Substrate utilization and durability during prolonged intermittent exercise in elite road cyclists

Twelve elite male road cyclists were recruited for the study (age, 23 years [21;25]; body mass, 71.5 kg [66.7;76.8]; height, 181 cm [178;185]; \(\dot{V}\)O₂peak, 73.6 ml kg⁻¹ min⁻¹ [71.2;76.0]; LT₁, 4.2 W kg⁻¹ [3.9;4.5]). The cohort consisted of cyclists representing a UCI continental team (n = 7) and cyclists competing in national and international races on the elite level (n = 5). The participants were fully informed of any potential risk associated with the experiments before obtaining their consents and the experiments were conducted in accordance with the standards of the Declaration of Helsinki. The present study utilized procedures used during the ordinary training of the cyclists. As such, the present protocol did not require ethical approval according to the local ethical committee.

The study was conducted in the final phase of the preparation period preceding the competitive part of the season (i.e., March and April). The cyclists visited the laboratory twice in the morning, separated by three to seven days (Fig. 1). The first visit involved preliminary tests including a submaximal step-test to obtain measurements of substrate utilization and gross efficiency, followed by a maximal 6 min-TT (i.e., fresh condition). The second visit (test day) consisted of a 4-h intermittent cycling protocol with a high CHO intake and repeated measurements of substrate utilization and peak power output (Fig. 1). Durability was examined through repeated 6 s peak power test during and after the 4 h of cycling as well as changes in maximal 6 min-TT conducted after the cycling protocol (i.e., fatigued state).

The participants arrived at the laboratory after fasting for four hours and after refraining from high intensity exercise for 24 h. First, height and body weight were obtained and subsequently, a graded submaximal test on a cycling ergometer was initiated [Schoberer Rad Messtechnik (SRM), 117 GmbH Julich, Germany]. The test involved a 10-min period of rest, followed by 4-min steps starting at 30W, and with an increase of 40W every third minute until the RER value remained above 1.00 for a full minute. \(\dot{V}\)O₂ and \(\dot{V}\)CO₂ were measured throughout the test and used to calculate substrate utilization and GE at each workload (see details below). During the final 30 s of each workload, capillary blood was drawn from a fingertip to determine blood lactate concetration.

After completing the graded submaximal test, the participants received an energy gel containing 20 g of CHO (OTE, Leeds, UK) and continued cycling for 15 min at 100W. Then a maximal 6 min-TT (i.e., fresh condition) was conducted at a self-selected cadence and with verbal encouragement throughout the test. Participants were instructed to pace themselves to achieve the highest possible average power output. \(\dot{V}\)O₂ and heart rate (HR) were recorded during the test. During each test, \(\dot{V}\)O₂peak (see analytical procedures) and average power output (MPO_6 min) was measured. HRpeak was defined as the highest measure of HR in each of the two MPO_6 min tests. The SRM ergometer was calibrated before each test.

The prolonged intermittent cycling protocol was performed on the same SRM ergometer as described above, with calibration performed before each test. The cyclists refrained from high intensity exercise during the 24 h preceding the visits. The participants initiated the protocol with a 10-min warm-up, which included 5 min at 25% followed by 5 min at 50% of MPO_6 min. The 4-h protocol was then started and comprised cycling at 50% of MPO_6 min interspersed with intervals of 1 min at 120% of MPO_6 min every 30 min (seven all in all) and maximal 6-s peak power tests every 60 min (five in total) (Fig. 1). After completing the 4 h protocol, subjects cycled for two minutes at 50% of MPO_6 min before the maximal 6 min-TT was conducted (i.e., same as preliminary test day but in a fatigued condition). The protocol was carried out at a self-selected cadence, except for the 6-s sprints, which were performed at a cadence of 100 RPM. HR was monitored throughout the protocol.

Except for the first sprint, all 6-s sprints were performed after 28 min of cycling at 50% of MPO_6 min. From each of the 6-s sprints, the highest 3-s running average was defined as PPO. Before each sprint, a 3-s countdown was initiated, and verbal encouragement was provided throughout the sprint. Participants were instructed to exert their maximal effort during the entire sprint while remaining seated. Every second 1-min interval was performed after 30–32 min of cycling at 50% of MPO_6 min, while the remaining 1-min intervals were performed after 30 min at 50% of MPO_6 min, a 6-s sprint and an additional 2 min at 50% of MPO_6 min (Fig. 1). Substrate utilization was measured for 10 min at 50% of MPO_6 min every hour, starting after 20 min, and then every hour thereafter (see below). Participants were weighed before and after the entire protocol to estimate the degree of dehydration.

Prior to the prolonged cycling protocol, participants were instructed to consume a CHO-enriched diet and to treat the nutritional preparation as they would before a real competition. Throughout the protocol, the participants were supplied with 100 g of CHO per hour, achieved by consuming an energy gel containing 20 g of CHO (OTE, Leeds, UK) and a 500 mL beverage containing 80 g of CHO (OTE, Leeds, UK) every hour (Fig. 1). The gels were ingested after every second 1-min interval (i.e., every hour), while consumption of the beverage was distributed across each 60-min period to prevent gastrointestinal issues (Fig. 1). In addition, the cyclists were allowed to consume water as desired. Importantly, all participants were accustomed to consume high amounts of carbohydrates during training and competition and none of the participants followed a “low carbohydrate − high fat” or other specific dietary strategies.

During each step of the graded submaximal test and during the 4-h cycling protocol, the rates of fat oxidation and CHO oxidation were calculated using the stoichiometric equations developed by Jeukendrup and Wallis: fat oxidation rate = (1.695 × \(\dot{V}\)O₂) − (1.701 × \(\dot{V}\)CO₂); CHO oxidation rate = (4.210 × \(\dot{V}\)CO₂) − (2.962 × \(\dot{V}\)O₂), under the assumption that urinary nitrogen excretion was negligible (Jeukendrup and Wallis 2005). Maximal fat oxidation (MFO) was defined as the highest 1-min measurement recorded during the graded submaximal test. Unfortunately, due to a leaky valve, substrate utilization data were excluded for one participant. The first lactate threshold (LT₁) was defined as the workload that resulted in the blood lactate concentration being 1 mmol L⁻¹ blood higher than at rest. The workload corresponding to LT₁ was identified by the workload-lactate relationship, by establishing a straight line through the lactate concentrations surrounding LT₁.

Gross efficiency was calculated as the ratio between the external load (i.e., power output) and the total energy expenditure (EE), calculated from \(\dot{V}\)O₂ and \(\dot{V}\)CO₂. For the calculation of total EE, the following equation was used: EE = (0.550 × \(\dot{V}\)CO₂) + (4.471 × \(\dot{V}\)O₂) (Jeukendrup and Wallis 2005).

Average \(\dot{V}\)O₂ in the last 60 s of each of the 9–10 submaximal steps (i.e., after attaining steady state \(\dot{V}\)O₂) of the graded submaximal test were used to establish the individual power-\(\dot{V}\)O₂ relationships. The anaerobic energy contribution was calculated based on the principle of determination of the accumulated oxygen deficit as described by Medbø et al. (1988). In brief, power output during the 6 min-TT’s in fresh and fatigued conditions was extracted in 10 s intervals and based on the power-\(\dot{V}\)O₂ relationship the corresponding \(\dot{V}\)O₂-demand was calculated. The oxygen deficit in each 10 s interval was then calculated as the difference between the calculated O₂ demand and the measured O₂ uptake and summarized for the whole 6-min exercise period.

One-way ANOVA tests with repeated measures were used to investigate changes during the 4-h intermittent cycling protocol and a paired Students t-test was employed to compare results from the two maximal 6 min-TTs. The strengths of relationships between variables were assessed by Pearson´s correlation coefficient and [confidence intervals (CI)]. Although power-to-weight ratio offers a fair comparison of cycling performance between individuals, some correlation analyses were performed using absolute power output to avoid inter-dependency between the dependent and independent variables by normalizing both variables to body mass. Results are presented as mean and 95% [CI]. P < 0.05 were considered significant, while P = 0.05–0.10 were considered as tendencies. All statistical analyses and figures were produced using GraphPad Prism 6.07 (GraphPad Software, LLC, San Diego, CA 92108 USA).

The average power output was 208 W [197;218] (2.9 W kg⁻¹ [2.7;3.1]) during the steady portion of the cycling protocol (50% MPO_6 min) and 498 W [472;523] (7.0 W kg⁻¹ [6.6;7.4]) during the 1-min intervals (120% MPO_6 min). On average, the intensity at 50% MPO_6 min corresponded to 70% of LT₁ [68;72]. Since \(\dot{V}\)O₂peak declined during the cycling protocol (see results later), the relative workload at 50% MPO_6 min increased from 54% [53;56] of pre \(\dot{V}\)O₂peak to 59% [57;62] of post \(\dot{V}\)O₂peak from the first to the fourth hour of exercise (mean effect: + 5% [2:8], P = 0.004).

HR increased from an average of 136 beats min⁻¹ [130;142] (71% [69;73] of Pre HRpeak) during the first hour of exercise to 146 beats min⁻¹ [138;153] (79% [75;82] of Post HRpeak) in the 4th hour of exercise (mean effect: + 7.5%-point [5:10], P < 0.0001) (2nd hour: 138 beats min⁻¹ [131;144]; 3rd hour: 143 beats min⁻¹ [135;150]). Body weight was slightly reduced following the cycling protocol (71.8 kg [66.7;76;8] to 71.3 kg [66.4;76.2], mean effect: − 0.4 kg [− 0.1; − 0.7], P = 0.003).

The substrate utilization changed during the 4 h of intermittent cycling, as indicated by a decline in the RER (1st hour: 0.94 [0.92;0.96]; 2nd hour: 0.93 [0.91;0.95]: 3rd hour: 0.92 [0.91;0.93]; 4th hour: 0.90 [0.89;0.92], overall time-effect: P = 0.01) (Fig. 2A). A post hoc analysis revealed that the RER was significantly reduced during the 3rd and 4th hours of exercise compared to the 1st hour (mean effect 1st to 4th hour: − 0.04 [− 0.01; − 0.06], P = 0.01). Accordingly, the FatOx rate increased during the cycling protocol (1st hour: 0.28 g min⁻¹ [0.16;0.39]; 2nd hour: 0.33 g min⁻¹ [0.23;0.43]; 3rd hour: 0.39 g min⁻¹ [0.34;0.44]; 4th hour: 0.47 g min⁻¹ [0.38;0.56], overall time effect: P = 0.01). In addition, a post hoc analysis revealed that the FatOx rate was significantly elevated during the 3rd and 4th hour of exercise in comparison to the 1st hour (mean effect 1st to 4th hour: + 0.20 g min⁻¹ [0.05;0.34], P = 0.01) (Fig. 2B). In line with this, the CHO oxidation was gradually reduced during the cycling protocol (1st hour: 2.9 g min⁻¹ [2.5;3.2]; 2nd hour: 2.7 g min⁻¹ [2.4;3.0]; 3rd hour: 2.6 g min⁻¹ [2.4;2.8]; 4th hour: 2.4 g min⁻¹ [2.2;2.6], overall time-effect: P = 0.002) (mean effect 1st to 4th hour: − 0.45 g min⁻¹ [− 0,19: − 0.72], P = 0.003) (Fig. 2C).

There was no significant change in GE during the cycling protocol, (1st to 4th hour: 20.9% [20.2;21.5] to 20.8% [20.3;21.3], mean effect: 0.0%-point [0.0;0.0], P = 0.75 (Fig. 2D) or \(\dot{V}\)O₂ (1st to 4th hour: 2.85 L O₂ min⁻¹ [2.68;3.01] to 2.88 L O₂ min⁻¹ [2.70;3.06], mean effect: 0.03 L O₂ min⁻¹ [− 0.05:0.11], P = 0.43).

Notably, there was a strong correlation (r = 0.76 [0.29;0.93], P = 0.007) between MFO in the fresh condition and the average FatOx during the prolonged cycling protocol, indicating that interindividual differences in fat oxidative capabilities were maintained despite a high CHO intake during the prolonged cycling protocol (Fig. 2E).

MPO_6 min was significantly reduced from 415 W [395;436] to 375 W [349;402] following the prolonged intermittent cycling protocol (mean effect: − 40 W [− 17; − 62], P = 0.003) (Fig. 3A). Furthermore, during the MPO_6 min-test in fatigued compared to fresh condition there was a 7% [3;11] decrease in \(\dot{V}\)O₂peak (5.25 L O₂ min⁻¹ [4.90;5.60] to 4.87 L O₂ min⁻¹ [4.57;5.16], mean effect: − 0.38 L O₂ min⁻¹ [− 0.16: − 0.61], P = 0.003) and a 3% [1;6] decrease in HRpeak (191 beats min⁻¹ [186;196] to 185 beats min⁻¹ [179;192], mean effect: − 6 beats min⁻¹ [− 2: − 9], P = 0.003) (Fig. 3B, C). While the average \(\dot{V}\)O₂ during the 6-min test was not different between fresh and fatigued conditions (4.51 L O₂ min⁻¹ [4.27;4.75] vs. 4.53 L O₂ min⁻¹ [4.26;4.80], mean effect: 0.02 L O₂ min⁻¹ [− 0.17:0.21], P = 0.88) (Fig. 3D) there were clear differences during the tests. Thus, when divided into 1-min intervals, a divergence in \(\dot{V}\)O₂kinetics appeared between conditions (Fig. 3D). Thus, \(\dot{V}\)O₂ was higher during the first minute in the fatigued state (2.92 L O₂ min⁻¹ [2.73;3.11] vs. 3.87 L O₂ min⁻¹ [3.64;4.11], mean effect: + 0.95 L O₂ min⁻¹ [0.78:1.12], P < 0.001), but lower during the 5th (5.05 L O₂ min⁻¹ [4.78;5.32] vs. 4.68 L O₂ min⁻¹ [4.37;4.99], mean effect: − 0.37 L O₂ min⁻¹ [− 0.11: − 0.63], P = 0.01) and 6th minute (5.12 L O₂ min⁻¹ vs. 4.73 L O₂ min⁻¹ [4.43;5.04], mean effect: − 0.4 L O₂ min⁻¹ [− 0.1: − 0.6], P = 0.006).

Calculations of the anaerobic energy contribution showed a reduction from 8.6% [7.5;9.7] in the fresh condition to 3.0% [2.3;3.8] (P < 0.0001) in the fatigued condition, meaning that the aerobic energy contribution increased from 91.4% [90.3;92.5] to 97.0% [96.7;97.2] (P < 0.0001). In absolute terms, the accumulated O₂-deficit during the 6 min-TT was reduced from 2.8 L min⁻¹ [2.3;3.3] to 1.0 L min⁻¹ [0.6;1.3] (P < 0.0001) after the cycling protocol.

Regarding PPO, there was a tendency towards an overall reduction during the 4 h of cycling (Fresh: 1064 W [994;1134]; 1st hour: 1048 W [977;1119]; 2nd hour: 1035 W [962;1108]; 3rd hour: 1024 W [949;1099]; 4th hour: 998 W [949;1099], overall time effect: P = 0.09). A post hos analysis revealed a significant reduction in PPO after 4 h of cycling compared to pre-exercise (mean effect: − 67 W [− 4; − 129], P = 0.04) (Fig. 4A, B).

We next investigated if the substrate utilization correlated with changes in MPO_6 min and PPO during the cycling protocol. The average (1st to 4th hour) or the time-specific FatOx rate at 1st, 2nd or 3rd hour did not show any strong correlations with ΔMPO_6 min and the data are most compatible with poor correlations (Fig. 5A–D). However, during the 4th hour a fair correlation was found with a confidence interval ranging from a poor to a good correlation (Fig. 5E), showing that higher FatOx rates were associated with larger declines in performance in the last portion of the cycling protocol (i.e., after 3 h of exercise).

The CHO oxidation during the cycling protocol was not associated with ΔMPO_6 min (average: r = 0.03, [− 0.58:0.62], P = 0.92; 1st hour: r = 0.17 [− 0.48:0.70], P = 0.62; 2nd hour: r = 0.04 [− 0.57:0.63], P = 0.90; 3rd hour: r = 0.12 CI [− 0.52:0.67], P = 0.73; 4th hour: r = − 0.36 [− 0.79:0.31], P = 0.28). Additionally, ΔPPO was not related to neither the average CHO oxidation during exercise nor the CHO oxidation during the first three hours of cycling (average: r = − 0.12 [− 0.57:0.62], P = 0.73; 1st hour: r = 0.11 [− 0.53:0.66], P = 0.75; 2nd hour: r = − 0.05 [− 0.63:0.57], P = 0.89; 3rd hour: r = − 0.14 [− 0.68:0.50], P = 0.68. However, CHO oxidation rates were negatively associated with ΔPPO during the 4th hour (i.e., high CHO oxidation associated with low reduction in PPO) (r = − 0.59 [− 0.88:0.00], P = 0.05).

Under the present conditions, correlation analyses unveiled that neither of the laboratory-based outcomes obtained in a fresh condition (i.e., MPO_6 min, \(\dot{V}\)O_2max, LT₁, MFO, and GE_230 W) were correlated with ΔMPO_6 min or ΔPPO (Tables 1, 2).

Besides the ability to withstand relative reductions in performance, the ranking in road cycling races is largely influenced by the absolute exercise capacity (i.e., W or W kg⁻¹) following prolonged accumulated intermittent exercise. Correlation analyses unveiled associations tending to statistical significance between MPO_6 min in the fatigued condition and MPO_6 min (r = 0.57 [0.00;0.86], P = 0.05) and LT₁ (r = 0.54 [− 0.04;0.85], P = 0.07) in the fresh condition (Fig. 6A, B), while no significant correlations were observed between MPO_6 min in the fatigued state and \(\dot{V}\)O₂peak (r = 0.31 [− 0.32;0.75], P = 0.33) and GE at 230 W (r = 0.37 [− 0.26;0.78], P = 0.24) in the fresh state (Fig. 6C, D). Finally, no significant correlation was observed between MFO measured in the fresh condition and performance during MPO_6 min in the fatigued state (r = − 0.45 [− 0.81;0.17], P = 0.14) (Fig. 6E).

Here, we designed a controlled prolonged intermittent cycling protocol simulating some of the demands of road cycling races, and by this protocol fatigue was induced, as demonstrated by 9% and 6% impairments in a 6-min TT and PPO, respectively. These reductions align with findings from previous studies involving elite cyclists subjected to prolonged exercise (Spragg et al. 2023; Valenzuela et al. 2023; Gejl et al. 2014). Notably, \(\dot{V}\)O₂peak was also reduced in the fatigued condition (Fig. 3D), and as expected this reduction displayed a strong correlation with the concurrent reduction in MPO_6 min (r = 0.79, P = 0.003) (Fig. 3D). Moreover, the reduction in \(\dot{V}\)O₂peak was accompanied by a reduction in HRpeak of 6 ± 5 bpm indicating that the reductions in \(\dot{V}\)O₂peak and MPO_6 min could, at least partly, be ascribed to a reduction in O₂ delivery. Unfortunately, the mechanisms underlying the reductions in \(\dot{V}\)O₂peak and HRpeak cannot be elucidated from the present study. While \(\dot{V}\)O₂peak was reduced, the average \(\dot{V}\)O₂ during the 6 min time-trials showed no alteration after the cycling protocol. Thus, given the decrease in average power, the reduction in MPO_6 min was likely explained by the observed reduction in the anaerobic energy contribution from 9.6 to 3.0% (change in accumulated O₂-deficit: 2.8 L min⁻¹ to 1.0 L min⁻¹). Another measure of performance at high exercise intensities is W´ (i.e., the amount of work that can be done above the CP), and reductions in W´ have previously been observed following prolonged intermittent exercise in both professional cyclists (Spragg et al. 2022) and recreationally active individuals (Bitel et al. 2023). Unfortunately, a robust measure of W´ could not be determined from the test used in the present study (i.e., 6 s PPO and 6 min-TT). Taken together, prolonged intermittent exercise apparently compromise the anaerobic capacity and W´, and eventually performance in elite cyclists.

The decline in \(\dot{V}\)O₂peak during exercise implied that the relative metabolic load at 50% MPO_6 min was increased from 54 to 59% of \(\dot{V}\)O₂peak from the first to the fourth hour of exercise. Consequently, the relative exercise intensity was at a higher level within the moderate intensity domain and for coaches and athletes such changes are likely important to consider to acquire a valid estimate of the training and competition load. In addition, prior research has demonstrated a drift in \(\dot{V}\)O₂ (i.e., \(\dot{V}\)O₂ slow component) during prolonged submaximal exercise, likely attributed to changes in substrate utilization and muscle fiber recruitment patterns, resulting in a loss of efficiency (Burke et al. 2017; Xu and Montgomery 1995; Passfield and Doust 2000; Hagberg et al. 1978; Pringle et al. 2003). In this regard, a recent study in endurance athletes demonstrated that prolonged moderate intensity exercise reduced the power output at the boundary between the moderate and heavy intensity domain due to reductions in GE and metabolic energy expenditure at this transition (Stevenson et al. 2022). Indeed, such drifts in \(\dot{V}\)O₂ or GE have previously been shown to be diminished in endurance trained individuals (Casaburi et al. 1987; Russell et al. 2002; MacDougall et al. 2022), and present data is in line with this, showing no changes in \(\dot{V}\)O₂ or GE despite a gradual shift towards a greater reliance on fat as an energy source.

The group of elite and professional cyclists exhibited a wide range of relative changes of MPO_6 min following the cycling protocol (− 31% to + 1%) and we aimed to investigate the relation between these changes and changes in substrate utilization during the cycling protocol, as well as MFO in the fresh condition. Considering the crucial role of muscle glycogen in sustaining high-intensity exercise performance (Gollnick et al. 1974; Balsom et al. 1999; Rockwell et al. 2003), the FatOx rate may have a substantial impact on the preservation of muscle glycogen. As muscle- and liver glycogen was not estimated directly in the present study, the utilization of glycogen can therefore only be estimated. Based on the average CHO oxidation of 2.63 ± 0.37 g min⁻¹ at 50% of MPO_6 min, and for simplicity neglecting the increased CHO oxidation during the high intensity intervals, the cumulative CHO oxidation was ~ 630 g during the cycling protocol. Assuming that 90 g of the ingested CHO was oxidized per hour (Podlogar et al. 2022), then approximately 270 g of CHO was derived from liver and muscle glycogen. As previously seen in less trained individuals (Fell et al. 2021), these estimations indicate that despite a high CHO intake there is a substantial use of muscle glycogen in active muscles of elite cyclists during this type of exercise, and consequently mobilization and oxidation of other fuel sources may eventually becomes important. Nonetheless, when employing correlation analysis, we could not demonstrate strong associations between measures of fat oxidation during exercise and durability in the present scenario. Moreover, MFO in the fresh and fasted condition (i.e., 4 h of fasting) did not display a correlation with durability.

As illustrated in Fig. 2B, the FatOx rate was relatively low for most cyclists during the first two hours of cycling (i.e., ≤ 0.4 g min⁻¹ and < 30% of total energy expenditure in 9 riders), meaning that the potential preservation of muscle glycogen during this phase was limited. In addition, a comparison of the cyclist with the highest average FatOx rate (0.53 g min⁻¹) and the one with the lowest average FatOx rate (0.23 g min⁻¹), and identical \(\dot{V}\)O₂ at 50% of MPO_6 min (~ 3.1 L min⁻¹), revealed a total difference in CHO oxidation of ~ 110 g and a difference of 18 g in the fourth hour of cycling (avg. total CHO oxidation: 632 g [573;691]. Therefore, it is possible that modest inter-individual variations in the FatOx rates may have prevented the detection of associations between FatOx rates and durability.

Only a limited number of studies have investigated predictors of durability in trained endurance athletes and our findings align with previous research failing to establish associations between declines in time-trial performance after prolonged cycling exercise and laboratory-based metrics measured under fresh conditions (e.g., \(\dot{V}\)O₂max, VT₁ and VT₂) (Valenzuela et al. 2023; Passfield and Doust 2000). Also, in a recent study of professional cyclists (Spragg et al. 2023) no association was found between FatOx capacity at 200 W and 300 W in a fresh condition and durability (r = 0.18, 95% CI [− 51:73] and r = − 0.08, 95% CI [− 68:58], respectively), which is consistent with findings of the present study (Table 1). However, they identified robust correlations between durability (i.e., relative reductions in CP) and laboratory-based measures obtained in a fresh condition (e.g., \(\dot{V}\)O₂max, VT₁, VT₂ and GE), which were not evident in the present study. In this regard, it should be noted that confidence intervals were wide in both studies, partly due to relatively low sample sizes, and there was a large overlap between confidence intervals of the correlations between durability and laboratory-based measures. Consequently, part of the observed differences between studies were probably due to uncertain estimates, emphasizing the need for larger sample sizes to obtain more precise estimates regarding the predictability of durability from laboratory-based metrics measured under fresh conditions.

Since the participants in the study by Spragg and colleagues were somewhat comparable to those in the present study (i.e., elite cyclists with similar \(\dot{V}\)O₂max), the contradictory findings could also be due to differences in methodological approaches, i.e., different duration and intensity of the fatiguing cycling protocols as well as differences in the CHO intake during exercise. We aimed to use a realistic protocol in terms of the duration of road cycling races, while the protocol used by Spragg and colleagues were somewhat shorter (i.e., 4 h vs. 2 h 12 min). On the other hand, the volume of high-intensity cycling was much higher in the latter study. Such differences in fatiguing protocols likely affect the contribution from mechanisms underlying fatigue and durability (e.g., glycogen utilization, fiber contractile properties, and central factors) (Jensen et al. 2020; Hvid et al. 2013; Millet and Lepers 2004), and thereby also laboratory-based predictors of durability. Furthermore, the determinants of durability may vary depending on the consumption of CHO during exercise. While it is generally accepted that CHO intake during prolonged exercise prevents reductions in performance by preserving liver glycogen and blood glucose levels and perhaps also muscle glycogen, the optimal dose of CHO is debated (Podlogar and Wallis 2022). Current recommendations do not advise a CHO intake above 90–100 g h⁻¹ since it seems that higher doses do not enhance endurance performance further (Thomas et al. 2016; Stellingwerff and Cox 2014). In the present study we adhered to these recommendations by providing 100 g CHO h⁻¹. Importantly, such a high CHO intake suppresses the FatOx rate in endurance athletes (Gejl et al. 2018; Fell et al. 2021), which may minimize the importance of a high FatOx capacity for performance and durability as long as sufficient glucose sources are present. Nonetheless, there was a significant correlation between MFO in the fasted state and the FatOx rate during the first hour of the cycling protocol, indicating a carry-over effect despite the high CHO intake. However, under the current conditions this was not positively associated with durability. The extent to which the FatOx capacity determines durability under conditions where the CHO availability may become limited should be further examined in the future. In this regard, we suggest that future studies in elite endurance athletes examine the interaction between carbohydrate availability and the FatOx rate and its importance for durability under various conditions (i.e., different types of endurance exercise and fueling strategies).

Based on the present findings, and most other cross-sectional studies, durability seems unrelated to traditional laboratory measures of endurance performance. In line, a recent study in untrained individuals showed that training induced changes in \(\dot{V}\)O₂max and power at LT₁ following 10 weeks of training were not associated with changes in durability (Matomäki et al. 2023). Future research in elite endurance athletes should investigate changes in durability in response to different training strategies and the extent to which changes in durability can be explained by changes in traditional measures of endurance performance.