Introduction

Exercise intensity determines the rate of ATP demand, biochemically governing the preferred fuel and metabolic energy system to best meet the specific energy requirements [24]. Thus, tailored nutritional strategies for different sports could be developed using exercise intensity-specific fuel use patterns. For instance, as fat is a major energy substrate for prolonged low-intensity exercise, the high fat, very low carbohydrate (CHO), and moderate protein ketogenic diet (KD) have drawn interest in endurance sport research over the past decade based on reports of significant increases in fat oxidation rates during exercise on these diets [35, 40].

Beyond endurance sport, team sports encompass another group of popular sports,, characterized by movement patterns of short-duration (< 10 s) supramaximal intensity sprints, interspersed with 60–300 s recovery, known as intermittent sprints [13, 30]. Yet, while the effects of KD on CHO- and fat metabolism during endurance exercise are well established by now, literature on a KD's impact on intermittent sprint metabolism is currently lacking. This hesitation to examine KD effects on high-intensity exercise metabolism may have its roots in the belief that CHO serves as the primary fuel for all forms of high-intensity exercise. The biochemistry of intermittent sprint metabolism, however, reveals that the phosphocreatine-ATP buffer (ATP-PCr) is the primary anaerobic energy source during these activities [16, 24, 32, 42], while reliance on anaerobic glycolysis is reduced after initial sprints [4, 16, 24, 32]. As a result, there is less dependence on CHO as a substrate during intermittent sprint exercise. So, in addition to the lack of KD research on intermittent sprint exercise, this unique fuel metabolism pattern where glycolysis is already bypassed, further serves as an encouraging biochemical rationale to explore the effects of CHO restriction during intermittent sprints.

The term keto-adaptation is used in the literature to describe the various physiological adaptations observed as a result of a KD [27]. Previous KD studies reported various adaptations in CHO and fat metabolism. However, while creatine is synthesized endogenously in healthy humans [20], endogenous ATP-PCr adaptations in muscle tissue to a nutritional intervention have, to our knowledge, not been investigated. Therefore, based on (a) the endogenous functioning of the ATP-PCr energy system and (b) the dominant contribution of this energy system during intermittent sprints, the investigation of potential keto-adaptation in this energy system is warranted.

The purpose of this study was therefore to test the hypothesis that keto-adaptation over a 6-week period may enhance energy derived through the ATP-PCr energy system during intermittent sprints.

Methods

Study Design

The study protocol was approved by the Institutional Health Research Ethics Committee (HREC) (S19/05/100). All testing and laboratory procedures were performed in accordance with the Declaration of Helsinki.

A within-subject repeated measures study design was adopted, similarly to previous studies investigating KD effects in sport [18, 39, 41]. Chrzanowski-Smith et al. [8] argued that study designs where each participant acts as his/her own control are recommended in exercise physiology research to address large inter-personal variation in participants’ physical and physiological characteristics. As adaptations in the ATP-PCr energy system were previously reported to be highly irreversible, the duration of an effective wash-out period cannot be defined [34] excluding the use of a crossover design in this study. Figure 1 depicts the timeline of the tests and measurements.

Fig. 1
figure 1

Study timeline and intermittent sprint test procedure. HD phase: habitual diet phase; KD phase: ketogenic diet phase

Full size image

Participants attended 10 laboratory sessions over 10 weeks. During the first session, they were informed of the purpose and possible risks of the study, and they completed a health questionnaire prior to signing the informed consent form. Hereafter, they were familiarized with the cycle sprint test. During weeks two and three, while maintaining their habitual diet (HD), participants completed baseline intermittent sprint tests (HD1 and HD2). The mean values of these results were considered the baseline values. After the second HD test, participants started the 6-week KD intervention and completed intermittent sprint tests during KD weeks two, four, and six.

During the alternate weeks when no exercise testing was performed, participants still had a check-in session to closely monitor their health and wellbeing, hydration, and blood ketone levels and attend to any challenges with the nutritional intervention. Participants were regularly reminded that they were free to withdraw from the study at any point if dietary compliance could not be maintained.

Participants

To promote dietary adherence, a self-selected sampling method was employed. Volunteers were required to habitually engage in physical exercise ≥ 3 times per week, should not have had any weight loss or weight gain goals, and had to habitually consume a non-restricted diet for at least 8 months before the study. Volunteers were excluded when they reported clinical conditions known to affect metabolism, used ergogenic aids or pharmaceuticals that may affect energy metabolism or performance assessments, and Illness or injury that would impair their exercise performance. Participant characteristics are depicted in Table 1.

Table 1 Participant characteristics
Full size table

Pre-exercise Testing Procedures

All exercise tests were conducted between 06:00 and 11:00. The ambient temperature in the laboratory was kept between 19 and 22 °C. Before each day of testing the metabolic analyser was calibrated according to the user manual (16% O2, 4% CO2, and the balance N2).

Upon arrival, participants completed a five-point wellness questionnaire [19] to closely monitor their wellbeing and general health and address any issues. They also provided a urine sample to obtain refractometer hydration status (Pocket Osmocheck, Vitech Scientific, UK) to ensure that no exercise testing was conducted in a dehydrated state and remind participants to stay adequately hydrated throughout the KD intervention. During the KD weeks, blood ketone levels were measured to monitor dietary compliance. Body mass for energy system calculations was recorded to the nearest 0.1 kg on a calibrated electronic scale (UWE BW-150, Brisbane, Australia). During the first session, a bike fit was performed on each participant, and measurements were recorded to ensure a consistent bike setup for all subsequent cycle tests.

To standardize the participant’s metabolic state during the test sessions, caffeine-containing drinks and alcohol were prohibited for at least 12 h before testing, they had to avoid unaccustomed- or vigorous exercise (exertion levels ≥ 12 on the Borg scale) 24 h before testing and they had to be sufficiently hydrated (refractometer measurement of < 800 mOsmols). Participants also consumed a standardized breakfast on the testing days. During the HD phase, breakfast consisted of a weighed portion of Cornflakes (2 g CHO per kg body mass), low-fat milk and fruit juice or a banana and during the KD phase, two to three eggs, fried or scrambled in coconut oil or butter, and green leafy vegetables. Breakfast was consumed two hours prior to reporting to the laboratory for testing.

Intermittent Sprint Testing

Figure 1 includes an illustration of the intermittent sprint test protocol with time points for metabolic measurements. The metabolic data were used in the calculation of the energy system contributions (described later).

Following fitment of the metabolic system’s facemask and heart rate monitor, the participant sat quietly on a chair while baseline oxygen consumption (VO2) was measured for three minutes. The participant then mounted the Velotron Dynafit Pro cycle ergometer (RacerMate, Seattle, USA), set according to their bike fit measurements, and performed a 5 min warm-up against a resistance of 80 W (women) or 100 W (men), followed by a 3 min passive rest period. During this period, the baseline blood lactate measurement was taken. A fingertip was cleaned with an alcohol swab and pricked with an Accu-Chek Soft Click (Roche diagnostics, Mannheim, Germany). The first droplet of blood was wiped away and the second was drawn into the capillary tube of the Lactate Pro 2 m (ARKRAY, Inc. Kyoto, Japan).

Then participants performed an intermittent Wingate test protocol consisting of 6 × 10 s all-out efforts, with 2 min passive recovery periods between sprints. Each sprint started with a count-down, “three, two, one, GO”. In the first 3 s the participant cycled against the inertia of the flywheel, after which a resistance of 0.075 kp per kg body mass was applied for the next 10 s [37]. Participants were verbally encouraged throughout the test and all efforts were made to replicate the amount of encouragement among all participants. On completion of the sprint protocol, the participant, still connected to the metabolic system, disembarked the bike and while sitting on a chair, gas exchange data collection continued for a 10 min recovery period. At the same time, blood lactate readings were taken every 2 min until a peak was identified.

Dietary Intervention

One week before the KD phase, participants had an individual information session with a sports nutritionist where they received an information booklet and instructions on how to implement a self-selected KD in order to meet the KD macronutrient targets of > 75% fat, 10%–15% protein, and < 50 g/day CHO (or < 10% of total macronutrient intake). The booklet contained sample meal plans for seven days, food lists with macronutrient breakdowns, KD recipes, as well as reputable internet resources where they could find more recipes and information on KD. Participants also received an electronic kitchen scale to weigh food portions.

Participants reported their daily food and drink intake on an online form on their smartphones for two separate two-week cycles. The first two-week meal logging cycle was the last week on the habitual diet and the first week of the KD phase, and the second two-week meal logging cycle was the last two weeks of the KD phase. Food and drinks were logged in as much detail as possible, including the quantity as measured with the kitchen scale provided. They also had to upload a picture of any new food item that they logged for the first time. Meal logging was analysed daily to record energy intake and macronutrient breakdowns and contact participants if adjustments were needed to their macronutrient intake during the KD intervention. Daily food and fluid consumption were analysed using the South African Medical Research Council (MRC) FoodFinder® software. If a food item was not detected on this database, the sport nutritionist uploaded it based on the submitted food label pictures. Blood ketone levels were measured once a week during the KD dietary phase using the Keto-Mojo device (952 School Street 212, Napa, CA 94559).

Energy System Calculations

The contributions of the aerobic, glycolytic and ATP-PCr energy systems to total chemical energy output during the intermittent sprint tests were calculated using various areas under the VO2 curve, blood lactate measurements, and the fast component of the excess post-exercise oxygen consumption according to existing equations and methods [1, 11, 21]. These indirect calculations were initially developed by Margaria et al. [25] from direct muscle biopsy measures and mathematically integrated with measures of VO2 and blood lactate. These mathematical equations were later revised based on experimental data obtained from more recent technologies [12].

The aerobic energy system contribution was derived from area under the curve of VO2 above rest (area under the exercise VO2 curve—3 min seated resting VO2) × duration of test (from the start of sprint 1 to the end of sprint 6) [12] and the energy- equivalent (20.92 kJ for each 1 L of utilized O2) in the equation: \({W}_{aer}\left(kJ\right)={VO}_{2}\;\left(mL\right)\times energy\;equivalent \;(kJ/{mL})\).

The anaerobic glycolytic energy system contribution was determined from the change (Δ) in blood [lactate] (peak post-test lactate—pre-test lactate), O2-lactate equivalent [3 (mL/kg)/(mmol/L)] [12], energy- equivalent and body mass with the equation [1, 11, 21]: \({W}_{lactate}\left(kJ\right)=\Delta \left[lactate\right]\left({mmol}/L\right)\times {O}_{2}-lactate\;equivalent \left(mL/{kg}/{mmol}/{L}\right)\times energy \, equivelant \left({kJ}/mL\right)\times body\;mass\;(kg)\).

The ATP-PCr energy system contribution was obtained from the fast component of excess post-exercise VO2 [12] after each sprint and the energy equivalent from the equation [1, 11, 21]: \({W}_{PCr}={VO}_{2fast}\;\left(mL\right)\times energy\;equivalent\;(k{J}/mL)\).

Data and Statistical Analysis

Data and statistical analysis were conducted using Excel (Microsoft Office 2016) and IBM SPSS (version 26). Descriptive statistics are presented as means ± SD and Cohen’s effect size (ES) ± 95% confidence interval (CI). Kolmogorov–Smirnov tests revealed that all outcome measures, except blood ketones, followed a Gaussian distribution. The Greenhouse–Geisser correction was applied in the analysis of the ketone data. In all cases results were considered statistically significant if P < 0.05.

Macronutrient composition for the HD and KD phases were reported as mean values for the meal logging weeks (HD = one week; KD = three weeks). These values were reported as calories and percentages of daily energy intake.

Mathematical modeling for detection of the VO2 fast component and VO2 under the curves were calculated using MATLAB®. Various exponential curve-fitting combinations were applied and corresponding goodness of fit (R2) statistics for each method were calculated. Upon inspection, the most accurate fit was the bi-exponential model, with a Trust-Region algorithm and set upper and lower coefficient constraints, in the Robust Bi-square setting and smoothed with a 10-point moving average.

Repeated measures ANOVA was used to determine whether statistically significant changes in fuel metabolism occurred during the KD intervention compared to the HD phase. In case of a statistically significant result, a Bonferroni post hoc correction was applied for pairwise comparisons.

Cohen’s effect sizes (ES) and 95% CI were calculated to compare the magnitude of change in outcome measures.

Results

Table 1 depicts the participant characteristics. Twenty-three recreationally active men and women (aged 18 to 39 years) volunteered to participate in the study. Fifteen participants (7 men, 8 women) completed the study and were included in the final data analysis. Reasons for dropout included: physical injury unrelated to the study (n = 1), withdrawal before the KD phase in fear of non-compliance (n = 2), KD non-compliance (n = 3), gastro-intestinal distress (n = 1) and illness unrelated to the study (n = 1).

Nutritional Intake and Body Composition Characteristics

The participants’ average CHO intake during the KD phase was less than the required 50 g per day (26.4 ± 9.3 g/day) and there was a statistically significant decrease in CHO intake compared to the HD phase (P < 0.001; ES = − 2.94) (Table 2). Mean blood [ketone] was significantly higher during the KD phase compared to baseline levels (0.8 ± 0.5 mmol/L vs. 0.3 ± 0.1 mmol/L; P = 0.002; ES = 1.29; CI = 0.47; 2.04). On average, the participants experienced a significant decrease in body fat percentage (P = 0.001; ES = − 0.34; CI = − 1.05; 0.39), body mass (P = 0.000; ES = − 0.25; CI = − 0.97; 0.47) and fat free mass (P = 0.011; ES = − 0.09; CI = − 0.80; 0.63).

Table 2 Dietary energy and -macronutrient intake and body composition changes
Full size table

Energy System Contributions During Intermittent Sprints

Figure 2 illustrates the absolute changes in energy system contributions at 2-week intervals throughout the study. Statistically significant changes were observed in the contributions of the ATP-PCr (P = 0.021) and anaerobic glycolytic (P = 0.008) energy systems. The contribution of the aerobic energy system remained unchanged over the study period (P = 0.85). There was a significant increase in the contribution of the ATP-PCr system to total energy expenditure following the 6-week KD (22.0 ± 43.2 J; P = 0.019; ES = 0.47; CI = − 0.27; 1.18). A statistically significant decrease in the contribution of the anaerobic glycolytic system was observed after two weeks on the KD (69.2 ± 27.4 J vs. 54.8 ± 24.8 J; P = 0.031; ES = − 0.51; CI = − 1.22, 0.23) and it remained low throughout the KD phase.

Fig. 2
figure 2

Absolute energy system contribution (mean ± SD) of the ATP-PCr. anaerobic glycolysis and aerobic systems (J). HD the mean of the habitual diet control phase, KD week 2–6 test sessions in week 2, 4 and 6 on the ketogenic diet. *Statistically significant difference from habitual die phase (P < 0.05)

Full size image

Figure 3 depicts peak blood lactate after the intermittent sprint tests over the study period. Corresponding to the anaerobic glycolytic response, peak post-intermittent sprint lactate levels reduced significantly at KD week 2 (− 2.92 ± 0.851 mmol; P = 0.004; ES = − 0.73; CI = − 1.45, 0.03) and KD week 4 compared to the HD phase (− 2.42 ± 1.074; P = 0.041; ES = − 0.58; CI = − 1.30, 0.16). A non-significant reduction in peak lactate concentration was observed from HD to KD week 6 (− 2.27 ± 1.092 mmol; P = 0.057; ES = − 0.55; CI = − 1.26, 0.19).

Fig. 3
figure 3

Peak blood lactate (mmol) with the intermittent sprints (mean ± SD). HD mean of the habitual diet control phase; KD week 2–6 test sessions in week 2, 4 and 6 on the ketogenic diet. *Statistically significant difference from habitual diet phase (P < 0.05)

Full size image

Cycle Performance

Cycle performance was maintained across sessions. No statistically significant differences in mean power output were detected at any timepoint (P = 0.356).

Discussion

To our knowledge, this was the first study to investigate keto-adaptations in ATP-PCr energy metabolism, by examining the effects of a 6-week KD on the contributions of each of the three main energy systems during intermittent sprint exercise. Regarding these metabolic adaptations, a steady increase in absolute ATP-PCr-derived energy was observed over the intervention period, suggestive of a novel site of keto-adaptation. Concurrently, we observed a decreased anaerobic glycolytic contribution on KD, with an associated decrease in peak blood [lactate] during the intermittent sprints. No significant change in the aerobic energy system was observed in response to the KD in this study. From an exercise performance perspective, the intermittent sprint mean power output was not compromised at any timepoint of the KD intervention compared to the HD phase.

While exogenous oral creatine supplementation is widely reported to improve high-intensity and intermittent exercise performance [9, 10, 15], creatine is also synthesized endogenously in the liver and kidneys of healthy humans [20]. While it is generally believed that the ATP-PCr energy pathway can only be improved by oral creatine supplementation, three previous studies, using exercise training as the stimulus, reported endogenous upregulation (i.e., in the absence of oral supplementation) of the ATP-PCr energy system [14, 22, 33]. Our finding of enhanced endogenous ATP-PCr functioning, therefore, suggests that similar to sprint interval training [14, 22, 33], KD may be a novel nutritional approach to endogenously enhance this important energy system.

The term keto-adaptation implies gradual changes in phenotypic expression in response to extended periods of gene transcriptional activation on prolonged KD. In line with this keto-adaptation phenomenon, the increase in ATP-PCr derived energy in this study occurred steadily, i.e., every two weeks across the 6-week KD. However, we observed a rapid decline in anaerobic CHO metabolism after only two weeks on KD. Similarly, Burke et al. [6] documented significant changes in CHO- and fat oxidation after only five days of KD and these changes were rapidly reversed within one day of CHO restoration. Therefore, acute changes in CHO- and fat metabolism upon initiation of KD may well be largely attributed to changes in fuel availability, rather than keto-adaptation.

Two potential molecular keto-adaptation mechanisms are suggested to explain the gradual endogenous upregulation of ATP-PCr observed in this study. Firstly, since creatine is largely synthesized endogenously [20], it is hypothesized that the restriction of CHO as anaerobic fuel for glycolysis may have upregulated biosynthesis of creatine to enlarge muscle phosphocreatine stores and compensate for the smaller muscle glycogen stores. Secondly, the mitochondrial creatine kinase enzyme (CKmito), responsible for re-phosphorylation of creatine after an exercise bout [24], may be upregulated. If true, team sport athletes may ‘recharge’ creatine to phosphocreatine in shorter recovery periods to provide fast rate energy in the subsequent sprint bout. In support of the CKmito keto-adaptation proposition, enhanced mitochondrial biogenesis and adaptations in other mitochondrial enzymes were previously reported following ketogenic interventions [5, 29].

In a sport setting, McSwiney et al. [27] previously compared changes in exercise performance in endurance-trained athletes after a 12-week KD-training intervention to a training-only control group. They reported significantly greater improvements in 6 s cycle sprint peak power relative to body weight in the KD group compared to the high-CHO diet group. The authors speculated that this finding implied the ATP-PCr energy system was not negatively affected by their 12-week KD-training intervention. Our finding of increased energy yield from the ATP-PCr energy system therefore affirms this speculation. Furthermore, our results may now also provide a metabolic explanation for previous reports of preserved exercise performances on KD during very short, maximal intensity strength and power activities [17, 23, 27, 31] where ATP-PCr is the main energy substrate.

It is known that ATP-PCr produces ATP at a faster rate than glycolysis and without the production of lactic acid [24] and that lactic acid-induced metabolic acidosis is associated with intermittent sprint fatigue [2, 3, 28]. For instance, a significant correlation was reported between plasma [H+] and decreased performance (greater power output decrements) in a repeated sprint exercise test (r = 0.66; P < 0.05) [3]. In another study, using the same repeated sprint test, NaHCO3 ingestion-induced metabolic alkalosis resulted in improved performance as seen in a significant increase in total work (P < 0.05) [2]. Milioni et al. [28] reported significant correlations between anaerobic glycolytic energy contribution and lowered repeated sprint performance (P < 0.05). Thus, our finding of a downregulation in anaerobic glycolysis with lowered peak blood [lactate] during intermittent sprints on the 6-week KD intervention may hold metabolic benefits for intermittent sprint sport athletes.

Regarding the aerobic energy system contribution, previous studies reported that KD interventions regularly result in increases in aerobic metabolism during endurance exercise, often indicated as lower respiratory exchange ratios (RER) [35, 40]. However, we did not observe significant changes in the aerobic energy system contribution during intermittent sprint exercise. We suspect that this discrepancy could be a consequence of the specific equation used in the current study. While RER is calculated from absolute VO2 values, we calculated the area under the curve for VO2 above rest, as previously described in the literature [1, 11, 21]. We presume that inclusion of resting VO2, may have resulted in similar increases in the aerobic energy system contribution as observed from lowered RER values on KD [35, 40].

Martin et al. [26] pointed out that dietary calorie restriction and weight-loss reduce an individual’s resting metabolic rate. As a lowering in metabolic rate could potentially negatively affect exercise metabolism negatively (i.e., our outcome measures of interest), our participants were asked not to purposefully limit their energy consumption. However, there was still a decrease in ad libitum energy intake by our participants on the KD (see Table 2). This was likely caused by the metabolic state of (a) elevated blood ketone levels [38] and (b) lowered insulin levels [36], both known to suppress appetite with carbohydrate restriction. Still, despite lower caloric intake and subsequent alterations in body composition, the ATP-PCr energy system contribution still increased significantly in this study.

A limitation of this study was the participants’ low response rate on their training logs (data therefore not reported). Nevertheless, given previous reports of higher perceptions of effort and fatigue with higher intensity exercise during the initial stages of KD interventions [6, 7], it is unlikely that our participants increased their voluntary training to stimulate additional adaptations in energy metabolism during the KD intervention. Another possible point of critique is that the exposure of the participants to repeated exercise tests may have elicited a training effect. However, Parra et al. [33] compared muscular adaptations between groups who either performed 14 consecutive days of sprint interval training, or 14 training sessions spread over six weeks. They reported that resting muscle stores of PCr and CK enzyme activity increased in the daily training group, while no significant changes were observed in the long programme training group. Thus, it can be inferred that ATP-PCr adaptations in response to sprint interval training requires frequent training stimuli, i.e., short recovery periods between sessions. It is, therefore, unlikely that the repeated measures in the current study (five sessions spread over ten weeks) contributed meaningfully to the changes that were observed in the energy system contributions.

In conclusion, while CHO is indeed a major substrate in sports consisting of higher-intensity continuous exercise, phosphocreatine mainly provides anaerobic energy during explosive, very short, maximal intensity sports, as well as intermittent sprints, typical of many team sports. Our metabolic findings of a simultaneous up-regulation of ATP-PCr energy metabolism and lowered peak blood [lactate] through anaerobic glycolysis, resulted in maintained intermittent sprint performance. These findings may have special use in team sports, as multiple, intermittent sprint clusters are performed during the course of a match. Specifically, lowered lactic acid accumulation is associated with lowered fatigue and better repeated sprint performance [28]. Also, in non-keto-adapted players limited muscle glycogen stores often impairs performance in later stages of matches. Therefore, reliance on creatine and fat stores as main energy substrates on KD, may aid performance maintenance towards later stages of matches. Herewith, KD research in team sport contexts pose an exciting field for future research.