Effects of exercise-induced arterial hypoxaemia and work rate on diaphragmatic fatigue in highly trained endurance athletes

Authors


Corresponding author I. Vogiatzis: Thorax Foundation, 3 Ploutarhou Str. 106 75, Athens, Greece. Email: gianvog@phed.uoa.gr

Abstract

Diaphragmatic fatigue occurs in highly trained athletes during exhaustive exercise. Since approximately half of them also exhibit exercise-induced arterial hypoxaemia (EIAH) during high-intensity exercise, the present study sought to test the hypothesis that arterial hypoxaemia contributes to exercise-induced diaphragmatic fatigue in this population. Ten cyclists (inline image: 70.0 ± 1.6 ml kg−1 min−1; mean ±s.e.m.) completed, in a balanced ordering sequence, one normoxic (end-exercise arterial O2 saturation (Sinline image): 92 ± 1%) and one hyperoxic (Finline image: 0.5% O2; Sinline image : 97 ± 1%) 5 min exercise test at intensities equal to 80 ± 3 and 90 ± 3% of maximal work rate (WRmax), respectively, producing the same tidal volume (VT) and breathing frequency (f) throughout exercise. Cervical magnetic stimulation was used to determine reduction in twitch transdiaphragmatic pressure (Pdi,tw) during recovery. Hyperoxic exercise at 90% WRmax induced significantly (P= 0.022) greater post-exercise reduction in Pdi,tw (15 ± 2%) than did normoxic exercise at 80% WRmax (9 ± 2%), despite the similar mean ventilation (123 ± 8 and 119 ± 8 l min−1, respectively), breathing pattern (VT: 2.53 ± 0.05 and 2.61 ± 0.05 l, f: 49 ± 2 and 46 ± 2 breaths min−1, respectively), mean changes in Pdi during exercise (37.1 ± 2.4 and 38.2 ± 2.8 cmH2O, respectively) and end-exercise arterial lactate (12.1 ± 1.4 and 10.8 ± 1.1 mmol l−1, respectively). The difference found in diaphragmatic fatigue between the hyperoxic (at higher leg work rate) and the normoxic (at lower leg work rate) tests suggests that neither EIAH nor lactic acidosis per se are likely predominant causative factors in diaphragmatic fatigue in this population, at least at the level of Sinline image tested. Rather, this result leads us to hypothesize that blood flow competition with the legs is an important contributor to diaphragmatic fatigue in heavy exercise, assuming that higher leg work required greater leg blood flow.

High-intensity whole-body endurance exercise (≥85%inline image) induces diaphragmatic fatigue in subjects with varying degrees of fitness levels, as evidenced by a substantial reduction in twitch transdiaphragmatic pressure (Pdi,tw) (Johnson et al. 1993). It has been hypothesized that exercise-induced diaphragmatic fatigue is dependent on several factors namely: (a) high diaphragmatic workload (Johnson et al. 1996) (b) arterial hypoxaemia (Babcock et al. 1995a; Gudjonsdottir et al. 2001) and (c) competition for blood flow between the diaphragm and the locomotor muscles associated with an increased production of metabolic byproducts by the active musculature (Fregosi & Dempsey, 1986; Johnson et al. 1996). More recently, Babcock et al. (2002) have shown that in healthy subjects diaphragmatic fatigue during high-intensity whole-body exercise depends on the relationship between the magnitude of work incurred by the diaphragm and the adequacy of its blood and/or oxygen supply.

In addition, there is evidence that diaphragmatic fatigue also occurs in highly fit athletes during exhaustive exercise (Babcock et al. 1996). Since during high-intensity exercise 40–50% of highly trained athletes exhibit exercise-induced arterial hypoxaemia (EIAH) (Powers et al. 1988; Dempsey & Wagner, 1999; Nielsen et al. 2002), the present study was designed to investigate whether near the maximal levels of exercise (WRmax), when cardiac output is maximal (Mortensen et al. 2005), arterial hypoxaemia per se or blood flow competition between the respiratory and locomotor muscles is more important in causing diaphragmatic fatigue in this highly fit population. To answer this question, highly trained cyclists exercised at two different work intensities (80% and 90% WRmax) under hypoxaemic and normoxaemic conditions, respectively. The experimental design aimed at comparing the degree of diaphragmatic fatigue under two exercise conditions, where arterial oxygen tension and leg work rate varied while maintaining respiratory muscle load similar. Since diaphragmatic power output contributes importantly to exercise-induced diaphragmatic fatigue (Johnson et al. 1996), control of respiratory load within the present experimental design was important in order to isolate the effect of hypoxaemia per se on diaphragmatic fatigue in highly trained athletes. We reasoned that under these conditions, if the hypoxaemic run (at lower leg work rate) produced more fatigue, this would point to arterial hypoxaemia as a prime cause for fatigue. However, if there was more fatigue in the normoxaemic run (at higher leg work rate), this would be compatible with the hypothesis that fatigue was related more to respiratory muscle blood flow limitations. This in turn is based on the presumption that a higher leg effort (90%versus 80% WRmax) would require greater leg blood flow (as the work of Mortensen et al. (2005) has recently demonstrated), potentially reducing flow available to the respiratory muscles (as the work of Harms et al. (1998) suggests may be the case).

Methods

Subjects

Ten male highly trained national team cyclists, with a maximal oxygen uptake inline image of 70.0 ± 1.6 ml kg −1 min−1 participated in the study. Physical characteristics of the subjects are presented in Table 1. Prior to participation in the study all subjects were informed of any risks and discomforts associated with the experiments, and signed an informed consent. The study was approved by the authors' University Ethics Committee and conducted in accordance with the guidelines of the Declaration of Helsinki.

Table 1.  Pulmonary function and maximal exercise data of the study population
  F inline image : 21% O2 F inline image : 50% O2
  1. Values are means ±s.e.m. Exercise data depict the results of the incremental exercise tests in room air (21% O2) and 50% O2. *Significant difference (P < 0.05) between the two Finline image levels. HR, heart rate.

Age (years)21 ± 1
Height (cm)175 ± 2 
Weight (kg)67 ± 2
FEV1 (l) 4.7 ± 0.2
FEV1 (% predicted)114 ± 3 
FVC (l) 5.6 ± 0.2
S inline image (%) at rest97 ± 197 ± 1 
WRmax (W)392 ± 15407 ± 16*
inline image (ml kg−1min−1)70.0 ± 1.675.3 ± 2.8*
HRmax (beats min−1)197 ± 3 187 ± 11 
RER at WRmax 1.20 ± 0.051.11 ± 0.07
inline image (l min−1)155.4 ± 7.7 137.7 ± 7.7* 
V Tmax (l min−1) 2.52 ± 0.122.54 ± 0.14
f max (breaths min−1)60 ± 354 ± 2*
S inline image (%) at WRmax92 ± 197 ± 1*

Experimental design

Experiments were conducted in three visits (Fig. 1). In visit 1, subjects underwent an incremental exercise test to the limit of tolerance. This test was carried out in room air. On the basis of the subjects' percentage oxygen saturation (%Sinline image) at the end of this test, EIAH was confirmed by a fall in %Sinline image of 4% below resting values (Dempsey & Wagner, 1999) (Table 1). In visit 2 (on a different day) subjects completed a second incremental test to the limit of tolerance, breathing a high fraction of inspired O2 (Finline image: 50% O2) gas mixture. This allowed maximal workload to be established both under normoxaemic and hypoxaemic conditions for all 10 subjects. In visit 3 (Fig. 1) subjects completed, in a balanced ordering sequence, two 5 min exercise tests separated by 90 min, either at an intensity initially set at 80% WRmax (LO) of the first incremental test in room air, or at an intensity initially set at 90% WRmax (HI) of the second incremental test while breathing a high Finline image (50% O2). In this manner all subjects completed a 5 min high-work rate normoxaemic test and a 5 min lower work rate hypoxaemic test in a well-balanced sequence assuring equal numbers of either sequence.

Figure 1.

Experimental design
Experiments were conducted in three visits. In visit 1, subjects underwent an incremental exercise test in room air. In visit 2, subjects completed a second incremental test, breathing a high fraction of inspired O2 gas mixture. In visit 3, subjects completed in a balanced ordering sequence two 5-min exercise tests, separated by 90 min, breathing either air or a high fraction of inspired O2 gas mixture.

Incremental exercise tests

The incremental exercise tests were performed on an electromagnetically braked cycle ergometer (Ergoline 800; Sensor Medics, Anaheim, CA, USA) starting at 20 W and increasing by 20 W every minute, with the subjects maintaining a pedalling frequency of 70–90 r.p.m. Tests were preceded by a 3 min rest period, followed by 3 min of unloaded pedalling. The following pulmonary gas exchange and ventilatory variables were recorded breath-by-breath (Vmax 229; Sensor Medics, Anaheim, CA): oxygen uptake inline image, carbon dioxide elimination inline image, respiratory exchange ratio (RER), minute ventilation inline image, tidal volume (VT), and breathing frequency (f). Heart rate (HR) and %Sinline image were determined using the R–R interval from a 12-lead online electrocardiogram (Marquette Max; Marquette Hellige GmbH, Germany) and a pulse oximeter (Nonin 8600; Nonin Medical, USA), respectively.

HI and LO exercise tests

During these tests, measurement of pulmonary gas exchange was performed as mentioned above. Arterial blood was taken every minute throughout the exercise tests, whereas continuous monitoring of oesophageal pressure (Poes), gastric pressure (Pga), and swings in transdiaphragmatic pressure (ΔPdi), averaged over 20 s breath samples every minute, was also performed throughout the exercise tests. Tests were always preceded by 5 min of warm-up cycling at 50%inline image in room air. In order to match the ventilatory requirement between the two 5 min exercise tests, the work rate of the second test was adjusted to produce the same inline image as during the first 5 min exercise test.

Assessment of diaphragmatic fatigue

To assess diaphragmatic fatigue, Pdi,tw was measured before and 10, 20, 40 and 60 min following the two exercise tests. Pdi,tw was obtained by subtracting Poes from Pga, which were assessed by two commercially available balloon catheters cut to 110 cm (Ackrad Laboratories, Inc, Crandford, NJ, USA). The balloons were inserted by nasal intubation following the application of 2% lidocaine anaesthetic gel to the nose and, with the assistance of continuous pressure monitoring, the two balloon tips were positioned in the oesophagus and stomach, respectively. Monitoring of the pressures was performed with the use of Validyne MP45 transducers (Validyne Corporation, Northridge, California, USA), and the pressure signals were recorded and analysed using a Direc Win recorder (model 218A).

At rest and during recovery, Pdi,tw was recorded during stimulation of the phrenic nerves at the neck according to recommended techniques (Hamnegard et al. 1996; Polkey et al. 1998; Mador et al. 2000, 2002; Man et al. 2004). The phrenic nerve roots were bilaterally stimulated with cervical magnetic stimulation (CMS) using a Magstim 200 magnetic stimulator with a circular (doughnut-shaped) 90 mm coil (9450-23-P12) with a maximum magnetic field of 2.3 T (Magstim Whitland, Dyfed, UK). To optimize the position of the coil and the accuracy of measurement, the neck was flexed. Several stimulations were conducted over the spinal processes at varying power outputs in the midline between C5 and C7, in order to determine correctly the position on the neck at which the maximal response could be elucidated (Hamnegard et al. 1996; Mador et al. 2002). Supramaximal stimulation was indicated by a plateau in Pdi,tw with increasing power output (Mills et al. 1995, 1996; Mador et al. 2000, 2002). Then, the position of magnetic stimulation on the neck was marked with an indelible marker and thereafter all measurements were conducted at the exact same position and at full stimulator output. During recovery the correct position on the neck was re-evaluated and confirmed by a plateau in Pdi,tw after supramaximal stimulation. To avoid twitch-on-twitch potentiation, adjacent measurements were performed 30 s apart (Wragg et al. 1994). The magnetic stimulation was performed with the subject in a seated position, with a nose clip on and with the mouth closed. Subjects were instructed to breathe quietly, then to perform a gentle expiratory effort to functional residual capacity (FRC) and to hold their breath while the magnetic stimulation took place. We relied on Poes as a measure of position in the respiratory cycle relative to FRC. Hence, Poes immediately before stimulation was always carefully evaluated to ensure constant lung volume (Mills et al. 1995, 1996; Hamnegard et al. 1996; Polkey et al. 1998; Mador et al. 2002). Once relaxation was achieved (as judged by leveling off of Pdi and Poes) the operator performed the stimulation. Twitch responses were rejected for analysis when Poes immediately before stimulation was more than 1 cmH2O pressure difference from that at FRC (Roussos et al. 1979). Continuous feedback of Poes was provided via the computer screen. A total of 10 measurements at each time point were conducted at full magnetic output and the average value of the five best measurements was used for the analysis.

Arterial blood gas measurements

Arterial tensions of O2(Pinline image), and CO2(Pinline image), pH and percentage arterial oxygen saturation (%Sinline image) were measured from 2 ml blood samples on a blood gas electrode system combined with a co-oximeter (ABL 625, Radiometer, Copenhagen, Denmark) within 10 min of collection. Blood samples, taken every minute during the 5 min tests and 10 and 60 min into recovery, were kept on ice prior to measurement. The blood gas analyser was auto-calibrated every 4 h throughout the day, and calibrating gases of known concentrations were run before each set of measurements. Blood gas measurements were corrected for the subject's tympanic temperature taken during withdrawal of each arterial blood gas sample.

Statistical analysis

All data are reported as mean ±s.e.m. Paired and independent t tests were employed to determine differences in baseline characteristics within and between groups, respectively. Two-way ANOVA with repeated measures was used to identify statistically significant differences across different time points between the HI and LO intensity exercise tests for each variable. When overall significance was obtained, differences between conditions were identified with the Tukey's post hoc test. With respect to Pdi,tw values at baseline and at the 10th and 20th minute of recovery, where for each subject the five best measurements within each time point and condition were considered, a mixed effect model (Laird & Ware, 1982) was used to account for the subject's variability in Pdi,tw. Linear regression analysis was performed using the least squares method. The level of significance was set at P < 0.05.

Results

Incremental exercise tests

Table 1 shows the subjects' responses to both incremental exercise tests. All subjects demonstrated a reduction in %Sinline image≥ 4% at the end of the incremental exercise tests in room air. At the end of the second incremental test while breathing the high inline image, WRmax, inline image, fmax and %Sinline image were significantly different compared to the test in room air (Table 1).

High workload (HI) and low workload (LO) exercise

The mean workload achieved in the two 5 min tests represented 80 ± 3% (LO) and 90 ± 3% (HI) WRmax achieved in the incremental exercise tests in room air. Mean workload throughout the HI test was significantly (P= 0.001) greater than the LO test (Fig. 2A). Subjects reached a plateau in inline image at the last 2 min of the LO and HI tests, with a mean inline image of 65.6 ± 2.6 and 72.0 ± 2.3 ml kg min−1, respectively, that corresponded to 94 ± 2 and 103 ± 3% of inline image of the incremental exercise test in room air, respectively (Fig. 2B). inline image was significantly higher in the HI compared to the LO test (P= 0.010), whereas HR was not different between the two tests, reaching, at the last minute of both exercise tests, 94 ± 2% of maximum (Fig. 2C) recorded during the incremental exercise test in room air.

Figure 2.

Exercise load and O2 delivery responses during the two 5-min exercise tests
Work rate (WR; A), Oxygen uptake (inline image; B), and heart rate (HR; C) throughout the low (80% WRmax (○)) and high (90% WRmax (•)) work rate tests. Values are means ±s.e.m. for 10 subjects. *Significant differences between the two conditions, P < 0.05.

Ventilatory response to the exercise

The ventilatory response to the exercise tests is shown in Fig. 3. inline image increased with exercise in both tests, reaching a maximum value at the end of exercise that corresponded to 85 ± 5% of the maximum value attained at the incremental test in room air (Fig. 3A). No significant differences were found in inline image between the two tests. Neither VT nor f differed between the two tests (Fig. 3B and C).

Figure 3.

Ventilatory responses during the two 5-min exercise tests
Minute ventilation (inline image; A), tidal volume (VT; B) and breathing frequency (f; C) throughout the low (80% WRmax (○)) and high (90% WRmax (•)) work rate tests. Values are means ±s.e.m. for 10 subjects.

Arterial blood sampling

Throughout both exercise tests %Sinline image, Pinline image  and Pinline image were significantly (P < 0.001) higher in the HI versus the LO tests (Fig. 4AC), whereas arterial lactate, pH and body temperature were not significantly different (Fig. 4DF). Arterial lactate and pH did not differ 10 min into recovery (Fig. 4E and F) and returned to baseline levels within 60 min of recovery. Changes in Pinline image throughout both HI and LO tests were not correlated with the degree of fall in post-exercise Pdi,tw.

Figure 4.

Blood variables during the two 5-min exercise tests
Arterial O2 pressure (Pinline image; A), arterial CO2 pressure (Pinline image; B), arterial O2 saturation (Sa,O2%; C), tympanic temperature (°C; D), arterial lactate concentration (Lactate conc.; E) and pH (F), at the end of 50%VO2,max warm-up (WU), throughout the low (80% WRmax (○)) and high (90% WRmax (•)) work rate tests and at the 10th minute of the recovery period (REC). Values are means ±s.e.m. for 10 subjects. *Significant differences between the two conditions, P < 0.05.

Diaphragmatic fatigue

Figure 5 shows the mean changes in ΔPdi during the HI and LO intensity exercise test. No differences were found in ΔPdi during the two exercise tests. Mean ΔPdi values during the HI and LO tests were 37.1 ± 2.4 and 38.2 ± 2.8 cmH2O, respectively.

Figure 5.

Response of swings in transdiaphragmatic pressure during the two 5-min tests
Changes in transdiaphragmatic pressure (Pdi) throughout the low (80% WRmax (○)) and high (90% WRmax (•)) work rate tests. Values are means ±s.e.m. for 10 subjects.

The time course of the percentage drop in Pdi,tw after exercise in the first and second exercise test (irrespective of which workload was undertaken first) is shown in Fig. 6A, showing lack of significant ordering effect. As shown in Fig. 6B, the major finding of our study was that the HI test induced a significantly (P= 0.022) greater fall in Pdi,tw 10 min after termination of exercise as compared to the LO test. The fall in Pdi,tw was still significant (P= 0.029) at the 20th minute of recovery only after the high work rate test (Fig. 6B).

Figure 6.

Response of twitch transdiaphragmatic pressure during recovery
Percentage fall in twitch diaphragmatic pressure (Pdi,tw) during recovery after A, the first (▴) and second (▵) exercise tests; B, the high (90% WRmax (•)) and low (80% WRmax (○)) work rate tests. Values are means ±s.e.m. for 10 subjects. *Significant difference between the two conditions, P < 0.05; †significant difference from baseline P < 0.05.

Note that 40 min after completion of the first exercise test, mean Pdi,tw had fully recovered to the pretesting levels (Fig. 6A). As a result, Pdi,tw at the baseline of the second test was at similar resting levels as for the first test. In addition, mean Pdi,tw 60 min after both HI and LO tests was not different from the pretesting levels (Fig. 6B). Accordingly, as shown in Fig. 7A, no significant (P= 0.41) effect of the order of the two intensity exercise tests was found on the highest percentage drop in Pdi,tw. However, work rate affected the degree of fatigue as the HI normoxaemic test (Fig. 7B and C) induced a significantly (P= 0.02) greater fall in Pdi,tw (−15%) compared to the LO test (−9.0%). The within-trial coefficient of variation in Pdi,tw assessed at the 10th min of recovery after the HI and LO tests was 7.1 ± 1.2 and 5.3 ± 1.0, respectively.

Figure 7.

Factors impacting on twitch transdiaphragmatic pressure during recovery
Effect of: A, exercise test order (first versus second exercise test); B, end-exercise Sa,O2% (low versus high Sa,O2%); and C, power output (low versus high work rate) on the individual (○) and mean ((•) (n= 10)) changes in percentage Pdi,tw from baseline. *Significant effect (P= 0.02) on the change in Pdi,tw was found for the higher power output at the high Sa,O2%.

The difference in the reduction in Pdi,tw at the 10th minute of recovery between the HI and LO tests was significantly correlated with the difference in mean work rate sustained during the two tests (r= 0.65, P= 0.041), but not with the difference in arterial pH (r= 0.34, P= 0.338), measured at the 10th minute of recovery following the two tests.

Discussion

There is evidence that diaphragmatic fatigue occurs in highly fit subjects during exhaustive exercise (Babcock et al. 1996). Additionally, it has been reported that approximately 50% of highly trained individuals exhibit arterial hypoxaemia during high-intensity exercise (Powers et al. 1988; Dempsey & Wagner, 1999) and that hypoxaemia worsens the degree of diaphragmatic fatigue (Babcock et al. 1995a). However, prior studies did not investigate individuals naturally demonstrating EIAH, and the employment of hypoxic challenge caused not only hypoxaemia but also a higher respiratory muscle effort because of the additional hypoxic stimulus to ventilation (Babcock et al. 1995a; Cibella et al. 1996; Gudjonsdottir et al. 2001). This makes it difficult to isolate the effect of hypoxaemia per se on fatigue. The purpose of this study was therefore to investigate whether, near the maximal levels of exercise, EIAH per se contributes to diaphragmatic fatigue in highly trained endurance athletes. The experimental design aimed at comparing the degree of diaphragmatic fatigue when ventilation was similar under conditions where Pinline image and leg work rate varied. We found that both HI (90% WRmax) and LO (80% WRmax) exercise tests induced significant diaphragmatic fatigue. However, exercise at the higher leg work rate (conducted in normoxaemia) resulted in a significantly greater reduction in Pdi,tw compared to the lower leg work rate test (conducted in hypoxaemia). This was found in spite of no differences in tidal volume or frequency between the two tests. These data suggest that EIAH per se is not the predominant causative factor of diaphragmatic fatigue, at least at the level of arterial oxygen saturation studied.

Respiratory muscle loading during the HI and LO tests

Factors leading to diaphragmatic fatigue have been suggested to include increased work by the diaphragm and reduced blood flow/oxygen availability (Babcock et al. 2002). It has been shown that hypoxaemia superimposed on high-intensity exercise hastens diaphragmatic fatigue during exercise (Babcock et al. 1995a, 1995b; Gudjonsdottir et al. 2001). However, hypoxaemia also stimulates ventilation, so that this strategy created not only hypoxaemia but also increased respiratory muscle load, making the role of hypoxaemia difficult to isolate. The key design feature of the present study was to control for respiratory muscle load by matching tidal volume, frequency and ventilation in normoxia and hyperoxia by adjusting leg work rate. Greater fatigue in hypoxaemic conditions would then point to the importance of Pinline image in diaphragmatic fatigue. Conversely, based on the suggestion that at high work rates there is competition for perfusion between the diaphragm and legs (Johnson et al. 1996), we reasoned that at the higher exercise intensity in normoxaemia, diaphragm blood flow might be compromised and lead to greater fatigue if blood flow and not Pinline image were more important, and that is what we found.

Effect of arterial hypoxaemia on diaphragmatic fatigue

In both exercise tests, we observed diaphragmatic fatigue as defined by a significant decline in Pdi,tw from pre- to post-exercise when CMS was applied (Similovski et al. 1989; Mador et al. 2000, 2002). The finding that the post-exercise percentage fall in Pdi,tw was higher after the normoxaemic test (Fig. 7B) suggests that arterial hypoxaemia per se is not a dominant factor in diaphragmatic fatigue, at least at the moderate level (Dempsey & Wagner, 1999) of arterial oxygen saturation tested (92 ± 1%) in the present study. At more severe levels of arterial hypoxaemia (83 ± 1%), diaphragmatic force-generating capacity has been shown to be more impaired during cycling at high altitude compared to equivalent work rates at sea level (Gudjonsdottir et al. 2001). The greater diaphragmatic fatigue observed by Gudjonsdottir et al. (2001) compared to the present study might therefore be attributed to the lower percentage Sinline image values attained in that study, resulting from the more severe hypoxic stimulus. Our conclusions should therefore be limited to conditions naturally causing desaturation to no greater a degree than seen in our study. We purposefully sought to investigate the effect of moderate arterial hypoxaemia, as the majority of endurance athletes who develop EIAH (Powers et al. 1988; Dempsey & Wagner, 1999; Nielsen et al. 2002) typically exhibit saturation values in the range seen in the present study (87–93%). Thus, our results seem appropriate to real-world conditions at sea level.

Effect of leg work rate on diaphragmatic fatigue

Our finding that the percentage drop in Pdi,tw was significantly greater following the HI than the LO tests is in line with the results of Johnson et al. 1993), showing that in healthy subjects with a variety of fitness levels, the higher the relative exercise intensity and percentage of inline image sustained during exercise, the greater the amount of work done by the diaphragm and the degree of diaphragmatic fatigue determined following exercise. However, in the work of Johnson et al. (1993), respiratory muscle load was higher the higher the power output. In the present study, respiratory effort throughout exercise was designed not to be greater in the high- than in the low-work rate tests. Our findings lead us to hypothesize that the higher degree of diaphragmatic fatigue following the higher leg work rate normoxaemic tests should be attributed to factors governing its blood flow (Babcock et al. 2002). It is reasonable to speculate that diaphragm blood flow would be compromised more during the high-leg-work rate tests, as the greater recruitment of the leg musculature presumably required a greater share of an already maximal cardiac output compared to the low leg work rate tests. Recent work by Mortensen et al. (2005) demonstrated that during incremental cycling, cardiac output increases linearly to 80% of peak power and then plateaus, whereas during constant-load cycling sustained at 85% of peak power, cardiac output reaches maximal values within 5 min. The notion of attainment of maximal cardiac output in both HI and LO tests is also supported by the finding that maximal cardiac output was similar between mild hypoxaemic (16%Finline image) and normoxaemic exercise conditions in fit subjects performing two-leg knee extension exercise, while the lower oxygen delivery in hypoxaemia led to lower whole-body oxygen uptake (Koskolou et al. 1997). Similarly, during maximal cycling under hyperoxia compared to normoxia, maximal cardiac output (Peltonen et al. 2001) has been reported not to differ, whereas oxygen delivery to the muscles was higher in hyperoxia leading to higher inline image (Knight et al. 1993). In the present study inline image was indeed significantly higher in the normoxaemic high-leg-work rate tests than in the hypoxaemic lower-leg-work rate tests, possibly because of the greater oxygen delivery (Knight et al. 1993; Koskolou et al. 1997) and leg blood flow (Mortensen et al. 2005). However, whole-body inline image reached near maximal values in both normoxaemic and hypoxaemic tests (103 and 94%inline image, respectively). Therefore, assuming that cardiac output was maximal and not different in the two tests, as implied by the similar heart rates, we reasoned that during the higher work rate normoxaemic tests, the locomotor muscles which in that condition produced higher work output, received a greater share of blood flow at the expense of other metabolically active tissues, including the respiratory muscles. Therefore, even if ΔPdi was similar, diaphragmatic blood flow was presumably decreased at the high-work-rate tests as the locomotor muscles would require more blood (Mortensen et al. 2005). Hence, one would expect a greater degree of diaphragmatic fatigue in the latter case due to a greater degree of competition for the available blood flow.

Further support for the hypothesis that blood flow limitation and not hypoxaemia is more important to developing diaphragmatic fatigue comes from the work of Harms et al. (1998, 2000) establishing a link between the amount of respiratory muscle work and limb blood flow at maximal levels of exercise. According to these investigators, increasing the amount of respiratory muscle work at maximal exercise significantly reduces leg blood flow and inline image, and increases leg vascular resistance, whereas a decrease in the amount of respiratory work increases blood flow and inline image and decreases vascular resistance. This has been attributed to an increased muscle sympathetic nerve activity and reflex vasoconstriction with respiratory muscle loading, and vice versa. According to Harms et al. (2000), these changes in sympathetic outflow are mediated by a muscle chemoreflex mechanism known to originate from type III and IV afferents in contracting muscles (Pickar et al. 1994) and in the diaphragm (Hussain et al. 1991). Accordingly, if such a regulatory reflex mechanism underlies the distribution of blood flow between the respiratory and locomotor muscles (Rowell & Cleary, 1990), keeping respiratory muscle work similar while increasing the work output by the legs, as occurred in the present study, could possibly evoke the same regulatory mechanism, but in the opposite direction. Therefore, the greater degree of diaphragmatic fatigue observed at the higher-leg-work rate in the present study could be the result of compromised blood flow to the respiratory muscles (Savard et al. 1989; Wetter et al. 1999; Harms et al. 2000).

Effect of metabolic by products on diaphragmatic fatigue

Acidosis and/or accumulation of metabolic byproducts, such as lactate, in the active musculature have been proposed as possible contributors to diaphragm fatigue (Fitzgerald et al. 1984; Fregosi & Dempsey, 1986; Babcock et al. 1995a). Previous work by Fregosi & Dempsey (1986) has shown that muscle lactate content in the diaphragm increases as blood lactate concentration increases after whole-body normoxic exercise, suggesting that the diaphragm may become progressively acidic, thus contributing to fatigue. In the present study, arterial lactate and pH were not significantly different between the two tests and therefore the difference found in diaphragmatic fatigue is unlikely to be due to either of the two factors. This is also confirmed by the lack of a significant correlation between the difference in post-exercise arterial pH measured 10 min into recovery after the HI and LO tests, and the difference in the reduction in Pdi,tw recorded at the 10th minute of recovery following the HI and LO tests.

Potential limitations of the study

Performing both 5 min exercise tests on the same day could have presented a limitation to the present study. However, the complexity of the experiment and the invasive techniques employed prevented us from testing the subjects on separate days. This is the reason why we performed the tests in a balanced ordering sequence to assure equal numbers of either load application sequence. In addition, the time length of the tests was designed to be the maximal sustainable in order to achieve high work rates and maximal values for inline image and cardiac output in our fit subjects (Wagner, 2000; Mortensen et al. 2005). Moreover in fit subjects who are susceptible to EIAH, an exercise test lasting 4–5 min has been reported by Dempsey & Wagner (1999) to be the ideal duration of exercise for the phenomenon of arterial hypoxaemia to be more profound.

Furthermore, we allowed a 90 min period for recovery between the two tests because in the paper by Johnson et al. (1993), the Pdi,tw values had recovered almost completely by an average time of 70 min in subjects exercising in normoxic conditions, while in the study of Babcock et al. (1995a)Pdi,tw had recovered by approximately 90 min following hypoxic exercise. It should be noted that by the 60th minute into recovery, pH and arterial lactate as well as, oxygen uptake and heart rate had returned to pre-exercise levels. It is therefore likely that restoration of blood flow and energy supplies during recovery, as well as removal of lactate from the circulation may explain the quick recovery of Pdi,tw in our study. Indeed, Pdi,tw had completely returned to baseline 60 min after the first test, therefore allowing the absolute value of Pdi,tw at the beginning of the second test to be at an identical level to that before the first test. Nevertheless, because we could not ensure that the diaphragm would be truly ‘back to normal’ but only that Pdi,tw would return to baseline after a 90 min period of rest, we applied a balanced allocation of the trials in our crossover protocol design in order to take into account the possible long-lasting (but not revealed by Pdi,tw measurement) effects of diaphragmatic fatigue.

To determine the degree of diaphragmatic fatigue in the present study we assessed post-exercise changes from baseline in Pdi,tw using CMS of the phrenic nerves. Although this technique coactivates extra-diaphragmatic musculature, it is well tolerated and compares favourably to bilateral phrenic nerve stimulation (BPNS) in terms of assessing diaphragmatic contractility and detecting diaphragmatic fatigue (Hamnegard et al. 1996; Mador et al. 2002; Man et al. 2004). Even though Pdi,tw is larger with CMS compared with BPNS, CMS is technically easier to perform, requiring fewer trial stimulations to confirm supramaximality (Man et al. 2004), which we thought would be advantageous for this study, in which comparative measurements had to be performed repeatedly at precise time intervals in highly trained athletes undergoing two exercise protocols on the same day. Furthermore, although we did not measure diaphragmatic EMG activity, but as in other studies (Hamnegard et al. 1996; Polkey et al. 1998; Mador et al. 2000; Mador et al. 2002) we relied only on Pdi,tw measurements to reproduce consistent CMS, the within-trial coefficient of variation in Pdi,tw assessed at the 10th minute of recovery after the HI and LO tests was 7.1 ± 1.2 0 and 5.3 ± 1.0, respectively, suggesting reasonably consistent stimulation from one series of measurements to another. This figure is in agreement with other reproducibility studies utilizing the same technique that showed that the within-occasion variability ranges between 5 and 8% (Mills et al. 1995, 1996; Mador et al. 2000, 2002).

Significance of EIAH on diaphragm fatigue in highly trained athletes

Our findings provide new insights into the functional effects of EIAH in highly trained endurance athletes. First, we have been able to quantify diaphragmatic fatigue induced by high-intensity (>90%inline image) normoxic exercise naturally causing moderate levels of arterial desaturation in this highly trained population. Interestingly, the degree of diaphragmatic fatigue in these elite athletes was within the range previously reported for healthy humans with a variety of fitness levels (Johnson et al. 1993, 1996), hence suggesting that EIAH does not exacerbate diaphragm fatigue in these athletes. Second, by preventing EIAH, we showed that the effect of work rate on diaphragm fatigue was greater than that of hypoxaemia, at least at the level of Sinline image tested. Hence, when leg work energy requirements are near maximal, blood flow to the diaphragm and not its O2 delivery per se, is more important in causing diaphragmatic fatigue in exceptionally fit athletes. The latter suggestion is in line with studies (Nielsen et al. 1998, 1999, 2002) that showed that there is not a significant effect of EIAH on muscle O2 delivery and that when hyperoxia restores arterial desaturation during exercise, muscle oxygenation is not different from that during exercise in normoxia.

In conclusion, the difference found in diaphragmatic fatigue between normoxaemia (at higher leg work rate) and hypoxaemia (at lower leg work rate) suggests that neither arterial hypoxaemia nor lactic acidosis are likely predominant causative factors in diaphragmatic fatigue in this population, at least at the level of Sinline image tested. Rather, this result leads us to hypothesize that blood flow competition with the legs is an important contributor to diaphragmatic fatigue in heavy exercise, assuming that higher leg work required greater blood flow.

Appendix

Acknowledgements

This work was supported by the Thorax Foundation

Ancillary