Introduction

Weight class sports require athletes to maximize performance while carefully controlling body weight to compete in a specific weight class with an optimum power-to-weight ratio. Competing at the upper-end of a weight class is advantageous, and most athletes therefore aim to transiently reduce body weight to make weight for competition. Targeted weight reduction can be accomplished by energy deficit (³⁴), but rapid weight loss strategies are also frequently used (¹⁵). Reducing weight is challenging, and weight-making strategies are not always effective for all athletes. Weight-making strategies can also result in impaired performance (⁴⁰), compromised lean body mass (LBM (¹²)), and deleterious health outcomes (^10,15). Strategies that allow athletes to effectively reduce body weight without compromising health or performance are therefore of importance.

Recent studies in athletic and nonathletic populations have shown weight loss without energy restriction using diets that reduce carbohydrate and increase fat intake (^{1,20,26,31,36,41}). There is also some evidence that weight loss arises from reductions in fat mass with a concomitant preservation of lean mass (^3,26,38,41). Low-carbohydrate, high-fat diets might therefore be a useful weight-making strategy for athletes competing in weight class sports. However, before such dietary strategies could be recommended, they would need to also demonstrate that they do not compromise performance. The impact of low-carbohydrate, high-fat diets on athletic performance has been explored in a number of studies (^20,26,41). Most studies have investigated aerobic performances in endurance-based sports. However, a few recent studies have examined the effects of low-carbohydrate, high-fat diets on strength and power. Sawyer et al. (³¹) showed maintenance in strength and power after short-term (7 days) carbohydrate restriction in resistance training men and women. In resistance training men, increases in strength and power were comparable between an 8-week low-carbohydrate, ketogenic diet (LCKD) group and a group consuming a high-carbohydrate, western diet (⁴¹). Paoli et al. (²⁶) reported a preservation of strength in elite gymnasts after a 4-week ketogenic diet. Collectively, these studies suggest that a low-carbohydrate, high-fat diet or LCKD might be useful for reducing body weight without compromising strength and power. However, with the exception of the study by Paoli et al. (²⁶), previous studies have used recreational athletes. Therefore, the efficacy of LCKD for performance and body composition among athletes competing in weight class sports that require maximal strength and power remains underexplored.

Little is known about the mechanisms underlying weight reduction and body composition changes during LCKD. Altering macronutrient intake could affect body weight through changes in fluid and fuel storage. Transitioning to an LCKD reduces fluid retention (^25,29) and stored glycogen (¹⁶) that would both contribute to reductions in weight. However, the reductions in weight associated with these initial adaptations to LCKD are modest, and additional factors seem likely. An LCKD alter the efficiency of metabolic pathways (¹¹), promote the oxidation of fatty acids (^37,39), and could alter resting and exercising energy expenditure (⁸), but additional studies are needed to better understand the contribution of these mechanisms to weight loss during LCKD.

The role of altered skeletal muscle metabolism during LCKD and the impact on training, adaptation, and performance are also unclear. Currently, sports dietary guidelines for strength and power performance emphasize a high-carbohydrate intake. This is related to concerns of the impact of glycogen depletion on fatigue and adaptation (^4,28), but the role of exogenous carbohydrate in this context has recently been challenged (^9,27). There is, for example, no evidence that carbohydrates are required for signaling of mammalian target of rapamycin complex 1 during muscle protein synthesis (⁶). Dietary carbohydrate also does not augment muscle protein synthesis when dietary protein is adequate (²⁷). Furthermore, an LCKD-induced metabolic shift toward fat oxidation (and glycogen sparing) could have favorable effects on ATP resynthesis during resistance exercise training (³⁹).

There is evidence that an LCKD might offer specific advantages for weight reduction without the negative impact on strength and power previously hypothesized to accompany carbohydrate restriction. To the best of our knowledge, no studies have explored the efficacy of an LCKD as a weight-making strategy for competitive weight class athletes by examining the impact on body weight, body composition, and performance. Therefore, the purpose of this study was to determine whether an LCKD could be used as an efficacious weight reduction strategy for athletes competing in the weight class sports of powerlifting and Olympic weightlifting. A secondary purpose was to examine metabolic changes in fuel utilization and energy expenditure during LCKD. We hypothesized that consuming an ad libitum LCKD would result in practically meaningful reductions in body mass with lifting performances comparable with usual diet (UD).

Methods

Experimental Approach to the Problem

To test the hypothesis, a randomized, crossover design was selected. In addition to reducing confounding variables at baseline, a crossover design provided sufficient power while recruiting competitive athletes for a long-term dietary intervention. Subjects consumed an ad libitum UD (>250 g daily intake of carbohydrates) and an ad libitum LCKD (≤50 g or ≤10% daily intake of carbohydrates) in random order, each for 3 months in a crossover design with a 2-week washout. Lifting performance, body composition, resting metabolic rate (RMR), blood glucose, and blood electrolytes were measured at baseline, 3 months, and 6 months at the university's laboratories. Subjects used online tools (MyFitnessPal, Baltimore, MD, USA and Qualtrics, Provo, Utah, USA) to log self-reported dietary intake, self-monitored blood glucose and blood ketones, and weekly training measures. Self-reported measures were logged 3 times per week on 2 weekdays and a weekend day in weeks 1, 4, 7, 10, and 12 in each study phase.

Subjects

Fourteen intermediate to elite level powerlifters and Olympic weight lifters (mean ± SD; age 35 ± 11 years, range 24-53 years, mass 78 ± 12 kg, body fat 17.5 ± 4.6%, n = 5 female) who competed at a local to national level were recruited to participate in this study (Figure 1: CONSORT flow diagram). Two subjects withdrew from the study; one due to illness and the other due to scheduling conflicts. Subjects had an average lifting experience of 6.2 ± 5.8 years and were free of limiting musculoskeletal injuries at the time of recruitment. Inclusion criteria included age between 18 and 55 years and a minimum 6 months competitive lifting experience. After being informed of the risks and benefits of the study, subjects provided written consent. The study procedures were approved by the Australian Catholic University Institutional Review Board (2016-76H). The trial was registered (ACTRN12618000035224). All subjects completed and submitted signed, informed consent prior to commencement.

Procedures

All subjects attended the university's laboratories to undertake testing at baseline, 3 months, and 6 months. Subjects arrived in the morning in a fasted state and were instructed to not consume caffeine or alcohol, or engage in intense training or exercise in the 24 hours before testing. Body composition, resting energy expenditure, and blood measures were conducted in the fasted state. After completing these tests, subjects were allowed to eat as part of their preparations for performance testing.

Diet

After baseline testing, subjects were randomly allocated to a diet phase. Diets were prescriptive with regards to carbohydrate intake but were ad libitum for total calories. During the LCKD phase, subjects were prescribed target macronutrient levels (70% fat, 20% protein, and ≤50 g or ≤10% carbohydrates) and were provided with nutritional counseling and resources to assist in adhering to the LCKD. Resources included print- and electronic-suggested daily meal plans, meal recipes, and lists of foods “encouraged to eat,” “eat in moderation,” and “foods to avoid.” The food lists encouraged a focus on eating unprocessed food, consisting of cruciferous and green leafy vegetables, raw nuts and seeds, eggs, fish, animal meats, dairy products, and plant oils and fats from avocados, coconuts, and olives (²³). Suggested meal plans and meal recipes were formulated to meet subject's micronutrient requirements (⁴³) with special attention to sodium intake because sodium losses have been shown to occur during LCKD (²⁹). Subjects and researchers used an LCKD community Facebook group during the study period to facilitate communication and sharing of LCKD meal recipes. No meals were provided to subjects. Some subjects in the LCKD phases initially expressed concern about the impact of high-saturated fat diet options on cardiovascular health and body composition. In these instances, subjects were given copies of recent articles demonstrating the lack of evidence for “unhealthy” effects of saturated fats (^19,23). During the UD phase, subjects were instructed to consume their UD. To mitigate the carry-over effect of being in the LCKD phase, subjects crossing over to UD were instructed to ensure they consumed >250 g daily intake of carbohydrates. This amount of carbohydrate was a conservative estimate based on the usual carbohydrate intake of subjects during the UD phase.

Total daily energy intake and macronutrient and micronutrient composition were self-reported through the online smartphone application, MyFitnessPal. Researchers administrated subject's MyFitnessPal user accounts and therefore had the ability to assess and modify the subject's macronutrient and micronutrient intake throughout the intervention. MyFitnessPal has previously been used in clinical trials to track dietary intake (¹⁷). Subjects used digital kitchen scales to measure food portions for total energy intake estimates. Subjects in this study were experienced at monitoring energy intake and macronutrient composition because they were accustomed to doing this as part of their usual practices in their respective weight class sports.

Training

Being competitive athletes, subjects had varying individual training and competition schedules during the study period. Subjects were instructed to maintain their normal training during both dietary phases of the study and continued to compete according to their normal competition schedule. Consequently, subjects were in various phases of their training cycle throughout the study. Because standardizing the competition and training schedule was not possible or desirable, we ensured that athletes' competitions did not coincide with the baseline, 3-month, and 6-month research testing. This was accomplished by altering either the research start date or negotiating changes to individual athletes competition schedules. Therefore, although training cycles varied during the study period, no athletes engaged in any other form of weight-reducing strategy within 2 weeks of research testing. Subjects recorded the quality and quantity of training undertaken in both study phases by reporting training session frequency, duration, and intensity using a 10-point scale.

Lifting Performance

One repetition maximum (1RM) was used as the primary performance variable. After a self-selected warm-up, subjects performed either one or all of their competition lifts: snatch and clean and jerk (Olympic weightlifting), squat, bench press, and deadlift (powerlifting). Subjects self-selected the lifts for performance testing. This approach was used to minimize the learning effect that would have occurred if subjects were required to perform unfamiliar lifts. We believe this is the optimal method for assessing performance among well-trained athletes. Lifting performances were assessed in a weightlifting facility using international competition standard bars and plates (Eleiko, Halmstad, Sweden) under the supervision of a researcher qualified to officiate powerlifting and Olympic weightlifting competitions. Performance testing mimicked a competition environment with calls to time to completion. Highest lifts were used for analysis. Athletes and researchers relied on established personal best lifting performances to gauge the reproducibility of lifting performances during research testing. All subjects knew their personal best lifting performances that could be verified from competition records.

Body Composition

Body mass was measured using electronic scales (SECA 813; Hamburg, Germany, ± 0.1 kg). Whole-body composition and estimates of fat and lean mass were measured using dual-energy X-ray absorptiometry (DXA; Medilink Medix DR, 2D-Fan beam, Montpellier, France). Post-test analysis was performed using manufacturer software (Medilink-Eazix Software). The coefficient of variation (CV%) in our laboratory was obtained after the scanning of 9 healthy university students twice, following repositioning. Whole-body CV (%) was 1.3 for lean mass and 1.5 for fat mass.

Resting Energy Expenditure and Fuel Utilization

Resting metabolic rate was estimated by indirect calorimetry using breath-by-breath gas analysis with a ventilated hood canopy (QUARK CPET; COSMED, Rome, Italy). After 30 minutes of rest in a supine position, expired gases were measured in awake subjects for 20 minutes. Room temperature was thermoneutral, between 22 and 25° C, and lights were dimmed. The first 5 minutes of data were discarded, and a CV of <10% for oxygen () and carbon dioxide (V̇co₂) was set as the criteria for a test to be accepted as valid, as per standard practice for RMR measurement (⁵). and V̇co₂ were used to calculate respiratory quotient (RQ) to determine resting fuel utilization.

Blood Sampling and Analysis

Fasting blood glucose and electrolytes (sodium and potassium) were measured in the laboratory at baseline, 3 months, and 6 months. A small capillary blood sample (95 µl) was obtain using dermal puncture and analyzed using a hand-held blood analyzer (I-STAT Chem 8 + cartridge and I-STAT1-300; Abbott Australasia, Macquarie Park, Australia). Sodium and potassium were monitored because electrolyte losses can accompany fluid losses during the transition to an LCKD, which can result in performance decrements (²⁹). To assist in verifying dietary compliance, subjects in the LCKD phase measured their own blood ketones (β-Hydroxybutyrate) and glucose using a portable analyzer (Freestyle Optimum Neo; Abbott Diabetes Care, Maidenhead, United Kingdom). At baseline, subjects were taught how to obtain a finger prick blood sample and use the analyzer. Subjects were then provided with a personal portable analyzer with which to perform these measures on waking and in a fasted state.

Statistical Analyses

Effect sizes from pilot data collected in our laboratory were used in an a priori power analysis for repeated-measures analysis of variance (ANOVA). The trial was designed to provide 80% power to detect meaningful changes in body weight between dietary phases. All data were checked for normal distribution and outliers at each time point using Shapiro-Wilk tests (p ≥ 0.05) and boxplots, respectively. Descriptive continuous data are presented as mean ± SD with paired-sample t-tests used to explore differences in independent variables between dietary phases. A 1-way repeated-measures ANOVA was used to test the overall hypothesis that the means of primary outcome variables: (a) body mass, (b) fat mass, (c) lean mass, and (d) 1RM strength performance, and secondary outcomes: (a) RMR and (b) RQ, were equal between baseline, UD, and LCKD. A between-subject factor of diet order (baseline, UD, and LCKD; baseline, LCKD, and UD) was included to adjust for the effect of diet sequence on within-subject factors. The assumption of sphericity was assessed by Mauchly's test of sphericity. Whenever the hypothesis was rejected and a significant main effect identified, we performed a post hoc pairwise comparison with Bonferroni adjustment. A p value <0.05 was considered statistically significant, and partial-eta squared (π²) was used to report effect size with 0.01 considered small, 0.06 medium, and 0.14 large effects (⁴). Two subjects were excluded from RMR analysis because their tests exceeded the maximum CV for gas variables. All statistical analyses were performed using IBM SPSS Statistics (V.24.0 for Windows).

Results

There were significant differences in carbohydrate (p = 0.001) and fat (p = 0.001) intake during LCKD compared to UD with no differences in total energy intake or protein intake (Table 1). Fasting blood ketones (β-hydroxybutyrate) were elevated during LCKD (0.4 ± 0.2 mmol·L⁻¹; range 0.2–1.7). There were no differences in training variables during LCKD and UD (Table 2).

Primary and secondary outcome variables are shown in Table 3. Table 4 shows the mean differences for outcomes with main effects. Within-subject factors in this table have been adjusted for the effect of diet sequence.

Diet had a significant main effect on body mass F (2, 20) = 5.449, p = 0.013, partial π² = 0.35 (Table 3). Post hoc analysis revealed that body mass was significantly lower at the end of the 3-month LCKD phase compared with the end of the 3-month UD phase with a mean difference adjusted for the effect of diet sequence of −3.26 ± 1.07 kg, (95% confidence interval [CI], −6.34 to −0.18; p = 0.038). Similarly, diet had a significant main effect on lean mass F (2, 20) = 8.217, p = 0.002, partial π² = 0.45 over time. Lean mass was significantly lower at the end of the 3-month LCKD phase compared with the end of the 3-month UD phase with a mean difference adjusted for the effect of diet sequence of −2.26 ± 0.64 kg, (95% CI, −4.10 to −0.42; p = 0.016). There were no other main effects in primary outcome variables (fat mass and 1RM strength) or secondary outcome variables (RMR and RQ).

Figure 2 presents individual and group mean changes in body composition and performance variables during the LCKD and UD phases compared with measures at baseline.

Discussion

To the best of our knowledge, this is the first study to explore the effect of an LCKD on body mass, body composition, and performance in athletes competing in weight class strength and power sports. Although many strategies are available for athletes wanting to reduce weight, weight-making strategies are not without risks and are not universally effective for all athletes. Powerlifting and Olympic weightlifting athletes in this study decreased body mass and achieved lifting performances that were comparable with their UD when consuming an ad libitum LCKD.

The LCKD resulted in meaningful reductions in body mass compared with both baseline and the end of the UD phase. The physiological basis for this mass loss is not entirely clear. Stored glycogen and the accompanying storage of water contribute up to 2 kg of weight loss during either energy or carbohydrate restriction (¹⁶). After accounting for weight loss likely to be associated with glycogen losses, the mass loss experienced by most subjects in this study was greater than expected based on subjects' energy intake and energy expenditure. Greater than expected mass loss during ad libitum LCKD has previously been reported. For a recent review, see Noakes et al. (²³). A reduced energy intake as a result of a greater satiating effect of LCKD has previously been proposed as a mechanism contributing to weight loss during LCKD (¹³). However, in this study, subjects' reported energy intakes were similar during the 2 dietary phases. There is growing interest in the metabolic effects of LCKD that could contribute to greater than expected weight loss. Several mechanisms that seem to facilitate weight loss may contribute to the efficacy of LCKD for weight loss (¹¹). The thermogenic effect of digesting greater amounts of dietary protein could account for the apparent greater than expected weight loss during LCKD (¹⁴). However, protein intakes in the current study were similar during the 2 dietary phases. The role of changes in RMR during LCKD have also been explored as possibly contributing to weight loss (^8,11). In our study, RMR was not different during the 2 dietary phases, but changes to RMR are expected to accompany the weight loss observed during the LCKD phase. Weight loss predictably reduces RMR (^7,22). This adaptive reduction in RMR is typically reported to occur among overweight or obese subjects when they lose weight; however, this effect has also been demonstrated in athletes losing similar amounts of weight to subjects in this study (²¹). Although our study may be underpowered to detect small intervention differences in RMR, the potential for LCKD to preserve RMR during weight loss is intriguing and is supported by emerging theories of weight loss and energy balance. In a recent crossover study of overweight and obese patients, a 10–15% reduction in total body weight resulted in decreases in RMR favoring weight regain that were greatest with the low-fat diet phase and least with the very low-carbohydrate diet phase (⁸).

In the current study, there was a significant decrease in LBM after LCKD relative to UD. Lean body mass losses are an undesirable consequence of weight reduction because of the potential negative effects on performance. Lean body mass losses can occur during energy restriction (¹⁰). During LCKD, a greater utilization of amino acids has been proposed to contribute to LBM losses (³⁵). Despite a shift toward fat utilization during carbohydrate restriction, some obligatory requirement for glucose remains. To fulfill this need, gluconeogenic pathways use amino acids to produce glucose and might therefore contribute to protein catabolism (²⁴). However, a number of studies have demonstrated a preservation of LBM during low-carbohydrate diets (^3,26,38,41). Furthermore, the measured LBM losses in this study were not accompanied by decrements in lifting performances. Recent studies have shown that DXA may overestimate losses in LBM during carbohydrate restriction and particularly in athletic populations (^2,35). Dual-energy X-ray absorptiometry measurement of LBM relies on estimates of the distribution of water between intracellular and extracellular compartments (³²), which are perturbed during carbohydrate restriction. This has limited our interpretation of the LBM losses observed in this study. A 4-compartment model of body composition such as that used by Wilson et al. (⁴²) could account for changes in total body water and would be a superior approach for assessing the impact of LCKD on body composition. Additional studies are needed to determine the effect of LCKD on LBM in resistance-trained subjects.

Our study found that neither training quality (the ability of athletes to maintain their usual training load) nor lifting performances were adversely affected during a relatively long exposure to an LCKD. This is supported by recent studies that have shown a preservation of strength in resistance training individuals consuming low-carbohydrate diets (^26,31,41). Sports dietary approaches for short-duration, resistance exercise have traditionally advised a high-carbohydrate intake. This is likely based on evidence supporting the importance of glycogen as both a fuel for exercise and regulator of skeletal muscle adaptation responses to training (^4,28). Thus, dietary strategies that involve carbohydrate restriction have been believed to compromise strength and power performances and longer-term adaptation as a result of diet-induced glycogen depletion (¹⁸). Yet, deleterious effects of restricted exogenous carbohydrate have not been demonstrated in resistance exercise (⁹). Creer et al. (⁶) studied human muscle cellular growth pathways during a low-carbohydrate diet and found no effect of either low exogenous carbohydrate or low muscle glycogen on signaling of mammalian target of rapamycin complex 1, a crucial step in muscle protein synthesis. Moreover, dietary carbohydrate does not seem to augment muscle protein synthesis when dietary protein is adequate (²⁷). It is also unclear whether glycogen depletion during LCKD persists beyond the initial 4–6 weeks of carbohydrate restriction. Volek et al. (³⁷) recently showed that muscle glycogen was not different in long-term adapted LCKD endurance athletes compared with matched controls consuming a high-carbohydrate diet. Besides exogenous carbohydrate intake, glycogen storage is influenced by provision of gluconeogenic substrates, insulinemic amino acids, glucose uptake into muscle, and insulin sensitivity (⁴). Although muscle glycogen was not assessed in this study, it is clear that exogenous carbohydrates are not obligatory for glycogen synthesis, and glycogen depletion may not be detrimental to muscle protein synthesis and therefore long-term muscle adaptations.

Low-carbohydrate diets result in a metabolic shift in fuel availability and utilization that include increases in fat oxidation (^37,39) and the production of ketones as an alternate fuel source when glucose is low (^30,37). These adaptations also do not seem to be detrimental for lifting performance. Waldman et al. (³⁹) recently hypothesized that an increased fat utilization could have favorable effects on ATP resynthesis during resistance exercise by increasing aceyl-CoA dependency on free fatty acids and sparing glycogen. Several studies have explored the role of ketones as a fuel during endurance exercise (^30,37), but the role of ketones as a fuel during resistance exercise is unknown. In this study, there was some evidence of a shift in fuel utilization during LCKD, but the findings were not clear. There was a trend toward greater fat oxidation at rest during LCKD, but differences were not significant. Fasting blood ketones (β-hydroxybutyrate) during LCKD were elevated (mean = 0.4 mmol·L⁻¹; range 0.2–1.7) to levels comparable with previous studies of very low dietary carbohydrate diets (^39,43), but in some individual subjects, ketone levels did not rise beyond levels expected during normal carbohydrate availability. Therefore, the role of altered fuel utilization in this study is unclear.

This study provides an important first step in informing evidence-based dietary approaches for weight class lifting athletes. Implementing dietary interventions in a real-world setting presents a number of challenges. To overcome some of these challenges, subjects in this study were prescribed ad libitum diets in both dietary phases that differed only in the relative proportions of macronutrients and were specifically instructed to not engage in energy restriction. Underreporting energy intake, a major challenge in energy restriction (³³), was therefore largely eliminated and thus overcame some of the limitations of self-reporting of dietary intake. Similarly, training approaches were not controlled in the current study. Standardized training programs are likely to accentuate a training effect (⁴¹). In the current study, experienced lifters followed individualized training programs and were assessed for performance changes using their usual lifts nullifying learning effects and reporting realistic and applicable performance outcomes. However, a number of limitations are needed to be recognized. To promote adherence and minimize the burden on subjects, laboratory testing was conducted at only 3 time points. Although diet order effects could be statistically controlled, additional data collection at the start of each new phase of the study could have strengthened comparisons by establishing new baselines before diet crossover. Furthermore, a 2-week washout period may have been insufficient to eliminate carry-over effects of the previous dietary phase. However, random allocation in the balanced crossover design somewhat diminished the influence of the short washout period. In this study, there were insufficient subjects to explore intervention sex differences in our mixed sex cohort. The ad libitum dietary design also resulted in 1 subject having a different energy intake during the 2 dietary phases. This subject's long-term UD involved a substantively restrictive caloric intake. After being instructed to switch to a 3-month ad libitum LCKD, this subject's total caloric intake increased significantly. Nevertheless, this effect would act as a disadvantage for weight loss during LCKD.

Practical Applications

Weight class sports often require athletes to transiently reduce body weight to make weight for competition. Energy restriction and rapid weight loss strategies used by athletes are not universally effective for reducing weight and can be associated with a number of negative side effects. In this study, a 12-week ad libitum LCKD resulted in practically meaningful reductions in body weight without compromising training or performance and therefore seems safe and suitable to resistance trained athletes who desire lower body mass. We have thus demonstrated an alternate weight-making strategy for weight class athletes involved in powerlifting and Olympic weightlifting. An LCKD might also be applicable to other popular weight class sports including combat sports, but this needs to be explored in future studies. Coaches and athletes should consider using an LCKD to achieve targeted weight reduction goals in favor of either energy restriction or rapid weight loss strategies.