It has been hypothesized that consuming a ketogenic diet may enhance performance among endurance athletes by promoting a shift in substrate utilization that enhances physiological training benefits [3, 18]. The present review explores this hypothesis by examining associations between EAKD consumption and VO2 max, a biomarker for endurance capacity . Two of the seven studies included in this review found a significant increase in VO2 max post-EAKD consumption [7, 12]. However, both articles reported significant VO2 max increases across all diets, and that outcomes were independent of dietary intervention. Interestingly, both studies were conducted among elite race walkers that self-selected their dietary intervention, and the athletes that self-selected into the EAKD had slightly higher average baseline and post-treatment VO2 max values [7, 12]. Furthermore, Burke et al., reported that VO2 max values for the high carbohydrate comparison group were significantly lower than EAKD or periodised carbohydrate groups at baseline and follow-up (p ≤ 0.02) . This suggests that other factors may influence athletes’ choice of diet and aerobic capacity concomitantly, such as genetic variation in trainability and/or chronic substrate utilization [19, 20]. A review conducted by Williams et al. revealed the potential for 97 genes to predict VO2 max trainability, suggesting that genetics may account for differing training outcomes among athletes . Certain dietary preferences, which both acutely and chronically influence substrate utilization, have also been linked to gene variations, highlighting the possibility for both dietary choices and training outcomes to be mediated by genetics [19, 21]. Randomized controlled trials and genome-wide association studies can be leveraged to control for, and explore the impact of, such factors in future studies of the EAKD.
Four of the seven studies reviewed reported non-significant VO2 max outcomes [14,15,16,17]. In a non-randomized trial, McSwiney et al. reported a VO2 max increase in both groups of male endurance athletes post-EAKD (EAKD: 53.6 ± 6.8 vs. 57.3 ± 6.7; HCD: 52.6 ± 6.4 vs. 57.2 ± 6.1) with a non-significant difference between groups (p = 0.968) . In a pre-posttest design, Phinney et al. reported a non-significant decrease in VO2 max from baseline among five elite male cyclists (pre- vs. post-EAKD: 5.10 ± 0.18 vs. 5.00 ± 0.20; p > 0.01) . In a case study, Zinn et al. reported a non-significant decrease among five recreational endurance athletes consuming the EAKD (− 1.69 ± 3.4; p = 0.63) . Finally, in a randomized crossover study, Shaw et al. reported no significant changes from baseline (59.4 ± 5.2) among male endurance athletes during either dietary intervention (p > 0.05) .
Heatherly et al. did not report VO2 max outcomes, instead providing the percentage of baseline VO2 max achieved at various race paces (i.e., 5 km, 10 km, 21 km, 42 km, sub-42 km) . The significantly greater percentages of baseline VO2 max achieved post-EAKD consumption at 10 km, 21 km, 42 km, and sub-42 km race paces demonstrate that the EAKD was negatively correlated with the athletes’ aerobic efficiency at these paces. This is corroborated by some of the secondary outcomes reported in Table 2, including reports of EAKD being associated with significantly higher RPE , and decreased TTE . Only one study reported significant positive secondary findings: a higher peak power in athletes post-EAKD compared to the standard, high carbohydrate diet . The authors of the study hypothesized that this outcome was likely due to an improved power to weight ratio among the EAKD athletes, who lost an average of 6 kg of body mass.
Despite the popularity of the diet as an ergogenic aid, this review provides evidence that EAKD consumption produces mixed results, in terms of endurance performance, when compared to a high carbohydrate diet. Several biological mechanisms may help to explain the potential for mixed and/or detrimental effects, including changes in fuel economy, production of certain metabolic byproducts, and reduced energy intake. For example, the EAKD significantly increases fat oxidation, requiring greater oxygen consumption due to the increased oxygen demands during fatty acid metabolism versus carbohydrate metabolism [12, 22]. This increased demand for oxygen reduces the beneficial impact of an increased VO2 max because a greater percentage of maximal oxygen uptake is now required to maintain any given race pace . Second, EAKD metabolites such as tryptophan and ammonia may promote fatigue by influencing the central nervous system [23, 24]. Finally, it has been shown that the EAKD leads to increased satiety and reduced energy intake . Reduced energy intake, and the accompanying weight loss, may be beneficial for some individuals but could also present a sustainability issue for highly active athletes. Substantial reductions in body weight may negatively impact mental, hormonal, and bone health, as well as recovery time and general exercise performance [26, 27]. Illustrating these mechanisms, Heatherly et al. reported that athletes exhibited greater oxygen consumption at race pace on the EAKD versus a high carbohydrate diet and that ad libitum EAKD consumption resulted in decreased intake of roughly 1000 kcal per day, leading to a 3 % loss of body mass over the study period .
In multiple studies, participant self-reports (e.g., interview data, training logs) suggested that the EAKD may have promoted perceived fatigue and decreased ability to train for certain athletes , particularly those training in summer months . This could be a combined result of the alterations in fuel economy, metabolism, and energy intake described above, though not all athletes reported experiencing negative side effects. Based on focus group results, one study reported that athletes had more positive than negative perceptions of the diet , suggesting that there may be additional unknown variables influencing EAKD outcomes across individuals and/or settings (e.g., temperature, humidity ).
One hypothesis for the variation in performance outcomes among studies might stem from the heterogeneity across the training/recovery protocols and fitness levels of the athletes . Both studies exhibiting a statistically significant increase in VO2 max examined the effects of EAKD consumption in professional race walkers with high base levels of aerobic capacity, a factor that has been associated with faster recovery times and greater positive adaptations to training [29,30,31]. Both studies also explicitly included a recovery protocol in their training prescription, which could impact the athletes’ training outcomes . Due to limited information on training/recovery protocols in many of these studies, strong conclusions cannot be generated regarding the impact of training versus diet on performance outcomes. However, based on previous evidence, it is reasonable to hypothesize that these protocol differences may have contributed to the diverse outcomes reported [6, 28, 32].
In examining the results, it is important to bear in mind that this review consists of just seven studies, only one of which was randomized . Carr et al., Burke et al., and McSwiney et al. were all prospective trials, however they allowed participants to choose their dietary intervention [7, 12, 14]. Although this self-selection method generally improves rates of adherence to the diets, it also introduces risk of bias in that those athletes who chose the EAKD may have other lifestyle or dietary tendencies that could affect their biological response to the diet. Heatherly et al. and Phinney et al. were pre-posttest studies, which are subject to threats to internal validity, such as the fact that passage of time results in natural decreases in VO2 max [13, 15]. Finally, Zinn et al. was a case study . Although the article provides a wealth of hypothesis generating observations, without a comparison group we cannot conclude whether the EAKD was more or less effective than the standard, high carbohydrate diet for athletes.
All studies had relatively small sample sizes, which reduced the statistical power of the analyses. It is possible that, with a larger sample size, the seven studies might have exhibited corroboratory results. The small sample sizes also exacerbated the problem of drop-out rates, which were considerable in one of the five studies. McSwiney et al. lost 18 participants in the EAKD group and nine in the comparison group, resulting in a participation rate of 33 and 55%, respectively .
At the review level, heterogeneity in dietary interventions, adherence measurements, VO2 max testing procedures, training protocols, and athlete types all introduced variation that made comparisons across studies difficult. For example, four studies measured VO2 max using a treadmill test [7, 12, 13, 16], while the other three studies used a cycle ergometer [14, 15, 17]. Previous reviews suggest that these two testing procedures produce inconsistent results, with higher VO2 max outcomes reported for treadmill as compared to cycle ergometer tests . Therefore, inter-article comparisons of the change in VO2 max by diet from baseline may be more reliable than inter-article comparisons of the absolute outcome values reported. Furthermore, research suggests that VO2 max may be an inaccurate predictor of endurance performance in runners, specifically due to variations in running economy and fatigue [34, 35]. Therefore, VO2 max may not be a strong indicator of endurance capacity in some sports, further complicating this measure as a comparison across heterogeneous groups of athletes.
In addition to VO2 max outcomes, Table 2 provides a matrix of secondary outcomes (i.e., TTE, race time, RPE, peak power), which can be used to complement the VO2 max findings from this review. For example, although all three diet groups in the study by Burke et al. experienced a significant increase in VO2 max from baseline, only the comparison groups (i.e., high carbohydrate, periodised carbohydrate) experienced faster 10 km race walk times. Furthermore, the EAKD group reported significantly higher RPE values compared to baseline during a graded economy test. Future research in this field can benefit from utilizing a variety of performance metrics, such as the ones discussed in this review, to triangulate overall effects of diet on athletic performance, limiting biases introduced from relying on one marker alone. Additionally, as this research area develops, it may be prudent to conduct reviews among athletes of a single type (e.g., runners only, cyclists only) to limit the heterogeneity among studies.
Because only two databases were used to identify articles for review, it is possible that other studies of EAKD and endurance performance do exist in the literature. However, exploratory investigations of other databases retrieved no additional articles that met inclusion criteria. It is noteworthy that six of seven studies included in this review were published within the last 5 years, suggesting that scientific attention to this topic is fairly recent. Due to the contemporary nature of the research question, it is also possible that yet-to-be-published research exists on this topic. Therefore, future reviews may eventually produce more conclusive evidence. Finally, the potential risk of reporting bias is always present. Unfortunately, it is difficult to assess publication bias because we cannot know the extent of the evidence that has gone unpublished. However, due to the controversial nature of this topic among scientists and lay people alike, it seems likely that both significant and null findings would be publishable.