Adaptation to a low carbohydrate high fat diet is rapid but impairs endurance exercise metabolism and performance despite enhanced glycogen availability

This study was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12618001974291p). It conformed to the standards set by the Declaration of Helsinki and was approved by the Human Research Ethics Committee of the Australian Institute of Sport (no. 20181203). Comprehensive details of the study protocol were explained orally and in writing prior to athletes providing their written informed consent.

This parallel‐groups designed study was conducted within a training camp that represented baseline preparation for the 2019 World Athletics (WA; formerly International Association of Athletics Federations) race‐walking season. The study took place at the Australian Institute of Sport (AIS) with athletes living in athlete residences and being supervised at all meals and training sessions. The study consisted of a 4‐week structured training block divided into three phases (see Fig. 1). Briefly, upon arriving at the training camp, athletes undertook tests (Test Block 1, Baseline) to identify baseline characteristics around aerobic capacity (V˙O2peak and exercise economy tests), performance (WA sanctioned 10,000 m race walk event, Race 1). This period involved a standardised diet of high energy and CHO availability during the first testing protocols, followed by a further 5‐day period with individualised achievement of the same high energy/CHO availability (HCHO, see below) to support structured training and prepare for the Baseline 25 km test. After the completion of this test, athletes were divided into two groups to commence a 5‐day dietary intervention (Adaptation) before repeating all testing elements. Test Block 2 (Adaptation) included a second 25 km walk test and economy/V˙O2peak tests while following the intervention diets (HCHO or ketogenic LCHF) before moving to the same standardised 24 h pre‐race dietary preparation undertaken for Race 1 (HCHO availability) prior to a second performance test (WA sanctioned 10,000 m race walk event, Race 2). To determine if the effects of the dietary interventions could be reversed, all athletes completed Phase 3, which involved 5 days on the HCHO diet and a third 25 km walk test (Test Block 3, Restoration). Since the absolute duration of each intervention involved 5 days of dietary intake and training, plus the period of the testing block (2–3 days) , we have simplified the study report to describe the intervention as a 6‐day period.

Fourteen male race walkers with international race experience were recruited for the study to fulfil our intention, based on resources and experience from previous studies of sufficient power to detect differences in metabolic variables when rigorous control of similar study intervention and testing protocols was implemented (Burke et al. 2017). Specific sample size estimation was calculated for performance measures using G Power software (Version 3.1, Bonn University, Bonn, Germany) based on our previous research in which the impact of LCHF on a 10,000 m race performance was investigated in a single camp (Burke et al. 2020). Based on such data, a sample size of seven athletes per group was considered appropriate (n = 7, critical t = 2.179; expected power = 0.939; P < 0.05). One athlete experienced visa delays and was unable to attend, leaving us with 13 athletes for this data set. The cohort ranged from world class athletes (e.g. Olympians, World Championship and WA Race Walking Team Championships medallists, national record holders) to highly trained athletes (e.g. training partners of world class athletes). Specifically, all athletes had competed internationally for their country during their career, and 11 of the 13 athletes included in this study were selected for at least one of the major events in the previous two seasons (2017 WA World Championships or the 2018 WA Race Walking Team Championships). Athletes were educated about the benefits and limitations of the dietary interventions and asked to nominate their preference for, or non‐acceptance of, each intervention, as described previously (Burke et al. 2017). We were able to allocate race walkers to a preferred dietary condition while achieving suitable matching between groups based on age, body mass, aerobic capacity and personal best for the 20 km race walk (Table 1). Seven athletes were previous participants in our earlier studies (Burke et al. 2017, 2020), with one choosing to repeat the LCHF intervention.

Assessment of fat free mass (FFM) and resting metabolic rate (RMR) was undertaken during Test Block 1 for use in the calculation of energy intake based on energy availability (EA) calculations. FFM was determined after an overnight fast and rest via dual X‐ray absorptiometry (DXA) using a standardised protocol previously described (Nana et al. 2015) and an iDXA (GE Healthcare, Milwaukee, WI, USA) with image analysis (enCore v16, GE Healthcare). Resting metabolic rate was assessed according to the standardised outpatient protocol described in full elsewhere (Bone & Burke, 2018)

All foods and fluids consumed during the study were provided and recorded by the research team, with menu construction and food preparation being undertaken by a professional chef, food service dietitians and sports dietitians. Meal plans were individually developed for each athlete as reported previously (Mirtschin et al. 2018), to integrate personal food preferences and nutrition requirements within the daily energy availability targets, planned energy expenditure of training and macronutrient goals. Daily checks of compliance to the training plan, dietary prescription and reporting requirements were undertaken, with food choices and portion sizes being adjusted at meals eaten in the later part of the day if actual training deviated from the plan.

Since a similar workload (training/testing protocols) was undertaken by each athlete in preparation for the first part of Test Block 1, we provided a standardised diet intended to provide high energy/CHO availability, based on our previous studies (Burke et al. 2017), of 56 kcal (225 kJ) kg⁻¹ body mass (BM) and 8.5 g CHO kg⁻¹ BM day⁻¹ over this short time period. Once estimations of FFM and RMR were completed, provision of daily energy intake was then undertaken according to EA estimations. Our goal was to provide diets with EA of ∼40 kcal (168 kJ) kg⁻¹ FFM day⁻¹ where EA was defined and implemented according to the original principle of Loucks (2004) as: Energy intake – (Energy cost of exercise; equal to net increase in energy expenditure above sedentary rates over the period of exercise)/FFM (kg)

This involved variation for each individual and each day according to the specific training load, including real‐time adjustments due to differences between planned and actual activities. The dietary treatments investigated in the present study are summarised below:

During Test Blocks 1 and 2 (Baseline and Adaptation), athletes completed an incremental exercise test to exhaustion on a custom‐built motorised treadmill (Australian Institute of Sport, Bruce, Australia) to determine exercise economy and V˙O2peak while race walking, as previously reported (Burke et al. 2017). Briefly, walking economy was assessed during four submaximal stages, each lasting 4 min and increasing in speed by 1 km h⁻¹ each stage. Starting speeds were selected at either 11 or 12 km h⁻¹ based on each individual's capacity; 20 km personal best times were compared with the 2019 WA World Championship qualifying standard of 1 h 22 min 30 s, with athletes faster than this mark commencing at 12 km h⁻¹ and increasing to 15 km h⁻¹ at the final stage. The speeds of the second and fourth stage corresponded approximately to each individual athlete's walking pace for the 50 and 20 km race walk event, respectively. This test commenced 2 h following the intake of the standardised test meal as outlined above.

Each stage was followed by 1 min rest for the collection of capillary (fingertip) blood samples to assess blood lactate (Lactate Pro 2, Akray, Japan), ketone bodies (β‐hydroxybutyrate (βHB); FreeStyle Optium Neo, Abbott Diabetes Care, Doncaster, Victoria, Australia) and glucose (FreeStyle Optium Neo) concentrations, as well as ratings of perceived exertion (RPE, 6–20 Borg Scale). Heart rate (HR) was measured continuously throughout the test (Polar Heart Rate Monitor, Polar Electro, Kempele, Finland). Expired gas was collected and analysed using a custom‐built indirect calorimetry system described previously (Saunders et al. 2010), with the final 60 s of gas collected accepted as steady state and used to calculate respiratory exchange ratio (RER) and O₂ uptake. A self‐selected warm‐up of 10 min duration preceded the test, which was maintained across trials.

Upon completion of the final submaximal walking stage, athletes rested for 5 min before completing a ramp (speed and then gradient) test to volitional fatigue. Treadmill speed was increased by 0.5 km h⁻¹ every 30 s until the speed corresponding to the individual's final submaximal stage was reached (14 or 15 km h⁻¹), with treadmill gradient increased by 0.5% every 30 s thereafter until exhaustion. Expired gas was collected and analysed throughout, maximal HR recorded, and capillary blood samples collected 1 min after completion.

During the Baseline and Adaptation Test Blocks, athletes competed in a 10,000 m race held on a synthetic 400 m outdoor athletics track (Canberra, ACT, Australia). To provide an incentive for a maximal effort, prize money was awarded to place getters as well as those athletes who achieved the highest percentage of their 20 km walking personal best when the times of the two races were combined. Each race commenced at 09.00 h and was conducted under WA rules, which involved officiating by technical judges, invitation for participation by competitors external to the study, a feed zone allowing water intake on the outside lanes of the track in hot conditions, the pit‐lane rule for technique infringements and photo‐finish electronic timing. Photo‐finish timing was used to provide official race times. Capillary blood samples were collected immediately before the start of the race and as each competitor completed the race.

The goals of the race nutrition plan were to standardise the individualised competition practices of each participant. Therefore, each athlete repeated a similar training load during the 48 h prior to each race (economy/aerobic capacity test schedule and personal training) and consumed the same HCHO diet over the 24 h pre‐race period, as well as the consuming the same CHO‐rich (2 g kg BM⁻¹) pre‐race meal 2 h prior to competition. The use of caffeine as a performance supplement by some participants was permitted under the following conditions: it was documented in pre‐race diaries, came from a reliable source (e.g. the same sports supplement) and was repeated in an identical protocol for both races. Due to the hot environmental conditions, the WA policy to allow a water station in an outside lane was implemented for both races. The athletes were able to choose freely to deviate from the inside track and obtain water or ice for consumption and cooling during the race; all made use of this resource.

Warm‐up for the 10,000 m race was undertaken according to our previous studies (Burke et al. 2017), whereby each athlete chose his own protocol, based on their real‐life experiences, and replicated this for each race. In the case of athletes allocated to the LCHF diet, the warm‐up was commenced in the laboratory to allow the impact of the 1‐day CHO restoration protocol on substrate utilisation and exercise economy during exercise to be measured. Between 90 and 45 min pre‐race (standardised for each athlete across races), these athletes completed a two‐stage incremental exercise test involving a 2 min warm‐up at 10 km h⁻¹ and two 4‐min stages (separated by 1 min at rest) at speeds corresponding to 50 and 20 km race walk events (i.e. 12 and 14 or 13 and 15 km h⁻¹ depending on ability). HR was measured continuously during the test, and expired gas was collected and analysed, with the final 60 s of gas collected accepted as steady state and used to calculate RER and O₂ uptake. Capillary blood samples were collected prior to warm‐up, and immediately after each stage to determine blood lactate, glucose and ketone concentration. Following completion of the test, athletes transferred to the athletics track (∼500 m walk) and completed the remainder of their race preparations, which were recorded and replicated across trials.

At the end of each of the test blocks, athletes completed a 25 km walk, 2 h after consuming a standardised meal that met the conditions of their dietary intervention. The training session was conducted as a hybrid laboratory‐field test, with 0–1, 6–7, 12–13, 18–19 and 24–25 km being undertaken on a treadmill in the laboratory and the remainder of the walk completed outdoors on a loop course (∼5 km) which included two aid stations to allow nutrition support to be received every ∼2 km as occurs in WA events. Athletes completed the treadmill portions of the walk at the speed corresponding to the second stage of the submaximal walking test (12 or 13 km h⁻¹), which approximated their 50 km race pace. Each negotiated an individualised but similar pace target (∼12 km h⁻¹) for the outside portion of the walk with the goal that each athlete would adhere to this pace for all three of his walks. Expired gas was collected during each treadmill segment for assessment of RER to determine rates of substrate oxidation and O₂ uptake. Capillary blood was collected for measurement of glucose, lactate and ketone concentrations immediately prior to beginning the session, and upon completion of each treadmill segment, using the protocols previously documented. HR and RPE were assessed at the end of each treadmill section.

The standardised pre‐exercise meal (previously described) was consumed 2 h prior to the commencement of the session. During the walk, handlers provided fluids and sports food choices that had previously been discussed with each participant to mimic race feeding practices. Athletes were offered sports gels and water to achieve an hourly intake of ∼600 ml of fluid and ∼60 g CHO in the Baseline and Restoration (both groups) and Adaptation (HCHO group). Meanwhile, in the LCHF Adaptation trial, athletes received non‐caloric fluid (electrolyte‐supplemented water) and LCHF cookies to match the energy intake from their other trial.

Prior to (fasted and post‐prandial pre‐exercise, ‘0 km’), during (1 km and 13 km), and immediately post (25 km) the 25 km‐long walk, blood was collected via an indwelling forearm cannula into 2.5 ml Vacuette SST tubes (Greiner Bio‐One, Kremsmünster, Austria), allowed to clot for 30 min then centrifuged at 1500 g, 4°C, 10 min. The collected serum was frozen and later analysed for free fatty acid (FFA) concentration using a colorimetric assay according to the manufacturer's instructions (NEFA‐c kit, Novachem Pty Ltd, Heidelberg West, Victoria, Australia).

Respiratory exchange ratio was calculated from steady‐state expired gases collected over 1 min periods during the economy test and maximal aerobic capacity (V˙O2peak) protocol. Rates of CHO and fat oxidation (g min⁻¹) were calculated from V˙CO2 and V˙O2 values using non‐protein RER values (Peronnet & Massicotte, 1991). These equations are based on the premise that V˙O2 and V˙CO2 accurately reflect tissue O₂ consumption and CO₂ production, and that indirect calorimetry is a valid method for quantifying rates of substrate oxidation in well‐trained athletes during strenuous exercise of up to ∼85% of V˙O2peak (Romijn et al. 1993). We did not correct our calculations for the contribution of ketone oxidation to substrate use in order to contextualise our findings with previous work performed by our group (Stellingwerff et al. 2006; Burke et al. 2017) as well as other reports of substrate utilisation in ultra‐endurance athletes who chronically consume LCHF diets (Volek et al. 2016; Webster et al. 2016). However, we acknowledge that there may be a small (but systematic) error in the use of conventional equations to calculate fat and CHO oxidation from gas exchange information (Frayn, 1983).

During each of the harmonisation and intervention periods, athletes were required to complete a 25–40 km‐long walk, interval training session (8–12 × 1 km on a 6 min cycle completed on a standard 400 m athletics track) and tempo hill session (14 km with ∼450 m elevation gain). The remaining training consisted of low intensity walking sessions (6–12 km each), and a strength training session. The daily energy cost of training (exercise energy expenditure; EEE), which represented the additional energy cost attributed to exercise during a training/test session period rather than the total energy expenditure during the period (Burke et al. 2018b), was calculated by subtracting resting metabolic rate from session energy expenditure.

Results are expressed as means (SD). Data from dietary intake, economy tests, long walks and race performances (Figs 2, 3, 4, 5, 6, Tables 2, 3) were analysed using a two‐way analysis of variance (ANOVA) within dietary groups. If significance was detected, a Bonferroni post hoc test was applied. Student's t test was used to analyse differences between subject groups (Table 1), and performances in Baseline and Adaptation races (Fig. 7). Significance was set at P < 0.05 (‘NS’ indicates not significant). All statistical analyses were performed, and graphs were produced using GraphPad Prism (version 8.3.1, GraphPad Software Inc., La Jolla, CA, USA). All data collected in this study are presented in the figures and tables; raw data are available on request to the corresponding author.

All athletes demonstrated compliance to their assigned dietary treatment and the monitoring of their food intake and training sessions. Table 2 summarises the results of the assessment of actual dietary intake, with the summary of mean daily intakes allowing the display of overall differences between diets. Assessments of exercise energy expenditure (EEE) showed a daily mean of ∼1600 kcal (6720 kJ) for HCHO group across all phases. Overall EEE was not different between groups, but the LCHF group incurred a reduced expenditure during the intervention compared with Baseline (P = 0.008) and Restoration (P = 0.18). There were no significant differences between groups for energy and macronutrient intake characteristics for the Baseline and Restoration phases (see Table 2), and each group was able to achieve the main targets for these phases: EA of ∼40 kcal (168 kJ) kg⁻¹ FFM day⁻¹, protein intake of ∼2 g kg⁻¹ BM day⁻¹, and CHO intake of 9–9.5 g kg⁻¹ BM day⁻¹ contributing ∼65% of energy. During the intervention phase, protein intake was maintained across both groups at the target of ∼2.2 g kg⁻¹ BM day⁻¹. The LCHF group achieved the targets of their dietary prescription, consuming <5% of energy and <0.5 g kg⁻¹ as CHO and 80% of energy from fat.

All race walkers participated in the graded economy and V˙O2peak test protocols. The results of these tests are summarised in Table 3 (BM, RER, absolute V˙O2 (L min⁻¹), HR and RPE data), Fig. 2 (relative V˙O2 (ml min⁻¹ kg⁻¹) and calculated substrate utilisation) and Fig. 3 (blood metabolites). There were no differences between groups at baseline (P = 0.419) or across the intervention period in absolute V˙O2peak (HCHO, P = 0.315; LCHF, P = 0.493). Both groups displayed an increase in RER, absolute V˙O2 (L min⁻¹), HR and RPE associated with the increase in exercise intensity across all four stages of both economy tests (all P < 0.0001). Within the LCHF group, there was a significant effect of the dietary intervention (Baseline vs. Adaptation) on metabolic cost and perceived effort, with changes to absolute V˙O2 (L min⁻¹) (Stages 1–4 higher, P < 0.005; V˙O2peakP = 0.493), RER (all stages lower following adaptation, P < 0.0001) and RPE (Stages 1–3 higher, P < 0.0001; Stage 4, P = 0.149). In contrast, there were no diet‐associated effects within the HCHO group for V˙O2 (L min⁻¹), RER, HR or RPE (Table 3).

These findings are supported by the change observed in relative V˙O2 (ml min⁻¹ kg⁻¹, Fig. 2A) and rates of CHO and fat oxidation (Fig. 2B and C). While there was a main effect for stage associated with increasing exercise intensity in the HCHO diet, no effect of diet was detected on relative V˙O2 (ml min⁻¹ kg⁻¹, P = 0.812) or fat oxidation rates (P = 0.139). However, the rate of CHO oxidation was lower across all four stages (P = 0.0004 and 0.0005 for Stages 1 and 3, P < 0.0001 for Stages 2 and 4). Within the LCHF diet, relative V˙O2 (ml min⁻¹ kg⁻¹) was higher across all submaximal stages following dietary adaptation (all P < 0.0001), indicating a decrease in exercise economy relative to Baseline when walking at the same absolute speed, with the magnitude of the increase being ∼6% and 4% at Stages 2 and 4, respectively. This increase in O₂ consumption was associated with the increased reliance on fat as a metabolic substrate, as fat oxidation rates increased across all four stages (P < 0.0001), peaking at 1.43 g min⁻¹ in the first stage (11–12 km h⁻¹), while CHO oxidation displayed a reciprocal decrease across all stages (P < 0.0001).

Figure 3 summarises circulating concentrations of metabolites (glucose, lactate and ketones) obtained via capillary blood samples measured during the economy test, representing resting values 2 h following a standardised meal, concentrations at the end of each stage and at the conclusion of the incremental test. There was a main effect of stage on blood glucose concentrations, reflecting exercise intensity, but this was unaltered by adaptation to diet in either the HCHO (P = 0.694) or LCHF (P = 0.308) group (Fig. 3A). Blood lactate concentrations increased with exercise intensity across both dietary treatments (P < 0.0001), with similar patterns in both pre‐ and post‐treatment trials for the HCHO group (Fig. 3B). In contrast, blood lactate concentrations at the two highest intensity stages were suppressed following Adaptation in the LCHF group (Stage 3, P = 0.030; Stage 4, P = 0.048). Blood ketone levels in the HCHO group were maintained at low concentrations (0.0–0.1 mmol L⁻¹), with no differences across the economy test for both pre‐ and post‐treatment. However, the LCHF group displayed significant increases in blood ketone concentrations following the dietary intervention (P = 0.008), which were highest at rest (1.2 (0.79) mmol L⁻¹). Despite declining over the subsequent economy stages, ketone concentrations remained elevated at all stages in comparison to the Baseline (all P < 0.0001).

All athletes performed the three hybrid laboratory‐field test 25 km walks, which are summarised in Table 4 (BM, RER, absolute V˙O2 (L min⁻¹), HR and RPE data), Fig. 4 (relative V˙O2 (ml min⁻¹ kg⁻¹) and substrate utilisation) and Fig. 5 (blood metabolites). In the HCHO group, there were no differences in time to complete the 25 km walk across the study (data not shown, ∼2:01 h, all P > 0.999), while BM decreased by an average of 0.96 (0.44) kg over each of the three ∼2 h training walks (P < 0.0001). The LCHF group followed a similar trend for BM changes in Walk 1 (0.92 (0.72) kg, P < 0.0001) and Walk 3 (0.69 (0.39) kg, P = 0.0002), but there was no significant change in post‐exercise body mass following Walk 2 (0.28 (0.45) kg, P = 0.110). Walk 2 undertaken with the LCHF diet was slightly but significantly slower than the other walks (2:11:44 (0:09:26), P = 0.0016 vs. Walk 1, 2:04:14 (0:05:34), P = 0.0005 vs. Walk 3, 2:03:22 (0:04:02); Walk 1 vs. Walk 3, P = 0.953).

Due to the prolonged nature of the exercise, both groups displayed a decrease in RER over the course of the 25 km walks (Table 4, main effect of distance, P < 0.0001), which was associated with a significant increase in both absolute and relative V˙O2 (Table 4 and Fig. 4A and D), respectively) and a gradual reduction in CHO (Fig. 4B, P = 0.001; Fig. 4E, P = 0.0001) and increase in fat oxidation (Fig. 4C, P < 0.0001; Fig. 4F, P < 0.0001). Within the HCHO group, there were no differences between Walks 1–3 in relative V˙O2 (Fig. 4A, P = 0.113) or CHO oxidation (Fig. 4B, P = 0.258); however, a main effect for testing block was detected for fat oxidation (Fig. 4C, P = 0.003), with Walk 2 (Adaptation) displaying higher overall rates compared to Walk 1 (Baseline, P = 0.03) and 3 (Restoration, P = 0.003). Similarly, RPE (P < 0.0001) and HR (P < 0.0001) increased across the test for the HCHO group, but this was not different between Walks 1–3 (P = 0.906).

Consistent with the intent of the dietary intervention, the LCHF group displayed substantially altered fuel utilisation in Walk 2, with fat oxidation being significantly increased in comparison to Walks 1 and 3 (P < 0.0001), and a reciprocal decrease in CHO oxidation (P < 0.0001). While HR (P < 0.0001) and RPE (P = 0.0005) increased across the session for all three walks in the LCHF group, this was more pronounced in the Adaptation test block (Table 4). When combined with the significant increase in the relative oxygen cost of walking in Walk 2 (Fig. 4D; P = 0.005 vs. Walk 1, P = 0.001 vs. Walk 3), this indicates an increase in the perceived effort and metabolic cost of exercise when relying on fat as the primary fuel. Critically, there were no differences in relative V˙O2 (Fig. 4D, P = 0.999) or substrate utilisation (Fig. 4E, P < 0.999, Fig. 4F, P = 0.850) between Walk 1 and Walk 3, indicating these effects were transient and the restoration period was sufficient to restore metabolic flexibility.

Changes in whole‐body metabolism were supported by the changes in circulating substrate availability measured in capillary and venous blood samples (Fig. 5). There were no differences in blood glucose concentrations across testing blocks (Walks 1–3) in the HCHO group (P = 0.355), but there was a main effect for distance as blood glucose increased across the 25 km trial consistent with the continuous intake of CHO (P = 0.0009). In contrast, the LCHF group displayed a main effect for testing block (P < 0.0001), with blood glucose concentrations (Fig. 5A) increasing in Walk 1 (Baseline) and to a lesser extent in Walk 3 (Restoration, P = 0.03 vs. Walk 1), but decreasing in Walk 2 (Adaptation, P < 0.0001 vs Walk 1 and 3). There was a main effect observed for both distance (P = 0.0046) and test block (P < 0.0001) in blood lactate responses in the LCHF group (Fig. 5B), as there was a decrease across the trial in Walks 1 and 3, while lactate increased across Walk 2 (P = 0.0007 vs. Walk 1, P < 0.0001 vs. Walk 3). Similarly, there was a robust increase in circulating blood β‐hydroxybutyrate in the Walk 2 in the LCHF group (P = 0.0004 vs. Walk 1 and Walk 3), peaking at 1.7 (0.9) mmol L⁻¹ towards the end of the trial, but there were no differences in the HCHO group. There was an increase in serum FFA concentrations across the duration of the walk (P < 0.0001) in both groups (Fig. 5D), but only the LCHF group displayed differences between testing blocks, with concentrations being significantly elevated at all time points in Walk 2 after the LCHF diet compared to Walks 1 and 3 (P < 0.0001).

To determine the effects of the 1‐day CHO restoration on substrate utilisation during exercise, athletes in the LCHF group reported to the laboratory on the morning of the 10,000 m race in the Baseline and Adaptation testing blocks and repeated Stages 2 and 4 of the submaximal exercise economy test within their pre‐race warm‐up (Fig. 6). There were no differences in calculated substrate utilisation between economy test and pre‐race measurements in the Baseline testing block. Following the 6‐day adaptation to LCHF and a 1‐day HCHO diet designed to enhance skeletal muscle glycogen availability, fat oxidation rates at Stages 2 and 4 were reduced by 40% (P = 0.011) and 59% (P = 0.0002), respectively, relative to the economy test performed following LCHF adaptation (Fig. 6B and D), but remained elevated compared to the first pre‐race test performance performed at Baseline (Stage 2; P = 0.0008, Stage 4; P = 0.002). While CHO oxidation rates at Stages 2 and 4 were increased by 197% (P = 0.0015) and 51% (P = 0.0007) compared to the LCHF adaptation trial, they remained suppressed relative to the pre‐race test performed at Baseline (Stage 2, P = 0.002; Stage 4, P = 0.003) (Fig. 5A and C).

Real world performance outcomes from these dietary interventions were tested by having athletes compete in WA sanctioned 10,000 m races. Environmental conditions for Baseline and Adaptation races, respectively, were 28.7 (1.2)°C and 22.8 (0.3)°C, 47.1 (3.5)% and 72.1 (1.2)% relative humidity, and wind speeds of 1.5 (0.4) m s⁻¹ (with gusts up to 2.6 m s⁻¹) and 1.2 (0.3) m s⁻¹ (gusts up to 2 m s⁻¹). Athletes raced competitively, with finishing times for the post‐intervention races representing 103 (3)% and 104 (8)% of the life‐time personal best times of walkers in the HCHO and LCHF groups, respectively, and four athletes recording times faster than their personal bests (three from the HCHO group under Baseline or Adaptation races and one from the LCHF group for Adaptation race). Performance results are presented in Fig. 7, with a summary of aerobic testing (Fig. 7A), individual race results (Fig. 7B) and difference in performances between races (Fig. 7C). There was a significant interaction between diet and time for race results (P = 0.02), and a significant difference between performance changes between the two groups (P = 0.009). Despite the lack of a statistically significant (P = 0.222) change in maximum aerobic capacity (relative V˙O2peak, Fig. 7A), all six athletes in the HCHO group improved performance by an average of 5.7 (5.6)%. In contrast, 6 of 7 LCHF athletes displayed impaired performance with a mean race time that was 2.2 (3.4)% slower (Fig. 7B), despite a small but significant increase in relative V˙O2peak (P = 0.025). The difference between changes in performance from Race 1 to Race 2 was significant (P = 0.009) (Fig. 7C).

None of the authors of this paper has a competing interest.

This study was conducted at the Australian Institute of Sport, Canberra, Australia. Conception and design of the experiments was undertaken by L.M.B., J.W., I.A.H., A.K.A.M., M.L.R.R. and A.P.S. Collection, assembly, analysis and interpretation of data was undertaken by L.M.B., J.W., I.A.H., M.L.R.R., N.T., S.G.F., R.H., A.K.A.M., A.M.W. and A.P.S. Drafting the article or revising it critically for important intellectual content was undertaken by L.M.B. and JW. All authors have read and approved the final version of this manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

This study was funded by a Grant from Australian Catholic University Research Fund (ACURF, 2017000034). J.W. is supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Postdoctoral Fellowship.