Hyperoxia improves 20 km cycling time trial performance by increasing muscle activation levels while perceived exertion stays the same

Eur J Appl Physiol (2007) 101:771–781
DOI 10.1007/s00421-007-0458-z
ORIGINAL ARTICLE

Hyperoxia improves 20 km cycling time trial performance by increasing muscle activation levels while perceived exertion stays the same

Ross Tucker · Bengt Kayser · Erin Rae · Laurie Rauch · Andrew Bosch · Timothy Noakes

Accepted: 23 March 2007 / Published online: 2 October 2007© Springer-Verlag 2007

Abstract Increasing inspiratory oxygen tension improvesexercise performance. We tested the hypothesis that this ispartly due to changes in muscle activation levels while per-ception of exertion remains unaltered. Eleven male subjectsperformed two 20-km cycling time-trials, one in hyperoxia(HI, FiO2 40%) and one in normoxia (NORM, FiO2 21%).Every 2 km we measured power output, heart rate, bloodlactate, integrated vastus lateralis EMG activity (iEMG)and ratings of perceived exertion (RPE). Performance wasimproved on average by 5% in HI compared to NORM(P < 0.01). Changes in heart rate, plasma lactate concentra-tion and RPE during the trials were similar. For the major-ity of the time-trials, power output was maintained in HI,but decreased progressively in NORM (P < 0.01) while itincreased in both trials for the last kilometre (P < 0.0001).iEMG was proportional to power output and was signiW-cantly greater in HI than in NORM. iEMG activityincreased signiWcantly in the Wnal kilometer of both trials(P < 0.001). This suggests that improved exercise perfor-mance in hyperoxia may be the result of increased muscleactivation leading to greater power outputs. The Wnding ofidentical RPE, lactate and heart rate in both trials suggeststhat pacing strategies are altered to keep the actual and per-ceived exercise stress at a similar level between conditions.

We suggest that a complex, intelligent system regulatesexercise performance through the control of muscle activa-tion levels in an integrative manner under conditions ofnormoxia and hyperoxia.

Keywords Hyperoxia · Pacing strategy · Fatigue · Exercise

Introduction

Endurance exercise performance is inXuenced by the pres-sure of oxygen in the inspired air. A decrease in oxygentension (hypoxia) leads to impaired performance and anincrease (hyperoxia) to an improvement in performancecompared to normoxia (Peltonen et al. 1995, 1997; Tayloret al. 1997). The impairment in performance in hypoxia isoften attributed to a decrease in oxygen availability andtherefore a decrease in sustainable aerobic metabolism,greater dependence on anaerobic metabolism and earlierdevelopment of muscle fatigue (Linossier et al. 2000; Tay-lor et al. 1997). Such an explanation, which attributesfatigue to a failure of the muscle to produce a required forcein hypoxia, has been termed a peripheral model for fatigue(Abbiss and Laursen. 2005; Noakes et al. 2001).

In support of this, Taylor et al. (Taylor et al. 1997) foundthat the integrated EMG activity (iEMG), an indirect mea-sure of skeletal muscle activation (Häkkinen. 1993), of oneof the active muscles was greater during sub-maximalcycling at a Wxed power output in hypoxic (FiO2 11.6%)than in normoxic conditions. This suggests that the force-generating capacity of the muscle was impaired in hypoxia,and an increase in muscle activation was required to main-tain power output (Taylor et al. 1997), supporting theperipheral model for fatigue.

R. Tucker (&) · E. Rae · L. Rauch · A. Bosch · T. NoakesMRC/UCT Research Unit for Exercise Science and Sports Medicine, Department of Human Biology, University of Cape Town, Sports Science Institute of South Africa, P.O. Box 115, Newlands 7725, South Africae-mail: [email protected]

B. KayserFaculty of Medicine, University of Geneva, Geneva, Switzerland

123

772 Eur J Appl Physiol (2007) 101:771–781

Conversely, endurance exercise performance isimproved by hyperoxia (Steinacker et al. 1986). Thisimprovement in performance has been attributed to anincrease in VO2 and associated ATP production (Wilberet al. 2003), a reduction in skeletal muscle and blood lactateaccumulation (Graham et al. 1987), maintenance of cere-bral oxygenation (Nielsen et al. 1999), attenuated arterialoxyhemoglobin desaturation and maintenance of restinglevels of ATP, ADP and total NADH levels (Linossier et al.2000), and to a protective eVect of hyperoxia on the con-tractile properties of the muscle as a result of reduced meta-bolic acidosis and lactate production (Linossier et al. 2000).

Recently, it has been suggested that oxygen-inducedchanges in metabolism and muscle function may not pro-vide the sole explanation for improved performances inhyperoxia and impaired performances in hypoxia. Kayseret al. (Kayser et al. 1994) found that when subjects weregiven hyperoxic air at the point of volitional exhaustionduring cycling exercise in hypoxic conditions, while at thesame moment the exercise load was increased, the subjectswere able to continue cycling at the higher power output.The continuation of exercise was associated with an imme-diate increase in the iEMG activity of one of the activemuscles (vastus lateralis). This suggests that volitionalfatigue in hypoxia occurred at a submaximal level of skele-tal muscle activation, and that the activation of skeletalmuscle was inXuenced by the FiO2. It was concluded that insuch conditions, the limiting factor to endurance exerciseperformance was not the development of contractile dys-function (Taylor et al. 1997), but rather the motor drive tothe active muscle (Kayser et al. 1994).

During self-paced exercise, alterations in performance inhypoxia and hyperoxia are associated with changes in thepacing strategy (Peltonen et al. 1997). Force output of max-imal-eVort rowing strokes was maintained throughouthyperoxic trials, but decreased progressively during trials innormoxia and hypoxia (Peltonen et al. 1997). This Wndingcould be attributed to a progressive reduction in the forcegenerating ability of the muscle in hypoxia and normoxia,as has been proposed (Linossier et al. 2000; Taylor et al.1997). However, Peltonen et al. (1997) found that theiEMG activity of the active muscles was lower during trialsin hypoxia than in normoxia, indicating that performance inhypoxia was impaired, at least in part, due to a centrally-mediated reduction in skeletal muscle activation. Further,Amann et al. (2006) found that central neural drive, mea-sured using EMG activity of the quadriceps muscle, wasreduced during 5 km cycling time-trials in hypoxia(FiO2 = 0.15) compared to hyperoxia (FiO2 up to 1.0). Theyhypothesized that performance ability was determined bycentral motor output to the active muscle, in order to pre-vent a critical threshold of peripheral fatigue (Amann et al.2006).

Thus, when exercise is performed in hyperoxia, it maybe that altered pacing strategies (Peltonen et al. 1997) arethe result of a combination of improved muscle contractil-ity (Linossier et al. 2000; Taylor et al. 1997) as a result ofthe eVects of hyperoxia on muscle metabolism, as well asan altered pacing strategy mediated by the brain in responseto changes such as metabolite accumulation (Amann et al.2006; Diamant et al. 1968; Jacobs et al. 1983; Jacobs andKaiser 1982; Karlsson and Saltin 1970) or phosphagendepletion (McLester Jr. 1997). The brain would then medi-ate anticipatory adjustments in the degree of skeletal mus-cle activation and power output in order to prevent thesechanges from becoming critical or harmful (Amann et al.2006; Ingen Schenau et al. 1991; Lambert et al. 2005; Noa-kes and St Clair Gibson 2004).

Accordingly, the present study Wrst examined the inXu-ence of hyperoxia on pacing strategy and skeletal muscleactivation during self-paced cycling time trials in normoxicand hyperoxic air. We hypothesized that exercise perfor-mance during a self-paced 20-km cycling time-trial wouldbe improved in hyperoxia (FiO2 0.40) compared to nor-moxia (FiO2 0.21), as a result of an altered pacing strategyand an increase in iEMG amplitude in the muscles respon-sible for force generation (i.e. quadriceps). We furtherhypothesized that the adjustments in power output andiEMG activity would occur within the Wrst few minutes ofexercise, as part of an anticipatory regulatory process.

We also examined the eVects of cycling at a constantworkrate in normoxia and hyperoxia on iEMG activity andphysiological measurements. This was performed to allowcomparisons to be made between the measured physiologi-cal variables in hyperoxia and normoxia when exerciseworkrate was free to vary versus when workrate wasimposed, to attempt to understand how any diVerences inpacing strategy between hyperoxia and normoxia may bemediated, as hypothesized in the Wrst part of the study.

Methods

Subject selection

Eleven well trained male subjects (Age 24 § 3 years,height 177 § 7 cm, mass 72 § 7 kg and peak power output(PPO) 395 § 33 W ) were recruited on the basis of perfor-mance in local cycling races and in previous laboratorystudies. We purposefully recruited experienced athletesused to racing on cycles in order to study pacing strategiesduring time trial like challenges. Subjects were fullyinformed of the risks associated with the study and gavewritten informed consent before participation. The studywas approved by the Research and Ethics Committee of theFaculty of Health Sciences of the University of Cape Town.

123

Eur J Appl Physiol (2007) 101:771–781 773

Preliminary testing

After recording stature and body mass using a precision sta-diometer and balance (Model 770, Seca, Bonn, Germany),subjects performed an incremental PPO test. All exercisetrials were conducted on a Kingcycle ergometer (KingcycleLtd., High Wycombe, UK), which allows the subjects toride their own bicycles in the laboratory(Tucker et al.2004). After completing a self-paced warmup, the testbegan at a power output of 2.5 W/kg body weight. Theworkload was increased by 20 W/min until exhaustion.Subjects were required to match a continuously increasingpower output displayed in analogue form on the computermonitor by increasing the speed of cycling through thechanging of gears and pedal rate. The test was terminatedwhen the subject was unable to match the required poweroutput. PPO was recorded as the average power outputachieved over the Wnal minute of the test.

Three to Wve days after the PPO determination, subjectsperformed a self-paced 20 km time-trial to become accus-tomed to the equipment and procedures during the actualexperimental testing. Testing was performed at the sametime of day for each subject’s four trials, and food intakeand training in the period of 24 h prior to the Wrst trial wasalso documented and then maintained for the remainingthree trials.

Experimental exercise trials

All subjects Wrst performed two 20-km time-trials: onetime-trial in normoxia (NORM-TT, FiO2 = 0.21) and onetime-trial in hyperoxia (HI-TT, FiO2 = 0.40). The FiO2 of40% was selected since ethical and safety requirementslimited the duration and exposure of subjects to high oxy-gen contents. Upon completing the time-trials, eight of the11 subjects performed cycling trials to fatigue at a constantworkload in both normoxia (NORM-CW) and hyperoxia(HI-CW). These trials were performed at a constant poweroutput which was set equal to each subject’s average poweroutput during their best 20 km time-trial performance. Sub-jects were instructed to ride for as long as possible at therequired power output.

Control of inspired oxygen tension

All trials were performed inside a Multi-place Class “A”18,000 l hyperbaric chamber (National Hyperbarics (Pty.)Ltd., Hull, UK), of length 3.5 m and diameter 2.5 m built toLloyd’s and ASME 1 PVHO speciWcations. The oxygenconcentration of the air was continuously monitored usingan oxygen sensor (Oxa 001, Scottish Anglo EnvironmentalProtection Ltd.) placed at the level of the subject’s headwhile in the riding position. When the FiO2 dropped 1.0%

below the required percentage, 100% oxygen was fed intothe chamber until the required FiO2 was achieved. Temper-ature and humidity were maintained between 19 and 21°Cand 60–70%, respectively during all trials, and pressurewas maintained equivalent to sea level for all trials.

The order of testing was randomized so that six subjectsperformed NORM-TT Wrst and Wve subjects performed HI-TT Wrst. The order of the constant workload trials, NORM-CW and HI-CW, was also randomized. Subjects wereblinded to the oxygen content of the air and were notinformed of the hypothesis of the study. Trials were sepa-rated by a minimum of three days to allow recoverybetween trials, and subjects were instructed to refrain fromheavy physical exercise the day before trials and to main-tain their training for the duration of their involvement inthe study.

Measurements

Maximal voluntary contraction and EMG activity

Prior to each experimental trial, subjects performed twomaximum voluntary contractions (MVCs) to allow normal-ization of the EMG signal obtained during the subsequentcycling bouts as previously described (Tucker et al. 2004).BrieXy, the electromyogram (EMG) activity of the vastuslateralis muscle of the athlete’s dominant leg was recordedduring maximal isometric contractions using an isokineticdynamometer (Bio-Dex dynamometer, Bio-Dex, UK). Theskin over the muscle was shaved, abraded and cleaned withethanol. Two electrodes (Medicotest A/S, Denmark) wereplaced over the belly of the muscle with an inter-electrodedistance of 20 mm and were connected to a transmitterwhich relayed a telemetric signal to an antenna connectedto an online computer (MyoResearch 2.02, Noraxon, USA)on which raw data were captured at 2,000 Hz and stored forsubsequent processing. Subjects were Wrmly strapped intothe dynamometer and the right leg attached to the arm ofthe dynamometer at a level slightly above the lateral malle-olus. The arm was set so that the knee was at a 60° anglefrom full leg extension. Following a warm-up consisting offour submaximal contractions, subjects performed two Wve-second maximal isometric knee extensions and the EMGactivity from the contraction producing the highest forcewas used for normalization of the EMG signal obtainedduring the subsequent cycling trials.

Normalising each subject’s EMG activity relative totheir own EMG signal before the trial reduces the eVect ofvariables such as electrode positioning, skin impedance anddiVerences in distribution of body fat, allowing compari-sons to be made between conditions and subjects. Thismethod was reported to be reliable for normalization of the

123


EMG signal obtained both during maximal (Hunter et al.2002, 2003) and sub-maximal cycling exercise (St ClairGibson et al. 2001; Tucker et al. 2004).

During the cycling trials, EMG activity was measuredfor a Wve second period every 2 km. EMG was measuredover a set time interval so that changes in pedal rate couldbe reXected in the iEMG, since at higher rates; a greaternumber of contractions would be included in this Wve-sec-ond measurement period. For subsequent analysis of theraw EMG signal, the root mean square (RMS) was calcu-lated over 50 ms periods and the signal was Wltered using a10–200 Hz Wlter. The iEMG amplitude was calculatedusing the MyoResearch software (MyoResearch 2.02,Noraxon, USA) and the value obtained from Wve seconds ofcycling at each measurement point was divided by thevalue obtained during the MVC.

Due to technical problems with the EMG measurementcomputer, EMG data for one subject were not captured dur-ing the constant workload trials. The EMG data are thuspresented for ten subjects during the time-trials and forseven subjects during the constant workload trials.

Plasma lactate concentration

A 20-G cannula (Nissho Corporation, Zaventem, Belgium)was inserted into a forearm vein prior to each trial for bloodsampling every 2 km during the trials. Blood was later ana-lysed for plasma lactate concentrations. Samples for analy-sis were stored on ice until centrifuged at 3,000g for 10 minat 4°C and the plasma stored at ¡20°C. Plasma lactate con-centrations were determined by the spectrophotometric(Beckman Spectrophotometer—M35) enzymatic assaymethod using a lactate kit (Lactate PAP, bioMérieux Kit,Marcey l’Etoile, France). Blood samples could not beobtained from 2 subjects during time-trials, and so lactateconcentrations are thus reported as means § SD for ninesubjects.

Power output and performance

During the trials, the power output was recorded every 200 mand used to calculate average power output for 2 km inter-vals. Performance was deWned as the time necessary to com-plete the time trial, and the duration of the constant load trial,respectively. Subjects were informed of the distance they hadcompleted every 2 km, when they were asked for a subjec-tive rating of perceived exertion (described below).

Heart rate

Average heart rate for each kilometre was measured using aPolar S410 heart rate monitor (Polar Electro OY, Kempele,Finland), and averaged for each 2 km interval.

Perceived exertion

A RPE was recorded every two kilometers using the Borgcategory-ratio scale (Borg 1982). This scale (0–10) wasexplained to the subjects on their Wrst visit to the labora-tory, during the familiarization session and prior to eachtrial. Upon a prompt from an experimenter the subjectswould look at the scale and indicate the perceived levelexertion which was then recorded.

Statistical analyses

Power output, iEMG amplitude, heart rate, plasma lactateconcentrations and ratings of perceived exertion were ana-lysed using a two-way analysis of variance (ANOVA) toexamine the interaction of oxygen concentration and time.Where a signiWcant eVect was detected, post-hoc compari-sons were made using a Tukey’s HSD for pairwise compar-isons. Performance times and average power output wereanalysed using a student’s paired t test. For the constantworkrate trials, subjects did not all complete equal dis-tances before volitional fatigue, and so data were analysedup to 14 km, which is the shortest distance completed by allnine subjects, as well as the Wnal measurement taken atvolitional fatigue. For all analyses signiWcance wasaccepted at P < 0.05. Data are presented as means § SD for11 subjects during time-trials, and for eight subjects duringconstant workrate trials, except where noted.

Results

Time-trial performance and power output

Twenty kilometre time-trial performance was signiWcantlyimproved in the hyperoxic condition (27 min 34 s in HI vs.28 min 8 s in NORM, P < 0.005). The average power out-put in HI was correspondingly higher; 292 § 36 W in HIcompared to 277 § 35 W in NORM (P < 0.001, Fig 1a).This corresponds to 74 and 70% of PPO for HI and NORM,respectively. Of the 11 subjects, ten subjects achieved theirbest performance in the hyperoxic condition, while onesubject performed similarly in both conditions. In that sub-ject, the EMG activity was reduced in HI compared toNORM (see later). The individual and mean average poweroutput recorded for each subject in HI and NORM areshown in Fig. 1a.

The average of the individual power outputs recordedduring each 2 km interval during the time-trials is shown inFig. 1b. The pacing strategy diVered between HI andNORM. In NORM, power output began to decrease from2 km onwards, and was signiWcantly lower than at the startof the trial from 10 km until 18 km (P < 0.05), before

123


increasing signiWcantly for the Wnal 2 km interval. In HI,power output was maintained throughout the trial, until theWnal two kilometers when it increased signiWcantly(P < 0.05). This resulted in a signiWcantly greater poweroutput in HI than in NORM from 12 km until the end of thetrial (P < 0.05). In HI, the highest power output during thetrial occurred in the Wnal 2 km, compared to NORM, wherethe highest power output was recorded in the Wrst 2 km andthe Wnal 2 km intervals (Fig. 1b).

During constant workload trials, the power output, basedon the average power output from the 20 km time-trials,was 292 § 40 and 293 § 41 W in HI and NORM, respec-tively. Hyperoxia resulted in improved performance duringthe constant workload trials to fatigue, with Wve out theeight subjects cycling for longer in hyperoxia, and threesubjects cycling similar distances in the two conditions(33.1 § 15 min vs. 29.9 § 12 min for HI vs. NORM,respectively). Five out of the eight subjects were also ableto cycle further than 20 km when cycling at a constantworkrate in hyperoxia.

iEMG activity

Cadence ranged between 85 and 100 revolutions per minuteand was not diVerent between conditions for each subject.A signiWcant time x trial interaction was found for iEMGduring the time-trials (Fig. 2a). That is, iEMG was main-tained during the trial in NORM, and increased slightly,though non-signiWcantly, over time in HI. The iEMG wasgreater in HI than in NORM at 2, 16 and 18 km (Fig. 2a).There was also a tendency for iEMG to be greater in HIthan in NORM at 14 km (P = 0.08). iEMG increased sig-niWcantly during the Wnal kilometer of trials in both condi-tions (P < 0.005), with peak values being recorded at 20 km(Fig. 2a).

During the constant workload trials, iEMG activity wasnot diVerent between HI and NORM (Fig. 2b). iEMG activity

increased signiWcantly in both HI and NORM, with valuesat fatigue signiWcantly greater than during the Wrst 8 km inboth conditions (P < 0.05, Fig. 2b).

Figure 3a and b show the calculated ratio of power out-put to iEMG during the 20 km time-trials (Fig. 3a) and con-stant workrate trials (Fig. 3b). During the time-trials, therewere no diVerences in this ratio between HI and NORM atany time, though the ratio decreased over time in both con-ditions (Fig. 3a). During constant work rate trials, the ratioof power output to iEMG activity also decreased progres-sively over time, and was similar between HI and NORM(Fig. 3a).

Plasma lactate concentrations

Figure 4 depicts the plasma lactate concentrations during20 km time-trials (Fig. 5a) and constant workrate trials(Fig. 5b). During the 20 km time-trials, plasma lactate con-centrations were not signiWcantly diVerent between condi-tions (Fig. 4a), and increased signiWcantly during the trials(P < 0.0005). During the constant workrate trials, plasmalactate levels were signiWcantly greater (P < 0.01) inNORM than in HI from 6 km until volitional fatigue(Fig. 4b).

Heart rate

Heart rate increased over time during both self-paced andconstant workrate trials, and there were no diVerencesbetween HI and NORM in any of the trials (P < 0.001,Fig. 5).

Ratings of perceived exertion (RPE)

Ratings of perceived exertion increased signiWcantly overtime during time-trials (P < 0.001, Fig. 6a) and during con-stant workrate trials (P < 0.001, Fig. 6b). There were no

Fig. 1 a Average power output for each subject during 20 km time-trials in HI (FiO2 0.4) and NORM (FiO2 0.21). b Average power output for 2 km intervals during time-trials in HI and NORM. Values are means § SD for 11 subjects. Dagger SigniW-cantly diVerent from starting power output in NORM (P < 0.05). *SigniWcantly diVer-ent from HI (P < 0.05). ‡ SigniW-cantly diVerent from receding values within the same condition (P < 0.05)

NORM HI200

220

240

260

280

300

320

340

360

Condition

Po

wer

ou

tpu

t (W

)

A

0 2 4 6 8 10 12 14 16 18 20200

220

240

260

280

300

320

340

360

380

400

NORMHI

* * *

*

*† ††††

‡

‡

Distance (km)

Po

wer

ou

tpu

t (W

)

B

123


diVerences between HI and NORM in any trials. The WnalRPE values during the 20 km time-trials in HI and NORMwere 10.0 § 0.9 and 10.0 § 0.6, respectively, while voli-tional fatigue during the constant workrate trials occurred atRPE values of 9.5 § 0.5 in HI and 10.0 § 0.7 in NORM.

Discussion

The Wrst important Wnding of the present study was thatboth self-paced 20 km cycling time-trial performances(Fig. 1a, b) and constant workrate performances wereimproved in hyperoxia; a FiO2 of 40% was thus suYcient toexert signiWcant eVects on cycling performance. In the self-paced 20 km time-trials, the improvement in performancewas associated with an alteration in pacing strategy, sincepower output decreased progressively from the start of thetrial in NORM, but was maintained throughout the trial in

HI (Fig. 1b). This was not associated with any diVerencesin heart rate, plasma lactate concentrations or ratings of per-ceived exertion between conditions (Figs. 4–6), but wasassociated with an increase in the iEMG activity, suggest-ing increased skeletal muscle activation in hyperoxia(Fig. 2a).

Thus, in NORM-TT, iEMG activity remained constantthroughout the trial, but increased progressively in HI-TT(Fig. 2a), resulting in signiWcant diVerences between condi-tions at 12, 16 and 18 km, the same time as the diVerence inpower output between HI and NORM reached statisticalsigniWcance (Fig. 1b). In contrast, during CW trials, theiEMG activity increased similarly during both HI andNORM, suggesting that similar levels of muscle activationwere necessary to produce the same power output, despitediVerences in the oxygen content of the inspired air.

Impaired endurance exercise performance has beenattributed to a progressive reduction in muscle force

Fig. 2 a Integrated electromyogram (iEMG) activity in the vastuslateralis (VLO) at 2 km intervals during 20 km time-trials in HI andNORM. Values are means § SD for ten subjects. *SigniWcantly lowerthan in HI (P < 0.05). ‡ SigniWcantly diVerent from preceding valueswithin HI and NORM (P < 0.005). b Integrated electromyogram activ-

ity in the VLO at 2 km intervals during constant workrate trials to fa-tigue in HI and NORM. Values are means § SD for seven subjects.*SigniWcant increase over time in both conditions (P < 0.05). No sig-niWcant diVerence between conditions

0 2 4 6 8 10 12 14 16 18 2015

20

25

30

35

40

45

50

NORM TT

HI TT

***

A

‡

Distance (km)

( G

ME

%V

M C

)

0 2 4 6 8 10 12 14 Final15

20

25

30

35

40

45

50

HI CW

NORM CW

*

Distance (km)

ME

( G

%M

V)

C

B

Fig. 3 a Ratio of power output to iEMG activity during 20 km time-trials in HI and NORM. Values are means § SD for ten subjects. * SigniWcant decrease over time in both conditions (P < 0.05). No signiWcant diVer-ence between conditions. b Ra-tio of power output to iEMG activity during constant work-rate trials to fatigue in HI and NORM. Values are means § SD for seven subjects. *SigniWcant decrease over time in both con-ditions (P < 0.05). No signiW-cant diVerence between conditions 0 2 4 6 8 10 12 14 16 18 20

4

6

8

10

12

14

16

18

NORM TT

HI TT

*

A

Distance (km)

Po

ew

r:E

Mr

Gat

io( W

/%M

VC

)

0 2 4 6 8 10 12 14 Final4

6

8

10

12

14

16

18

HI CW

NORM CW

*

B

Distance (km)

oP

wer

:g

mE

r at

io( W

/%M

VC

)

123


producing ability caused by increased metabolite levels(Haseler et al. 1998; Hettinga et al. 2006; Hogan et al.1999; Linossier et al. 2000; Taylor et al. 1997). For exam-ple, it has been found that improved performances with

small muscle groups in hyperoxia are associated withreduced Pi accumulation (Haseler et al. 1998). It is possiblethat large muscle group dynamic exercise is aVected in asimilar way. However, it is signiWcant that the measured

Fig. 4 a Plasma lactate concentrations during 20 km time-trials in HIand NORM. Values are means § SD for nine subjects. ‡ SigniWcant in-crease over time in both conditions (P < 0.001). No diVerences be-tween conditions. b Plasma lactate concentrations during constant

workrate trials to fatigue in HI and NORM. Values are means § SD foreight subjects. *SigniWcantly diVerent from HI (P < 0.01). ‡ SigniW-cant increase over time in both conditions (P < 0.001)

0 2 4 6 8 10 12 14 16 18 200

2

4

6

8

10

12

14

16

HI TT

NORM TT

Distance (km)

Ltca

taoc e

ecnnt

rtaio

( nm

)M ‡

A

0 2 4 6 8 10 12 14 Final0

2

4

6

8

10

12

14

16

HI CW

NORM CW

**

***

*

Distance (km)

Lac

tate

co

nce

ntr

atio

n (

mM

)

‡

B

Fig. 5 a Heart rate responses during 20 km time-trials in HI and NORM. Values are means § SD for 11 subjects. *SigniWcant increase over time in both conditions (P < 0.001). No diVerences between condi-tions. b Heart rate responses dur-ing constant workrate trials to fatigue in HI and NORM. Val-ues are means § SD for eight subjects. *SigniWcant increase over time in both conditions (P < 0.001). No diVerences be-tween conditions

0 2 4 6 8 10 12 14 16 18 20130

140

150

160

170

180

190

200

HI TT

NORM TT

Distance (km)

Hea

rar tte

m( in

1-) *

A

0 2 4 6 8 10 12 14 Final130

140

150

160

170

180

190

200

HI CW

NORM CW

*

Distance (km)

Hea

rar tte

m( in

1 -)

B

Fig. 6 a Ratings of perceived exertion measured during time-trials in HI and NORM. Values are means § SD for 11 subjects. *SigniWcant increase over time in both conditions (P < 0.001). b Ratings of perceived exertion measured during constant work-rate trials to fatigue in HI and NORM. Values are means § SD for eight subjects. *SigniWcant increase over time in both condi-tions (P < 0.001)

0 2 4 6 8 10 12 14 16 18 200

2

4

6

8

10

HI TT

NORM TT

*

A

Distance (km)

PR

E

0 2 4 6 8 10 12 14 Final0

2

4

6

8

10

HI CW

NORM CW

*

B

Distance (km)

PR

E

123


reductions in power output and skeletal muscle activationlevels in NORM-TT (Figs. 1b, 2a) were reversible, sinceactivation levels increased during the Wnal 2 km in both HIand NORM, allowing power output to return to the levelmeasured during the Wrst 2 km of the time trial. Hence,peripheral muscle fatigue alone cannot explain our results.If the greater reduction in power output in NORM-TT(Fig. 1b) were solely the result of a metabolite-inducedperipheral inhibition or impairment of muscle contractilitycompared to HI, then it would not be possible to increasepower output at the end of exercise, unless the impairedforce production could suddenly be overridden in theaVected muscle Wbres.

Instead, we suggest that the increase in power output inthe Wnal 2 km is the result of an increase in muscle activa-tion levels, as occurs during cycling time-trials in the heat(Tucker et al. 2004), and has been found to occur atincreased levels of oxygen content (Amann et al. 2006). Byextension, the ability to maintain a greater power outputduring hyperoxia than normoxia (Fig. 1b) may be the resultof an increased capacity for the activation of motor units inhyperoxia, as suggested by the increased iEMG activity inthat condition (Fig. 2a). This Wnding is thus in agreementwith that of Amann et al. (2006), who in addition, usedmagnetic stimulation of the quadriceps muscle to assessperipheral fatigue, and found that the diVerent levels ofinspired oxygen had no signiWcant eVect on force produc-tion. This suggests that peripheral fatigue was similaracross conditions, and it was hypothesized that the brainregulated motor unit activation diVerently in hyperoxia andhypoxia to prevent the excessive development of ‘periphe-ral fatigue’ (Amann et al. 2006).

On the other hand, when peripheral muscle fatiguedevelops, power output can only be maintained by activat-ing more muscle. Our Wnding that the power-iEMG rela-tionship decreased in all trials (Fig. 3a, b) supports thenotion that the peripheral changes impair muscle force-pro-ducing ability during all trials. Interestingly, however, wefound that the iEMG activity was similar between NORM-CW and HI-CW when power output was the same betweenconditions (Fig. 2b). If there was some oxygen-sensitivefactor that acted peripherally to impair muscle force pro-duction, it would be expected that the iEMG activity wouldbe greater in normoxia than in hyperoxia in order to pro-duce the required power output, yet this was not the case.As a result, the Power/EMG ratio was not diVerent betweenHI and NORM for time-trials (Fig. 3a) or constant work-load trials (Fig. 3b).

The ratio of force output to EMG activity has been usedpreviously to examine the relative changes in muscle func-tion during exercise (Taylor et al. 1997). It was found thatthe force/EMG ratio was lower in hypoxia than in nor-moxia, which was interpreted as evidence that hypoxia

impaired muscle function (Taylor et al. 1997). Similarly,Nummela et al. (1992) found that EMG activity in theactive muscles during a 400 m sprint increased signiWcantlyover the course of the run, even though speed decreased,and it was concluded that additional motor units were beingactivated to compensate for the progressive reduction inmuscle force production as a result of metabolic acidosis inthe muscle (Nummela et al. 1992). Finally, Hettinga et al.(2006) have found that EMG activity increases during4,000 m cycling time-trials, even when power output isdecreasing, and suggested that peripheral fatigue wasoccurring. These studies (Hettinga et al. 2006; Nummelaet al. 1992) may have only limited application to the pres-ent study, since they involved short (less than 5 min), highintensity exercise, compared to endurance exercise in thepresent study.

The present results therefore support the notion thatperipheral changes in the muscle during cycling exercisemay necessitate an increase in skeletal muscle activation tomaintain a given force output. However, using the samereasoning, the Wnding of a similar Power output/EMG ratioin HI and NORM during both self-paced and constant workrate cycling (Fig. 3) suggests that hyperoxia does notdecrease the rate of development of peripheral musclefatigue compared to normoxia. This is a novel Wnding.

We do acknowledge that the measurement and interpre-tation of EMG during dynamic exercise is diYcult and maybe inXuenced by experimental and physiological bias. Nev-ertheless, in the absence of alternative ways to quantifymuscle activation levels during long duration dynamicexercise, changes in iEMG amplitude can be attributed tochanges in either centrally-mediated motor unit recruitmentor to changes in the motor unit Wring rate and, in experi-mental conditions like ours, are a reXection of changes inmuscle activation (Farina et al. 2004).

The possibility that changes in iEMG activity are theresult of altered motor neuron Wring rates must be acknowl-edged. These changes can be mediated either centrally(Bilodeau et al. 2003) or peripherally, in response to fatigue(Marsden et al. 1983). That is, Wring rates decline to com-pensate for a reduction in muscle relaxation time and con-traction speed (Gandevia. 2001; Marsden et al. 1983),secondary to peripheral changes in the muscle (Hepple.2002; MacLaren et al. 1989; Westerblad et al. 2002). Thus,the diVerence in iEMG activity between HI-TT andNORM-TT may be attributed to this phenomenon. It is notclear, however, how the iEMG would then increase signiW-cantly in the Wnal 2 km of both TT if this phenomenon wereoccurring, since peripheral changes in the muscle, whichare presumably responsible for the reduction, are still pres-ent and often maximal at the end of exercise. Conversely,the elevated iEMG activity during HI-TT may be due toincreased motor unit Wring rates, possibly as a result of

123


reduced peripheral metabolite levels, or to altered centralactivation patterns.

During fatiguing isometric exercise, the median fre-quency (MF) of the EMG signal is often measured since itprovides a measure of changes in motor unit Wring rate withfatigue (Bigland-Ritchie 1981; Bilodeau et al. 2003). How-ever, during dynamic exercise, this method is probably notreliable since muscle Wbre conduction velocity and henceMF may be inXuenced by temperature changes in the mus-cle (Bigland-Ritchie 1981) and the continuous changes inmuscle length and position during such dynamic exercise.Other promising techniques such as functional MRI, multi-modal imaging and Transcranial Magnetic Stimulation are,at least for now, not feasible because of the large movementcomponent during dynamic exercise, leaving only iEMG asan indirect means to measure activation levels during exer-cise bouts. Thus, while we must interpret our iEMG datawith caution, our Wnding that pacing strategy is altered inassociation with changes in iEMG activity suggests thataltered skeletal muscle activation, in combination withperipheral changes in the muscle, is responsible for thealtered exercise performance.

If such a hypothesis is correct, it is not clear from thepresent data how this diVerential regulation of muscle acti-vation and exercise performance in the hyperoxic conditionis achieved. Much debate exists regarding the eVects of theaccumulation of metabolites on muscle function and exer-cise performance (Foster et al. 2003; Linossier et al. 2000;Nummela et al. 1992; Taylor et al. 1997). Previous studieshave shown that when exercise is performed in hyperoxia,fatigue occurs at similar arterial pH levels and lactate con-centrations compared to normoxia (Linossier et al. 2000).Nielsen et al. have suggested that elevated inspiratory O2

fraction improves exercise performance by maintainingcerebral oxygenation (Nielsen et al. 1999) and it is possiblethat our subjects presented this phenomenon. Similarly, lac-tate accumulation (Hogan and Welch 1984; Linnarsson et al.1974), ventilation (Peltonen et al. 1999), the �-adrenergicresponse to exercise (Howley et al. 1983), reductions inarterial oxygen saturation levels (Nielsen et al. 1999; Pelto-nen et al. 2001a, b), and reductions in skeletal muscle ATP,ADP and total NADH levels (Linossier et al. 2000) are allattenuated in hyperoxia. Thus, it has been proposed thathyperoxia improves performance to fatigue by reducing thebiochemical and physiological disturbances to homeostasis(Wilber et al. 2003), which would allow exercise to continuefor longer before ‘limiting’ levels are reached.

Indeed, we found that when exercise was performed atthe same constant power output in HI and NORM, plasmalactate concentrations were lower in HI compared toNORM (Fig. 4b). However, when exercise was self-paced,lactate levels were similar (Fig. 4a), presumably becausethe exercise workrate was higher in hyperoxia. Foster et al.

(Foster et al. 2003) have suggested that athletes are sensi-tive to proton accumulation and that the pacing strategyduring high intensity exercise is regulated to ensure that aharmful metabolic acidosis does not occur until the end ofthe event. While we do not have measurements of lactate inthe active muscle, it seems reasonable to suggest thathyperoxia attenuated the degree of metabolic acidosis at agiven power output (Fig. 4b), and so during the self-pacedtime-trials, a greater power output could be maintained inbefore a harmful acidosis could occur.

Other variables, including enhanced rates of energyrelease from muscle (Eiken et al. 1987), ventilation (Pelto-nen et al. 1999), �-adrenergic activation (Howley et al.1983), alveolar gas exchange and ventilation (Mateika andDuYn 1994), reductions in systemic (Peltonen et al. 1999)and cerebral arterial saturation levels (Nielsen et al. 1999),and the reduction in skeletal muscle ATP, ADP and totalNADH levels (Linossier et al. 2000), all of which are pro-posed to limit exercise performance, may all be similarlyregulated by the presently proposed regulatory systemwhen exercise is self-paced and the workrate is free vary.The present hypothesis therefore holds that changes inpower output and skeletal muscle recruitment are adjustedto ensure that these variables, which are often implicated as“limiting” factors for exercise performance, are instead thevery variables that are regulated similarly in the two condi-tions, as has previously been suggested by Amann et al.(2006).

The novel aspect of this model is that the changes inworkrate and muscle activation occur in advance of theattainment of critical changes in the physiological systemsand processes they are ultimately responsible for regulat-ing. Changes in pacing strategy and performance are thusthe markers of the integration of aVerent sensory feedbackwith a feed forward, anticipatory adjustment in power out-put. This has previously been shown during exercise in theheat, where exercise intensity and iEMG activity arereduced in advance of diVerences and excessive increasesin core temperature compared to cool conditions (Marinoet al. 2004; Tucker et al. 2004), and the major contributionof the present study is to suggest that a similar phenomenonoccurs during exercise in hyperoxia, possibly due to theregulation of metabolite accumulation.

A Wnal interesting observation in the present study wasthat Wve out of eight subjects were able to cycle for morethan 20 km when power output was held constant at thesame average power output as they achieved during self-paced exercise. This resulted in greater work being doneduring the constant workload trials compared to the time-trials in hyperoxia, and suggests that those subjects did notperform optimally during the time-trials. This Wnding maybe due to the neural control system which we propose regu-lates the characteristic pacing strategy that is observed

123


during self-paced time-trials, both in the present study andin previous research (Foster et al. 2003; Marino et al. 2004;Tucker et al. 2004).

Exercise intensity is maintained at sub-maximal levelsfor most of the trial, before a signiWcant increase in poweroutput occurs in the Wnal kilometer (Fig. 1b). The mainte-nance of this ‘reserve’ during the self-paced trials does nottherefore result in optimal performance, though factorssuch as training may alter the pacing strategy that isadopted. Further research is required to examine the regula-tion of pacing strategy and its implications for performance.

In conclusion, self-paced exercise performance isimproved in hyperoxia as a result of diVerences in pacingstrategy compared to normoxia. Power output was main-tained in HI but decreased progressively in NORM, untilthe Wnal 2 km. This characteristic pacing strategy, togetherwith the associated changes in iEMG activity, suggests thathyperoxia improves exercise performance through a combi-nation of peripheral eVects and an enhanced ability to acti-vate skeletal muscle hence power output. Exerciseperformance is therefore regulated in a complex system,sensitive to peripheral changes and to the oxygen content ofthe inspired air.

Acknowledgments Funding for this experiment was provided byMedical Research Council of South Africa, the University of CapeTown Harry Crossley and Nellie Atkinson StaV Research Funds, Dis-covery Health and the National Research Foundation of South Africathrough the THRIP initiative.

References

Abbiss CR, Laursen PB (2005) Models to explain fatigue during pro-longed endurance cycling. Sports Med 35:865–898

Amann M, Eldridge MW, Lovering AT, Stickland MK, Pegelow DF,Dempsey JA (2006) Arterial oxygenation inXuences central mo-tor output and exercise performance via eVects on peripheral loco-motor muscle fatigue in humans. J Physiol 575:937–952

Bigland-Ritchie B (1981) EMG/force relations and fatigue of humanvoluntary contractions. Exerc Sports Sci Rev 11:75–117

Bilodeau M, Schindler-Ivens S, Williams DM, Chandran R, SharmaSS (2003) EMG frequency content changes with increasing forceand during fatigue in the quadriceps femoris muscle of men andwomen. J Electromyogr Kinesiol 13:83–92

Borg GA (1982) Psychophysical bases of perceived exertion. Med SciSports Exerc 14:377–381

Diamant B, Karlsson J, Saltin B (1968) Muscle tissue lactate after max-imal exercise in man. Acta Physiol Scand 72:383–384

Eiken O, Hesser CM, Lind F, Thorsson A, Tesch PA (1987) Humanskeletal muscle function and metabolism during intense exerciseat high O2 and N2 pressures. J Appl Physiol 63:571–575

Farina D, Merletti R, Enoka RM (2004) The extraction of neural strat-egies from the surface EMG. J Appl Physiol 96:1486–1495

Foster C, de Koning JJ, Hettinga F, Lampen J, La Clair KL, Dodge C,Bobbert M, Porcari JP (2003) Pattern of energy expenditure dur-ing simulated competition. Med Sci Sports Exerc 35:826–831

Gandevia SC (2001) Spinal and supraspinal factors in human musclefatigue. Physiol Rev 81:1725–1789

Graham TE, Pedersen PK, Saltin B (1987) Muscle and blood ammoniaand lactate responses to prolonged exercise with hyperoxia. JAppl Physiol 63:1457–1462

Häkkinen K (1993) Neuromuscular fatigue and recovery in male andfemale athletes during heavy resistance exercise. Int J Sports Med14:53–59

Haseler LJ, Richardson RS, Videen JS, Hogan MC (1998) Phosphocre-atine hydrolysis during submaximal exercise: the eVect of FiO2. JAppl Physiol 85:1457–1463

Hepple RT (2002) The role of O2 supply in muscle fatigue. Can J ApplPhysiol 27:56–69

Hettinga FJ, de Koning JJ, Broersen FT, Van GeVen P, Foster C (2006)Pacing strategy and the occurrence of fatigue in 4000-m cyclingtime trials. Med Sci Sports Exerc 38:1484–1491

Hogan MC, Welch HG (1984) EVect of varied lactate levels on bicycleergometer performance. J Appl Physiol 57:507–513

Hogan MC, Richardson RS, Haseler LJ (1999) Human muscle perfor-mance and PCr hydrolysis with varied inspired oxygen fractions:a 31P-MRS study. J Appl Physiol 86:1367–1373

Howley ET, Cox RH, Welch HG, Adams RP (1983) EVect of hyper-oxia on metabolic and catecholamine responses to prolongedexercise. J Appl Physiol 54:59–63

Hunter AM, St Clair GA, Lambert M, Noakes TD (2002) Electromyo-graphic (EMG) normalization method for cycle fatigue protocols.Med Sci Sports Exerc 34:857–861

Hunter AM, St Clair Gibson A, Lambert DL, Nobbs L, Noakes TD(2003) EVects of supramaximal exercise on the electromyo-graphic signal. Br J Sports Med 37:296–299

Ingen Schenau GJ, Jacobs R, de Koning JJ (1991) Can cycle powerpredict sprint running performance? Eur J Appl Physiol OccupPhysiol 63:255–260

Jacobs I, Kaiser P (1982) Lactate in blood, mixed skeletal muscle, andFT or ST Wbres during cycle exercise in man. Acta Physiol Scand114:461–466

Jacobs I, Tesch PA, Bar-Or O, Karlsson J, Dotan R (1983) Lactate inhuman skeletal muscle after 10 and 30 s of supramaximal exer-cise. J Appl Physiol 55:365–367

Karlsson J, Saltin B (1970) Lactate, ATP, and CP in working mus-cles during exhaustive exercise in man. J Appl Physiol 29:596–602

Kayser B, Narici M, Binzoni T, Grassi B, Cerretelli P (1994) Fatigueand exhaustion in chronic hypobaric hypoxia: inXuence of exer-cising muscle mass. J Appl Physiol 76:634–640

Lambert EV, St Clair Gibson A, Noakes TD (2005) Complex systemsmodel of fatigue: integrative homoeostatic control of peripheralphysiological systems during exercise in humans. Br J SportsMed 39:52–62

Linnarsson D, Karlsson J, Fagraeus L, Saltin B (1974) Muscle metab-olites and oxygen deWcit with exercise in hypoxia and hyperoxia.J Appl Physiol 36:399–402

Linossier MT, Dormois D, Arsac L, Denis C, Gay JP, Geyssant A, La-cour JR (2000) EVect of hyperoxia on aerobic and anaerobic per-formances and muscle metabolism during maximal cyclingexercise. Acta Physiol Scand 168:403–411

MacLaren DP, Gibson H, Parry-Billings M, Edwards RH (1989) A re-view of metabolic and physiological factors in fatigue. ExercSport Sci Rev 17:29–66

Marino FE, Lambert MI, Noakes TD (2004) Superior performance ofAfrican runners in warm humid but not in cool environmentalconditions. J Appl Physiol 96:124–130

Marsden CD, Meadows JC, Merton PA (1983) ‘Muscular wisdom”that minimizes fatigue during prolonged eVort in man: peak ratesof motoneuron discharge and slowing of discharge during fatigue.Adv Neurol 39:169–211

Mateika JH, DuYn J (1994) The ventilation, lactate and electromyo-graphic thresholds during incremental exercise tests in normoxia,

123


hypoxia and hyperoxia. Eur J Appl Physiol Occup Physiol69:110–118

McLester JR Jr (1997) Muscle contraction and fatigue. The role ofadenosine 5�-diphosphate and inorganic phosphate. Sports Med23:287–305

Nielsen HB, Boushel R, Madsen P, Secher NH (1999) Cerebral desat-uration during exercise reversed by O2 supplementation. Am JPhysiol 277:H1045–H1052

Noakes TD, Peltonen JE, Rusko HK (2001) Evidence that a centralgovernor regulates exercise performance during acute hypoxiaand hyperoxia. J Exp Biol 204:3225–3234

Noakes TD, St Clair Gibson A (2004) Logical limitations to the“Catastrophe” models of fatigue during exercise in humans. Br JSports Med 38:648–649

Nummela A, Vuorimaa T, Rusko HK (1992) Changes in force produc-tion, blood lactate and EMG activity in the 400-m sprint. J SportsSci 10:217–228

Peltonen JE, Rantamäki J, Niittymäki SPT, Sweins K, Viitasalo JT,Rusko HK (1995) EVects of oxygen fraction in inspired air onrowing performance. Med Sci Sports Exerc 27:573–579

Peltonen JE, Rantamäki J, Niittymäki SPT, Sweins K, Viitasalo JT,Rusko HK (1997) EVects of oxygen fraction in inspired air onforce production and electromyogram activity during ergometerrowing. Eur J Appl Physiol 76:495–503

Peltonen JE, Leppavuori AP, Kyro K-P, Makela P, Rusko HK (1999)Arterial haemoglobin oxygen saturation is aVected by FiO2 at

submaximal running velocities in elite athletes . Scand J Med SciSports 9:265–271

Peltonen JE, Tikkanen HO, Ritola JJ, Ahotupa M, Rusko HK (2001a)Oxygen uptake response during maximal cycling in hyperoxia,normoxia and hypoxia. Aviat Space Environ Med 72:904–911

Peltonen JE, Tikkanen HO, Rusko HK (2001b) Cardiorespiratory re-sponses to exercise in acute hypoxia, hyperoxia and normoxia.Eur J Appl Physiol 85:82–88

St Clair Gibson A, Schabort EJ, Noakes TD (2001) Reduced neuro-muscular activity and force generation during prolonged cycling.Am J Physiol Reg Int Comp Physiol 281:R187-R196

Steinacker JM, Marx TR, Marx U, Lormes W (1986) Oxygen con-sumption and metabolic strain in rowing ergometer exercise. EurJ Appl Physiol 55:240–247

Taylor AD, Bronks R, Smith P, Humphries BB (1997) Myoelectricevidence of peripheral muscle fatigue during exercise in severehypoxia: some references to m. vastus lateralis myosin heavychain composition. Eur J Appl Physiol 75:151–159

Tucker R, Rauch L, Harley YXR, Noakes TD (2004) Impaired exerciseperformance in the heat is associated with an anticipatory reduc-tion in skeletal muscle recruitment. PXugers Arch 448:422–430

Westerblad H, Allen DG, Lannergren J (2002) Muscle fatigue: lactic acidor inorganic phosphate the major cause? News Physiol Sci 17:17–21

Wilber RL, Holm PL, Morris DM, Dallam GM, Cannon J (2003) EVectof FiO2 on Physiological Responses and cycling performance andmoderate altitude. Med Sci Sports Exerc 35:1153–1159

123

Documents

Hyperoxia improves 20 km cycling time trial performance by increasing muscle activation levels while perceived exertion stays the same