Post‐Exercise R‐1,3‐Butanediol did not Improve Markers of Fatigue or Recovery During a Multi‐Stage Mountain Bike Race

ABSTRACT

Evidence suggests that exogenous ketosis (EK) induced by ketone monoester (KME) supports recovery and attenuates physiological stress during endurance exercise. However, the potential benefits of the ketone precursor R‐1,3‐butanediol (BDO) during real‐world multi‐stage cycling remain unknown. We investigated the effects of post‐exercise BDO supplementation on physiological strain, dietary intake, hydration, cognitive function, sleep, and subjective recovery during the 2025 Absa Cape Epic mountain bike race. In this double‐blind, placebo‐controlled study, 18 trained mountain bikers (10 teams) completed an 8‐day multi‐stage race. Participants received either 87.5 mL ketone IQ (BDO) or a placebo (CON) twice daily (post‐stage and pre‐sleep). Heart rate, energy intake, sleep metrics, cognitive performance, and subjective stress‐recovery were monitored throughout the race. BDO induced mild post‐stage ketosis (∼1.0 mM β‐hydroxybutyrate) from stage 1 onwards. Across the race, both groups demonstrated lower exercise heart rate, increased resting heart rate, increased energy and carbohydrate intake, and slower reaction time by stage 6, indicating accumulating physiological strain. BDO did not alter any of these effects compared with placebo. These findings suggest that the modest ketosis achieved with BDO was not effective under real‐world racing conditions, highlighting the importance of formulation‐specific and ecologically valid evaluations of exogenous ketosis strategies in future studies.

Highlights

• Post‐exercise and pre‐sleep BDO supplementation produced mild exogenous ketosis during an 8‐day MTB race.

• BDO supplementation did not influence cardiovascular strain, nutrition intake, cognition, or sleep.

• Under the current study conditions, ketone precursor supplementation was insufficient to improve recovery during multiday cycling.

1. Introduction

Multi‐stage cycling races place substantial physiological and cognitive demands on participants (Engelbrecht and Terblanche 2018; Reinpõld et al. 2022). The Absa Cape Epic, one of the most demanding mountain bike (MTB) stage races, spans ∼600–680 km and ∼15,000–17,000 m of climbing over eight days. Riders must sustain prolonged high‐intensity efforts under cumulative fatigue, resulting in progressive physiological strain (Engelbrecht and Terblanche 2018; Reinpõld et al. 2022). Previous research has shown that multi‐day cycling events are associated with reduced exercise heart rate, elevated resting heart rate, and a shift towards lower exercise intensities (Greeff 2014; Wirnitzer and Kornexl 2008). Although the cardiovascular demands vary according to stage characteristics and pacing strategy, the progressive changes in heart rate responses observed across a multi‐day event are consistent with the accumulation of physiological fatigue. During the 2014 Absa Cape Epic, riders' peak heart rates dropped ∼11 bpm by day 3, with an even more pronounced drop in average heart rate (∼18 bpm) (Greeff 2014). Riders spent progressively more time in the easy heart rate intensity zone and progressively less time in the moderate and hard intensity zones, based on the three‐intensity zone model, previously used to describe exercise intensity during multi‐day mountain bike events (Greeff 2014; Wirnitzer and Kornexl 2008; Seiler and Kjerland 2006).

Optimal recovery is essential for sustaining high‐performance levels in events such as the Absa Cape Epic. Single and multiday endurance events have shown substantial alterations in autonomic modulation (Rundfeldt et al. 2018; Belinchon‐deMiguel and Clemente‐Suárez 2018; Clemente‐Suárez 2015), highlighting the stressfulness of these events and their potential to negatively impact sleep quality and quantity (Chennaoui et al. 2015). These events are further characterized by energy deficits (Burke 2002; Black et al. 2012; Robberechts, Stalmans, et al. 2026), dehydration (Sawka et al. 2007), and impaired cognitive function (Perrotta et al. 2022; Van Cutsem et al. 2017; Cona et al. 2015), all of which may compromise performance and recovery.

Exogenous ketosis (EK) has been proposed as a strategy to mitigate these stressors (Egan 2025). Importantly, most supportive evidence to date has been derived from studies employing ketone monoesters (KME). In this regard, KME‐induced EK has been shown to enhance post‐exercise recovery and attenuate physiological stress responses (Poffé and Hespel 2020; Robberechts and Poffé 2024), maintain endurance exercise output and appetite and energy intake (Poffé et al. 2019) during excessive exercise training, and preserve sleep quality during periods of heavy exercise (Robberechts et al. 2023). Recent studies have also suggested that EK helps preserve cognitive function during mentally fatiguing activities (M. Evans and Egan 2018; Poffé et al. 2023; Mosquera‐Lopez et al. 2026).

While these findings highlight the potential utility of EK, it is critical to acknowledge that ketone supplements differ substantially in their absorption kinetics, peak β‐hydroxybutyrate (βHB) concentrations, and physiological effects (Stubbs et al. 2017; Falkenhain et al. 2024). Consequently, the positive outcomes observed with KME cannot be directly extrapolated to other ketone formulations. R‐1,3‐butanediol (BDO), a ketone precursor that induces EK via hepatic metabolism, has received considerably less attention in exercise contexts than ketone monoesters (Shaw et al. 2019; Scott et al. 2019; Stalmans et al. 2026). Nonetheless, preliminary evidence suggests that acute BDO ingestion prior to exercise can lower the fatigue index during repeated high‐intensity exercise and significantly improve anaerobic performance in moderately trained adults (Soto et al. 2025). Additionally, post‐exercise BDO supplementation has been shown to increase vascular function, angiogenic signalling and circulating erythropoietin concentration (Stalmans et al. 2026). Furthermore, BDO tastes slightly better and is cheaper than KME, and it induces nutritional ketosis for a longer period (Falkenhain et al. 2024).

Accordingly, we conducted a field study during the 2025 Absa Cape Epic to investigate the impact of BDO‐induced EK on (i) physiological strain, (ii) energy intake, (iii) mental fatigue, (iv) exercise performance, and (v) sleep quality and quantity. Given the distinct metabolic kinetics of R‐1,3‐butanediol (BDO), this study specifically evaluated the ecological effectiveness of a ketone precursor under real‐world racing conditions, rather than extrapolating the findings from KME‐based laboratory studies.

2. Materials and Methods

2.1. Participants

Ten teams (2 riders per team) of healthy, trained mountain bikers who entered the 2025 Absa Cape Epic volunteered for this study. Riders were included if they had a valid entry, volunteered as a team, were ≥ 18 years of age, and had a downloadable heart rate monitor. Exclusion criteria included adherence to a ketogenic or low‐carbohydrate diet in the preceding six months, smoking, failure to complete any stage, or a difference in lactate threshold more than 30 W between teammates. The study was approved by the Health Research Ethics Committee of Stellenbosch University (S22/08/156) and conformed to the principles of the Declaration of Helsinki. Following a thorough explanation of the study's objectives, procedures, and potential risks, the participants provided written consent.

2.2. General Study Design

This study followed a double‐blind, placebo‐controlled, within‐team, randomized design (Figure 1). Riders competed in pairs, with one rider randomly assigned to the ketone IQ (BDO) group and the other to the control (CON) group. Randomization was performed by an independent investigator who was not involved in the study. Pre‐testing was conducted in the week before the event, and post‐testing the day after.

FIGURE 1.

Schematic overview of the study design. In a double‐blind randomized, parallel design, participants (CON: n = 10, BDO: n = 8) participated in the Absa Cape Epic 2025 edition (8‐day mountain bike race). Immediately after each stage and just before sleep, the participants received either a control or BDO supplement. REST‐Q sport: Recovery Stress Questionnaire for athletes, VAS, visual analog scale, BDO, (R)‐1,3‐ butanediol.

Eighteen riders (3 female and 15 male) completed the race and were included in the final analyses. One participant withdrew due to injury, and another withdrew after Stage 3, both in the BDO group. The physical characteristics of the participants included in the analysis were as follows: CON (n = 10) versus BDO (n = 8); mean ± SD: age, 36.1 ± 13.8 years versus 41.8 ± 13.3 years; height, 1.76 ± 0.11 m versus 1.78 ± 0.07 m; body mass, 72.3 ± 13.7 kg versus 74.7 ± 12.1 kg; body mass index, 22.5 ± 1.9 kg/m² versus 23.3 ± 3.9 kg/m²; maximal oxygen uptake rate ($\dot{\dot{V}}$O₂max), 56.2 ± 8.1 mL.kg⁻¹.min⁻¹ versus 52.0 ± 8.4 mL.kg⁻¹.min⁻¹; power output at the second lactate threshold (LT2 PO), 254 ± 36 W versus 265 ± 36 W; peak power output (PPO), 295 ± 58 W versus 313 ± 33 W. In the final analyses, CON included two females (using contraceptives, Copper T (hormone‐free contraceptive device) and Mirena (hormonal intrauterine device), both in the luteal phase), and BDO included one female participant (menstruation started on the third day of the race).

2.3. Preliminary Testing and Familiarization

In the weeks (2–4) preceding the event, participants visited the laboratory twice for preliminary testing and familiarization sessions. During visit one, participants completed a maximal incremental cycling test on a calibrated cycling ergometer (Avantronic Cyclus II, Leipzig, Germany) to determine $\dot{\dot{V}}$O₂max and heart rate intensity zones. The protocol began at 120 W for men and 80 W for women, increasing by 30 W every 150 s until volitional exhaustion (Engelbrecht and Terblanche 2018). Gas exchange was measured continuously (Cosmed Quark CPET; Rome, Italy) and blood lactate (Lactate Pro 2, Arkray, Japan) samples were collected from the earlobe in the final 30 s of each stage. $\dot{\dot{V}}$O₂max was defined as the highest mean oxygen uptake recorded over 30 s. Heart rate intensity zones were defined by blood lactate concentrations: Zone 1 (< 2 mM), Zone 2 (between 2 and 4 mM), and Zone 3 (≥ 4 mM) (Engelbrecht and Terblanche 2018; Wirnitzer and Kornexl 2008). During their second visit, participants were familiarized with the cognitive testing battery (CANTAB, Cambridge Cognition; Cambridge, UK; details below) and received a LifeQ‐enabled smartwatch to monitor sleep during the race.

2.4. The Cape Epic Mountain Bike Race

The 8‐day mountain bike stage race was held in the Western Cape, South Africa. The 2025 edition of the race covered a distance of 567 km with 15,300 m of elevation. The metrics and weather conditions for each stage are presented in Table 1. The latter information was obtained from the closest weather station to the stage location and at a time when the riders were on the course. Following the prologue, the riders began each stage in batches at 10‐min intervals based on their overall general classification.

TABLE 1.

Characteristics of the 2025 Absa Cape Epic.

	Prologue	Stage 1	Stage 2	Stage 3	Stage 4	Stage 5	Stage 6	Stage 7
Distance (km)	26	92	66	78	80	98	87	40
Total climbing (m)	750	2750	950	2450	1950	2850	2550	1050
Temp high (oC)	27.0	26.4	31.0	43.0	29.1	29.0	26.0	24.0
Temp low (oC)	16.6	16.4	16.0	16.0	14.4	15.4	15.5	15.8
Wind (km/h)	4,3	18	15,6	17,2	8,4	8	7,2	19
Humidity range (%)	68,8	84	76	54	66	74	94	83
Rainfall (mm)	0	0	0	0	0	3,1	1,4	4
Average race time (h:mm:ss)	1:31:36	6:00:25	3:20:55	6:05:41	4:45:50	6:46:14	6:20:01	2:22:51

Heart rate was recorded during each stage using participants' personal devices. Capillary blood samples were collected before each stage and within 30‐ to 45‐min consuming the post‐stage supplement to determine βHB concentration (Keto Mojo GKI, Taidoc Technology Corporation/Foracare Inc.; Napa, USA).

2.5. Nutrition Intervention

Participants consumed either BDO (R‐1,3‐butanediol) or a non‐isocaloric taste‐ and viscosity‐matched placebo (98.35% w/v water, 0.71% w/v bitter type flavour, 0.04% w/v Monk fruit extract, 0.03% w/v Tasteva, 0.61% w/v citric acid, 0.02% w/v natural flavour blend, 0.1% w/v natural bitter masking type flavour, and 0.01% w/v natural passion fruit type flavour, all identically present in the BDO drink) provided in identical tubes. Both supplements were provided by Health Via Modern Nutrition (HVMN, Miami, USA). Two 87.5 mL doses (equivalent to 25 g BDO) were consumed daily: immediately after each stage and before sleep.

Riders managed their own food and drinks, which were recorded after each stage. An additional 175 kcal/day was added to the BDO group's total energy intake to account for the supplement dosage. Participants were instructed to avoid all food and fluid intake for 30 min after consuming their assigned supplement to minimise interference with supplement absorption. Compliance with this instruction could not be strictly controlled following each race stage under field conditions, and because we did not have access to the rider recovery area, we could not determine how long riders refrained from eating their post‐stage meal after consuming their respective supplements. However, compliance was likely greater for the pre‐bedtime supplementation because participants were no longer engaged in post‐race recovery and logistical activities.

2.6. Experimental Sessions

Pre‐testing was conducted in the week leading up to the race, and post‐testing was conducted the day after the race. Upon arrival, they completed the appetite visual analogue scale (VAS) and the Recovery‐Stress Questionnaire for Athletes (RESTQ‐76 Sport Questionnaire), followed by a cognitive test battery. Below is a more thorough explanation of the procedures.

2.7. Appetite and Nutrition Intake

Appetite was assessed upon arrival at the laboratory and before each stage using a validated 0‐10 Likert VAS (Woods et al. 2018), with the following four questions: “How hungry do you feel?”, “How full do you feel?”, “How satisfied do you feel?”, “How much do you think you can eat now?” To assess nutritional intake, a 24‐h food recall was performed at the end of each stage, supported by wrappers and photographs of meals. If needed, we asked participants to clarify or provide additional information on their food and drink consumption. During the food recall, participants were asked to report any gastrointestinal (GI) symptoms. Nutritional analysis was done using the South African Medical Research Council (MRC) FoodFinder software (version 3), during which the caloric content of the BDO supplement was added to the BDO group's data.

2.8. Stress‐Recovery State and Subjective Markers of Fatigue

Participants completed the RESTQ‐76 Sport questionnaire during the laboratory visits. This self‐report method assesses the participants' physical, subjective, behavioural, and social aspects of stress and recovery through 10 stress subscales and nine recovery subscales. The total stress and recovery scores were calculated as the sum of the respective subscales' scores. The total stress score was then subtracted from the total recovery score to determine the recovery‐stress balance (Kellmann and Kallus 2001). A high score on the stress‐oriented subscale indicates a high subjective strain, whereas a high score on the recovery‐oriented subscale indicates sufficient recovery.

Before each stage, the participants completed a 0‐10 Likert VAS on muscle soreness, perceived recovery, and perceived mental alertness. At the end of each stage, the participants rated their perceived exertion using the 6‐20 Borg scale.

2.9. Cognitive Function

Cognitive performance was evaluated using three CANTAB tasks (CANTAB, Cambridge Cognition, Cambridge, UK): reaction time (RTI), rapid visual information processing (RVP), and spatial working memory (SWM) (Poffé et al. 2023). Tests were performed at baseline and after Stages 3 and 6. The test battery lasted ∼14 min and was conducted on iPads (iPad, ninth generation, iOS 18.3.2, Apple Inc., CA, USA) while the participants remained seated. The participants were instructed to use the index finger of their dominant hand to perform each test. Detailed descriptions of each test are available at www.cambridgecognition.com/cantab/cognitive‐tests/. The following outcome measures were assessed: (i) RTI: median reaction time (RTIMRT) and median movement time (RTIMMT); (ii) RVP: sensitivity to the target sequence (0.00–1.00; bad to good) (RPVA), median response latency (RVPMDL), number of false alarms out of 54 (RVPTFA), and number of correct responses (RVPTH); (iii) SWM: strategy score (SWMS) and number of errors (SWMBE).

2.10. Sleep and Resting Heart Rate

Each participant wore a LifeQ‐enabled device on the opposite wrist of their own smart watches. The device has demonstrated moderate agreement with polysomnography (Cohen's κ = 0.63; overall accuracy 71%, data provided by the manufacturer). The participants were instructed to wear the devices while sleeping to monitor sleep quantity (total sleep duration, wakefulness after sleep onset (WASO), and sleep quality (rapid eye motion (REM) and deep and light sleep). To establish baseline measurements, participants wore the device during sleep in the week preceding the race. Participants did not receive any sleep prescriptions but were asked to avoid sleeping pills or any sleep remedies.

2.11. Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics (v. 30) (IBM Corp., Chicago, USA). Data are presented as mean ± standard deviation (SD). G*Power 3.1.9.7 was used to perform an a priori sample calculation using repeated measures ANOVA with the significance level set at 5%, power at 0.80, effect size at 0.679 (Cohen's F), number of groups at 2, and number of measurements at 8. The calculation was based on three outcome variables (energy intake, power output, and average heart rate) from a previous study (Poffé et al. 2019), and the required sample size was estimated to be 6 pairs.

Linear mixed‐effects models (LMM) were used with group, stage, and their interactions as fixed effects. Random intercepts were specified for participants and teams to account for repeated measurements across stages and pairing of riders within teams. When group differences were present at baseline, the baseline value of the outcome variable was included as a covariate and specified as a fixed effect in the LMM to adjust for pre‐race imbalances. LMM was selected because of its robustness in handling unbalanced data, missing observations and ability to model both fixed and random effects. In the case of a significant group, time, or interaction effect, Sidak post hoc analysis was performed. Fixed effects were quantified using standardized regression coefficients derived from linear mixed‐effects models. p < 0.05 was considered statistically significant.

3. Results

3.1. Blood Ketone Concentration

Pre‐stage blood [βHB] were similar between the groups (F(1, 9.92) = 2.36, P _group = 0.186), and the participants were not in ketosis (0.3 ± 0.1 mM). Post‐stage [βHB] was significantly higher in BDO than CON (F(1, 8.64) = 41.54, P _group < 0.001; Figure 2). The estimated marginal means were 1.02 ± 0.09 mM for BDO and 0.37 ± 0.08 mM for CON, corresponding to a mean difference of −0.65 mM (95% CI: −0.88 to −0.42). A significant effect of time was also observed (F(7, 105.67) = 5.35, P _time < 0.001), with [βHB] increasing above prologue values during stages 1–2 (p = 0.001), stages 5–6 (p < 0.001), and stage 7 (p = 0.005), peaking at approximately 1.5 mM during stage 6 in BDO (Figure 2). In contrast, post‐stage [βHB] remained unchanged across the race in CON (0.4 ± 0.3 mM; range 0.1–1.1 mM; p > 0.05).

FIGURE 2.

Blood ketone response. The rise in blood βHB concentration 30 minutes after ingestion. #* indicates a significant time and group effect.

3.2. Cape Epic Race Data

Resting heart rate increased significantly above baseline with 4–7 bpm from Stage 1 onwards (p < 0.05, all Stages) (Figure 3a). Peak heart rate differed between groups (F(1, 4.34) = 9.13, P _group = 0.035), with estimated marginal means of 172.5 ± 3.9 bpm in CON and 161.8 ± 4.4 bpm in BDO, corresponding to a mean difference of 10.7 bpm (95% CI: 1.2–20.3). However, there was no group × time interaction (P _interaction = 0.968), indicating that both groups exhibited a similar decline in peak heart rate over the course of the race (Figure 3b). A significant effect of time was observed (F(7, 76.04) = 10.69, P _time < 0.001), with peak heart rate decreasing by approximately 16 ± 12 bpm below Prologue values from Stage 2 onwards (Stage 2, p = 0.002; Stage 3, p = 0.030; Stage 4–7, p < 0.001). From Stage 4 onwards, peak heart rate was also approximately 14 bpm lower than Stage 1 (Stage 4 and Stages 7, p < 0.001; Stage 5, p = 0.030; Stage 6, p = 0.004).

FIGURE 3.

Effect of BDO on heart rate and RPE during the race. Data are presented as mean ± SD (n = 18) for (a) resting heart rate, (b) maximal heart rate (HR max), (c) average heart rate, (d) RPE, and (e) heart rate intensity zones. # indicates a significant difference from the prologue (p < 0.05), x indicates a significant difference from Stage 1 (p < 0.05). #b indicates a significant difference from Baseline (p < 0.05).

Average heart rate was unaffected by BDO (P _interaction = 0.978) but declined significantly over time (F(7, 76.33) = 20.89, P _time < 0.001; Figure 3c). Across both groups, average heart rate was approximately 23 ± 21 bpm lower than the Prologue from Stage 1 onwards (all p < 0.001) and a further ∼12 ± 10 bpm lower from Stage 4 onwards compared with Stages 1 and 2 (all p < 0.05).

Ratings of perceived exertion (RPE) were unaffected by BDO supplementation (F(7, 118) = 1.53, P _interaction = 0.164) but changed significantly over the course of the race (F(7, 118) = 6.00, P _time < 0.001). Compared to stage 1, RPE was lower during Stage 4 (MD = −2.35, 95% CI −3.83 to −0.88) and Stage 7 (MD = −3.22, 95% CI −4.75 to −1.69) (Figure 3d).

Time spent in each heart rate zone also changed significantly throughout the race (Zone 1: F(7, 85) = 15.81, P _time < 0.001; Zone 2: F(7, 85) = 4.69, P _time < 0.001; ZONE 3: F(7, 73.66) = 12.01, P _time < 0.001), with no group × time interactions for any zone (all P _interaction > 0.05). Time spent in Zone 1 increased progressively, rising from approximately 24% during the Prologue to 75% by the later stages, and was significantly greater from Stage 3 onwards than during the Prologue (Stage 3, p = 0.011; Stage 4–7, p < 0.001). From Stage 4 onwards, riders spent significantly more time in Zone 1 than in Zones 2 and 3 (all p < 0.001). Conversely, time spent in Zone 2 declined compared with the prologue, with lower values during Stages 5 (approximately −20%; p = 0.028) and 6 (approximately −33%; p < 0.001). Time spent in Zone 3 also declined progressively across the race, decreasing from approximately 45% during Stage 1 to approximately 2% by the final stage (all p < 0.001) (Figure 3e).

3.3. Appetite and Nutrition Intake

Total daily energy (F(7,97.57) = 6.79, p _time < 0.001) and carbohydrate intake (F(7, 97.30) = 7.98, p _time < 0.001) increased progressively over the course of the race (Figure 4a and 4b). Carbohydrate intake during each stage remained ∼60 g·h⁻¹(F(6, 96.01) = 0.46, P _time = 0.835) (Figure 4c). Protein (F(7,97.11) = 2.54, P _time = 0.019) and fat intake (F(7,97.85) = 3.49, P _time = 0.002), also increased over time. Although post hoc analysis did not reveal at which time point these changes occurred (Figure 4d and 4e). Total daily fluid intake also increased significantly over time (F(7, 97.68) = 10.41, P _time < 0.001), with higher intakes from Stages 1–6 than during the Prologue (Stage 1–5, p < 0.001; Stage 6 p = 0.003) compared to the Prologue (Figure 4f).

FIGURE 4.

Effect of BDO on nutrition and fluid intake. Data are presented as mean ± SD for (a) total daily energy intake, (b) daily carbohydrate intake, (c) stage carbohydrate intake per hour, (d) protein intake, (e) fat intake, and (f) daily fluid intake. # indicates a significant difference from the prologue (p < 0.05).

BDO supplementation had no effect on any dietary intake variable (all p > 0.05), and appetite ratings remained unchanged between groups and across the race (all p > 0.05). No participant reported gastrointestinal distress at any time during the race.

3.4. Stress‐Recovery State and Subjective Markers of Fatigue

At baseline, the BDO group exhibited a more favourable stress–recovery profile than the CON group, with a lower ‘total stress’ score (53 ± 26 vs. 78 ± 21) and a higher ’total recovery’ score (150 ± 30 vs. 124 ± 22). However, after adjusting for baseline differences, no significant effects of group, stage, or group × stage interaction were observed for total stress or the recovery–stress balance (all p > 0.05), indicating that overall perceived stress and recovery remained comparable between groups throughout the race.

A significant group × stage interaction was observed for perceived recovery (F(7, 122) = 2.28, P _interaction = 0.032); however, this interaction was attributable solely to higher recovery scores in the BDO group before the prologue, prior to the commencement of supplementation. Consequently, the interaction is unlikely to reflect the effect of the intervention. Irrespective of group, perceived recovery declined over the race (F(7,122) = 2.68, P _time = 0.013), with lower values at Stages 4 and 6 compared with the Prologue (Stage 4 p = 0.032; Stage 6, p = 0.020) (Figure 5a). Muscle soreness also increased progressively from Stage 2 onwards in both groups (all p < 0.001; Figure 5b).

FIGURE 5.

Markers of perceived fatigue. Data are presented as mean ± SD for (a) perceived recovery, (b) perceived muscle soreness, and (c) perceived mental alertness. # indicates a significant difference to prologue (p < 0.05), indicates an interaction effect (p < 0.05), and * indicates a group difference (p < 0.05). A higher score in perceived muscle soreness indicates a higher level of muscle soreness, while higher scores in perceived recovery and perceived mental fatigue indicate less fatigue.

Perceived mental alertness was higher overall in the BDO group than in the CON group (group effect: F(1, 122) = 7.92, p _group = 0.006), with an estimated mean difference of 1.64 points (95% CI: 0.49–2.79). However, the difference was present before supplementation started and is therefore not a treatment effect. Both groups demonstrated a decline in alertness at Stage 6 compared with the Prologue (Stage 6, p = 0.012) (Figure 5c).

3.5. Cognitive Function

Reaction time changed significantly over the course of the race (F(2, 46) = 4.16, P _time = 0.022), remaining stable from pre‐race to Stage 3 before slowing by approximately 6.5% after Stage 6 compared with Stage 3 (p = 0.018; Figure 6a). BDO supplementation had no effect on reaction time (F(2, 46) = 1.12, P _interaction = 0.335), with no difference between the groups (F(1,46) = 0.32, P _group = 0.57). Movement time was similarly unaffected by either race progression or BDO supplementation (all p > 0.05; Figure 6b). No significant changes were observed in any of the other cognitive measures (Table 2).

FIGURE 6.

Effect on cognitive function. Data are presented as mean ± SD for (a) reaction time and (b) movement time. # indicates a significant difference from stage 3 (p < 0.05).

TABLE 2.

Cognitive function scores. Data presented as mean ± SD.

	CON (n = 10, F = 2)			BDO (n = 8, F = 1)			p‐value
	Baseline	Stage 3	Stage 6	Baseline	Stage 3	Stage 6	Time	Group	T × G
Reaction time (RTI)
RTIMRT (ms)	391 ± 48	391 ± 47	408 ± 41#	391 ± 39	368 ± 22	403 ± 33#	0.022##	0.574	0.335
RTIMMT (ms)	285 ± 92	316 ± 116	275 ± 47	241 ± 59	259 ± 72	275 ± 76	0.354	0.419	0.313
Rapid visual processing (RVP)
RVPA	0.88 ± 0.06	0.91 ± 0.06	0.8 ± 0.09	0.91 ± 0.05	0.92 ± 0.06	0.91 ± 0.07	0.147	0.114	0.454
RVPMDL (ms)	473 ± 54	451 ± 64	500 ± 138	438 ± 40	428 ± 68	439 ± 44	0.351	0.129	0.634
RVPTFA	24.9 ± 40.5	24.4 ± 41.4	26.4 ± 42.4	20.1 ± 50.9	19.4 ± 45.6	26.9 ± 64.9	0.652	0.359	0.653
RVPTH	34.1 ± 9.4	38.0 ± 10.5	34.9 ± 10.5	37.9 ± 12.6	39.6 ± 13.0	39.1 ± 14.9	0.201	0.129	0.592
Spatial working memory (SWS)
SWMS	7.6 ± 3.2	7.6 ± 3.6	6.6 ± 3.4	7.0 ± 2.6	6.0 ± 3.4	5.9 ± 2.7	0.134	0.359	0.857
SWMBE	15.2 ± 11.2	16.9 ± 10.2	13.8 ± 13.0	12.3 ± 7.8	11.0 ± 6.3	10.1 ± 11.5	0.343	0.384	0.974

Abbreviations: T × G, interaction effect; RTIMRT, median reaction time; RTIMMT, median movement time; RVPA, sensitivity to the target sequence (0.00–1.00; bad to good), RVPMDL: median response latency; RVPTFS, total number of false alarms; RVPTH, total number of correct responses; SWMS, strategy score; SWMBE, number of errors.

## indicates a significant main time effect; # indicates a significant time effect compared to Stage 3 (p < 0.05).

3.6. Sleep

Total sleep duration, light sleep, and WASO remained unchanged throughout the race and did not differ between groups (all p > 0.05). In contrast, REM sleep decreased during Stage 5 (34 ± 27 min) and Stage 6 (43 ± 29 min) compared with baseline (69 ± 37 min) (p = 0.038) (Table 3). Deep sleep was slightly higher in the CON than in BDO (F(1, 6.01) = 6.42, P _group = 0.044), with an estimated mean difference of 18.76 min (95% CI: 0.65–36.87); however, no group × stage interaction was observed (F(7, 6.20) = 1.42, P _interaction = 0.212; Table 3).

TABLE 3.

Sleep quality and quantity. Data presented as mean ± SD. (CON = 10, F = 2; BDO = 8, F = 1).

	Baseline	Prologue	Stage 1	Stage 2	Stage 3	Stage 4	Stage 5	Stage 6	p‐value
									Time	Group	T x G
Total sleep duration (min)
CON	372 ± 88	358 ± 38	404 ± 99	396 ± 49	363 ± 112	339 ± 104	318 ± 117	416 ± 11	0.110	0.906	0.677
BDO	414 ± 39	369 ± 85	442 ± 32	371 ± 42	339 ± 65	381 ± 36	306 ± 122	331 ± 144
Deep sleep (min)
CON	78 ± 41	72 ± 55	102 ± 60	53 ± 16	86 ± 27	96 ± 37	86 ± 54	63 ± 12	0.362	0.044**	0.212
BDO*	86 ± 26	53 ± 31	74 ± 51	77 ± 62	39 ± 23	32 ± 11	49 ± 35	32 ± 25
Light sleep (min)
CON	225 ± 64	262 ± 58	243 ± 69	273 ± 73	230 ± 89	223 ± 81	189 ± 102	300 ± 36	0.774	0.562	0.317
BDO	230 ± 25	279 ± 53	243 ± 50	246 ± 44	247 ± 45	271 ± 56	294 ± 69	248 ± 147
REM sleep (min)
CON	70 ± 30	44 ± 35	73 ± 51	70 ± 28	58 ± 28	39 ± 36	33 ± 27#	53 ± 30#	0.017##	0.953	0.517
BDO	54 ± 38	58 ± 8	84 ± 34	70 ± 33	60 ± 50	65 ± 27	45 ± 26#	27 ± 22#
WASO (min)
CON	32 ± 15	56 ± 32	31 ± 19	29 ± 11	36 ± 28	61 ± 30	37 ± 29	31 ± 14	0.545	0.381	0.274
BDO	41 ± 12	46 ± 12	47 ± 21	39 ± 7	47 ± 19	40 ± 11	40 ± 20	70 ± 55

Abbreviations: REM: rapid eye motion; T x G: interaction effect; WASO: wakefulness after sleep onset.

** indicates a significant main group effect; * indicates a significant group effect (p < 0.05); ## indicates a significant main time effect; # indicates a significant time effect compared to baseline (p < 0.05).

3.7. Identification of the Experimental Condition

During post‐testing, participants were asked to identify the supplement they believed they had received. Four participants correctly identified their group allocation (three in the BDO group, each reporting 70% certainty, and one in the CON group, reporting 100% certainty). These findings suggest that blinding was maintained throughout the study.

4. Discussion

During multi‐stage mountain bike races, riders aim to optimize their recovery and reduce fatigue symptoms. EK, induced by KME, has shown promise for attenuating physiological stress responses and supporting recovery in controlled environments, with some evidence indicating the potential of BDO (Stalmans et al. 2026). However, it remains unclear whether alternative EK formulations confer similar benefits during real‐world racing. Therefore, we investigated the effects of post‐exercise ketone supplementation on various aspects of fatigue during the Absa Cape Epic. BDO produced mild post‐stage ketosis from Stage 1 onwards, but did not significantly influence markers of cardiovascular strain, nutrition intake, cognitive performance, or sleep compared with placebo. Although the race was associated with clear signs of physiological strain and modest changes in fatigue‐related measures across stages (e.g., reduced exercise heart rate and slower reaction time), BDO‐based EK did not mitigate these responses during the race.

The absence of clear functional benefits may be explained by the magnitude, timing, and formulation of the induced ketosis. Unlike KME, which is rapidly hydrolysed to βHB and BDO, BDO requires hepatic conversion to βHB, resulting in slower and more prolonged hyperketonaemia response and lower peak concentrations (Lowder et al. 2023). KME studies demonstrating recovery and performance benefits with post‐exercise supplementation produced [βHB] between 2 and 5 mM (Poffé et al. 2019; Robberechts et al. 2023; E. Evans et al. 2023; Vandoorne et al. 2017), whereas BDO supplementation in the present study induced only mild ketosis (1.1 ± 0.5 mM) from Stage 1 onwards, consistent with previous BDO studies (Soto et al. 2025; Lowder et al. 2023). We could not measure [βHB] at later time points, which prevented the confirmation of peak concentrations following supplementation. Circulating [βHB] was not measured following the pre‐sleep dose, during which [βHB] likely had been 1–2 mM. Although they used KME, blood [βHB] was slightly higher during sleep compared to post‐postexercise (4.0 ± 0.4 mM vs. 2.8 ± 1.2 mM) (Robberechts et al. 2023). Participants were instructed to refrain from eating for 30 min post‐supplementation, although adherence could not be strictly controlled under race conditions, and most consumed a substantial post‐stage meal (∼92 g carbohydrate and ∼ 45 g protein) shortly after finishing. While a standard meal (31 g carbohydrate, 2.5 g, 13 g protein) has previously been shown not to impact BDO absorption (Soto et al. 2025), the larger meals in the present study may have attenuated the hyperketonaemia response. Supporting this, Furthermore, A recent study using a higher and repeated BDO dose achieved βHB concentrations more than 2 mM and reported a beneficial outcome (Stalmans et al. 2026), further suggesting that the dosing strategy and timing used in the present study may have limited its effectiveness. Collectively, these findings suggest that the formulation, timing and duration of exogenous ketosis are critical determinants of functional outcomes during multi‐stage endurance events.

The ecological context of the present study provides a further explanation for the absence of measurable benefits. Previous studies demonstrating beneficial effects of KME supplementation were conducted under controlled laboratory conditions and involved longer periods of supplementation or intensified training (Poffé et al. 2019; Robberechts, Bekhuis, et al. 2026). In contrast, the study examined BDO supplementation during an 8‐day real—world mountain bike event, where environmental conditions, nutrition, recovery opportunities, and training status could not be standardised. The participants were also well‐trained cyclists specifically prepared for the demands of the multi‐day race, potentially leaving less scope for an additional recovery benefit than in the laboratory studies involving recreationally trained participants or periods of deliberate overreaching. Thus, the lower and potentially delayed ketosis achieved with BDO, combined with the shorter intervention and highly trained ecologically valid race setting, may have limited the potential for supplementation to produce measurable functional benefits.

Another explanation concerns the duration of the race. The race may have been too short to elicit chronic benefits of EK, as a 4% KME‐induced improvement in endurance performance was only observed after 7 weeks of training, with no group difference at the midway point (Robberechts, Bekhuis, et al. 2026). Additionally, KME was only able to blunt the symptoms of overreaching after the third week of excessive training (Poffé et al. 2019). These findings suggest that depending on the exercise intensity, a minimum of 3 weeks of EK is needed to observe chronic benefits. Furthermore, participants in the KME group increased their energy intake in response to the higher training load, which may have contributed to the observed findings. In contrast, the rise in energy intake and carbohydrate consumption was similar between the groups in the present study. These differences suggest that both the shorter intervention period and the higher training status of the present cohort may have limited the potential for BDO supplementation to elicit measurable recovery benefits.

Multi‐day endurance events are associated with physiological responses consistent with functional overreaching, including a reduced exercise heart rate and an elevated resting heart rate (Faulkner et al. 2025). Across the race, riders progressively spent less time in high‐intensity zones and peak heart rate declined by ∼16 bpm by Stage 2, with further drops in both peak and average heart rate at Stage 4, indicating cumulative physiological strain. This decline was not attenuated by BDO supplementation, despite the heart being a major consumer of ketone bodies (Cotter et al. 2013). Resting heart rate increased across the race but did not differ between groups, consistent with previous KME studies reporting no effect on resting HR after long‐term intake (Prins et al. 2026). High environmental temperatures during stages 2 and 3 (31°C–43°C) likely contributed to the observed cardiovascular responses (Cheuvront et al. 2010).

Energy intake increased across the race, particularly carbohydrate intake from Stage 3 onwards, and both groups consumed ∼60 g·h⁻¹ carbohydrates during stages. Although hunger ratings increased across the race and may have driven increased intake, appetite remained similar between the groups. These findings contrast with those of a laboratory (overreaching) study reporting spontaneous increases in energy intake with ketone supplementation (Poffé et al. 2019), suggesting that real‐world race environments may override the potential appetite‐related effects. However, the benefit was only seen in the third week (Poffé et al. 2019), suggesting that any appetite‐modulating effects of ketone supplementation may require a longer intervention than the duration of the present race.

The total stress scores (RESTQ‐76 Sport) at the end of the race were slightly higher than those observed following an 8‐week endurance training programme (Robberechts, Bekhuis, et al. 2026), but lower than the scores observed following an overtraining study (Poffé et al. 2019). Despite the physiological demands of the race, BDO supplementation did not improve subjective recovery or attenuate perceived stress. Our findings are consistent with previous reports that KME supplementation did not alter stress‐recovery state during training and overtraining interventions, despite differences in study design and exercise context (Poffé et al. 2019; Robberechts, Bekhuis, et al. 2026). Similarly, evidence that EKS attenuates exercise‐induced muscle soreness remains limited. In the present study, BDO did not reduce muscle soreness scores although soreness remained relatively low throughout the race. Whether the higher and more sustained β‐hydroxybutyrate concentrations typically achieved with ketone monoesters would produce different perceptual responses during prolonged stage racing remains to be determined.

Cognitive function remained stable at mid‐race, but reaction time slowed in both groups by Stage 6. This decline coincided with reductions in perceived recovery and mental alertness. Although these responses may be related, the present study was not designed to determine whether changes in perceived recovery contributed to the observed decline in cognitive performance. These findings differ from previous studies reporting preserved cognition during exercise with KME supplementation (M. Evans and Egan 2018; Poffé et al. 2023; Mosquera‐Lopez et al. 2026). These differences may reflect the lower magnitude of ketosis achieved with the present BDO protocol, as discussed above. In the present study, cognitive testing occurred immediately after supplementation when blood [βHB] remained < 1 mM, and the use of a ketone precursor may further explain the absence of cognitive benefits. Overall, these findings suggest that mild ketosis induced by post‐exercise BDO supplementation is insufficient to preserve cognitive performance during prolonged stage racing.

Sleep quantity and quality remained largely stable across the race. Apart from a reduction in REM sleep during the later stages of the race and a modest overall difference in deep sleep between groups, no meaningful changes were observed in objective sleep parameters. BDO supplementation did not affect any objective sleep outcomes. Previous evidence suggests that exogenous ketone supplementation may improve sleep primarily under conditions in which sleep is compromised by strenuous exercise or sleep restriction. In the present study, however, objective sleep quantity and quality were generally well maintained, leaving limited scope for BDO supplementation to confer additional benefits.

5. Limitations

The field‐based nature of this study presented several logistical challenges. Restricted access to the finish area occasionally delayed supplement administration and cognitive testing following stages 3 and 6 was conducted in the noisy finish section area rather than in a controlled environment. Although this may have increased variability in the cognitive outcomes, it also reflects the conditions experienced during real‐world competition. Participants were instructed to refrain from eating for 30 min after supplementation; however, adherence could not be strictly controlled because post‐stage meals were readily available. This may have introduced variability in BDO absorption and contributed to the modest ketosis observed. Dietary intake was assessed using 24‐h recalls supported by meal photographs and food wrappers although some items may have been missed. Nevertheless, variability in dietary intake was similar between the groups. The BDO group consumed an additional 175 kcal·day⁻¹ from the supplement. Given that daily energy expenditure during multi‐stage cycling events often exceeds 4000–6000 kcal, it is unlikely that the supplement's energy content materially influenced recovery outcomes. Sleep was assessed using a validated wrist‐worn device rather than polysomnography; consequently, subtle changes in sleep architecture may not have been detected. However, the reliability of the device was reported by the company, and the data have not been published; therefore, these findings should be interpreted with caution. Given the small sample size, our study may have been underpowered to detect effects for secondary outcomes; therefore, null findings should be interpreted as exploratory and should not necessarily be taken as evidence of no effect. Finally, only 3 women (1 in BDO) completed the study, limiting the generalisability of the findings to female cyclists.

6. Conclusion

The present study demonstrates that post‐exercise and pre‐sleep supplementation with the ketone precursor BDO induced mild ketosis throughout the Absa Cape Epic but did not meaningfully influence cardiovascular responses, subjective recovery, cognitive function, sleep, or nutritional intake compared with placebo. These findings suggest that the magnitude of ketosis achieved with this formulation and dosing strategy was insufficient to confer measurable recovery benefits during a multi‐day mountain bike stage race. Future research should determine whether alternative ketone formulations, higher and more sustained circulating [βHB], or different dosing strategies improve recovery during prolonged endurance events. By evaluating ketone supplementation under the demands of real‐world competition, the present study provides important ecologically valid evidence that the level of ketosis achieved with post‐exercise and pre‐sleep BDO supplementation did not translate into measurable functional recovery benefits.

Author Contributions

L.E., C.P., K.K. and E.T.: Concept and design of the study. L.E.: Data collection and data analysis. L.E. and E.T.: Interpretation of data and manuscript drafting. All authors critically revised the manuscript and approved the final version.

Funding

Chiel Poffé has a FWO senior postdoc grant (12B0E24 N).

Ethics Statement

The study was approved by the Stellenbosch University Health Research Ethics Committee (S22/08/156).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.