Journal Logo


Exercise-induced Fatigue in Severe Hypoxia after an Intermittent Hypoxic Protocol


Author Information
Medicine & Science in Sports & Exercise: December 2017 - Volume 49 - Issue 12 - p 2422-2432
doi: 10.1249/MSS.0000000000001371


Whole-body exercise tolerance in severe acute hypoxia (AH) is markedly impaired (4,14). The reduction in exercise tolerance in AH is a concern for not only mountaineers but also military forces, where tactical necessity can result in rapid ascent of service personnel to high altitude and result in debilitating reductions in physical operational capabilities (28). At task failure after constant-power cycling in AH, neuromuscular fatigue is evident as a reduction in the ability to produce maximal isometric force, with a clear central contribution (4,17,19). Studies using transcranial magnetic stimulation (TMS) before and after a fatiguing motor task in AH have shown that at least some of the resulting loss of force originates at or upstream of the motor cortex (17,19,36). This decrease in cortical voluntary activation (VATMS) occurs alongside pronounced cerebral deoxygenation, and as such, exercise in AH is considered to be limited primarily by the hypoxic central nervous system (CNS) (12,47). Further evidence for this is provided from studies using an increase in the partial pressure of inspired O2 (PIO2) at volitional exhaustion, where the capacity to resume whole-body exercise occurs too rapidly to be due to a reversal of metabolic disturbance in the locomotor muscles (e.g., Ref. [4]).

Initial evidence suggests that the mechanisms of exercise-induced fatigue in AH can be modulated by the physiological adaptations associated with acclimatization (3,19). A 14-d exposure to high altitude (PIO2 76 mm Hg) alleviated exercise-induced central fatigue, and this occurred alongside improvements in indices of systemic and cerebral O2 availability (3,19). However, chronic exposure to hypoxia involves substantial logistical demand, immunological consequences (27), and risk of acute mountain sickness (AMS) (15). As such, intermittent hypoxia (IH; repeated exposures to sustained hypoxia lasting minutes to hours) via a decrease in PIO2 has been investigated as a means of promoting physiological adaptations without a prolonged stay at high altitude (or confinement to a hypoxic chamber) (28).

After acclimatization, a number of mechanisms may be responsible for an improved exercise tolerance in severe hypoxia (41). Hemoglobin concentration [Hb] is higher due to a reduction in plasma volume, and O2 carrying capacity is improved via erythropoiesis (measured as an increase in total hemoglobin mass (Hbmass) [37]). Furthermore, hemoglobin saturation (SpO2) is increased during hypoxic exercise (3). These mechanisms contribute to an increase in arterial O2 content (CaO2) during constant-power cycling in chronic hypoxia in comparison to AH (3). However, it is not necessarily the systemic improvement in CaO2 that results in an alleviation of central fatigue in chronic hypoxia, but may be the resulting improvement in cerebral O2 delivery (CD˙O2) (19). CD˙O2 may be improved in the face of an unchanged cerebral blood flow (CBF), where CBF is subject to the opposing influences of hypocapnia-induced cerebral vasoconstriction (via hyperventilation) and hypoxia-induced cerebral vasodilation (1).

At least some of the adaptations that compensate for a reduced PIO2 in chronic hypoxia may be achievable with an IH protocol. The principal beneficial response to IH that mimics acclimatization is considered an increase in CaO2 via early respiratory changes related to an increase in hypoxic chemosensitivity (8,10,28). In contrast, the evidence is largely unsupportive of an increase in [Hb] or Hbmass with IH protocols (26), likely due to an insufficient total duration of hypoxic exposure (32). However, alterations are more rapid in severe hypoxia (37) and may be possible with an IH protocol involving exercise training (34).

Few studies have investigated exercise tolerance in severe hypoxia after IH protocols conducted at the same PIO2, and some have shown improvements in whole-body exercise tolerance in the severe-intensity domain (e.g., Refs. [5,8]). However, neither of these examples included a control group. To our knowledge, no study has investigated central and peripheral fatigue after an IH intervention. Therefore, the aims of this study were to determine whether (i) exercise tolerance in severe hypoxia could be improved after an IH protocol in comparison to a control protocol in normoxia and (ii) exercise-induced central fatigue would be alleviated after an IH protocol. It was hypothesized that (i) an IH protocol would result in an improvement in exercise tolerance in severe hypoxia and (ii) the central contribution to neuromuscular fatigue would be alleviated after an IH protocol.



Twenty-one participants were fully informed of the procedures and risks involved and provided written consent to participate. All participants were male, nonsmokers, free from contraindications to experimental procedures including any history of cardiorespiratory or neurological disease, and lowlanders and had not visited altitudes of ≥1000 m in the 3 months preceding the study. Participants were regularly physically active at a recreational level and were instructed to refrain from strenuous training for the duration of the experimental protocol. Participants were matched for normoxic peak O2 uptake (V˙O2peak) and randomly assigned (GraphPad Software Inc, La Jolla, CA) to one of two treatment groups: IH or control. Nineteen participants completed all trials (IH: n = 11, V˙O2peak 3.32 ± 0.42 L·min−1, age 23 ± 2 yr, height 180 ± 6 cm, body mass 76.4 ± 13.7 kg; control: n = 8, 3.48 ± 0.36 L·min−1, 22 ± 4 yr, 180 ± 6 cm, 83.0 ± 5.5 kg). The study was approved by the university research ethics committee and was performed according to the Declaration of Helsinki.

Experimental Design

For each participant, all tests were performed at the same time of day ± 1 h. During two familiarization visits, participants were accustomed to (i) the neuromuscular assessment and cycle ergometer, and (ii), the optimized carbon monoxide rebreathing (oCOr) method (Fig. 1A). On a third visit, V˙O2peak was obtained from a normoxic incremental cycling test (5 W per 12 s from 80 W, preceded by a 3-min prior exercise at 50 W). Peak power (peak) was derived as the highest power averaged for 30 s (IH: 309 ± 23; control: 313 ± 20 W). All subsequent cycling trials and both interventions were performed inside a large, purpose-built chamber (The Altitude Centre, London, UK; ambient temperature 19°C ± 1°C, relative humidly 40% ± 2%, barometric pressure 760 ± 2 mm Hg). Normobaric, poikilocapnic hypoxia was delivered and maintained at a PIO2 of 82 ± 1 mm Hg via nitrogen enrichment (equivalent to ~4700 m above sea level).

Overview of the experimental protocol (A), the constant-power cycling trial visits (B), and the neuromuscular assessment (C). Fam1 and Fam2, familiarization sessions 1 and 2; Max, measurement of maximal O2 uptake and peak power. *Denotes exposures on which 30-min exercise was performed. †Indicates that this trial did not differ from TTF-Pre and was treated as an isotime trial.

All participants performed a constant-power cycling trial to task failure (TTF) before the intervention (≥48 h after the incremental exercise test) and after the intervention (48 h after the final exposure). The preintervention TTF is referred to as TTF-Pre. After the intervention, there was no significant change in the performance of the control group (see the Results section). As such, the postintervention TTF was treated as an “isotime” trial (ISO). In the IH group, there was a significant increase in TTF postintervention (TTF-Post). The IH group performed an additional isotime trial (ISO = TTF-Pre) 48 h after TTF-post. Hbmass was measured before the intervention and 72 h after the final exposure.

The Intervention

For 14 ± 2 d, participants in the IH group performed 10 hypoxic exposures of 2-h duration. The control group completed a matched protocol in normoxia (PIO2 149 ± 1 mm Hg). During exposures 1, 5, and 10, participants remained seated for 2 h. During the remaining sessions, participants undergoing IH were seated for 90 min and cycled for 30 min at 25% peak. During the first exercise bout in the control group, the power was adjusted to produce the same heart rate (HR) as for the IH group (131 bpm). This power was fixed for the remaining exercise bouts (38% ± 2% peak). Arterial O2 saturation was estimated using a pulse oximeter (SpO2), with a fingertip sensor placed on the participant’s right index finger (PalmSAT 2500 and 8000AA; Nonin Medical Inc., Plymouth, MN). HR (Polar Electro, Tampere, Finland) and SpO2 were recorded at 10-min intervals at rest and 5-min intervals during exercise. Symptoms of AMS were assessed at 10-min intervals using the Lake Louise Questionnaire (LLQ) (33), with the sleep questions removed.

During the intervention, participants were naïve to the aims of the study and blinded to the O2 levels inside the chamber, their HR, SpO2, and power during cycling. To assess the blinding procedures after completion of the final visit, participants were asked to complete a brief questionnaire to indicate whether (i) their exposures were in “severe hypoxia (more than 4000 m above sea level)” or “normoxia (sea level),” and (ii) if they were “certain,” “fairly sure,” or “uncertain” about their answer. In response to item i, 53% of participants answered correctly and in response to item ii, 74% indicated that they were “uncertain” about their answer.


The oCOr was used to derive Hbmass, blood volume, and plasma volume (38). A detailed description of the equipment and specific methods used has previously been described (45). Test–retest reliability was evaluated before the study with eight recreationally active men (Hbmass: coefficient of variation (CV) 3.9%, intraclass correlation coefficient (ICC) 0.94, typical error of measurement (TEM) 36 g; plasma volume: CV 3.3%, ICC 0.92, TEM 43 mL; blood volume: CV 3.1%, ICC 0.94, TEM 66 mL).

Constant-Power Cycling Trials

Upon arrival to the laboratory, participants provided a midstream urine specimen. Euhydration was accepted as a urine specific gravity of <1.020 g·mL−1 and osmolality of <700 mOsm·kgH2O−1. After 3 min at 50 W, power output was increased to 60% peak (IH: 187 ± 14 W; control: 188 ± 12 W). Trials were performed on an electromagnetically braked cycle ergometer (SRM High Performance Ergometer; Schroberer Rad Meβtechnik, Jülich, Germany), with body position and self-selected cadence determined during familiarization and replicated for the duration of the study. Cadence (87 ± 4 rpm) was the only real-time feedback participants received and verbal instructions were given should participants drift by ≥4 rpm for ≥5 s. The nature and frequency of verbal encouragement were replicated across trials. Task failure was defined as a drop to ≤70% of self-selected cadence for >5 s, despite strong verbal encouragement.

Within-exercise measures

Transcranial Doppler sonography was used to measure blood flow velocity in the left middle cerebral artery (MCAv). The same experimenter isonated the motor cortex segment of the MCA over the left temporal window using a 2-MHz probe on each visit. Signal quality was optimized and the probe was fixed in position within an adjustable headpiece. Continuous traces of the maximal velocity envelope were recorded and processed offline for determination of beat-by-beat mean velocity. Pulmonary ventilation and gas exchange were measured using a breath-by-breath system (Metalyzer 3B; Cortex Biophysik, Leipzig, Germany). After orientation to the scales, RPE and breathlessness were obtained using the Borg RPE Scale and the modified Borg CR10 Scale, respectively. RPE, breathlessness, HR, and SpO2 were recorded at 1-min intervals. Resting [Hb] in combination with SpO2 was used to estimate CaO2 at rest and throughout exercise in all conditions using the equation CaO2 = [Hb] × 1.39 × (SpO2/100). Subsequently, an index of CD˙O2 was calculated as the product of MCAv and estimated CaO2. The estimation of CD˙O2 in this study is based on the assumption that MCAv is a valid surrogate of MCA blood flow. There are uncertainties about the constancy of MCA diameter which prelude this assumption (see Ref. [1]), and as such, the data should be interpreted with due caution.

Neuromuscular assessment

Neuromuscular data were captured using a data acquisition system and analyzed offline using custom-made macroinstructions (PowerLab 26T and LabChart Vv7; ADInstruments Ltd., Oxford, UK). Before the neuromuscular assessment, optimal coil position and resting motor threshold (rMT) were determined (Fig. 1B, described in the TMS section). The neuromuscular assessment (Fig. 1C) was completed in 205 s pre- and postexercise and was performed within the chamber. The first maximal voluntary contraction (MVC) was performed ≤40 s after task failure.

Knee-extensor force

Participants sat upright on a custom-built chair with the knees and hips at 90° of flexion and were secured using straps across trunk and shoulders. Knee-extensor force was measured using a calibrated load cell (Tedea Huntleigh 615; Vishay, Basingstoke, UK) positioned directly behind the point of applied force and connected to a noncompliant cuff attached around the participant’s right leg, 1–2 cm superior to the ankle malleoli. After a set of preparatory contractions (2 × 50%, 2 × 75%, and a minimum of 2 × 100% “efforts”), MVC were performed for 3–5 s with strong verbal encouragement and visual feedback of force. MVC force was measured as the highest 500-ms plateau.

Electrical stimulation of the femoral nerve

Single electrical stimuli (200-μs pulse width) were delivered to the right femoral nerve via 5-cm2 surface electrodes (CF5050B; Nidd Valley Medical, Hampshire, UK) and a constant-current stimulator (DS7AH; Digitimer Ltd., Hertfordshire, UK). The cathode was positioned over the femoral nerve, high in the femoral triangle. The anode was placed midway between the greater trochanter and the iliac crest. The site of stimulation that produced the highest mechanical twitch force and peak-to-peak M-wave amplitude in the vastus lateralis (VL) was located. Stimulations were delivered at increasing intensity (10 mA + 20 mA) until no further increase in either twitch force or M-wave amplitude could be elicited. The plateau intensity was increased by 30% to ensure supramaximality (254 ± 53 mA). Femoral nerve stimulation (FNS) was delivered during and within 2 s of MVC to quantify M-waves, potentiated twitch force (Qtw,pot), and VAFNS. VAFNS was calculated using the interpolated twitch technique where the amplitude of the superimposed twitch (SIT) was normalized to the corresponding Qtw,pot using the equation VAFNS (%) = (1 − (SIT/Qtw,pot)) × 100 (25).


Single TMS pulses (1-ms duration) were delivered with a 110-mm diameter concave double-cone coil powered by a monopulse magnetic stimulator (Magstim 2002; The Magstim Company Ltd., Whitland, UK) with a posteroanterior intracranial current flow. Optimal coil position (location eliciting the largest motor-evoked potential (MEP) in the VL and a concurrent small MEP in the biceps femoris (BF) with stimulations delivered at 70% maximal stimulator output (MSO)) was measured relative to the vertex and clearly marked on the scalp with indelible ink. rMT was determined at the beginning of each TTF visit using the modified relative frequency method (20), defined as the intensity that elicited a VL MEP of ≥0.05 mV in three of six trials (42). Stimulations were delivered at 130% rMT (18), which corresponded to 70% ± 11% MSO (VL MEP/Msup area of 60%–70% at 50%–75% MVC). The BF MEP response was <20% of the VL response at all contraction strengths. Participants performed three sets of contractions (100%, 75%, and 50% MVC, separated by 10-s rest) with 20-s rest between sets (Fig. 1C). The estimated resting twitch (ERT) was calculated by taking the y-intercept of a linear regression (baseline, r = 0.95 ± 0.04; fatigued state, r = 0.95 ± 0.05) of the SIT–voluntary force relationship (43). VATMS was subsequently quantified using the equation VATMS (%) = (1 − (SIT/ERT)) × 100. Where regressions were not adequately linear (defined as r < 0.85 [16]), or an individual set or contraction was problematic, it was excluded and the mean of the two sets was used to estimate ERT. This occurred in <5% of evaluations.

Electromyographic responses

Surface EMG was recorded from the right VL and the lateral head of the BF. The skin was shaved, abraded, and cleaned with an alcohol wipe. Two single-use Ag/AgCI electrodes (33 × 22 mm; Kendall H59P, Mansfield, MA) were placed in a bipolar configuration (interelectrode distance of 20 mm) over the muscle belly. The reference electrode was placed over the patella. Electrode placement was marked with indelible ink to ensure consistent positioning between visits. Raw EMG signal was amplified (×1000), digital band-pass filtered (20 Hz–2 kHz), and sampled at 4 kHz. The peak-to-peak amplitude and area under the curve of the VL Mmax and Msup, and VL and BF MEP at each contraction strength were quantified. VL MEPs were normalized to the corresponding maximal M-wave at rest (Mmax) or during an MVC (Msup). The cortical silent period (CSP) was measured during three MVC cases from stimulus artifact to the continuous resumption of EMG, determined by the same experimenter using visual inspection of the EMG trace.

Statistical Analyses

Data are presented as mean ± SD throughout. All statistical procedures were completed using IBM SPSS v22. Before ANOVA, homogeneity of variance was confirmed using Levene test. Data were checked for sphericity using Mauchly test, and if violated, the Greenhouse–Geisser correction was applied. Two-way mixed ANOVA were performed to determine differences in HR, SpO2, and RPE during the intervention protocol (exposure–group) and differences in exercise time and Hbmass before and after the intervention (trial–group). For neuromuscular and within-exercise data, three-way mixed-design ANOVA was performed (trial–time–group), where trial denotes TTF-Pre versus ISO. For within-exercise data, a resting baseline, minutes 1–4, and the final minute of constant-power cycling were included in the statistical analysis given the duration of the shortest exercise time. After a significant overall interaction, two-way repeated-measures ANOVA was performed in each group (trial–time). To account for the difference in exercise time after the IH intervention, two-way repeated-measures ANOVA was performed in the IH group alone (trial–time), where trial denotes TTF-Pre versus TTF-Post. After a significant interaction effect, post hoc analysis was conducted using Tukey HSD tests to localize differences. Statistical significance was set at P < 0.05. Effect sizes are presented as partial eta squared (ηp2) for main and interaction effects and Cohen dav for pairwise comparisons.


Exercise Time

After an interaction of trial–group, exercise time did not differ significantly before and after the control intervention (535 ± 124 vs 557 ± 131 s, P > 0.05; d = 0.18), but improved by 35% ± 18% after the IH protocol (535 ± 213 to 713 ± 271 s, P < 0.05; d = 0.73; Fig. 2).

TTF (s) with constant-power cycling in severe hypoxia before (pre) and after (post) a protocol of IH or a matched protocol in normoxia (control). *P < 0.05, vs post. The mean is plotted as the closed circle.

Hematological Measures

Because of a technical issue, nine participants in the IH group and four participants in the control group completed the postintervention oCOr measurement 72 h after the final exposure. No differences were found for any Hbmass, plasma volume, blood volume, or [Hb] before and after the intervention (all P < 0.05) (see Table, Supplemental Digital Content 1, Hematological measures before and 72 h after an intervention of intermittent hypoxic or control protocol in normoxia,

Maximal Voluntary Force

After an interaction of trial–time–group, MVC was reduced from pre- to postexercise in the control group (time: P < 0.001; ηp2 = 0.92), with no differences before and after the intervention (trial–time: P = 0.69; ηp2 = 0.02). The exercise-induced decrease in MVC in the IH group was alleviated at isotime after the intervention (trial–time: P < 0.001; ηp2 = 0.78, postexercise MVC: P < 0.05; d = 0.67). When cycling continued to task failure in TTF-Post, the resulting decrease in MVC (Fig. 3) did not differ before and after the intervention (trial–time: P = 0.16; ηp2 = 0.19).

Neuromuscular measures made pre (open bars) and post (closed bars) constant-power cycling in severe hypoxia, before and after a protocol of IH (left column) or a matched protocol in normoxia (control, right column). *P < 0.05 vs Post. †P < 0.05, postexercise in TTF-Pre.

Central Fatigue and Corticospinal Excitability

After an interaction of trial–time–group, VATMS was reduced postexercise in the control group (time: P < 0.001; ηp2 = 0.92), with no differences before and after the intervention (trial–time: P = 0.52; ηp2 = 0.06). As presented in Figure 3, the exercise-induced decrease in VATMS (i.e., central fatigue) was alleviated at isotime after the IH intervention (trial–time: P = 0.012; ηp2 = 0.49, postexercise VATMS: P < 0.05; d = 1.20). When cycling continued to task failure, the decrease in VATMS did not differ before and after the intervention (trial–time: P = 0.296; ηp2 = 0.11). The overall pattern in regard to VAFNS was analogous to that described for VATMS (Table 1). No differences were found for corticospinal excitability parameters (all P > 0.05), and data for 100% MVC are presented in Table 1.

Neuromuscular parameters pre and post constant-power cycling to task failure, before and after a protocol of IH or a matched control protocol in normoxia.

Peripheral Fatigue and Neuromuscular Transmission

After an interaction of trial–time–group, Qtw,pot was reduced postexercise in the control group (time: P < 0.001; ηp2 = 0.93), and the level of peripheral fatigue did not differ after the normoxic protocol (trial–time: P = 0.053; ηp2 = 0.44). In contrast, the exercise-induced decrease in Qtw,pot was alleviated at isotime after the IH intervention (trial–time: P = 0.004; ηp2 = 0.57; postexercise Qtw,pot: P < 0.05; d = 0.94). However, the level of peripheral fatigue reached at task failure did not differ before and after the IH intervention (trial–time: P = 0.079; ηp2 = 0.28). No differences were found for parameters relating to neuromuscular transmission (all P > 0.05), and data are presented in Table 1.

Within-Exercise Measures

Data for HR, RPE, breathlessness, and pulmonary gas exchange are presented in Table 2.

Within-exercise data at rest (after a 15 min wash-in period to severe hypoxia) and in the final minute of exercise during TTF before and after an IH protocol or a matched control protocol in normoxia.

Pulmonary ventilation

After an interaction of trial–time–group, minute ventilation (E) was found to be higher at isotime after the IH intervention (P < 0.05; d = 0.27). A higher E was also observed in TTF-Pre versus TTF-Post (P = 0.004; ηp2 = 0.42), such that E was 17% ± 21% higher at task failure after the intervention (P < 0.05; d = 0.43).

Hemoglobin Oxygen Saturation

After an interaction of trial–time–group, the profile of decreasing SpO2 during exercise in severe hypoxia (Fig. 4) did not differ before and after the intervention in the control group (trial–time: P = 0.45; ηp2 = 0.12). This is in contrast to the IH group (trial–time: P = 0.001; ηp2 = 0.35), where SpO2 was higher at isotime after the intervention (P < 0.05; d = 1.57). However, when exercise time was prolonged to task failure, end-exercise SpO2 reached the same level as that in TTF-Pre (P > 0.05; d < 0.1).

Hemoglobin oxygen saturation (SpO2, top row), CBF velocity (MCAv, middle row), and CD˙O2 index (bottom row) in the IH group (left column) and the control group (right column) during TTF-Pre (open circles) and isotime (gray circles). *P < 0.05 vs ISO.

Arterial oxygen content

After an interaction of trial–time–group, in the control group, the profile of CaO2 did not differ before and after the intervention (trial–time: P = 0.741; ηp2 = 0.08). This is in contrast to the IH group (trial–time: P < 0.001; ηp2 = 0.36), where CaO2 was higher at isotime: 16.8 ± 1.1 versus 15.9 ± 0. 8 mL O2·dL−1 (P < 0.05; d = 0.90). However, at task failure, although the interaction of trial–time was significant (P = 0.041; ηp2 = 0.20), no differences were found when post hoc analysis was conducted.

CBF Velocity and CD˙O2

Due to an inadequate signal, one participant was removed from the analysis of MCAv and the derived estimate of CD˙O2 (IH, n = 11; control, n = 7). No significant differences were found in the profile of MCAv or CD˙O2 (i.e., interaction effects) before and after the interventions (all P > 0.05). Data are presented in Figure 4.

The Intervention Protocols

No incidence of AMS (LLQ score ≥3) was observed. After an interaction of exposure–group, in the IH group, resting SpO2 improved during exposures 9 and 10 (e.g., 79% ± 6% vs 83% ± 4% for exposure 1 vs 10, P < 0.05; d = 0.59). Exercising SpO2 in IH improved from the first to the last exercise bout (74 ± 4 vs, 78 ± 4, P < 0.05; d = 1.0). SpO2 was 98% ± 1% at rest and 97% ± 1% during exercise in the control group and did not differ from that during exposures 1 to 10 (all P > 0.05). No differences were found between groups in exercising HR or RPE (all P > 0.05). During the first session of exercise, HR was 131 ± 15 and 131 ± 6 bpm in IH and control, respectively. By exposure 10, exercising HR reduced to 122 ± 14 and 124 ± 5 bpm in IH and control, respectively. In both groups, RPE was 12 ± 1 on exposure 1 and 11 ± 1 on exposure 10.


The aim of the present study was to assess both whole-body exercise tolerance and the mechanisms of neuromuscular fatigue in severe hypoxia after an IH protocol. Exercise tolerance in severe hypoxia improved after an IH protocol completed in the same severity of hypoxia, but not in a control group who completed a matched protocol in normoxia. This is the first study to show that the development of neuromuscular fatigue with whole-body exercise in severe hypoxia is attenuated after an IH protocol. In particular, central fatigue was lower at the same exercise time achieved before the IH intervention.

The IH protocol and exercise tolerance in severe hypoxia

The IH protocol involved 10 exposures to a low PIO2 (82 mm Hg) at rest (2-h duration) with 30 min of moderate-intensity exercise (25% peak) within seven of the sessions. Selecting an optimal hypoxic dose that is both suitable for practical application and capable of eliciting beneficial adaptations is challenging because protocols vary considerably in the literature, depending on the aim of the specific study. Nevertheless, in the design of the present study, we considered the characteristics of IH protocols (in relation to the combination of exposure and training, the training intensity, and duration and the total hypoxic dose) and exercise performance in severe hypoxia as reviewed by Muza (28) in 2007 and investigated in later studies (e.g., Refs. [6,8–10,24]). In the initial IH sessions of the present study, there was an anticipated and pronounced arterial hypoxemia at rest (SpO2 <80%), which improved in the last two sessions (~83%). During the first exercise bout in IH, SpO2 decreased further to <75%. This improved significantly by the final exercise bout alone (~78%). Comparable improvements in SpO2 during hypoxic exercise have been reported during constant-power exercise after an IH protocol (e.g., Ref. [8]), although not consistently (6).

Constant-power cycling in AH resulted in task failure in 8.9 ± 2.9 min. During the intervention, power output was not equivalent in IH and control because of the increase in relative exercise intensity at the same absolute power output in severe hypoxia. To detect changes due to the intervention, the exercise intensity in normoxia was adjusted to match HR recorded during the first exercise bout in the IH group (~131 bpm). In matching the cardiovascular demand of the exercise, it is noted that exercise was also performed at the same perceived exertion (RPE 12), where SpO2 was significantly greater for the control group (98%). It was therefore hypoxic exposure and/or hypoxic exercise that elicited an improvement in exercise tolerance in severe hypoxia in the IH group. Although adaptations cannot be attributed to one of these stressors alone, seven sessions of matched-intensity exercise in normoxia did not improve exercise tolerance in severe hypoxia. Indeed, there was no statistical difference from baseline exercise time in the control group, where the mean difference (36 s) was deemed too small to warrant and justify from an ethical standpoint, a further trial in severe hypoxia. The improvement in exercise time in the IH group was substantial (~35%) and systematic (range, 1.1–5.2 min). Given the successful blinding to the intervention and the blinding of all real-time feedback during the cycling trials, we consider it to be highly unlikely that the improvement in exercise tolerance was due to an anticipated benefit of IH (i.e., a placebo effect).

Neuromuscular fatigue in severe hypoxia

Constant-power cycling induced neuromuscular fatigue in AH, indicated by a ~20% decrease in the ability to produce maximal voluntary force in the knee extensors. After the IH protocol, the reduction in MVC force was less pronounced at isotime (~12%) but reached preintervention levels when exercise continued to task failure. One of the primary limitations to whole-body exercise tolerance in severe hypoxia is exacerbated central fatigue (4,17). The decrease in VATMS from pre- to postexercise in severe hypoxia before the intervention was similar to that reported previously in AH (17,19). Goodall et al. (19) found a 12% decrease in VATMS after constant-power cycling in AH. After acclimatization, central fatigue was alleviated at isotime, and this was attributed, in part, to an improved cerebral O2 availability. However, because of the constraints of the wider project protocol, an improvement in exercise tolerance was not confirmed. In the present study, the IH protocol resulted in an alleviated central fatigue at isotime, where the decrease from baseline was no longer significant. When exercise was permitted to continue to task failure after the IH intervention, central fatigue (and indeed the overall decrease in MVC force) ultimately reached the same levels that coincided with task failure before the intervention (12% decrease in VATMS). We therefore propose that the alleviation of the central contribution to neuromuscular fatigue is an important, though not necessarily the singular, mechanism by which exercise was prolonged after the IH protocol.

The mechanisms for the alleviation of central fatigue may be related to improved cerebral oxygenation secondary to an improved CD˙O2 (12). A number of researchers have substantiated a link between a reduced CD˙O2 and the impairment to whole-body exercise that occurs in severe hypoxia (e.g., Ref. [48]). However, a challenge in the research area is isolating the influence of a systemic improvement in oxygenation (i.e., SpO2) from an improvement in cerebral oxygenation. Isolating the effect of systemic and cerebral oxygenation requires innovative experimental procedures such as CO2 clamping, which increases CBF and therefore CD˙O2. Interestingly, increasing CD˙O2 in AH has not been shown to improve exercise tolerance (11). However, the method is problematic during whole-body exercise (e.g., because of increased respiratory muscle work) and has only been performed with a neuromuscular assessment in a single-limb model (36). In the present study, although the estimate of CaO2 was higher at isotime after the IH intervention (because of the improvement in SpO2 and not [Hb]), CBF and CD˙O2 were not, despite the improvement in exercise time. The relationship between CD˙O2, exercise tolerance and central fatigue in severe hypoxia remains unresolved (39). However, the eventual use of O2 in mitochondrial oxidation depends not only on CD˙O2 but also on the capillary O2 tension, the O2 conductance from capillary to mitochondria, and the cerebral metabolic rate of O2 (31). Under normoxic conditions, there is a tight coupling of the cerebral metabolic rate of O2 and CBF (30). During physiological increases in neuronal activity (e.g., synaptic transmission and firing rate), there is an uncoupling of these variables such that CBF largely exceeds the consumption of O2 in tissue (13). It may be that the signal for reduced central motor output depends on a step that is uncoupled with CD˙O2 during whole-body exercise in severe hypoxia, and this may be altered with an IH protocol. However, this is speculative and further studies on the relationship between cerebral O2 metabolism and central fatigue in severe hypoxia are warranted.

Neither the CSP, as a representation GABA(B) receptor-mediated inhibition of cortical excitability (40), or the MEP, used to assess changes in the state of excitability in the corticospinal system (7), was modulated pre- to postexercise exercise in severe hypoxia, or by the intervention itself. Previous findings indicate time-dependent increases in corticospinal excitability (at rest) with severe hypoxia (3,35). However, both studies used continuous exposures (3 h and 14 d, respectively), and it may be that the discontinuous nature of an IH protocol (i.e., the wash-out in normoxia) masked any transient neurophysiological alterations over the time course of an IH protocol. This warrants further investigation, given the therapeutic potential of IH (29,46).

As evidenced by an increase in E and a decrease in partial pressure of end-tidal carbon dioxide after the IH intervention, the IH protocol used conferred a level of ventilatory adaptation to hypoxic exercise. This resulted in an increase in SpO2 during constant-power cycling after the IH protocol. A proposed threshold of arterial hypoxemia where hypoxia-sensitive mechanisms originating in the CNS are thought to override other inhibitory influences on central motor output (i.e., afferent feedback from the exercising limb) during whole-body exercise is less than 70%–75% (4,12). We note that at isotime after the IH intervention, SpO2 was 79% (vs 70% at task failure in TTF-Pre). A number of previous studies have shown similar ventilatory adaptations to IH, which resulted in an increased E and SpO2 during hypoxic exercise, augmented by increased hypoxic chemosensitivity (8,10,28). Limited studies have investigated exercise tolerance in severe hypoxia, but some have shown improvements in cycle time-trial performance comparable to chronic hypoxia (5,8). More recently, in a study that used four exposures of 4 h to PIO2 of ≈92 mm Hg, increased exercise E and SpO2 were observed without changes in cerebral oxygenation or constant-power cycling to task failure in hypoxia (10). A beneficial effect of IH on exercise tolerance in severe hypoxia is not a consistent finding (6,9), and the disparity may be due to differences relating to protocol and hypoxic dose.

In the present study, we found no change in Hbmass, plasma volume, or [Hb]. Hbmass is a measure of O2 carrying capacity that is not subject to vascular volume shifts. A recent study reported an increase in Hbmass after an IH protocol in moderate hypoxia (34). This is surprising given the evidence suggesting that the total duration of hypoxic exposure required to produce a change in Hbmass is equivalent to more than ~7-d continuous exposure (32).

The peripheral contribution to neuromuscular fatigue

In AH, peripheral locomotor muscle fatigue was identified as a decrease in potentiated quadriceps twitch force by ~20%. The results of earlier studies indicate that in the severe-intensity domain, the level of peripheral fatigue at task failure in AH is less than that reached at task failure in normoxia (>30%) (e.g., Ref. [4]). In severe hypoxia, task disengagement occurs before metabolic disturbance reaches levels attained at the end of the same task performed in normoxia. For this reason, we do not consider peripheral fatigue to be a major limitation to whole-body exercise under these specific and extreme conditions. Nevertheless, after the IH protocol, the decrease in Qtw,pot was less prominent at isotime (~16%). This is in contrast to the findings after a 14-d exposure to high altitude, where the development of peripheral fatigue was not alleviated (3). The mechanisms for this are equivocal but may be related to a lower limb blood flow and therefore limb O2 delivery, an important regulator of peripheral fatigue (2). Speculatively, in the present study, an unchanged limb blood flow coupled with an increase in CaO2 is one explanation for the reduction in peripheral fatigue. Alternatively, an exercise–hypoxia interaction may have induced skeletal muscle adaptations (21) that delayed the development of peripheral fatigue such that it was lower at isotime, but ultimately reached the same levels at task failure.

The decision to disengage from a task can be described as an internally coordinated response to internal and/or external stimuli, that is, a behavior (23). It would be simplistic to ascribe this solely to lower-level neurophysiological properties (22). The perception of the sensations associated with hypoxic exercise is an important consideration, and both perceived limb discomfort and breathlessness (notably, despite a higher E) were also lower at isotime. We further acknowledge that task failure in severe hypoxia is likely to have a cognitive component (44). Disentangling the relative contributions of these complex and interactive processes is a major challenge for exercise scientists and, in hypoxic physiology, warrants further consideration.

In summary, the novel findings of this study were that whole-body exercise tolerance in severe hypoxia was prolonged after a protocol of IH involving exposure and exercise, but not in a control group who performed a matched protocol in normoxia. At isotime after the IH intervention, the central contribution to neuromuscular fatigue was alleviated. These alterations occurred alongside an augmented ventilatory response to hypoxic exercise which improved the pronounced arterial hypoxemia induced by hypoxic exercise in the severe-intensity domain.

The authors wish to acknowledge Dr. Jamie Pringle and the English Institute of Sport for loan of equipment and technical support related to the oCOr.

No conflicts of interest, financial or otherwise, are declared by the authors. The authors declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.


1. Ainslie PN, Subudhi AW. Cerebral blood flow at high altitude. High Altitude Med Biol. 2014;15(2):133–40.
2. Amann M, Calbet JA. Convective oxygen transport and fatigue. J Appl Physiol (1985). 2008;104(3):861–70.
3. Amann M, Goodall S, Twomey R, Subudhi AW, Lovering AT, Roach RC. AltitudeOmics: on the consequences of high-altitude acclimatization for the development of fatigue during locomotor exercise in humans. J Appl Physiol (1985). 2013;115(5):634–42.
4. Amann M, Romer LM, Subudhi AW, Pegelow DF, Dempsey JA. Severity of arterial hypoxaemia affects the relative contributions of peripheral muscle fatigue to exercise performance in healthy humans. J Physiol. 2007;581(1):389–403.
5. Beidleman BA, Muza SR, Fulco CS, et al. Intermittent altitude exposures improve muscular performance at 4,300 m. J Appl Physiol (1985). 2003;95(5):1824–32.
6. Beidleman BA, Muza SR, Fulco CS, et al. Seven intermittent exposures to altitude improves exercise performance at 4300 m. Med Sci Sports Exerc. 2008;40(1):141–8.
7. Beidleman BA, Muza SR, Fulco CS, et al. Intermittent hypoxic exposure does not improve endurance performance at altitude. Med Sci Sports Exerc. 2009;41(6):1317–25.
8. Bestmann S, Krakauer JW. The uses and interpretations of the motor-evoked potential for understanding behaviour. Exp Brain Res. 2015;233(3):679–89.
9. Debevec T, Amon M, Keramidas ME, Kounalakis SN, Pisot R, Mekjavic IB. Normoxic and hypoxic performance following 4 weeks of normobaric hypoxic training. Aviat Space Environ Med. 2010;81(4):387–93.
10. Debevec T, Mekjavic IB. Short intermittent hypoxic exposures augment ventilation but do not alter regional cerebral and muscle oxygenation during hypoxic exercise. Respir Physiol Neurobiol. 2012;181(2):132–42.
11. Fan JL, Kayser B. The effect of adding CO2 to hypoxic inspired gas on cerebral blood flow velocity and breathing during incremental exercise. PLoS One. 2013;8(11):e81130.
12. Fan JL, Kayser B. Fatigue and exhaustion in hypoxia: the role of cerebral oxygenation. High Alt Med Biol. 2016;17(2):72–84.
13. Fox PT, Raichle ME, Mintun MA, Dence C. Nonoxidative glucose consumption during focal physiologic neural activity. Science. 1988;241(4864):462–4.
14. Fulco CS, Rock PB, Cymerman A. Maximal and submaximal exercise performance at altitude. Aviat Space Environ Med. 1998;69(8):793–801.
15. Gallagher SA, Hackett PH. High-altitude illness. Emerg Med Clin North Am. 2004;22(2):329–55.
16. Girard O, Bishop DJ, Racinais S. Neuromuscular adjustments of the quadriceps muscle after repeated cycling sprints. PLoS One. 2013;8(5):e61793.
17. Goodall S, González-Alonso J, Ali L, Ross E, Romer L. Supraspinal fatigue after normoxic and hypoxic exercise in humans. J Physiol. 2012;590(11):2767–82.
18. Goodall S, Romer LM, Ross EZ. Voluntary activation of human knee extensors measured using transcranial magnetic stimulation. Exp Physiol. 2009;94(9):995–1004.
19. Goodall S, Twomey R, Amann M, et al. AltitudeOmics: exercise-induced supraspinal fatigue is attenuated in healthy humans after acclimatization to high altitude. Acta Physiol (Oxf). 2014;210(4):875–88.
20. Groppa S, Oliviero A, Eisen A, et al. A practical guide to diagnostic transcranial magnetic stimulation: report of an IFCN committee. Clin Neurophysiol. 2012;123(5):858–82.
21. Hoppeler H, Klossner S, Vogt M. Training in hypoxia and its effects on skeletal muscle tissue. Scand J Med Sci Sports. 2008;18(1 Suppl):38–49.
22. Krakauer JW, Ghazanfar AA, Gomez-Marin A, MacIver MA, Poeppel D. Neuroscience needs behavior: correcting a reductionist bias. Neuron. 2017;93(3):480–90.
23. Levitis DA, Lidicker WZ, Freund G. Behavioural biologists don’t agree on what constitutes behaviour. Anim Behav. 2009;78(1):103–10.
24. Mekjavic IB, Debevec T, Amon M, Keramidas ME, Kounalakis SN. Intermittent normobaric hypoxic exposures at rest: effects on performance in normoxia and hypoxia. Aviat Space Environ Med. 2012;83(10):942–50.
25. Merton PA. Voluntary strength and fatigue. J Physiol. 1954;123(3):553–64.
26. Millet GP, Roels B, Schmitt L, Woorons X, Richalet JP. Combining hypoxic methods for peak performance. Sports Med. 2010;40(1):1–25.
27. Mishra KP, Ganju L. Influence of high altitude exposure on the immune system: a review. Immunol Invest. 2010;39(3):219–34.
28. Muza SR. Military applications of hypoxic training for high-altitude operations. Med Sci Sports Exerc. 2007;39(9):1625–31.
29. Navarrete-Opazo A, Mitchell GS. Therapeutic potential of intermittent hypoxia: a matter of dose. Am J Physiol Regul Integr Comp Physiol. 2014;307(10):R1181–97.
30. Raichle ME, Grubb RL Jr, Gado MH, Eichling JO, Ter-Pogossian MM. Correlation between regional cerebral blood flow and oxidative metabolism. Arch Neurol. 1976;33(8):523–6.
31. Rasmussen P, Dawson EA, Nybo L, van Lieshout JJ, Secher NH, Gjedde A. Capillary-oxygenation-level–dependent near-infrared spectrometry in frontal lobe of humans. J Cereb Blood Flow Metab. 2006;27(5):1082–93.
32. Rasmussen P, Siebenmann C, Díaz V, Lundby C. Red cell volume expansion at altitude: a meta-analysis and Monte Carlo simulation. Med Sci Sports Exerc. 2013;45(9):1767–72.
33. Roach RC, Bartsch P, Hackett PH, Oelz O. The Lake Louise Acute Mountain Sickness Scoring System. In: Sutton JR, Houston CS, Coates G, editors. Hypoxia and Molecular Medicine. Burlington (VT): Queen City Printers. 1993. pp. 272–4.
34. Robach P, Bonne T, Flück D, et al. Hypoxic training: effect on mitochondrial function and aerobic performance in hypoxia. Med Sci Sports Exerc. 2014;46(10):1936–45.
35. Rupp T, Jubeau M, Wuyam B, et al. Time-dependent effect of acute hypoxia on corticospinal excitability in healthy humans. J Neurophysiol. 2012;108(5):1270–7.
36. Rupp T, Mallouf Tle R, Perrey S, Wuyam B, Millet GY, Verges S. CO2 clamping, peripheral and central fatigue during hypoxic knee extensions in men. Med Sci Sports Exerc. 2015;47(12):2513–24.
37. Ryan BJ, Wachsmuth NB, Schmidt WF, et al. AltitudeOmics: rapid hemoglobin mass alterations with early acclimatization to and de-acclimatization from 5260 m in healthy humans. PLoS One. 2014;9(10):e108788.
38. Schmidt W, Prommer N. The optimised CO-rebreathing method: a new tool to determine total haemoglobin mass routinely. Eur J Appl Physiol. 2005;95(5–6):486–95.
39. Siebenmann C, Rasmussen P. Does cerebral hypoxia facilitate central fatigue? Exp Physiol. 2016;101(9):1173–7.
40. Stetkarova I, Kofler M. Differential effect of baclofen on cortical and spinal inhibitory circuits. Clin Neurophysiol. 2013;124(2):339–45.
41. Subudhi AW, Bourdillon N, Bucher J, et al. AltitudeOmics: the integrative physiology of human acclimatization to hypobaric hypoxia and its retention upon reascent. PLoS One. 2014;9(3):e92191.
42. Temesi J, Gruet M, Rupp T, Verges S, Millet GY. Resting and active motor thresholds versus stimulus-response curves to determine transcranial magnetic stimulation intensity in quadriceps femoris. J Neuroeng Rehabil. 2014;11(1):40.
43. Todd G, Taylor JL, Gandevia SC. Measurement of voluntary activation of fresh and fatigued human muscles using transcranial magnetic stimulation. J Physiol. 2003;551(Pt 2):661–71.
44. Torres-Peralta R, Losa-Reyna J, Morales-Alamo D, et al. Increased PIO2 at exhaustion in hypoxia enhances muscle activation and swiftly relieves fatigue: a placebo or a PIO2 dependent effect? Front Physiol. 2016;7:333.
45. Turner G, Pringle JS, Ingham SA, Fudge BW, Richardson AJ, Maxwell NS. The influence of carbon monoxide bolus on the measurement of total haemoglobin mass using the optimized CO-rebreathing method. Physiol Meas. 2014;35(2):N11–9.
46. Verges S, Chacaroun S, Godin-Ribuot D, Baillieul S. Hypoxic conditioning as a new therapeutic modality. Front Pediatr. 2015;3:58.
47. Verges S, Rupp T, Jubeau M, et al. Cerebral perturbations during exercise in hypoxia. Am J Physiol Regul Integr Comp Physiol. 2012;302(8):R903–16.
48. Vogiatzis I, Louvaris Z, Habazettl H, et al. Frontal cerebral cortex blood flow, oxygen delivery and oxygenation during normoxic and hypoxic exercise in athletes. J Physiol. 2011;589(Pt 16):4027–39.


Supplemental Digital Content

© 2017 American College of Sports Medicine