Combining heat stress and moderate hypoxia reduces cycling time to exhaustion without modifying neuromuscular fatigue characteristics

Eleven unacclimatized, physically active males (mean ± SD; age 29.0 ± 1.5 years, body weight 78.4 ± 2.0 kg, stature 179 ± 8 cm, training frequency 6.3 ± 0.7 h week⁻¹) volunteered to participate in the study. Participants were instructed to avoid caffeine for 12 h and strenuous exercise for 48 h preceding each trial. This research project conformed to the standards set by the latest revision of the Declaration of Helsinki and was approved by the scientific committee of the Research and Education Centre and by the Ethics Committee of Aspetar—Qatar Orthopaedic and Sports Medicine Hospital. All subjects gave written informed consent prior to the commencement of the study, once they have received the explanation about the experimental procedures, associated risks, and potential benefits of participation.

Each participant completed a familiarization session and four experimental trials during which they cycled to exhaustion in four different environmental conditions [temperate/sea level (control), hot/sea level (hot), temperate/moderate altitude (hypoxia) and hot/moderate altitude (hot + hypoxia)]. Temperate and hot conditions were 22 °C/30 % rH and 35 °C/40 % rH, respectively. Sea level (FiO₂ 0.21) and moderate altitude (FiO₂ 0.15) corresponded to a simulated altitude of ~0 and ~2,500 m, respectively. The trials were randomized, separated by at least 5–7 days, and performed at the same time of the day (±2 h) with subjects wearing shorts and t-shirts. As depicted in Fig. 1a, each experimental session was conducted as follows: (1) rest in a seated position for 30 min inside the climatic chamber, while participants were instrumented; (2) 10-min warm-up on a computer-controlled electrically braked cycle ergometer (Excalibur Sport, Lode, Netherlands) at 75 W (pedalling rate 70–80 rpm); (3) neuromuscular tests (pre-tests; ~10 min); (4) 5-min rest; (5) time trial to the limit of exhaustion at a fixed workload, equal to 66 % of the power output associated with their maximal oxygen uptake (180.5 ± 23.6 W; pedalling rate 80–90 rpm); (6) 5-min recovery including 90 s of low-intensity (50 W, 60–70 rpm) pedalling followed by 3-min rest (time for the participants to be seated on the test ergometer) and (7) neuromuscular tests (post-tests; ~10 min). Cycling exhaustive bouts began exactly 20 min after the end of the warm-up and was preceded by a 2-min low-intensity phase at the same work rate as the one used for warm-up. Constant visual and vocal feedbacks were given to the subjects to avoid variations in pedal cadence. Exercise was terminated when pedal cadence dropped below 60 rpm for >5 s (exhaustion). The participants were unaware of the experimental hypotheses and naïve to the purpose of the study. All tests (cycling and neuromuscular assessment) were performed in an environmental chamber (Tescor, Warminster, PA, USA).

Thermoregulatory (Tcore and Tskin), physiological (HR and SpO₂) and perceptual (RPE) responses were determined at the onset of exercise (1 min after the start), every 5 min from the fifth minute until exhaustion, and at the point of volitional exhaustion. The rate of change in these variables was estimated using linear regressions from exercise data collected at 5-min interval.

During the MVCs, torque production was recorded during a 1-s plateau prior to delivering the motor nerve stimulation. Mechanical responses to resting twitches were analysed for maximal amplitude (Pt). Voluntary activation was defined as follows: (VA, %) = [(1 − (superimposed twitch/potentiated twitch)] × 100). For voluntary contractions, the soleus or tibialis anterior root mean square (RMS) was calculated over the 1-s period of the maximal torque. Raw soleus RMS values were normalized to M _SUP (RMS/M _SUP ratio). The following parameters were determined from the electrically evoked contractions: peak-to-peak amplitude of maximal H-reflex (at rest: H _MAX; during submaximal and MVC: H _SUP), maximal M-wave (at rest: M _MAX; during submaximal and MVC: M _SUP), and V-wave (during MVC: V _SUP). The H _MAX/M _MAX, H _SUP/M _SUP (during submaximal and MVC) and V _SUP/M _SUP ratios were then calculated to assess the proportion of motor units activated by the Ia afferents (Duclay and Martin 2005). For all MVCs or electrically evoked waves and associated twitches, the values of three or six trials, respectively, were averaged for subsequent analysis.

Data are presented as mean ± SD. Normal distribution of the data was tested using the Kolmogorov–Smirnov test. Sphericity (homogeneity of covariance) was verified by the Mauchly’s test. When the assumption of sphericity was not met, the significance of F-ratios was adjusted according to the Greenhouse–Geisser procedure. A three-way analysis of variance (time × temperature × altitude) with repeated measures was used to compare exercise data. Neuromuscular variables were compared by a three-way analysis of variance (exercise × temperature × altitude) for repeated measures. Statistical analyses were undertaken using the SPSS statistical package (version 15.0, SPSS, Chicago, IL, USA). Statistical significance was accepted at P < 0.05.

Temperature (P < 0.05) and altitude (P < 0.05) both reduced time to exhaustion (Fig. 2), while there was also a significant interaction effect (P < 0.05) between these two factors. Hot + hypoxia was shorter than hot or hypoxia alone (P < 0.05). There was an antagonistic effect of either a hot (−22 ± 4 % from control to hot vs. −11 ± 3 % from hypoxia to hot + hypoxia; P < 0.05) or an altitude (−24 ± 6 % from control to hypoxia vs. −13 ± 5 % from hot to hot + hypoxia; P < 0.05) exposure when the other stressor was already present.

Significant reductions occurred from pre- to post-exercise for MVC torque, VA and Pt (−8.6 ± 1.5, −3.9 ± 1.5 and −6.1 ± 0.8 %, respectively; all trials compounded; P < 0.05), independently of the environmental conditions (P < 0.05; Fig. 4). Table 2 presents EMG-related variables determined during the neuromuscular assessment performed before and +5 min after the fatigue protocol in various conditions. Raw H _MAX (P < 0.05) and H _MAX/M _MAX ratio values were significantly (P < 0.05) reduced from pre- to post-exercise, with no temperature or altitude main effects. Compared to temperate environments, the amplitude of both M-waves (at rest, during submaximal contraction and MVC) and V-waves, but not H-reflexes, together with raw soleus RMS activity was significantly reduced (P < 0.05) in hot conditions (Table 2). There was neither a significant main effect of altitude nor a significant interaction between exercise, altitude and/or temperature for any normalized EMG-related variables.

It is already well documented that the exercise capacity for endurance-type efforts deteriorates in hot (Ftaiti et al. 2010; Galloway and Maughan 1997; Mitchell et al. 2013; Rowland et al. 2007) and hypoxic (Amann et al. 2007; Goodall et al. 2012) conditions. A novel finding, however, was that combining those stressors further compromised locomotor performance since exercise capacity was reduced by 50 % compared with similar intensity exercise in hot or hypoxic environment alone. These data show for the first time that combining heat stress and moderate hypoxia reduces cycling time to exhaustion with no additional effect on neuromuscular fatigue characteristics. Other factors (not evaluated in this study) such as increased de-oxygenation trends (Goodall et al. 2012; Rasmussen et al. 2010) or cardiovascular strain due to impairments in stroke volume and/or cardiac output (Gonzalez-Alonso et al. 2008) may have played a pivotal role in earlier exercise cessation in challenging environmental conditions.

Our results confirm that exercising in hot or hypoxic conditions has an impact on the rate of changes in physiological, thermoregulatory and perceptual responses (Amann et al. 2007; Galloway and Maughan 1997; Périard et al. 2011; Racinais and Girard 2012). Increases in HR and RPE were faster in both environments compared to control, while hot environment alone speeded the exercise-induced rise in Tcore and hypoxic exposure only accelerated the SpO₂ decrease. While the absence of statistical difference in Tcore might be surprising, it has to be acknowledged that the current exercise model leads to shorter time to exhaustion in hot trials (i.e. the rate of increase in Tcore was higher in hot than temperate trials). We also report that the level of blood lactate concentration, as measured 5-min post-exhaustion, was elevated with altitude exposure despite nearly identical levels of muscle fatigue (see below). In line with previous time trial findings (Amann et al. 2006), these results indicate a greater anaerobic energy release under hypoxic conditions (McLellan et al. 1990).

It has recently been reported that combining environmental stressors accentuated the physiological stress (HR and RPE values) during a moderate-intensity (60 min at 40 % of maximal oxygen uptake on a treadmill) in different hot (environmental temperature = 23 vs. 35 °C) and hypoxic (FiO₂ = 20.9 vs. 16.5 %) conditions (Buono et al. 2012). The current data showed that if the constant-load exercise is performed until exhaustion, Tcore, HR and SpO₂ values recorded at exhaustion with heat or altitude alone did not differ from a combination of both stressors.

In this study, strength losses in PF were relatively modest (~9 %) as compared to the losses induced by running exercises (~10–40 %; Girard et al. 2012; Millet et al. 2011; Perrey et al. 2010; Racinais et al. 2007) but were in accordance with the losses previously observed after a maximal cycling exercise (~8 %; Racinais and Girard 2012). Maximal EMG activity and twitch interpolation recordings obtained during MVC were used to assess the completeness of plantar flexors activation capacity. Our results displayed a ~7 % reduction in maximal, raw soleus RMS activity, whereas reductions (even though of similar extent) for the normalized EMG signal (RMS/M _SUP ratio) did not reach significance. In addition, VA decreased consistently from ~93 to 89 % across trials. In line with previous studies showing that plantar flexor muscles are subjected to decrements in VA in response to sustained, complex functional movements (e.g. running: Millet et al. 2011; Racinais et al. 2007; cycling: Racinais and Girard 2012; tennis match play: Girard et al. 2011), our results confirm that the neural input reaching the muscle—at least measured during MVCs—decreases with fatigue. However, surface EMG and twitch interpolation techniques do not quantify the descending drive to the lower motoneurons nor do they take into account the source of this drive (De Haan et al. 2009). Thus, other neurostimulation techniques were implemented to shed more light on the potential sites and mechanisms underlying neural adjustments to exhaustive cycling (Gruet et al. 2013).

Previous studies have reported no alteration in voluntary activation during the first seconds of an MVC (Nybo and Nielsen 2001), suggesting that participants could override central fatigue during brief MVCs. However, it is difficult to discriminate the relative role of the temperature and the exercise in these previous data (Racinais and Oksa 2010) as passive hyperthermia has been shown to alter even brief MVCs (Morrison et al. 2004; Thomas et al. 2006; Racinais et al. 2008). Notwithstanding, the muscle group investigated in the current study (i.e. PF) has been shown to be altered due to both exercise and hot ambient conditions, even during brief MVCs performed 5 min after the end of an exhaustive cycling exercise (Racinais and Girard 2012).

In our study, the reduction in resting H-reflexes amplitude from pre- to post-exercise, occurring independently of temperature or altitude, would suggest that the balance between excitation and inhibition affecting the α-motor neuron pool was altered with fatigue. Because the nature of spinal reflex modulations during passive and active conditions may differ (Racinais et al. 2013), H-reflex and V-wave responses were assessed during voluntary efforts/background contractions. In addition to reflecting both reflex excitability and presynaptic inhibition of Ia afferents (spinal processes), V-wave response is also indicative of the level of neural drive in descending corticospinal pathways (supraspinal mechanisms) (Del Balso and Cafarelli 2007). Few existing studies have shown concomitant decrements in H-reflex and V-wave responses during MVC in response to various whole-body exercises such as a 90-min run (Racinais et al. 2007) and a 3-h tennis match play (Girard et al. 2011). There are also reports of decreased H-reflex and V-wave amplitudes in response to passive hyperthermia (Racinais et al. 2008), while the impact of hypoxia on those reflex responses has not been documented yet. In this study, however, reflex responses as measured during voluntary efforts remained unchanged, indicating that neither fatigue nor heat and hypoxia had a significant impact on spinal loop modulation mechanisms.

The present findings confirm that maximal soleus M-wave amplitudes are reduced in hot compared to temperate conditions (Racinais et al. 2008; Racinais and Girard 2012), whereas healthy subjects exposed to acute hypoxia rarely experience muscle excitability (ionic disturbances) perturbations (Perrey and Rupp 2009). This may in turn explain the reduced soleus RMS (VA values were not sensitive to heat exposure) values in hot conditions, as changes in neuromuscular transmission and/or muscle membrane propagation characteristics are likely to affect the raw EMG signal. Noteworthy, no change in soleus M-wave properties—as measured both at rest and during voluntary contractions—but a decrement in the maximal amplitude of evoked twitch torque occurred from pre- to post-exercise, which places the site of peripheral perturbations distant to the muscle membrane.

Although our experimental techniques did not allow investigating further the likely cause of this exercise-induced impeded calcium handling (i.e. excitation–contraction uncoupling; for a review see Allen et al. 2008), a remarkable finding was that the degree of environmental stress modifies the rate of peripheral fatigue development (i.e. identical magnitude of peripheral fatigue incurred at exhaustion across all four trials). When nine male subjects cycled to exhaustion at a fixed work rate (90 % of VO_2max) in normoxia and severe hypoxia (FiO₂ = 0.21 and 0.13, respectively), Romer et al. (2007) also reported a very similar (−34 vs. −39 %) end-exercise reduction in quadriceps twitch force, while endurance capacity was reduced by more than two-third with altitude exposure. This equally attenuated twitch force occurring, in this study, independent of the temperature or altitude and/or the marked differences in exercise capacity is in line with the hypothesis claiming that peripheral fatigue is a carefully regulated variable—i.e. exercise is voluntarily terminated once a specific level of peripheral fatigue (sensory tolerance limit) has been reached during whole-body cycling (Amann 2011).

It has also recently been postulated that subjects stop exercising (exhaustion) in the face of extreme hot (core temperature >39 °C; Racinais and Girard 2012) and hypoxic (SpO₂ <75 %; Amann et al. 2006, 2007) conditions before substantial levels of peripheral fatigue are incurred. Under these circumstances, substantial levels of muscle/cerebral de-oxygenation and hypoexcitability of the cortical circuits may outweigh the limiting effects of peripheral fatigue and associated inhibitory feedback (Amann 2011; Nybo 2008). However, none of these direct sources of inhibition of central motor drive have been explicitly examined in this study, therefore preventing us to draw definitive conclusions. Although similar to what many athletes and military personnel may experience on the field, however, the heat stress (core temperature <~38.8 °C) and degree of hypoxia (SpO₂ >~87 %) were relatively modest in the conditions of the present study. It is therefore unlikely that these ‘direct’ neural influences may have played a significant role in curtailing peripheral fatigue development when exercising in hot and/or hypoxic conditions.

A few limitations of our study must be noted. Firstly, because significant recovery occurs in skeletal muscle function within the first 2 min after intense exercise, it cannot be ruled out that the 5-min period separating exhaustion to post-exercise neuromuscular tests may have caused some recovery and hence underestimated neuromuscular function alterations (Froyd et al. 2013). Of note, however, not only the extent, but also the nature of the neuromuscular fatigue induced by the constant-load exercise was similar at the termination of exercise. Moreover, the possible priming effect of ~40 min of intervention, potentially affecting the neuromuscular system before pre-tests of muscle function and cycling time trial, should not be overlooked.

Secondly, it is unclear whether fatigue mechanisms identified during maximal isometric contractions of a single muscle group [i.e. plantar flexors with EMG analysis of one ankle extensors (soleus) and one ankle flexor (tibialis anterior) muscle] and during a complex functional movement (i.e. cycling to the limit of tolerance at a submaximal constant work rate) are comparable.

Thirdly, we chose to study PF because fatigue-induced neuromuscular adjustments have previously been documented in this particular muscle group after intense cycling (Racinais and Girard 2012), hence allowing a careful evaluation of reflex responses. Nevertheless, whether similar or more pronounced alterations in neuromuscular fatigue characteristics occur in other muscle groups largely involved in cycling (e.g. knee extensors) still needs to be researched. Furthermore, EMG recordings during cycling would be needed to confirm that the relative level of central motor output required for a fixed cycling work rate was exaggerated in the environment where heat and hypoxia were combined. This evaluation needs to be extended to other mono-articular (e.g. vastus lateralis, vastus medialis and gluteus maximus) and several bi-articular muscles (e.g. rectus femoris, semimembranosus, gastrocnemius lateralis and medialis) to carefully evaluate the pattern of muscle activation.

Lastly, the results of the present study have been obtained with a moderately trained population performing laboratory-based tests and therefore should not be generalized to highly trained athletes. Future studies adopting more face-valid tests of performance are warranted to reflect what happens on the field. Finally, the potential role of respiratory muscle fatigue in exercise capacity limitation under hypoxic conditions cannot be ruled out and needs to be investigated in future studies, since hypoxia is known to exacerbate both diaphragm and abdominal muscle fatigability (Verges et al. 2010).