To test the hypothesis that hypoxia centrally affects performance independently of afferent feedback and peripheral fatigue, we conducted two experiments under complete vascular occlusion of the exercising muscle under different systemic O2 environmental conditions. In experiment 1, 12 subjects performed repeated submaximal isometric contractions of the elbow flexor to exhaustion (RCTE) with inspired O2 fraction fixed at 9% (severe hypoxia, SevHyp), 14% (moderate hypoxia, ModHyp), 21% (normoxia, Norm), or 30% (hyperoxia, Hyper). The number of contractions (performance), muscle (biceps brachii), and prefrontal near-infrared spectroscopy (NIRS) parameters and high-frequency paired-pulse (PS100) evoked responses to electrical muscle stimulation were monitored. In experiment 2, 10 subjects performed another RCTE in SevHyp and Norm conditions in which the number of contractions, biceps brachii electromyography responses to electrical nerve stimulation (M wave), and transcranial magnetic stimulation responses (motor-evoked potentials, MEP, and cortical silent period, CSP) were recorded. Performance during RCTE was significantly reduced by 10–15% in SevHyp (arterial O2 saturation, SpO2 = ∼75%) compared with ModHyp (SpO2 = ∼90%) or Norm/Hyper (SpO2 > 97%). Performance reduction in SevHyp occurred despite similar 1) metabolic (muscle NIRS parameters) and functional (changes in PS100 and M wave) muscle states and 2) MEP and CSP responses, suggesting comparable corticospinal excitability and spinal and cortical inhibition between SevHyp and Norm. It is concluded that, in SevHyp, performance and central drive can be altered independently of afferent feedback and peripheral fatigue. It is concluded that submaximal performance in SevHyp is partly reduced by a mechanism related directly to brain oxygenation.
- central nervous system
- transcranial magnetic stimulation
- electrical stimulation
- near-infrared spectroscopy
performance during prolonged whole-body exercise (e.g., cycling) or low-intensity sustained or repeated local muscle contractions is reduced in hypoxia (2, 10, 14, 27, 30). The origin of this performance reduction has for years been attributed to the effect that this decreased O2 content has on muscle contractility (20). For example, using femoral nerve magnetic stimulation, studies have confirmed that decreasing systemic O2 transport increases the rate of peripheral fatigue during dynamic (5, 32) or static (22) knee-extensor contractions. Although such altered muscle contractile capacities may be valid in normoxia (Norm) when subjects are affected by exercise-induced arterial hypoxemia (33), there is also no doubt that central drive is lowered with hypoxia (1).
Evidence supporting the involvement of the central nervous system (CNS) as a mediator of performance reductions under hypoxia originates from experiments reporting improvements in power output almost immediately following a shift from hypoxia to hyperoxia (Hyper) or Norm at the end of exhausting hypoxic cycling exercise (6, 7, 23). Further support for its involvement arises from studies using transcranial magnetic stimulation (TMS) over the motor cortex. In fact, lower supraspinal maximal voluntary activation (%VA) was detected in severe hypoxia (SevHyp, FiO2 = 0.10) after unilateral knee-extensor repeated contractions to exhaustion (16).
Although there is compelling evidence to suggest an altered CNS drive with hypoxia, the underpinning mechanisms are poorly understood. One theory suggests that accumulation of metabolites such as H+ (due to lower O2 delivery) triggers sensory feedback to the CNS via group III and IV afferents, and this may explain the reduced central command (measured through surface electromyography, EMG) and power output shown in hypoxia (2). The increased net discharge of group III and IV afferents could result from a higher baseline firing frequency as well as an additional increase in firing frequency evoked by the hypoxia-induced accumulation of muscle metabolites (3, 18). In other words, central fatigue might originate from exacerbated muscle metabolic changes in hypoxia. How exactly the magnitude of fatigue and associated metabolic milieu in the peripheral muscle might be sensed and projected to higher brain areas to inhibit central motor drive is, at present, not known (1, 19).
Alternative arguments in favor of a direct influence of hypoxia on the brain have also been proposed for the following reasons. First, peripheral fatigue (i.e., quadriceps twitch force response to magnetic femoral nerve stimulation) has been shown to be attenuated after constant-load cycling exercise to exhaustion in SevHyp (FiO2 = 0.10) but not in moderate hypoxia (ModHyp, FiO2 = 0.15) (6). Even if the time (i.e., recovery) between exhaustion and the measurements may have impacted the results of this study (6), the findings likely reflected a difference in the level of peripheral fatigue at exhaustion. Second, cerebral oxygenation has been shown to be reduced during exercise in hypoxia either at the prefrontal (34, 36, 38) or premotor and motor cortex regions of the frontal cortex (38). Third, as mentioned above, lower supraspinal %VA has been shown with SevHyp (FiO2 = 0.10) (16). Interestingly, these authors reported that the decreased supraspinal %VA in SevHyp, when the subjects were exhausted, occurred despite an attenuated peripheral fatigue. The effects of hypoxia on corticospinal excitability or inhibition determined by EMG responses to TMS have been assessed, but the results are contradictory (16, 28, 39, 40).
Although the reduction in exercise performance under severe hypoxic conditions has been suggested to result directly from a decrease in CNS motor command (3, 6), there is a lack of data showing such a direct inhibitory hypoxic effect (i.e., not mediated by peripheral changes). We recently proposed a novel study design to directly test the influence of hypoxia on central mediators of exercise performance (27). The method involved the complete occlusion of blood flow to the knee extensor muscles so that the exercising muscles maintained the same metabolic and functional conditions irrespective of the systemic O2 environment. Such a method should, in theory, result in a similar group III and IV afferent feedback and similar muscle status of the contractile apparatus regardless of the environmental conditions (i.e., hypoxia or Norm). The most important finding of our previous study (27) was that the performance of the occluded leg (i.e., the number of contractions in a submaximal repeated isometric contraction task) was slightly (∼15%) but significantly lower in hypoxic compared with normoxic conditions. This suggests the existence of a small direct effect of hypoxia on central drive.
However, several limitations were identified from the previous experiment that require further investigation to confirm a direct effect of hypoxia on central motor drive. First, it should be examined whether the local muscle metabolic and functional conditions are the same despite different FiO2. In our previous study (27), this was assumed because peripheral fatigue (tested with electrical stimulation) did not differ between hypoxic and normoxic conditions with cuff occlusion. However, measurements of muscle oxygenation and blood volume using near-infrared spectroscopy (NIRS) provide the potential to directly confirm the match of the local metabolic state between different environmental conditions. Second, in our previous study, subjects breathed hypoxic air for only 1 min before the exercise and the arterial O2 saturation (SpO2) measured using pulse oximetry was not stable. Thus further data obtained under exercise conditions starting with a stable SpO2 are needed. Third, it has been suggested that the direct influence of hypoxia on central command may depend on the level of hypoxia (6). Hence, different levels of hypoxia (i.e., ModHyp and SevHyp) need to be examined. Finally, as our previous study did not include any direct measurements of the brain response to hypoxia, the present study used TMS and NIRS to assess the corticospinal excitability/inhibition (16, 40) and the level of prefrontal cortex oxygenation (36, 38), respectively.
The aims of the present study were to determine 1) whether hypoxia has a direct influence on central drive during exercise, independently of the factors within the working muscles, and 2) whether there is a relationship between the level of SpO2 and the amplitude of this direct effect on central command. We hypothesized that 1) hypoxia would induce a lower level of performance despite peripheral fatigue and muscle oxygenation being similar at isotime (thus inferring similar neural inhibition from the muscle type III and IV afferent fibers), and 2) the level of arterial/cerebral oxygenation (due to conditions of SevHyp vs. ModHyp vs. Norm vs. Hyper) would affect performance via neural command. We also aimed to determine whether the altered central drive in hypoxia was related to modifications in corticospinal excitability/inhibition. Overall, we aimed to further test the hypothesis that submaximal exercise performance in hypoxia may be limited, not only by peripheral, but also by central origins.
MATERIALS AND METHODS
Two experiments using healthy male subjects, who performed regular physical activity (e.g., tennis, running, basketball), were conducted. Experiment 1 (Exp 1) recruited 12 subjects (age: 30.9 ± 8.3 yr; height: 1.77 ± 0.06 m; mass: 76.1 ± 9.8 kg), whereas experiment 2 (Exp 2) recruited 10 (age: 33.8 ± 7.3 yr; height: 1.77 ± 0.05 m; mass: 75.0 ± 6.2 kg). Subjects were asked to refrain from strenuous exercise for at least 48 h before each session. The experiments were conducted according to the Declaration of Helsinki. Participants were fully informed of the procedure and risks involved and gave their written consent to participate. Approval for the two experiments was obtained from the Edith Cowan University Human Research Ethics Committee.
The aim of Exp 1 was to assess submaximal performance under full arm occlusion in four environmental conditions (SevHyp to Hyper) and to ascertain, by measuring force and NIRS responses, that the exercising muscles were in a comparable metabolic state while the brain was affected by the experimental conditions. In Exp 2, we compared EMG responses with TMS under full arm occlusion in Norm and SevHyp.
Participants reported to the laboratory 1 wk before their first trial to become familiarized with the experimental procedures. Particular attention was paid to familiarizing them with the maximal voluntary contractions (MVC) of the elbow flexors and electrical stimulation measurements, and subjects repeated the procedures until they were able to produce consistent values. All subjects participated in the four experimental sessions with four different FiO2 conditions (SevHyp, ModHyp, Norm, Hyper, as shown below) separated by at least 1 wk in a randomized, counterbalanced order. One of the subject's arms was used for each session, but both arms were used alternately so that the same arm was used twice.
As shown in Fig. 1A, elbow flexor strength through MVC was first measured on a Cybex6000 isokinetic dynamometer (Lumex, Rankonkoma, NY) connected to a PowerLab (ADInstruments, Bella Vista, Australia). Two MVCs were performed, with each contraction separated by 30 s of recovery. The cuff and inflator (SC 5–6 cm × 83 cm, TD 312; Hokanson, Bellevue, WA) were then placed on the upper part of the subject's arm and inflated to a level of 300 mmHg to ensure complete vascular occlusion. Simultaneously, subjects were submitted to one of the four FiO2 conditions (see below) at rest. The time from cuff inflation to exercise commencement was 5 min, referred to as the wash-in period (16), and it was determined from our pilot studies that showed that this time corresponded to the delay needed to stabilize SpO2 for the most severe level of hypoxia (i.e., FiO2 = 9%) without causing excessive pain for the subject. Under complete vascular occlusion, subjects performed repeated intermittent isometric contractions of the elbow flexor to exhaustion (RCTE) as explained below. Air was inhaled from a 500-l Douglas bag via an air-tight hose connected to a mouthpiece while subjects wore a nose clip. FiO2 was set at 9% (SevHyp), 14% (ModHyp), 21% (Norm), or 30% (Hyper). Both the subjects and the researcher that encouraged the subjects during RCTE were blinded to the FiO2 being administered.
During the RCTE, subjects were requested to perform a task that consisted of repeated sustained 40% MVC isometric contractions for 5 s, followed by 5 s of rest, and to continue to repeat the sequence until exhaustion, defined as an inability to sustain 40% MVC for 2 s. Task failure was determined by an investigator who was blinded to the experimental conditions. Real-time visual feedback of torque signals were displayed on the PowerLab computer monitor, and the value corresponding to 40% of MVC was shown on the monitor. Subjects were strongly encouraged to achieve the required task to their full potential during the RCTE. High-frequency (100 Hz) paired-pulse electrical stimulations (PS100) were delivered to the biceps brachii motor point every two submaximal contractions during the 5-s relaxation period. The numbers of completed 5-s contractions at 40% MVC and mechanical impulse (force × time) until failure were considered the performance metrics for all sessions.
In Exp 2, participants reported to the laboratory 1 wk before the first trial to become familiarized. Subjects were tested on two separate occasions under different systemic O2 environmental conditions (Norm: FiO2 = 21%, SevHyp: FiO2 = 9%). Each session was separated by at least 1 wk, in a randomized, counterbalanced order. The RCTE (described previously) was performed by subjects with their left arm for each FiO2 condition using the same occlusion procedure and wash-in period as described in Exp 1. Before the RCTE and every three submaximal contractions during the RCTE, MVC of the elbow flexor muscles was measured. MVC of the elbow extensor was also measured for maximal EMG of the triceps brachii. Air was inhaled from the same 500-l Douglas bag as used in Exp 1 via an air-tight hose connected to a mouthpiece while the subject wore a nose clip. Again, both the subjects and the researcher who encouraged subjects to perform maximally during the RCTE test were blinded to the FiO2 condition.
As shown in Fig. 1B, subjects were requested to maintain 25% of elbow flexor MVC for 5 s followed by 5 s of rest during the RCTE. A lower percentage of MVC was used for Exp 2 because subjects performed 2-s MVCs regularly during RCTE. A single TMS was superimposed on each MVC, while single pulse nerve stimulations were also evoked on the relaxed muscle every 3 submaximal contractions so that the compound muscle action potentials (M waves, see Fig. 1B) were obtained. The protocol was continued until exhaustion, defined as the inability to sustain 25% of MVC for 2 s. Similar to Exp 1, task performance was quantified as the number of 5-s contractions completed and mechanical impulse (force × time) until failure. During all MVCs that lasted ∼2–3 s, visual feedback and verbal encouragement were provided.
Dependant Variables and Data Analysis
Force measurements (Exp 1 and Exp 2).
For Exp 1 and Exp 2, the same warm-up protocol, consisting of 15 progressive submaximal voluntary isometric contractions, was performed before the MVC measurements. Subjects were positioned on a seated preacher arm curl bench by securing the shoulder angle at 45° flexion with a supinated forearm position, and with their elbow aligned with the axis of rotation of the Cybex6000 isokinetic dynamometer lever arm that set the elbow joint angle at 90° (180°: full extension). The lever arm of the dynamometer was secured to the subject's wrist using a strap; thus there was minimal movement of the shoulder and trunk on the bench. Torque signals were collected from the Cybex6000 isokinetic dynamometer using the PowerLab at a sampling rate of 200 Hz.
Arterial saturation (Exp 1 and Exp 2).
SpO2 and heart rate (HR) were measured continuously using a pulse oximeter (Tuffsat; Datex, Ohmeda, Finland) placed on the nonexercised arm forefinger. SpO2 and HR values were averaged over 5-s periods 1) before cuff, 2) every minute of wash-in, and 3) every two submaximal contractions.
Muscle and cerebral NIRS parameters (Exp 1).
A NIRO-200 oximeter (Hamamatsu Photonics, Hamamatsu, Japan) was used to measure muscle and cerebral NIRS signals. The two probe units of the NIRO-200 have two silicon photodiodes as photodetectors on one side and three laser-emitting diodes (775, 810, and 850 nm) on the other side separated from each other by a distance of 4 cm. The NIRO-200 provides estimates of absolute concentration changes (from an arbitrary baseline of zero) in oxygenated-hemoglobin (ΔO2Hb), deoxygenated-Hb (ΔHHb), and total-Hb (tHb = O2Hb + HHb, expressed in μM·cm), and an absolute measure of O2Hb saturation represented as the tissue oxygenation index (TOI = O2Hb/tHb, expressed in %). TOI reflects the dynamic balance between O2 demand and O2 supply in the tissue microcirculation, whereas tHb reflects blood volume changes.
One probe unit was firmly attached to the skin at the medial side over the mid-belly of the exercising biceps brachii muscle, parallel to the major axis of the arm using double-sided adhesive tape. The other probe unit was firmly attached over the skin of the forehead (in a region corresponding to either Fp1 or Fp2 according to the 10–20 International EEG system) contralateral to the exercising arm with double-sided adhesive tape. Each probe unit was covered with a soft black cloth, and all wires were taped down to minimize movement during exercise. The positions of the NIRS probes were marked on the skin with a semi-permanent ink marker to obtain consistent measures over subsequent testing sessions. NIRS signals were sampled at 6 Hz by the NIRO-200 and were collected simultaneously with torque data onto the PowerLab system and stored on the computer for subsequent analysis. Sampling of muscle and cerebral NIRS signals began upon completion of subject setup on the exercise testing apparatus and continued uninterrupted until the end of each testing session. NIRS signals were averaged over 5-s periods 1) before cuff, 2) every minute of wash-in, and 3) during every submaximal contraction.
Electrical muscle stimulation force (Exp 1).
Square-wave paired-pulse electrical stimulations (width = 200 μs) were evoked using a high-voltage (maximal voltage 400 V) current-constant stimulator (DS7AH; Digitimer, Hertfordshire, UK) applied to the biceps brachii motor point. The cathode electrode (20-mm diameter, Type 0601000402; Contrôle Graphique Medical, Brie-Comte-Robert, France) was taped midway between the anterior edge of the deltoid and the elbow crease with the elbow flexed at 90°. The anode, a 5 cm × 5 cm gel pad electrode (Compex SA, Ecublens, Switzerland), was located over the bicipital tendon (2–3 cm proximal to the elbow). PS100 were obtained at supramaximal intensity, i.e., the stimulation intensity corresponding to ∼120% of the optimal intensity. Optimal intensity was defined as the stimulus intensity at which the maximal twitch force was reached. The procedure for determining the optimal intensity was as follows: during the familiarization session, the optimal intensity was roughly determined using 10-mA increments. Then, for each experimental session, accurate measurements starting at ∼30 mA below the familiarization session-determined intensity was assessed more accurately using 5-mA increments. PS100 was the highest value of torque production recorded.
Electrical nerve stimulation and EMG response (Exp 2).
Single square-wave electrical stimulations (width = 200 μs) were delivered to the brachial plexus using the same stimulator used for the electrical muscle stimulations (Digitimer DS7). The cathode electrode (20 mm diameter, Type 0601000402; Contrôle Graphique Medical) was positioned in the supraclavicular fossa (Erb's point) and the anode (a 5 cm × 5 cm gel pad Compex electrode) on the acromion. The stimulation intensity was supramaximal, i.e., 120% of the stimulus intensity at which the maximal M wave was reached. The supramaximal intensity was determined at rest at the beginning of each session. Peak-to-peak amplitude and duration of the biceps brachii M wave were determined on the relaxed muscle.
Transcranial magnetic stimulation parameters (Exp 2).
A magnetic stimulator (Magstim 200; Magstim, Dyfed, UK) was used to stimulate the motor cortex. Single TMS pulses were delivered via a circular coil (13.5 cm outside diameter) positioned over the vertex of the scalp and oriented to preferentially activate the right motor cortex (contralateral to the left arm). A mark was positioned over the scalp to ensure reproducibility of the stimulation conditions for each subject throughout the entire experiment. The stimulation intensity was adjusted (up to 90% of stimulator output) to elicit large motor-evoked potentials (MEPs) of the biceps brachii (minimum amplitude 60% of maximal M wave) and only small MEPs of triceps brachii (amplitude <15% of maximal M wave) during brief MVCs of the elbow flexor muscles performed at the beginning of each session (42). Stimulation intensity remained unchanged during the course of each session. MEPs were normalized to peak-to-peak amplitude of the biceps brachii M wave. The duration of the cortical silent period (CSP) was determined manually from the time of stimulation to return of continuous voluntary EMG (42). The inter-day intraclass correlation coefficient calculated on MEP·M−1 and CSP before cuff and gas exposure was found to be 0.71 and 0.72, respectively. We were not able to measure the superimposed twitch correctly because stimulation was not always evoked on the force plateau. Also, because of the experimental protocol used in the present study, the measurement of cortical voluntary activation was not easy because several series of activation are needed to determine this parameter (42); such an outcome is difficult to achieve under full occlusion.
EMG during voluntary contractions (Exp 2).
Myoelectric activity was recorded continuously from the biceps brachii and triceps brachii using Ag-AgCl surface electrodes (Type 0601000402, Contrôle Graphique Medical). After swabbing the skin with alcohol, we placed the electrodes in a muscle belly-to-tendon configuration. EMG signals were amplified and band-pass filtered (5 Hz-1 kHz) using the PowerLab system, recorded at a sampling rate of 2 kHz, and stored on a computer for subsequent analysis. The biceps brachii and triceps brachii root mean square values were calculated for each submaximal contraction over the 5 s (including onset and offset) and normalized to the peak-to-peak amplitude of the biceps brachii M wave. Coactivation was calculated as the ratio of the EMG triceps brachii during submaximal elbow flexors contraction divided by the maximal EMG triceps brachii determined at the beginning of each session (see Fig. 1B).
Data were screened for normality of distribution and homogeneity of variances using a Skewness-Kurtosis normality test and the Levene's test, respectively. When the conditions of application were met, a two-way (condition × time) ANOVA with repeated measures was first performed for each dependent variable. Time measurements considered in the analysis were all data with the minimum number of contractions. This number depended on the variable being considered. For instance, there were 13 measurements for SpO2 in Exp 1, i.e., before cuff, every minute of wash-in (5 points), and during contractions 1, 3, 5, 7, 9, and 11. Newman-Keuls post hoc tests were applied to determine a difference between two mean values if the ANOVA revealed a significant main effect or interaction effect. If the condition applications for a two-way ANOVA were not met, the Friedman ANOVA (condition) was used. When the Friedman analysis was significant, differences among conditions were identified at exhaustion using a Wilcoxon test with P values corrected by the number of comparisons (i.e., P < 0.0167). Two additional comparisons were also performed using one-way ANOVA (Exp 1) or t-test (Exp 2). First, each subject's final contraction in hypoxia was compared with the final contraction in Norm. Second, each subject's final contraction in hypoxia was compared with the contraction performed at the same time in Norm. For all statistical analyses, an α level of 0.05 was used as the cut-off for significance. Time effects were omitted for clarity. All descriptive statistics presented in the text are mean values ± SD. Data presented in Figs. 2 through 5 are expressed as means ± SE.
Reproducibility of Maximal and Submaximal Forces Among Sessions
For Exp 1, preexercise MVCs were similar (Hyper: 73.7 ± 14.7; Norm: 75.4 ± 16.3; ModHyp: 74.6 ± 15.5; SevHyp: 74.5 ± 16.4 Nm) and the forces sustained over the submaximal contractions were consistent (Hyper: 40.2 ± 0.8% MVC; Norm: 40.0 ± 0.7% MVC; ModHyp: 39.7 ± 0.7% MVC; SevHyp: 40.3 ± 0.9% MVC) between sessions. The forces were not significantly different between arms. As in Exp 1, preexercise MVCs (Norm: 77.2 ± 13.7; SevHyp: 77.8 ± 12.8 Nm) and the forces sustained over the submaximal contractions were also very consistent across sessions (Norm: 25.4 ± 0.6% MVC; SevHyp: 25.7 ± 0.5% MVC) in Exp 2.
The number of muscle contractions achieved in each condition and the mechanical impulse until failure during RCTE are presented in Table 1. No significant differences were detected between Hyper, Norm, and ModHyp. The number of contractions (performance) and the mechanical impulse in SevHyp were slightly but significantly lower than that in the other conditions in Exp 1 and Exp 2 (∼10% and ∼15%, respectively).
HR, SpO2, and Cerebral NIRS Parameters
HR responses were affected by the conditions of O2 breathing (Fig. 2A). In particular, HR was significantly higher in SevHyp compared with Norm. As expected, there was a large difference in SpO2 among the four FiO2 conditions (Fig. 2B), with the two hypoxic conditions significantly lower than Norm. A similar pattern was observed for cerebral oxygenation (Fig. 2C), but the difference between ModHyp and Norm did not reach significance. No significant difference was shown between conditions for cerebral tHb (Fig. 2D). When considering the difference (Δ) between Norm and the hypoxic conditions (moderate and severe) at the end of RCTE, there was a significant correlation between Δcerebr TOI and ΔSpO2 (R = 0.77; P < 0.001), but no significant correlation was found between Δcerebr TOI (or ΔSpO2) and Δperformance. In Exp 2, changes in HR and SpO2 values were similar to those shown in Exp 1 for the SevHyp and Norm conditions and were significantly different between the two conditions (data not shown).
Muscle NIRS and Electrical Muscle Stimulation
As shown in Fig. 3, A and B, the changes in muscle TOI and tHb were not different across the four FiO2 conditions, supporting the fact that the biceps brachii muscle was in the same state of oxygenation irrespective of the environmental condition. This was further confirmed by the peripheral fatigue measure (PS100 reductions), which were similar among the FiO2 conditions (Fig. 3C).
MVC, EMG, TMS, and Electrical Nerve Stimulation
There was a tendency toward lower MVC values in SevHyp compared with Norm (Exp 2, Fig. 5A, MVCs 1 to 5), but the difference did not reach significance. However, when comparing each subject's final contraction in SevHyp to the one at the same time in Norm (Fig. 5A), MVC was found to be lower in SevHyp (Cohen's d = 0.67). M wave peak-to-peak amplitude (Fig. 4A) and duration (data not shown), measured on the muscle in the relaxed state, were not significantly different between the two conditions. EMG measured during 25% MVC repeated contractions for agonist (biceps brachii) and antagonist (triceps brachii coactivation) muscles are presented in Fig. 4, B and C, respectively. No significant difference was observed between SevHyp and Norm for the increase in biceps brachii EMG activity and coactivation. However, EMG at exhaustion was lower (P < 0.05) in SevHyp. MEP (or MEP normalized to M wave amplitude) and CSP were not significantly different between Norm and SevHyp (Fig. 5, B and C, respectively).
The main finding of the present experiment was that, under the condition of complete vascular occlusion, the number of isometric contractions of the elbow flexors (i.e., performance) was significantly reduced by 10–15% in SevHyp (FiO2 = 9%) compared with ModHyp (FiO2 = 14%), Norm (FiO2 = 21%), and Hyper (FiO2 = 30%). This reduced performance in SevHyp (Exp 1 and Exp 2) existed despite similar 1) metabolic (as shown by muscle NIRS parameters in Exp 1) and functional (as shown by PS100 in Exp 1 and M wave responses in Exp 2) muscle conditions and 2) MEP and CSP TMS responses (Exp 2), suggesting comparable motor corticospinal excitability and inhibition between SevHyp and Norm. Our data show that performance in SevHyp was reduced by a mechanism related to brain oxygenation levels. As shown by the lower EMG at exhaustion, SevHyp leads eventually to an earlier CNS-mediated termination of exercise, independent of muscle metabolic status (in terms of altered muscle contractility and group III and IV afferent-mediated inhibition of central drive) or changes in corticospinal excitability and spinal and cortical inhibition.
It has traditionally been considered that performance decrements with hypoxia are related to local muscle metabolic and functional alterations, particularly in response to an increased rate of H+ accumulation when O2 arterial content was reduced (e.g., Ref. 14). Indeed, it is known that muscle metabolic and functional consequences are exaggerated in hypoxia compared with Norm for a given exercise load during both dynamic (5) and intermittent isometric (20, 22) exercise conditions. Using new techniques, such as cerebral NIRS, TMS, transcranial Doppler ultrasonography, measurements of cerebral mitochondrial O2 tension or nuclear magnetic resonance, previous studies (16, 29, 31, 34, 43) have examined the idea that a reduction in brain oxygenation may induce centrally mediated reductions in exercise performance through the inhibition of cortical activation of efferent motoneurones. The reductions in cerebral oxygenation shown while performing exercise at high altitudes has lead some authors (21, 35, 36) to suggest that cerebral hypoxia might also be a factor limiting exercise performance at altitude. In a review on the topic, Amann and Calbet (1) highlight the uncertainty around how exactly the magnitude of fatigue and associated metabolic milieu in the peripheral muscle might inhibit central motor drive with hypoxia. The aim of the present paper was to address, through two complementary experiments, the direct influence of hypoxia on central drive and performance during repeated submaximal contractions. By direct influence, we refer to the fact that we isolated the known inhibitory hypoxic effects (peripheral fatigue) on the CNS mediated by group III and IV muscle afferents. We used an original approach that involved the complete occlusion of blood flow to the exercised muscles that maintained a similar metabolic state within the elbow flexor muscles, irrespective of the specific environmental (i.e., arterial O2 content) condition. In an attempt to determine a causal effect of the specific interference of somatosensory input to the CNS from the working locomotor muscle (in Norm), some authors have blocked afferent feedback (4, 19, 24). Kjaer et al. (24) used epidural anesthesia to block afferent feedback from the leg muscles during high-intensity cycling exercise to exhaustion in SevHyp (FiO2 = 0.078). The intervention had minimal effects on rating of perceived exertion and performance, suggesting for the first time that afferent feedback was not the main factor causing the termination of exercise in SevHyp. In the present experiment, afferent feedback was not blocked, but we can confidently assume a comparable effect between the environmental conditions with our cuff-occlusion method.
Comparison of Metabolic and Functional Muscle Conditions Between Different Environmental Conditions
In the present experiment, it was necessary to control the local muscle conditions so that they were comparable across trials. This was monitored using muscle NIRS parameters (TOI and tHb) and evoked response to electrical stimulation (PS100, as an index of peripheral fatigue) on a relaxed muscle in Exp 1. As shown in Fig. 3, muscle NIRS parameters were similar regardless of the FiO2 used in Exp 1. No muscle NIRS was measured in Exp 2, but the same cuff pressure was used, meaning it is likely that the cuff method applied had the same effect in the second experiment. The changes in tHb are influenced by the intramuscular pressure during muscle force production (as shown in Fig. 3B), and, although tHb is a measure of muscle blood volume changes, it can also be considered an indirect measure of changes in local muscle blood flow/O2 supply (15). It is not surprising that muscle oxygenation (i.e., TOI) was also comparable between the four environmental conditions because 1) the forces sustained over the submaximal contractions were very consistent throughout conditions, 2) muscle oxygenation reflects the dynamic balance of O2 supply by the muscle microcirculation (arterioles, capillaries, and venules) and O2 consumption/demand by the muscle (17), and 3) no tHb difference existed among conditions.
The agreement between the different FiO2 conditions examined was also demonstrated in the muscle functional consequences. In line with our previous study (27), the mechanical responses to electrical stimulation (PS100, Exp 1, Fig. 3C) were not different across the four FiO2 conditions. This was further confirmed by comparable M wave responses to single electrical nerve stimulation in Exp 2 (Fig. 4A). One might have expected an attenuated peripheral fatigue in SevHyp, i.e., a higher final value for PS100, as found by Amann et al. (6) in SevHyp after cycling to exhaustion and in our previous study (27) after repeated knee-extensor muscle contractions with cuff occlusion. In the present study, the PS100 reduction during exercise was linear over the initial repetitions but tended to plateau near exhaustion so that the final values were not different compared with the preceding contraction values. This may explain the lack of PS100 difference among the environmental conditions at exhaustion. Another difference between the present study and that of Amann et al. (6) is that these authors evoked their twitches ∼2–4 min after the end of exercise; our study performed this procedure immediately after the end of exercise. As also found by Amann et al. (6), lower EMG values in the final submaximal contraction were found in the present study in SevHyp compared with Norm. Similarly, MVC was found to be lower in SevHyp vs. Norm when comparing each subject's final contraction in SevHyp to the contraction performed at the same time in Norm. These findings support a difference in central drive at the time of task failure and for a given load, respectively.
At rest (i.e., during the wash-in), cerebral oxygenation was reduced in hypoxia (Fig. 2C) as reported in previous studies (29, 34, 36, 38). In the present study, we only measured prefrontal cortex oxygenation using a two-channel NIRS system. However, using multichannel NIRS, Subudhi et al. (38) showed that the pattern and magnitude of deoxygenation were correlated between prefrontal, premotor, and motor regions during hypoxic exercise.
During exercise in hypoxia, cerebral oxygenation has been reported to decrease (21, 29, 35) or slightly increase (16, 34) compared with resting levels. Such differences likely relate to the volume of muscle mass utilized (i.e., decreases for whole body exercise vs. increases for unilateral repeated or sustained contractions). In line with previous studies on unilateral exercises (16, 34), no additional changes in cerebral TOI were evident in the present study during the RCTE (Fig. 2B) despite the fact that activation of the motor cortex increases during muscle contraction (25). Instead, slight increases in SpO2, cerebral TOI, and tHb were observed near exhaustion (Fig. 2), probably mainly as a result of increases in ventilation (not measured) and cardiac output, as indicated by the elevated HR (Fig. 2A). Despite the likely decrease in partial pressure of CO2 attributable to (moderate) hyperventilation and the resultant cerebral vasoconstriction, the net result was an increase in cerebral oxygenation near exhaustion, as found elsewhere during unilateral knee-extensor exercise (16). The relationship between changes in cerebral TOI (or SpO2) and changes in RCTE performance in Norm and SevHyp was also not significant. This suggests that the decreases in cerebral (or arterial) oxygenation observed did not directly explain the decreases in performance attributable to hypoxia. Furthermore, it is important to note that a reduction in oxygen delivery does not necessarily mean that brain oxygen consumption is lowered because an increase in brain O2 extraction is possible to preserve brain V̇o2. Further studies are needed to resolve this question.
TMS over the motor cortex can elicit short-latency excitatory responses called MEPs (41). When TMS is delivered during a voluntary contraction, MEP is followed by a period of EMG silence called CSP, which reflects intracortical inhibition mediated by GABAB receptors (41). When the duration is less than 100 ms, spinal and cortical mechanisms contribute to the silent period. It has been suggested that inadequate cerebral oxygenation may depress CNS activity and excitability of cortical neurons, not only attributable to hypoxia-induced neuronal depolarization and reductions in ion channel and pump activity of neurones, but also attributable to the extracellular release of neurotransmitters affecting the postsynaptic membrane (3). A reduced Hoffmann reflex (H reflex) response has been evidenced in acute hypoxia, suggesting an inhibitory effect and/or a reduced excitability of the α-motoneurone pool (44). Because it was found that the α-motoneurone itself is relatively insensitive to hypoxia (13), the authors have concluded that there is a significant inhibitory effect on spinal α-motoneurones via descending supraspinal influences. However, these conclusions are rather speculative. In addition, increased rather than a reduced H reflex has been reported in hypoxia (11). In the present study, neither MEPs nor CSPs were affected by SevHyp (Fig. 5). These results are in line with previous studies at different levels of hypoxia (FiO2 = 10 to 16%), which have reported comparable cortical responses between Norm and acute hypoxia (16, 40). On the basis of Fig. 5C, however, we cannot exclude the possibility that CSP measurements may have lacked sufficient sensitivity to identify the changes in intracortical inhibition that could have lead to reduced performance. Further studies should use other methods, such as the paired-pulse TMS technique, to confirm that cortical inhibition was not involved in the submaximal exercise cessation in SevHyp.
It has been suggested that the determinants of fatigue during exercise shift from predominantly a peripheral origin in Norm and ModHyp to a hypoxia-sensitive central component under the condition of SevHyp (6). Amann et al. (6) have proposed that CNS fatigue dominates over peripheral muscle fatigue below 70–75% SpO2, a value similar to that reached in the present study for the severe hypoxic condition in both Exp 1 (Fig. 2B) and Exp 2 (data not shown). Because no direct effects of SevHyp on MEPs and CSPs were observed in the present study, our results are in agreement with the proposal that central fatigue probably involves a direct hypoxic brain effect on effort perception under the condition of SevHyp (6, 26). Nevertheless, we acknowledge that this comment is somewhat speculative because we did not assess perception of effort in the present study. Although our study excludes the potential role of muscle chemoreceptors (either on motor drive or effort perception), we cannot exclude the possibility that hypoxemia per se, independent of metabolites, may have increased the discharge rates of group III-IV muscle afferents (3). Nonetheless, such an effect of hypoxemia per se is unlikely to have a major influence because maximal force producing capacity has been shown to be preserved at altitude (3).
Other mechanisms should also be considered to explain our findings. For instance, fatigue perception is influenced by many sensory inputs that may have played a role in the decision to stop exercise earlier (37). The question of performance decrement in hypoxia is complex because the lower arterial O2 content can potentially affect, not only the working muscles and the CNS, but also other organs such as the cardio-respiratory system. Indeed, the higher cardio-respiratory requirement in hypoxia may induce sympathetic vasoconstrictor activity and thus alter blood flow distribution (12). However, this did not affect our results because we used an exercise model with circulation occlusion in the active muscles. As shown in Fig. 2A, HR was significantly higher in SevHyp compared with the Norm condition. Unfortunately the present experimental design cannot be reproduced during dynamic whole body exercise where the difference between SevHyp and Norm in terms of hyperventilation, Pco2, vasoconstrictor activity, and cerebral oxygenation during exercise is likely to be exaggerated. Because of higher oxygen delivery combined with a higher perfusion pressure during exercise with a small compared with a large muscle mass, Calbet et al. (8) recently suggested that the risk of mismatch between O2 delivery and demand in the myocardium, respiratory muscles, and CNS may be minimized under small muscle mass conditions (i.e., elbow flexor muscles). Thus it is not known whether the same 10–15% reductions in performance would be witnessed during SevHyp with whole body exercise. Also, the present results do not allow us to speculate further around potential causes of the hypoxic effect on the brain that may have exacerbated the perception of effort. It has been suggested that hypoxia itself (i.e., without exercise) may alter the turnover of several CNS neurotransmitters, particularly in the basal ganglia, which play an important role in the limbic-to-motor neural network link that drives motivation (3, 6). Thus the basal ganglia nuclei seem particularly vulnerable to hypoxia, and alterations in neurotransmitter balance may cause dysfunction at this level of the neural network (9). Finally, the present experiment used full occlusion to test the direct effect of hypoxia on central command during sustained submaximal contractions, but the effects of cuff-induced pain on the findings are not known. Although this represents another limitation of the study, the pain induced by the cuff was thought to be quite minor.
In conclusion, under complete vascular occlusion, which allowed similar metabolic and functional muscle states, submaximal isometric exercise performance of the elbow flexors was significantly reduced by 10–15% in SevHyp (SpO2 = ∼75%). Despite greater reductions in cerebral oxygenation under the condition of SevHyp, comparable MEP and CSP TMS responses were shown between SevHyp and Norm. These findings support the view that performance in SevHyp but not ModHyp is reduced by a mechanism related to brain oxygenation.
No conflicts of interest, financial or otherwise, are declared by the authors.
Author contributions: G.Y.M. and M.M. conception and design of research; G.Y.M., M.M., M.J., P.B.L., and K.N. performed experiments; G.Y.M. analyzed data; G.Y.M., M.M., and M.J. interpreted results of experiments; G.Y.M. prepared figures; G.Y.M. drafted manuscript; G.Y.M., M.M., M.J., P.B.L., and K.N. edited and revised manuscript; G.Y.M., M.M., M.J., P.B.L., and K.N. approved final version of manuscript.
The authors extend their appreciation to Andrew Dingley and Jane Dundas for their assistance with data collection.
- Copyright © 2012 the American Physiological Society