In contrast to short-term exposure to an LCHF diet, which reduces exercise capacity by depleting liver and muscle stores of glycogen without producing a compensatory increase in fat oxidation [27, 28], longer-term adherence to this dietary regimen causes a range of adaptations to enhance the breakdown, transport, and oxidation of fat in skeletal muscle . Several different approaches have been investigated.
Ketogenic High-Fat Diets
According to recent reviews [5, 6], historical observations of considerable exercise stamina in explorers who followed traditional Inuit diets almost devoid of carbohydrate (energy contribution: 85 % fat, 15 % protein) led to a laboratory investigation of this phenomenon in the 1980s [30, 31]. In this study by Dr. Stephen Phinney, carefully conducted in a metabolic ward, five well-trained cyclists were tested following 1 week of a carbohydrate-rich diet (~57 % of energy) and again following 28 days of a severely carbohydrate-restricted (<20 g/day) but isoenergetic diet with energy contributions of 85 % fat and 15 % protein (Table 2). This diet was associated with ketosis, as demonstrated by increased blood concentrations of beta-hydroxybutyrate from <0.05 to >1 mmol/L after a week, and this was maintained thereafter. Exercise was monitored by a time to exhaustion cycling test at ~63 % of maximal aerobic capacity (VO2max) under conditions of low carbohydrate availability (overnight fast and water intake during the ride) , with the mean result being a maintenance of exercise capacity (see Fig. 1). Despite the negligible intake of carbohydrate, resting muscle glycogen stores were not depleted but rather reduced to ~45 % of values seen on the high-carbohydrate phase (76 vs. 140 mmol/kg wet weight muscle). Furthermore, in both trials, at the cessation of exercise, muscle glycogen depletion was seen in type 1 fibers with a fourfold reduction in its contribution to fuel use in the LCHF trial. Blood glucose contribution to fuel use was reduced threefold, with gluconeogenic contributions from glycerol released from triglyceride use as well as lactate, pyruvate, and certain amino acids preventing hypoglycemia during exercise as well as allowing glycogen storage between training sessions. Lipid oxidation was increased to make up the fuel substrate for the exercise task.
The researchers’ insights into the results of their study were that “metabolic adaptation to limit CHO [carbohydrate] oxidation can facilitate moderate submaximal exercise during ketosis to the point that it becomes comparable to that observed after a high CHO diet.” Furthermore, they noted that “because muscle glycogen stores require many days for repletion, whereas even very lean individuals maintain appreciable caloric stores as fat, there is potential benefit in this keto-adapted state for athletes participating in prolonged endurance exercise over two or more days”. However, they also commented on the results of VO2max tests undertaken during each dietary phase with respect to the ketogenic diet: “… the price paid for the conservation of CHO during exercise appears to be a limitation of the intensity of exercise that can be performed … there was a marked attenuation of respiratory quotient [RQ] value at VO2max suggesting a severe restriction on the ability of subjects to do anaerobic work”. Their explanation for this observation was that “the controlling factor does not appear to be the presence or absence of substrate in the fiber. Rather it is more likely a restriction on substrate mobilization or fiber recruitment. The result, in any case, is a throttling of function near VO2max”.
The researchers were clear that their ketogenic diet did not, as is popularly believed, enhance exercise capacity/performance, noting that, at best, endurance at sub-maximal intensities was preserved at the expense of ability to undertake high-intensity exercise. However, examination of the design and outcomes call for further caution. Although excellent dietary control was achieved in this study, few details were provided of the training protocols followed by the cyclists. It is curious in light of the order effect in the study design (all subjects undertook the ketogenic exercise trial 4 weeks after their carbohydrate trial), that no benefit to exercise capacity was derived from an additional training period. Furthermore, it should be recognized that the exercise task was undertaken under conditions that should have favored any advantage to being adapted to low carbohydrate availability (moderate-intensity exercise, overnight fast, no intake of carbohydrate during exercise). However, and most importantly, the focus on the mean outcomes of the trial in a small sample size hides the experiences of the individual cyclists. As shown in Fig. 1, the published interpretations of the results of this study are largely skewed by the experience of a single subject who showed a large enhancement of exercise capacity after the ketogenic diet (and additional training period). Indeed, statistical analysis of the same data using a magnitude-based inferences approach  reveals an unclear outcome, with the chances of a substantially positive, trivial, and substantially negative outcome being 32, 32, and 36 %, respectively (Stellingwerff, personal communication).
Non-Ketogenic High-Fat Diets
A number of studies have been undertaken in trained individuals involving exposure for ≥7 days to a diet high in fat and restricted in carbohydrate content without achieving ketosis [33–37]; much of this work was driven by Dr. Vicki Lambert and Professor Tim Noakes from the University of Cape Town. Two studies in which carbohydrate and fat intake was manipulated in trained populations have not been included in this summary since the dietary changes were not sufficient to meet the criteria of >60 % fat intake or <25 % carbohydrate intake [38, 39]. The summarized literature (Table 3) includes one study that focused on titrating the carbohydrate content of the diet in modestly trained female cyclists  and four studies that specifically set out to adapt their subjects to a high-fat diet [34–37], although in one case, the smaller degree of carbohydrate restriction resulted in a failure to create clear differences in muscle glycogen content between treatments . Again, the diets provided within studies were isoenergetic and aimed at maintaining energy balance.
In the case of studies specifically focused on adapting athletes to a high fat intake, the rationale of increasing dietary fat involved increasing IMTG stores , restricting carbohydrate to reduce muscle glycogen content [34–36] and allowing sufficient exposure for adaptations to occur to retool the muscle to alter fuel utilization patterns during exercise to compensate for altered fuel availability [34–37]. The avoidance of ketosis was chosen to remove its confounding effect on the relationship between respiratory exchange ratio and substrate utilization during exercise, thereby preventing a true measurement of changes in carbohydrate and fat oxidation during exercise . A range of adaptive responses to the LCHF diet was observed or confirmed in the trained individuals.
As summarized in Table 3, the effect of exposure to the LCHF diets on exercise capacity/performance was tested under a range of different exercise scenarios and feeding strategies. This includes a series of exercise protocols undertaken sequentially  or within a single exercise task , as well as dietary strategies that would either further increase fat availability [33, 36, 37], increase carbohydrate availability [35–37], or deliberately decrease carbohydrate availability against current guidelines or common practices . In some cases, different dietary strategies were implemented before and during the exercise protocols for the high carbohydrate and LCHF trials, making it difficult to isolate the effects of the fat adaptation per se [36, 37]. This variability in study design makes it difficult to make a single and all-encompassing assessment of the effect of LCHF on exercise, as is popularly desired. Theoretically, however, it offers the opportunity to identify conditions under which adaptation to a high-fat diet may be of benefit or harm to sports performance. Unfortunately, the small number of studies and the small sample sizes in the available literature do not allow this opportunity to be fully exploited. The learnings from these studies have been incorporated into the summary at the end of this section. In the meantime, attention is drawn to two important observations from this body of literature:
Evidence of reduced utilization of muscle glycogen as an exercise fuel following adaptation to LCHF cannot be considered true glycogen ‘sparing’ since the observations are confounded by lower resting glycogen concentrations, which are known to reduce glycogen use per se . Only scenarios in which muscle glycogen concentrations are matched prior to exercise can allow the specific effect of fat adaptation on muscle glycogen utilization as an exercise fuel to be measured.
The period required for adaptation to the non-ketogenic LCHF is shorter than previously considered. According to the time course study of Goedecke et al. , whereby muscle fuel utilization was tracked after 5, 10, and 15 days of exposure to the LCHF diet, a substantial shift to increase fat oxidation and reduce carbohydrate utilization was achieved by 5 days without further enhancement thereafter. Of course, it should be noted that a shift in respiratory exchange ratio during exercise, marking shifts in substrate utilization can reflect the prevailing availability of substrate rather than a true adaptation in the muscle.
However, other studies have confirmed the presence of a robust change in the muscle’s substrate use via observations of alterations in the concentrations or activity of proteins or metabolites that regulate fatty acid availability, as well as the persistence of increased fat oxidation in the face of abundant carbohydrate supplies. Such evidence is discussed later.
Importantly, the observation from this series of studies—that retooling of already trained muscle to optimize muscle utilization of fat as an exercise fuel can be achieved in a conveniently short period—led in part to the next phase of investigation, in which attempts were made to enhance sports performance by separately optimizing the muscle’s capacity for lipid and carbohydrate utilization.
Fat Adaptation and Carbohydrate Restoration
In the absence of finding clear benefits from adapting to a high-fat diet on exercise performance, attention was drawn to a tactic of dietary periodization in which a short-term adaptation to an LCHF diet might be followed by glycogen restoration (‘carbohydrate loading’) with 1–3 days of a carbohydrate-rich diet with [1, 36, 41–44] or without  additional carbohydrate intake pre- and during subsequent exercise. Such strategies were aimed at promoting simultaneous increases in fat and carbohydrate availability and utilization during exercise. Indeed, studies that directly compared fuel utilization during submaximal exercise under controlled conditions after the fat adaptation protocol and then again after carbohydrate restoration practices [41, 42, 45] showed that the muscle re-tooling was robust enough to maintain an increase in fat utilization during exercise in the face of the practices that supported plentiful carbohydrate availability (Fig. 2).
As discussed in the previous section, a range of permutation and combinations of dietary strategies and exercise protocols can be investigated in combination with the fat adaptation and carbohydrate restoration strategies to test the effect of such dietary periodization on exercise capacity/performance. The available literature is summarized in Table 4 and includes multiple studies from the author’s own laboratory as well as from the University of Cape Town. However, within this group of investigations, only one fully published study  attempted to investigate an exercise test that bears any real resemblance to a sporting competition; its characteristics include a sole focus on performance rather than a hybrid of metabolism and performance, self-pacing, and a protocol interspersing passages of high-intensity exercise against a background of moderate-intensity work to reflect the stochastic profile of many real-life events. This study , which prompted the 2006 editorial about which this review revolves, merits special reflection before a general summary of the literature is provided.
Havemann et al.  had well-trained cyclists undertake either a 6-day LCHF diet followed by a 1-day high-carbohydrate diet or 7 days of high-carbohydrate diet before undertaking a laboratory-based cycling protocol designed to test some of the features of endurance sporting events. Specifically, cyclists were required to undertake a series of sprints throughout the self-paced 100-km trial: 4-km sprints undertaken at ~78–84 % peak power output and 1-km sprints undertaken at >90 % peak power output (see Fig. 3). Overall, differences in the performance times for the 100-km time trial (TT) were not statistically significant, although the mean performance on the high-carbohydrate trial was 3 min 44 s or ~2.5 % faster (153 min, 10 s for high-carbohydrate trial and 156 min, 53 s for LCHF adapted, p = 0.23). While there was no difference between trials with regard to the 4-km sprint times, performance of the 1-km sprints was significantly impaired in the LCHF-adapted trial in all subjects, including the three subjects whose overall 100-km TT performance was faster than in their high-carbohydrate trial. The authors stated that although adaptation to the LCHF diet followed by carbohydrate restoration increased fat oxidation during exercise, “it reduced high-intensity sprint power performance, which was associated with increased muscle recruitment, effort perception and heart rate”.
Although the mechanisms associated with the compromised performance in this study were unclear, speculations by the authors included “increased sympathetic activation, or altered contractile function and/or the inability to oxidize the available carbohydrate during the high intensity sprints”. Indeed, evidence for this latter suggestion was provided by data from this author’s own laboratory collected contemporaneously. In an investigation of possible mechanisms to explain the performance outcomes associated with the LCHF-adaptation and carbohydrate-restoration model, we examined muscle metabolism at rest, during sub-maximal exercise, and after an all-out 1-min sprint following the usual dietary treatment (Fig. 4) . In comparison with the control trial (high-carbohydrate diet), we found that adaptation to the LCHF diet and subsequent restoration of muscle glycogen was associated with a reduction in glycogenolysis during exercise, and a reduction in the active form of pyruvate dehydrogenase (PDHa) at rest, during submaximal cycling, and during sprint cycling. Explanations for the down-regulated activity of this enzyme complex responsible for linking the glycolytic pathway with the citric acid cycle included the observed post-sprint decrease in concentrations of free adenosine monophosphate (AMP) and adenosine diphosphate (ADP) and potentially an up-regulation of PDH kinase (PDK) activity, which has previously been observed in association with a high-fat diet . This study provided evidence of glycogen ‘impairing’ rather than ‘sparing’ in response to adaptation to an LCHF diet and a robust explanation for the impairment of key aspects of exercise performance as a result of this dietary treatment.
Summary of Learnings from the Literature: 1999–2006
Key interpretations by this author from the literature on adaptation to an LCHF conducted up until 2006 are summarized below:
Exposure to an LCHF diet in the absence of ketosis causes key adaptations in the muscle in as little as 5 days to retool its ability to oxidize fat as an exercise substrate. Adaptations include, but are not limited to, an increase in IMTG stores, increased activity of the hormone-sensitive lipase (HSL) enzyme, which mobilizes triglycerides in muscle and adipose tissue, increases in key fat-transport proteins such as fatty acid translocase [FAT-CD36] and carnitine-palmitoyl transferase (CPT) (for extended review, see Yeo et al. ). Together, these adaptations further increase the already enhanced capacity of the aerobically trained muscle to utilize endogenous and exogenous fat stores to support the fuel cost of exercise of moderate intensity. Rates of fat oxidation during exercise may be doubled by fat-adaptation strategies.
These muscle-retooling activities stimulated by fat adaptation are sufficiently robust that they persist in the face of at least 36 h of aggressive dietary strategies to increase carbohydrate availability during exercise (e.g., glycogen supercompensation, pre-exercise carbohydrate intake, high rates of carbohydrate intake during exercise). Although the increased carbohydrate availability reduces rates of fat oxidation compared with fat adaptation alone, fat utilization remains similarly elevated above comparative rates in the absence of fat adaptation. The time course of the ‘washout’ of retooling is unknown.
In addition to up-regulating fat oxidation at rest and during exercise, exposure to an LCHF diet down-regulates carbohydrate oxidation during exercise. Direct [34, 42, 45] and indirect  techniques of measuring the source of changes in substrate utilization show that changes in utilization of muscle glycogen, rather than blood glucose or exogenous glucose, account for the change in carbohydrate use. The reduction in glycogen use persists in the face of glycogen supercompensation  and high-intensity exercise , noting that it is robust and independent of substrate availability. A down-regulation of PDH activity explains at least part of the impairment of glycogen utilization as an exercise fuel , representing a decrease in metabolic flexibility.
Despite the enhanced capacity for utilization of a relatively limitless fuel source as an exercise substrate, fat-adaptation strategies with or without restoration of carbohydrate availability do not appear to enhance exercise capacity or performance per se. Several inter-related explanations are possible for the failure to observe benefits:
Type II statistical error: failure to detect small but important changes in performance due to small sample sizes , individual responses [42, 45], and poor reliability of the performance protocol. While this explanation often looks attractive , in some cases, further exploration and enhanced sample size increases confidence in the true absence of a performance enhancement .
Benefits are limited to specific scenarios: characteristics of conditions under which fat-adaptation strategies appear to be more likely to be beneficial include protocols of prolonged sub-maximal exercise in which pre-exercise glycogen is depleted and/or no carbohydrate is consumed during exercise (e.g., low-carbohydrate availability).
Benefits are limited to specific individuals: characteristics of individuals who may respond to fat-adaptation strategies include carbohydrate-sensitive individuals who are subjected to scenarios in which carbohydrate cannot be consumed during exercise.
The experience of athletes, at least in the short-term exposure to LCHF diets, is of a reduction in training capacity and increase in perceived effort, heart rate, and other monitoring characteristics, particularly in relation to high-intensity/quality training, which plays a core role in a periodized training program .
Fat-adaptation strategies may actually impair exercise performance, particularly involving shorter high-intensity events or high-intensity phases during a longer event, which require power outputs or intensities of 85–90 % maximum level or above. This is likely to be due to the impairment of the muscle glycogen utilization needed to support high work rates, even in scenarios where strategies to achieve high carbohydrate availability are employed.
On the basis that conventional competitive sports generally provide opportunities to achieve adequate carbohydrate availability, that fat-adaptation strategies reduce rather than enhance metabolic flexibility by reducing carbohydrate availability and the capacity to use it effectively as an exercise substrate, and that athletes would be unwise to sacrifice their ability to undertake high-quality training or high-intensity efforts during competition that could determine the outcome of even an ultra-endurance sport, this author decided to abandon a research and practical interest in fat-adaptation strategies. A meta-analysis published about the same time on the effect of the carbohydrate and fat content of athletic diets on endurance performance  summarized that the heterogeneity around their findings that high-carbohydrate diets (defined as >50 % of energy from carbohydrate) have a moderate (effect size 0.6) benefit on exercise capacity compared with high-fat diets (defined as >30 % of energy from fat) showed that “a conclusive endorsement of a high-carbohydrate diet is hard to make”. However, this heterogeneity speaks to the limitations of undertaking a meta-analysis with such a broad and undefined theme as well as the problem of the ‘black and white’ thinking that is discussed in the conclusion to this review.