycling with CHO rather than fat dependence. These workers also reported that subjects perceived a fixed, submaximal exercise task to be easier after several days of consuming a high CHO diet than when the preceding diet was CHO‐restricted and high in fat.

The application of surgical techniques to exercise biochemistry in the 1960s (Bergstrom & Hultman, 1966) made it possible to obtain direct measurement of substrate storage and utilisation from skeletal muscle biopsy samples and precipitated a more sophisticated understanding of the role of dietary CHO in determining exercise capacity (Bergstrom et al. 1967) and endurance performance (Karlsson & Saltin, 1971). Largely as a result of these studies, sport nutrition guidelines in the early 1990s promoted a high CHO diet, particularly around exercise sessions, to maximise muscle glycogen stores, and exogenous CHO fuels for both training and competition (Coyle, 1991). However, recent guidelines (Burke et al. 2004, 2011; Thomas et al. 2016) have evolved to acknowledge that a universal recommendation of absolute CHO intake is unsuitable because (1) the highly variable nature of exercise/sporting activities dictates that the quantity and timing of dietary CHO should be prescribed according to the anticipated fuel needs of exercise (‘CHO availability’), (2) high CHO availability is beneficial for competition sessions and for specific training sessions where high intensity exercise and performance outcomes are required, but it may be unnecessary when sustained intense exercise is less important, and (3) the quantity and timing of CHO intake should be personalised to the athlete and periodised within the various micro‐ and macro‐cycles of training and competition requirements. Furthermore, there is recognition that a periodised programme that includes some training sessions deliberately undertaken with low endogenous and/or exogenous CHO availability (‘train low’; Hawley & Burke, 2010; Bartlett et al. 2015) or a delay in replacing muscle glycogen after a session (‘sleep low’) may promote greater cellular adaptations to training (Lane et al. 2015) and enhance performance (Marquet et al. 2016) to a greater magnitude than undertaking all sessions with high CHO availability.

An alternative view on the optimal fuel support for sports performance, however, is to maximise the contribution of fat as a substrate for exercising muscle by following a low CHO, high fat diet (Noakes et al. 2014; Volek et al. 2015). A key argument in favour of this notion is that even the leanest athlete has an abundance of endogenous lipid stores in comparison to limited CHO reserves. Diets high in fat upregulate the release, transport, uptake and utilisation of fat in the muscle, even in endurance athletes whose training would have been expected to maximise such adaptations (Spriet, 2014; Burke, 2015). This concept has been explored in trained individuals in various formats over the past 40 years. Strategies have included chronic exposure to either a ketogenic (< 20 g day⁻¹ CHO), high fat (80% of energy) diet (Phinney et al. 1983) or a restricted CHO (15–20% of energy), high fat (60–65% of energy) diet (Lambert et al. 1994; Goedecke et al. 1999) as well as short‐term adaptation to a high fat diet and 1 day of high CHO availability (Burke et al. 2000, 2002; Carey et al. 2001; Havemann et al. 2006). Although these iterations of high(er) fat, low(er) CHO diets result in increased rates of fat oxidation during exercise of varying intensities, evidence that this substrate shift translates to a clear enhancement of sports performance in athletic populations is lacking (Burke & Kiens, 2006; Burke, 2015).

Despite this, there has been a resurgence of interest in the published literature and social media (Burke, 2015) in a specific application of the high fat diet for athletes; the chronic consumption of a very low (< 50 g day⁻¹) CHO, moderate protein, high fat (75‐80% of energy) diet. The dual intent of this strategy, termed LCHF or the ‘keto‐diet’, is to increase the utilisation of fat as a muscle fuel and to expose the body to high levels of circulating ketones (Noakes et al. 2014; Volek et al. 2015). To date, the only published investigations of this diet in endurance athletes are a single intervention study (Phinney et al. 1983) and two recent cross‐sectional studies comparing ultra‐endurance runners/triathletes who have chosen this eating style with similar athletes following higher CHO diets (Volek et al. 2016; Webster et al. 2016). Since none of these investigations measured the effects of LCHF on sports performance or were undertaken using exercise of an intensity that is relevant to the majority of competitive endurance athletes, there is a need to systematically investigate claims that LCHF enhances endurance performance (Burke, 2015). Accordingly, we determined the effects of a 3‐week adaptation to an LCHF diet during a period of intensified training on exercise metabolism and performance of world‐class endurance athletes. Comparisons of interest included both the traditional sports nutrition guidelines of training with sustained high CHO availability and the contemporary ideas around periodised CHO availability. We hypothesised that an LCHF diet would result in increased rates of fat utilisation during exercise, but would impair exercise economy and fail to translate into a performance benefit.

This is the first study of the effect of a ketogenic low‐CHO, high fat diet consumed during a period of intensified training in elite athletes in which exercise characteristics across the range of intensities at which endurance and ultra‐endurance athletes train and compete were investigated. The novel findings were as follows. (1) Three weeks of intensified training and mild energy deficit in already well‐trained athletes increased peak aerobic capacity, independent of dietary support. (2) An LCHF diet was associated with the highest rates of whole‐body fat oxidation ever reported across exercise of varying speeds or intensities. (3) The cost of a shift in substrate oxidation (from CHO‐based to fat‐based fuels) during exercise was an increased oxygen demand for a given walking speed across velocities that translate to real‐life race performance in elite race walkers. (4) An improvement in performance was achieved by training and competing while consuming diets providing periodic or chronic high CHO availability, but not with the LCHF diet.

Our study was conducted over an ∼4‐week period, which included a 3‐week period of intensified training. Previous experience of undertaking training studies in well‐trained but lesser calibre endurance athletes (Lindsay et al. 1996; Yeo et al. 2008a; Cox et al. 2010) has shown this to be a sufficient time frame to induce detectable perturbations in muscle metabolism and performance. The exposure to the LCHF diet in the current investigation comfortably exceeded the time frame shown to produce robust cellular adaptations to ‘retool’ the muscle to increase its capacity for fat oxidation (Burke et al. 2000, 2002; Carey et al. 2001; Cameron‐Smith et al. 2003). For example, in the case of a non‐ketogenic high fat diet, peak rates of fat oxidation occur in as little as 5 days (Goedecke et al. 1999). However, in the case of the ketogenic LCHF diet, where further adaptation occurs to allow near‐excusive reliance on fat‐derived fuels, it has been suggested that several weeks of exposure is required before the fatigue and general loss of well‐being associated with such an extreme dietary shift abates (Phinney et al. 1983). With regard to the periodisation of CHO availability, we have recently shown in well‐trained but sub‐elite triathletes that a 3‐week intervention of this type was able to produce performance improvements compared with a more traditional diet of sustained high CHO availability (Marquet et al. 2016). The current study was limited to a tightly controlled intervention and rigorously collected non‐invasive measures of whole‐body metabolism and performance to allow us to work with elite athletes in a training camp environment with field and laboratory activities. Collectively, these factors give our results ‘real‐world’ credibility while also offering valuable insights into this topical area of exercise science.

Our data show that the LCHF group experienced changes in blood metabolites and substrate utilisation as a result of their dietary treatment. There was a general decrease in blood glucose concentrations at rest and during exercise, while blood ketone (β‐hydroxybutyrate) concentrations were generally elevated within the range of 0.8–2.0 mmol l⁻¹. Respiratory gas exchange measurements showed a substantial increase in whole‐body rates of fat oxidation and a concomitant reduction in CHO utilisation during the graded economy test across the range of walking speeds, as well as the sustained intensity 25 km training session. Although we did not investigate the mechanism(s) underpinning the increases in utilisation of fat during exercise, previous studies on high fat diets from our group and others have reported changes including increases in intramuscular triglyceride (Yeo et al. 2008b), an increase in hormone‐sensitive lipase (Stellingwerff et al. 2006), increases in the expression of fatty acid translocase FAT/CD36 protein (Cameron‐Smith et al. 2003) and carnitine palmitoyl transferase (Goedecke et al. 1999). Collectively, these changes suggest increases in fat availability, mobilisation and transport activities within the complex regulation of fat utilisation by the muscle.

Rates of fat oxidation were calculated during the 25 km training session to allow comparison with results from other studies of adaptation to low‐CHO, high fat diets. Our previous research on short term (5 days) adaptation to a non‐ketogenic low(er) carbohydrate (< 20% of energy), high(er) fat (65% of energy) diet found whole‐body rates of fat oxidation of ∼60 and ∼70 μmol kg⁻¹ min⁻¹ in well‐trained cyclists during training sessions at 85% and 65% of V˙O2 peak , respectively (Stepto et al. 2002). These figures approximate to absolute rates of fat oxidation of ∼1.3–1.5 g min⁻¹ and represented a doubling of fat utilisation in these cyclists compared with their substrate use at the same exercise intensity on a high CHO diet. Meanwhile, in the only other study involving the ketogenic LCHF diet in endurance athletes (Phinney et al. 1983), cyclists who were exposed to 4 weeks of a ketogenic (< 20 g day⁻¹ CHO) high fat (85% of energy) diet showed mean rates of fat oxidation of ∼1.5 g min⁻¹ while cycling at 62–64% V˙O2 max following an overnight fast. Other available data on the ketogenic LCHF diet come from recently published cross‐sectional studies of ultra‐endurance athletes who have self‐selected and self‐reported long‐term (> 6 months) adherence to this dietary strategy. In one investigation (Webster et al. 2016), cyclists habituated for ∼13 months to an LCHF diet (< 50 g day⁻¹ or 10% of energy CHO; ∼70% of energy from fat) sustained mean rates of fat oxidation of 1.2 g min⁻¹ during 2 h of cycling at ∼70% V˙O2 max compared with a gradual drift to 0.5 g min⁻¹ by a matched group of athletes who consumed diets providing higher CHO availability. However, Volek and co‐workers (2016) reported the highest peak rates of fat oxidation in the literature in elite ultra‐distance triathletes/runners with a mean adherence of 20 months to LCHF nutrition (∼10% energy from CHO, ∼70% energy from fat). These rates (1.54 ± 0.18 vs. 0.67 ± 0.14 g min⁻¹ occurring at ∼70% vs. 55% V˙O2 max ) were measured during a graded protocol involving short (2 min) exercise periods and attributed to the duration of the adherence to the LCHF diet. These same athletes were found to have fat oxidation rates of 1.21 vs. 0.76 g min⁻¹ over 3 h of submaximal treadmill running at 64% V˙O2 max , following a pre‐exercise meal based on their habitual dietary intake (Volek et al. 2016). In the current study, we observed sustained rates of fat oxidation of ∼1.5 g min⁻¹ in our elite race walkers, reaching a peak of 1.57 ± 0.32 g min⁻¹ towards the end of 2 h of exercise undertaken at ∼80% V˙O2 peak (50 km race pace). These rates represent a 2.5‐fold increase on the pre‐treatment values of 0.62 ± 0.32 g min⁻¹. In some individuals we observed peak fat oxidation rates exceeding 1.9 g min⁻¹, and note they were achieved by adaptation to the LCHF diet for only 3 weeks, in concert with a high fat pre‐exercise meal and further fat intake during the session. Our findings confirm that remarkable alterations in substrate utilisation can be achieved in well‐trained athletes; indeed, to the best of our knowledge, these are the highest rates of fat oxidation reported. Moreover, they may address the potential criticism that the current study was too brief since it appears to have achieved the most hallmarked feature of athletes who have ‘keto‐adapted’ for much longer periods.

Concomitant with the increased rates of fat oxidation, there was a decrease in rates of CHO oxidation in the LCHF group, as reported previously in studies in which CHO intake was substantially reduced (Goedecke et al. 1999; Burke et al. 2000; 2002; Carey et al. 2001), or restricted (Phinney et al. 1983; Volek et al. 2016; Webster et al. 2016). Mechanisms for the down‐regulation of CHO metabolism include the reduced availability of CHO substrate (e.g. reduced muscle glycogen stores, lower plasma glucose concentrations and the absence of exogenous intake of CHO during exercise). However, we have previously found, at least in the case of adaptation to a non‐ketogenic low CHO diet, a reduction in glycogenolysis during exercise, and a reduction in the active form of pyruvate dehydrogenase (PDHa) at rest and during exercise of both moderate‐ and supra‐maximal exercise, thus reducing the capacity for an oxidative fate of CHO disposal even when the supply is adequate (Stellingwerff et al. 2006). The reliance on measurements of whole‐body metabolism in the current study prevents a more mechanistic investigation of the changes in CHO storage and utilisation. However, the maintenance of blood glucose concentrations, albeit at reduced levels, and the increases in blood lactate concentrations during the graded economy and aerobic capacity tests indicate the presence of endogenous CHO stores despite minimal CHO intake. Indeed, due to the importance of glucose as a substrate for many tissues and a source of carbon for biosynthesis and anaplerosis, humans can adapt to conditions of food or CHO deprivation by synthesising glucose from a variety of substrates and reducing CHO oxidation (Soeters et al. 2012).

While one study of ultra‐endurance athletes habituated to an LCHF diet found little difference between that and a high‐CHO diet with regard to the storage and subsequent utilisation of muscle glycogen while running (despite paradoxical reductions in the rates of CHO oxidation during exercise) (Volek et al. 2016), the majority of observations of keto‐adapted athletes have reported modest reductions in resting muscle glycogen stores and lower rates of utilisation during exercise (Phinney et al1983; Webster et al. 2016). Tracer technology and muscle biopsies allowed Phinney and co‐workers (1983) to estimate that glycogen utilisation was reduced 4‐fold, and blood glucose utilisation, 3‐fold, during moderate intensity (62–64% V˙O2 max ) cycling when cyclists adapted to an LCHF diet, and they hypothesised that gluconeogenesis (from glycerol, lactate, pyruvate and gluconeogenic amino acids) was able to maintain blood glucose concentrations and permit glycogen restoration during recovery from exercise. Webster et al. (2016) measured glucose kinetics at rest and during 2 h of cycling at a slightly higher intensity (∼70% V˙O2 peak ) and showed that endogenous glucose production at rest and during exercise in keto‐adapted ultra‐endurance athletes was lower compared with athletes consuming higher CHO diets. They hypothesised that this reduced endogenous glucose production was due to a reduction in liver glycogen breakdown that was not compensated for by an absolute increase in gluconeogenesis (Webster et al. 2016) while noting that although absolute rates of gluconeogenesis were similar to those seen in fasted athletes with a habitually high CHO intake, they now contributed a greater proportion of body CHO stores, and possibly came from different substrates (glycerol, preferentially in the case of the LCHF group and lactate in the case of the high CHO group). In the current study we can only speculate on the exact nature of changes in CHO utilisation and storage with the LCHF diet. Although we recognise that plasma metabolite concentrations can mask changes in the rates of appearance and disappearance, high plasma lactate concentrations were observed after the 10 km race and the test of peak aerobic capacity, suggesting that the glycolytic fate of body CHO stores was preserved in the face of significant reductions in CHO oxidation.

In the present study, all groups of race walkers improved their aerobic capacity by 3–7% following the intensified training programme, independent of dietary intervention. Despite this, and their enhanced capacity to oxidise fat during higher intensity exercise, the race walkers in the LCHF diet failed to gain the improvement in 10 km race performance achieved by the HCHO and PCHO groups. This is a crucial finding for both competitive athletes and sports scientists, and the mechanisms underpinning the impairment of performance merit further discussion. Other factors such as increased perception of effort associated with exercise or effects of the reduced quality of workouts during the training programme may have contributed to the failure of the LCHF group to improve race performance, but the effect of the LCHF on the economy of race walking at speeds required for competitive performance is of major interest. The increased oxygen cost of ATP production from fat vs. CHO oxidation is self‐evident from stoichiometry and has been known empirically for a century (Zuntz & Schumburg 1901; Krogh & Lindhard, 1920). Indeed, since the oxidative phosphorylation yield is higher when NADH is the electron donor (3 coupling sites) compared with FADH₂ (2 coupling sides), and CHO metabolism produces a greater ratio of the reducing equivalents NADH/FADH2 than β‐oxidation, CHO is able to produce a greater ATP yield per unit of oxygen consumption despite the greater ATP production per unit of substrate from fat (Leverve et al. 2007). However, to the best of our knowledge, this notion appears to have been largely ignored in the recent discussions regarding LCHF and sports performance.

In sports such as running, cycling, swimming and race walking, the economy or efficiency of energy transfer to the speed of movement is a key determinant of performance (Joyner & Coyle, 2008). For example, although many elite runners present with a high aerobic capacity, within a homogeneous group, running economy (defined as the relationship between oxygen consumption and running velocity) at specific race paces is better correlated to performance than V˙O2 peak (Daniels & Daniels, 1992). In the present study, the oxygen cost of race walking at speeds approximating race pace in both 20 km events and the 50 km event was significantly increased with the LCHF intervention. Indeed, in the LCHF group, this negated the benefit of the training‐induced increase in peak aerobic capacity. In contrast, the athletes training and competing with high CHO availability achieved a reduction in the relative oxygen cost of walking, notably at the higher velocity of the shorter event on their race programmes. Others have also shown that the energy expenditure of cycling is reduced, and gross efficiency increased, following short‐term consumption of a high‐CHO compared with a low‐CHO diet (Cole et al. 2014). When exercise is undertaken at a modest fractional utilisation of an athlete's aerobic capacity, there is opportunity to compensate for a loss of economy by increasing oxygen uptake. However, during activities conducted at or around an individual's lactate threshold, such compensation is not possible and reduced ATP production from the available oxygen supply will limit exercise capacity.

The previous studies of LCHF in trained populations (Phinney et al. 1983; Volek et al. 2016; Webster et al. 2016) have examined exercise capacity at moderate exercise intensities (60–70% V˙O2 max ) that are unrelated to race pace in competitive endurance athletes (Hawley & Leckey 2015). We have previously noted that the discussion around LCHF and sports performance in both lay and scientific media often shows a poor understanding of the exercise characteristics that are important to competitive athletes (Burke, 2015; Hawley & Leckey 2015). For example, sports such as multi‐stage road cycling, triathlons and marathons are often described as endurance and ultra‐endurance events conducted at sub‐maximal exercise intensities; in fact, for competitive athletes at least, the terrain, pacing strategies and tactical elements in these events mean that brief but critical parts of the race, which often determine the outcomes (e.g. breakaways, hill climbs, surges, sprint finishes), are conducted at higher and often near maximal pace (Fernandez‐Garcia et al. 2000; Bentley et al. 2002; Tucker, 2016 2016). In addition, for many athletes, even the ‘background’ pace from which these brief bursts are performed in endurance sports such as the marathon requires high exercise economy and a sustained use of a very high percentage of maximal aerobic intensity (Joyner et al. 2011; Tucker, 2016 2016). The V˙O2 peak values (65–75 ml kg⁻¹ min⁻¹) measured in our elite race walkers are not exceptionally high in relation to values often cited for elite endurance runners and cyclists, although consideration must be given to the time of the season, the race walking test protocol and moderate altitude of 600 m of the testing facility. Nevertheless, they are comparable to values reported in other studies of international level male race walkers (Chwała et al. 2014) and we note that our group included several of the world's best athletes in this Olympic sport. Importantly, our studies show that performance in their events requires them to compete (and undertake substantial amounts of training) at a high percentage of their aerobic capacity where CHO oxidation is the major fuel‐generating pathway (Hawley & Leckey, 2015).

The 20 km race walk requires mean velocities of 15.6 and 14.3 km h⁻¹ to achieve the world record of 76:36 min and qualifying standard for the 2016 Rio Olympics of 84 min, respectively. Indeed, even after the period of intensified training, which improved the relative economy of the HCHO and PCHO groups, walking at a velocity approximating an individual's personal best race pace for the 20 km event within the graded economy protocol required a fractional oxygen utilisation of ∼85–90% V˙O2 peak . The corresponding data for the 50 km event (world record of 3:32.33 (h:min.s) and Olympic qualifying standard of 4:06 h are a mean race pace of 14.1 and 12.2 km h⁻¹, respectively. Among our participants, walking at a velocity of 12–13 km h⁻¹ to approximate the 50 km race pace of at least the middle calibre athletes in our group required ∼75% of their increased post‐training peak aerobic capacity in the graded economy test and 25 km standardised walk for the higher CHO groups. However, it remained at a mean of ∼80% V˙O2 peak of the LCHF group in the post‐treatment walk. It should also be noted that the mean 50 km race pace for the top competitors in our group is higher than the stage of the economy test at which we recorded these results and therefore mean race pace is conducted at a higher percentage of peak aerobic capacity, and key parts of the race at even faster paces.

Other sources support the concept that adaptation to a high fat diet impairs capacity/performance of sustained higher intensity exercise via a down‐regulation of CHO oxidation. Indeed, Phinney et al. (1983) noted that although chronic adaptation to an LCHF diet improved fat oxidation to the point that capacity for moderate submaximal exercise was ‘comparable to that observed after a high CHO diet’, ‘the price paid for such extreme conservation of CHO during exercise appears to be a limitation on the intensity of exercise that can be performed’. Havemann and colleagues (2006) found that a 5 day fat‐adaptation followed by restoration of CHO availability on the day before and during a performance permitted similar overall performance of a 100 km cycling time‐trial compared with a trial undertaken following a chronic high CHO availability approach, including similar performances of 4 km sprints (∼80% peak power output) interspersed within the time trial. However, when required to undertake 1 km sprints at > 90% of peak power output (∼95% of V˙O2 max ) within the time‐trial, there was a significant decrease in power output despite the maintenance of perceived effort and heart rate and an increased attempt to recruit muscle fibres (Havemann et al. 2006).

Another factor that has contributed to confusion about the LCHF diet and sports performance is a misunderstanding of the goal and outcomes of endurance training. An increase in fat utilisation during exercise is the most frequently discussed outcome of endurance training in relation to muscle fuel (Holloszy & Coyle 1984). While there is a training‐induced increase in rates of fat oxidation during exercise undertaken at the same absolute intensity, it is perhaps not well appreciated that the fractional contribution of fat to the new (higher) workload corresponding to the same relative intensity of the athlete's increased peak aerobic capacity remains essentially the same (Bergman & Brooks, 1999). Indeed, another goal of endurance training is to increases an athlete's absolute capacity to use CHO as a fuel substrate, via both oxidative and oxygen‐independent pathways (Hawley & Leckie, 2015), and a more holistic understanding of the ideal training programme is that it should enhance an athlete's capacity to utilise and integrate all of the body's range of energy‐producing pathways, with special focus on ways in which they become specifically limiting for performance.

The training intervention in the current study is representative of the periodised programme undertaken by endurance athletes to include a variety of workouts with different goals within a training phase. We attempted to expand on this principle, by including a separate arm of the study in which we manipulated CHO intake around specific training sessions to better achieve the adaptation and performance goals of the workout. For example, within the weekly training cycle we matched sessions featuring moderate intensity ‘recovery’ work with practices such as training in a fasted state or training with low glycogen concentrations to promote mitochondrial biogenesis and up‐regulation of fat oxidation (for review, see Bartlett et al. 2015). We also coupled practices to enhance CHO availability with sessions requiring good performance of higher intensity exercise as well as opportunities to practice race day CHO intake and to increase capacity for gut tolerance and absorption of CHO (Stellingwerff, 2013). However unlike the recent study in which we undertook such a manipulation in a 3‐week training programme with well‐trained triathletes (Marquet et al. 2016), we failed to see a benefit of this CHO periodisation over the intervention in which all sessions were undertaken with strategies aimed at achieving high CHO availability. It is uncertain whether the lack of additional benefit relates to the calibre of our athletes, the involvement of our intervention in the base phase of the training programme (where the increased training stimulus may have already maximised the adaptive response and reduced the capacity for differences due to CHO availability) or the large training volumes (which might have depleted CHO stores even in the face of the high CHO intakes around training such that the actual differential between the HCHO and PCHO interventions was reduced).

There are a number of other elements raised in this study that warrant further investigation. These include a titration of exercise intensity and exercise economy to find the pace at which the negative effects of the LCHF on exercise economy become detectable, thus differentiating the real‐life sporting events and athletes for which this represents an unsuitable practice. The potential for keto‐adaptation as a periodised rather than chronic activity, and the time course of some elements of the ‘de‐adaptation’ should also be considered. The effect of a low carbohydrate intake on gut absorption of CHO is worthy of merit since it is likely to be reduced, in the same way that an increased intake can improve the capacity for muscle use of exogenous CHO sources consumed during long exercise (Cox et al. 2010). This is of importance to the longer endurance events in which CHO intake during the race provides a substantial fuel source and is associated with enhanced performance (Stellingwerff & Cox, 2014), and is relevant since popular advice to athletes contemplating the use of LCHF practices includes the possibility that some may still choose to consume CHO during their competitive events. The role of ketones as a muscle fuel source was not investigated in the current study but is also of interest (Cox et al. 2016).

In conclusion, the results of the present study showed that despite achieving substantial increases in the capacity for fat oxidation during intense exercise, chronic adaptation to a ketogenic low‐CHO, high fat diet impaired exercise economy and negated the transfer of training‐induced increases in aerobic capacity into improved performance of a real‐life endurance event in elite athletes. In contrast, training with a diet rich in carbohydrate and which provided either high or periodised carbohydrate availability around training sessions was associated with improved race outcomes.

Incredible commitment from the larger ‘Supernova’ team was required for the successful completion of this study; these include the research crew of Rebecca Hall, Emma Lindblom, Carolina Castellanos, Michelle Minehan, Bronwen Charlesson, Taryn Richardson (Food team), Jill Leckey, Rachel McCormick, Peter Peeling, David Pyne, Philo Saunders, Alannah McKay, Julia Bone, Bronwen Lundy, Reid Reale, Victor Vuong, Damian Arazello (Laboratory team), Craig Hilliard, Daniel Coleman, Alison Campbell (Athletics Australia), Stephan Praet, Sacha and James Teh (Sports Medicine and Soft Tissue team), Masashi Kasahara (language assistance) and Lynne Mercer (administrative assistance). The support of colleagues and the environment of the Australian Institute of Sport were crucial in allowing a study of this magnitude to be tackled, but its inception and implementation would have been impossible without the support and insights of Jared and Claire Tallent. Finally, the generosity and commitment to sports science of the Supernova race walkers cannot be underestimated. This study would not have happened without their blood, sweat, tears and good humour.