Effect of a Ketogenic Diet on Submaximal Exercise Capacity and Efficiency in Runners

David M. Shaw; Fabrice Merien; Andrea Braakhuis; Ed Maunder; Deborah K. Dulson

Disclosures

Med Sci Sports Exerc. 2019;51(10):2135-2146. 

In This Article

Results

Participants

Two participants were excluded during the study because of poor dietary reporting, poor dietary compliance to the KD, and low training load, thus giving a sample size of n = 8 (two marathoners, four ultramarathoners, and two long-distance triathletes; age, 29.6 ± 5.1 yr; body mass, 73.1 ± 6.9 kg; height, 1.81 ± 0.05 m; BMI, 22.4 ± 1.7 kg·m−2; V̇O2max, 59.4 ± 5.2 mL·kg−1·min−1; hours training per week, 10.4 ± 1.6 h; years training, 9.0 ± 3.5 yr).

Dietary and Training Compliance

There was no diet–adaptation interaction for body mass (P = 0.083); however, main effects for diet (P = 0.011) and adaptation (P < 0.001) were observed. Body mass was lower in the KD compared with HD (P < 0.002) and post- compared to pre-adaptation (P = 0.015). There was a significant diet–adaptation interaction for Σ8 skinfolds (P = 0.05), with a significant moderate reduction at post- compared with pre-KD (Δ = −5.2 ± 2.7 mm; P = 0.028; d = −0.60 ± 0.35). There was no difference for preintervention dietary macronutrient and EI between groups (all, P > 0.05; d = unclear or trivial; data not shown). Dietary intake during each intervention is summarized in Table 1. There was no difference in daily EI between diets (P = 0.49; d = 0.35 ± 0.92). Protein intake did not differ between diets (P = 0.99, d = 0.01 ± 1.08). However, in the KD compared with HD, there was a significant large reduction in CHO intake (Δ = −302 ± 49 g·d−1; P < 0.001; d = −3.44 ± 0.64) and a significant extremely large increase in fat intake (Δ = 152 ± 35 g·d−1; P < 0.001; d = 4.57 ± 1.20). Urinary AcAc and capillary blood D-βHB concentrations are summarized in the Supplemental Content (Figs. A and B, Supplemental Digital Content 3, Ketone body concentrations during keto adaptation, http://links.lww.com/MSS/B598). During the KD, all participants demonstrated positive urinary ketones by day 3 and for 95.5% of participant days. For blood D-βHB, all participants had elevated concentrations of ≥0.3 mmol·L−1 by day 3 and ≥0.5 mmol·L−1 by day 7. Participants demonstrated blood D-βHB concentrations ≥0.5 mmol·L−1 on 87.5% of participant days from day 3 onward. Training load did not differ between diets as determined by total TRIMP (P = 0.39; d = 0.25 ± 0.51) and total session RPE (P = 0.59; d = −0.17 ± 0.56), which is summarized in the Supplemental Content (Table, Supplemental Digital Content 4, Summary of accumulated training load during the 31-d adaptation periods, http://links.lww.com/MSS/B599). However, there was a trend toward a small increase in total running TRIMP during the KD compared with HD (Δ = 279 ± 220; P = 0.08; d = 0.27 ± 0.25), despite total running distance not differing (253.7 ± 112.8 vs 249.7 ± 104.9 km; P = 0.71; d = 0.03 ± 0.16).

Cardiorespiratory variables, perceived exertion, and substrate oxidation during the metabolic test. A summary of the outcomes from the metabolic tests can be viewed in the Supplemental Content (Table, Supplemental Digital Content 5, Summary of metabolic variables and perceived exertion during the metabolic test, http://links.lww.com/MSS/B600 Figs. A–D, Supplemental Digital Content 6, Summary of oxygen uptake and exercise efficiency during the metabolic test, http://links.lww.com/MSS/B601). For the post-KD test, blood D-βHB concentration was 0.94 ± 0.45 mmol·L−1 on arrival to the laboratory. There were no diet–adaptation interactions or main effects for V̇O2max, EEaero-max, VT2, or HRmax (all, P > 0.05; all d = trivial or unclear). However, there were diet–adaptation interaction for vV̇O2max (P = 0.01), with a significant moderate reduction in the post- compared with pre-KD test (Δ = −1.0 ± 0.8 km·h−1; P = 0.025; d = −0.92 ± 0.52). There were no interactions or main effects for HR or RPE (all, P > 0.05). A significant diet–adaptation–intensity interaction was observed for absolute V̇O2 (P = 0.003), with a small nonsignificant increase at 12 km·h−1 (Δ = 0.13 ± 0.11 L·min−1; P = 0.22; d = 0.34 ± 0.31) and a significant moderate increase at 13.5 km·h−1 (Δ = 0.29 ± 0.10 L·min−1; P = 0.011; d = 0.66 ± 0.27) in the post- compared with pre-KD test. Relative to body mass, there was a significant diet–adaptation–intensity interaction for V̇O2 (P = 0.011), with a trend toward a moderate increase at 12 km·h−1 (Δ = 2.7 ± 1.5 mL·kg−1·min−1; P = 0.082; d = 0.70 ± 0.47) and a significant large increase at 13.5 km·h−1 (Δ = 5.0 ± 1.6 mL·kg−1·min−1; P = 0.009; d = 1.30 ± 0.52) during the post- compared with pre-KD test.

There were significant diet–adaptation interactions for RER during the metabolic test (P = 0.001) and RER at V̇O2max (P = 0.004). In the post- compared with pre-KD test, there was a significant reduction in RER across all intensities (P < 0.001) and a very large reduction in RER at V̇O2max (Δ = −0.14 ± 0.04; P = 0.002; d = −2.84 ± 1.07). No differences were observed between the pre- and post-HD tests for RER (all, P > 0.05). There were significant diet–adaptation interactions for MFO and FATmax (all, P < 0.001), with a significant extremely large increase in MFO (0.57 ± 0.10 vs 1.12 ± 0.10 g·min−1; P < 0.001; d = 4.95 ± 1.07) and FATmax (43% ± 5% vs 70% ± 4% V̇O2max; P < 0.001; d = 5.05 ± 0.82) in the post- compared with pre-KD test, whereas the HD intervention had no effect on MFO (0.57 ± 0.11 vs 0.60 ± 0.12 g·min−1; P = 0.43; d = 0.20 ± 0.25) or FATmax (45% ± 7% vs 44% ± 3% V̇O2max; P = 0.42; d = −0.18 ± 0.39). All participants exhibited MFO rates >1.0 g·min−1 and FATmax >65% V̇O2max in the post-KD test.

Exercise efficiency during the metabolic test. In the post-KD test, there was a significant adaptation–intensity interaction for predicted compared with measured V̇O2 (P < 0.001), with a significant large increase in measured V̇O2 at 13.5 km·h−1 (Δ = 4.4 ± 1.6 mL·kg−1·min−1; P = 0.01, d = 1.25 ± 0.53; Figure C, Supplemental Digital Content 6, Summary of oxygen uptake and exercise efficiency during the metabolic test, http://links.lww.com/MSS/B601). The shift in RER from the pre- to post-KD trial explained 14% ± 8% of the increase in V̇O2 at 13.5 km·h−1. Furthermore, there were significant diet–adaptation–intensity interactions for absolute EE (P < 0.001) and EE relative to body mass (P = 0.011). There was a significant small increase in absolute EE at 13.5 km·h−1 (Δ = 5.1 ± 2.2 kJ·min−1; P = 0.026; d = 0.56 ± 0.28) and, when relative to body mass, a small nonsignificant increase at 12 km·h−1 (Δ = 0.02 ± 0.02 kJ·kg−1·min−1; P = 0.18; d = 0.56 ± 0.47), and a significant moderate increase at 13.5 km·h−1 (Δ = 0.04 ± 0.02 kJ·kg−1·min−1; P = 0.018; d = 1.14 ± 0.73) in the post- compared with pre-KD test.

Environmental and dietary conditions during the run-to-exhaustion trial. Trials were performed within standard laboratory conditions of 17°C ± 1°C and 45% ± 3% humidity. There was no difference in running speed between the HD and KD trials (12.9 ± 0.7 vs 12.9 ± 0.8 km·h−1, P = 0.89), and all participants commenced each trial with urine-specific gravity values <1.025. The exercise-induced reduction in body mass did not differ between trials (all, P > 0.05, all d = trivial or unclear; Table 2). Macronutrient compositions of the pre-HD and post-HD trial meals were identical (148 ± 14 g CHO, 7 ± 1 g fat, and 23 ± 4 g protein) and similar to the pre-KD trial meal (146 ± 14 g CHO, 7 ± 1 g fat, and 23 ± 4 g protein). However, CHO and fat contents were lower and higher, respectively, in the post- compared with pre-KD trial meal (8 ± 1 g CHO and 67 ± 6 g fat), whereas protein content was similar (26 ± 6 g). There was no difference in CHO ingestion rate during the pre-HD, post-HD, and pre-KD trials (55.7 ± 6.0, 53.4 ± 2.8, and 55.3 ± 8.4 g·h−1, respectively; all P > 0.05; all d = trivial or unclear). Coconut oil was ingested during the post-KD trial at a rate identical to the EI in the pre-KD trial (26.0 ± 4.0 g·h−1).

Submaximal Exercise Capacity

There were no diet–adaptation interactions or main effects for TTE (P = 0.557) or distance-to-exhaustion (DTE; P = 0.508; Figure 1A and B). Furthermore, there was no difference in mean change between diets for TTE (P = 0.56; d = 0.25 ± 0.60) or DTE (P = 0.51; d = 0.26 ± 0.60). The range within the 90% CI for TTE and DTE increased approximately twofold in the post- compared with pre-KD trial, whereas there was a reduction from the pre- to post-HD trial ([pre-HD, 112–263 min vs post-HD, 211–252 min and pre-KD, 223–254 min vs post-KD, 188–250 min] and [pre-HD, 45.8–56.4 km vs post-HD, 45.7–53.7 km and pre-KD, 47.8–54.6 km vs post-KD, 40.5–53.1 km]).

Figure 1.

Submaximal exercise capacity values presented as mean ± SD and individual TTE (A) and mean ± SD and individual DTE (B).

D-βHB, glucose, and lactate concentration during the run-to-exhaustion trial. There was a significant diet–adaptation–time interaction for blood D-βHB concentration (P < 0.001), with significant large to extremely large increases in the post- compared with pre-KD trial for all time points (all, P < 0.05; Figure 2A). No differences in blood D-βHB concentration were observed between the pre- and post-HD trials (all, P > 0.05; all, d = trivial or unclear). There was a significant diet–adaptation–time interaction for serum glucose concentration (P = 0.032); however, post hoc analysis could only locate a moderate nonsignificant increase at preexercise (P = 0.21, d = 0.92 ± 0.62) and a moderate nonsignificant reduction at 2-h exercise (P = 0.49, d = −0.93 ± 0.82) in the post- compared with pre-KD trial (Figure 2B). Serum glucose concentration was elevated from 1-h exercise to 1-h postexhaustion compared with preexercise in all trials (all, P < 0.05), except for the post-KD trial, for which only exhibited an increase from preexercise to 1-h exercise (P = 0.005). There was no diet–adaptation–time interaction for blood lactate concentration (P = 0.061). However, there was a significant effect for time (P = 0.03), with blood lactate concentrations lower at 2-h exercise compared with exhaustion (P = 0.02; Figure 2C).

Figure 2.

Capillary blood D-βHB (A), serum glucose (B), and capillary blood lactate (C) concentrations during the run-to-exhaustion trials. Values are mean ± SD. The individual gray responses in panel A are individual post-KD D-βHB values for participants, and those in panel C are post-KD lactate values for the participant who reduced TTE from 263 to 145 min. Significantly different in the post- compared with pre-KD trial (*P < 0.01; **P < 0.001). ES (d): #moderate, ##large, ###very large, and ####extremely large.

Substrate oxidation during the run-to-exhaustion trial. There was a significant diet–adaptation interaction for the rate and percentage contribution to total EE of CHO and fat oxidation (all, P < 0.001; Figure 3A–D). CHO oxidation was ~2.5-fold higher in the pre- compared with post-KD trial, and fat oxidation was ~3.5-fold higher in the post- compared with pre-KD trial (all, P < 0.001). In the post-KD trial, fat oxidation rates ranged between 0.88 and 1.51 g·min−1. There were no differences in substrate oxidation between the pre- and post-HD trials (all, P > 0.05; all, d = trivial or unclear).

Figure 3.

The contribution of substrate to EE presented as rate of CHO oxidation (A) and percentage contribution of CHO to total EE (C), and rate of fat oxidation (B) and percentage contribution to total EE (D). Diet–adaptation interaction; significant effect of diet (a P < 0.001). ES: #small, ##large, ###very large, and ####extremely large.

Cardiorespiratory variables and perceived exertion during the run-to-exhaustion trial. Table 2 and Figure 4A–D provide a summary of the cardiorespiratory variables and perceived exertion data during the run-to-exhaustion trials. There was a significant diet–adaptation–time interaction for absolute V̇O2 (P < 0.001), with significant small increases in the post- compared with pre-KD trial during the first (P = 0.025; d = 0.46 ± 0.26) and second hours (P = 0.016; d = 0.48 ± 0.23; Table 2). These effects increased when accounting for body mass, resulting in significant moderate increases in V̇O2 occurring during the first (P = 0.009; d = 0.83 ± 0.37) and second hours (P = 0.005; d = 0.84 ± 0.23) and a significant small increase at exhaustion (P = 0.046; d = 0.50 ± 0.39; Figure 4A). No effects were observed between the pre- and post-HD trials for absolute and relative V̇O2 (all, P > 0.05; all, d = trivial or unclear). Despite a significant diet–adaptation–time interaction for exercise intensity relative to V̇O2max, post hoc analysis could not locate specific differences between trials (all, P > 0.05; all, d = trivial or unclear; Figure 4B). There was a significant diet–adaptation interaction for HR (P = 0.011), with an increase in the post- compared with pre-KD trial (P < 0.001; Table 2). HR did not differ between the pre- and post-HD trials (P = 0.85).

Figure 4.

Exercise efficiency during the run-to-exhaustion trial presented as oxygen uptake relative to body mass alongside predicted values for the post-KD trial (A), oxygen uptake relative to V̇O2max (B), EE relative to body mass (C), and EE relative to EEaero-max (D). Significantly different in the post- compared with pre-KD trial (*P ≤ 0.05). Trending toward significantly different in the post- compared with pre-KD trial (ΩP = 0.06). ES (d): #small and ##moderate.

Exercise efficiency during the run-to-exhaustion trial. In the post-KD trial, there was a significant adaptation–time interaction for predicted compared with measured V̇O2 (P = 0.001; Figure 4A), with trends for small increases in measured V̇O2 during the first hour (Δ = 2.2 ± 1.4 mL·kg−1·min−1; P = 0.06; d = 0.46 ± 0.33) and second hours (Δ = 2.2 ± 1.2 mL·kg−1·min−1; P = 0.06; d = 0.48 ± 0.30). The shift in RER from the pre- to post-KD trial explained 55% ± 32% of the increase in V̇O2 during the first 2 h. There was a significant diet–adaptation–time interaction for absolute EE (P = 0.001), with trends for small increases the first hour (Δ = 3.0 ± 2.0 kJ·min−1; P = 0.076; d = 0.33 ± 0.25) and second hours (Δ = 3.1 ± 1.7 kJ·min−1; P = 0.056; d = 0.36 ± 0.23) in the post- compared with pre-KD trial (Table 2). When accounting for body mass, there was also a significant diet–adaptation–time interaction (P < 0.001), with significant moderately higher rates of EE during the first hour (Δ = 0.06 ± 0.03 kJ·kg−1·min−1; P = 0.022; d = 0.65 ± 0.36) and second hours (Δ = 0.06 ± 0.02 kJ·kg−1·min−1; P = 0.014; d = 0.67 ± 0.31) in the post- compared with pre-KD trial (Figure 4A). Despite a significant diet–adaptation–time interaction for the intensity relative to EEaero-max (P = 0.001), post hoc analysis could not locate specific differences between trials (all, P > 0.05; Figure 4D).

Exercise capacity and lactate accumulation based on RER at V̇O2max. Comparisons for exercise capacity and blood lactate concentrations were made between pre- and post-KD trials based on RER at V̇O2max (<1.0 (n = 4) and >1.0 (n = 4)). In the group with RER <1.0 at V̇O2max, TTE significantly declined from pre- to post-KD (237 ± 31 min vs 174 ± 24 min; P = 0.04; d = 1.49 ± 1.04). In contrast, for the group with RER >1.0 at V̇O2max, there was no effect of the KD on TTE (241 ± 27 min vs 265 ± 21 min; P = 0.15; d = 0.67 ± 0.83). Using an unadjusted Student's unpaired t-test, there was a large significant reduction for mean change in TTE from prediet to postdiet in the group with RER <1.0 compared with >1.0 at V̇O2max (−63 ± 38 min vs 24 ± 21 min; P = 0.009, d = 1.22 ± 0.86). Furthermore, using an unadjusted Student's unpaired t-test for each time point, the group with RER <1.0 at V̇O2max exhibited a significant large increase in the difference for mean lactate concentration at exhaustion from prediet to postdiet compared with the group with RER >1.0 at V̇O2max (P = 0.015, d = 1.45 ± 0.86). These TTE and lactate data are presented in the Supplemental Content (Figs. A–B, Supplemental Digital Content 7, Time-to-exhaustion and blood lactate concentration based on RER at V̇O2max in the post-KD trial, http://links.lww.com/MSS/B602). There were no differences in exercise intensity relative to VT2, V̇O2max, or vV̇O2max, or rates of substrate oxidation or blood D-βHB and serum glucose concentrations between the two groups (data not shown).

processing....