Diaphragm fatigue during exercise at high altitude: the role of hypoxia and workload

Subjects

Eight healthy male subjects (age 38±2 yrs) born and living at SL, not acclimatized to HA, gave their consent to perform the study.

Measurements

Flow (V') was measured with a mass flow sensor (Sensormedics, Yorba Linda, CA, USA) and volume (V) was obtained by integrating the flow signals. The mass flow sensor was calibrated with a 3 L syringe before each test.

Oesophageal (P_oes) and gastric (P_ga) pressure changes were measured as previously reported 8. Airway opening pressure (P_ao) was measured via a side-port placed on a two-way breathing valve (720068, Jaeger, Würzburg, Germany) connected to the mass flow sensor. Transpulmonary (Pl) and transdiaphragmatic (P_di) pressures were obtained by subtracting P_oes from P_ao and P_ga, respectively. Pl, V' and V data were used to calculate dynamic lung elastance (E_L,dyn) and lung resistance (Rl) 9.

Maximal oesophageal and transdiaphragmatic pressures were measured with the Mueller 10 (P_oes,max and P_di,max, respectively) and the sniff 10 (P_oes,sniff and P_di,sniff, respectively) manoeuvres. End-expiratory lung volume (EELV) before sniffs was monitored using end-expiratory P_oes 11, 12. To assess diaphragmatic length, P_ga was continuously monitored 11. Criteria for accepting and correcting P_oes,sniff for abdominal muscle contractions (P_{oes,sniff corr}) were those described by Kyroussis et al. 12. The sniff showing the greatest pressure deflection in each condition and fulfilling all criteria 12 was selected for analysis. Figure 1⇓ shows a representative record of sniffs during the protocol.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Representative recordings of maximal inspiratory pressures measured with the sniff manoeuvre in subject 6 at SL. a-d) maximal inspiratory oesophageal (P_oes,sniff), e-h) gastric (P_ga,sniff) and i-l) transdiaphragmatic (P_di,sniff) pressures, measured at baseline (a, e, i), maximal exercise (b, f, j), and at the 1st min (c, g, k) and 60th min of recovery (d, h, l).

To estimate the diaphragmatic and respiratory muscle effort, the pressure-time product for diaphragm (PTP_di) and respiratory muscles (PTP_oes) over 1 min were computed 8. Since diaphragm external work rate (W'_di) has been shown to be a better index of diaphragm effort than PTP_di when inspiratory flow is not constant 13, the former was also computed, by multiplying PTP_di by the actual mean inspiratory flow 13.

Oxygen consumption (V'_O2), and carbon dioxide production (V'_CO2) were measured breath-by-breath (Vmax29; Sensormedics, Yorba Linda, CA, USA). Gas analysers were calibrated before each test. Heart rate (HR) was measured from a 12-lead electrocardiogram.

Arterialized venous blood 14 was obtained to estimate P_a,CO2 (ABL 300; Radiometer, Copenhagen, Denmark) and measure lactate (2300 STAT; Yellow Spring Instrument, Yellow Spring, OH, USA). P_a,CO2 and barometric pressure (Pb) were used to calculate the alveolar oxygen partial pressure (P_A,O2) 15. A finger oximeter (BIOX 3760; Datex-Ohmeda, Division Instrumentarium Corp., Helsinki, Finland) measured arterial O₂ saturation (S_a,O2).

Protocol

Measurements were performed at baseline, during an incremental exercise test to volitional tolerance on cycle-ergometer (Ergo-metrics 800S; Sensormedics, Yorba Linda, CA, USA) and during the hour after exercise. The protocol was performed both at 3,325 m of altitude (Rifugio Torino, Monte Bianco, Italy) and at SL. At HA, Pb and resting P_A,O2 were 66.93±0.14 and 7.60±0.27 kPa respectively. At HA, the subjects were studied within a few hours of the ascent (cableway), sitting on the cycle-ergometer with the thorax kept in the same position throughout the study.

Additional experiments

On a separate day, four subjects repeated the exercise at SL using cervical magnetic stimulation (CMS) to obtain twitch maximal transdiaphragmatic pressure (P_di,tw) and the degree of voluntary neural activation during maximal voluntary efforts by the twitch occlusion technique 17. Right and left diaphragmatic electromyograms (EME_di) were obtained using surface electrodes (Ref. 9013L0202; Dantec Medical, Tonsbakken, Denmark) connected to an electromyograph (Mystro+ MS20, Medelec, Woking, Surrey, UK) 18, amplified and band pass filtered (band width 20 Hz–5 kHz).

CMS was performed by a Magstim 200 stimulator equipped with a circular (doughnut shaped) 90 mm coil with a maximum magnetic field of 2.3 Tesla (Magstim, Whitland, Dyfed, UK) 18, 19. Stimulation amplitude was 100% of the maximal output of the stimulator. Supramaximal stimulation was verified according to Similowski et al. 19. Stimuli were delivered at end-expiration with the airways closed.

P_di,tw and, consecutively, twitch occlusion were performed firstly 10 min after end-exercise to avoid twitch potentiation, and then after 60 min of recovery.

Data and statistical analysis

All data are reported as mean±sem. A paired t-test 20 was used to compare data, i.e. maximal WR (WR_max), iso-WR data at SL and HA, and for comparison between P_di,max and P_di,sniff. One-way analysis of variance (ANOVA) 20 was used for determination of differences in mean values for P_di,sniff, lactate and V'_E during baseline, exercise and recovery, and, when allowed by the F-value, the significance between measures was computed using Fisher's Protected least significant difference (PLSD) test. Significance was set at p<0.05.

Maximal exercise

Table 2⇓ shows WR, metabolic parameters, HR, and breathing pattern at maximal exercise. WR_max was similar at SL and HA, even though three subjects reached a lower WR at HA. At SL the subjects reached predicted maximal V'_O2 (V'_O2,max) and HR (HR_max). At HA both V'_O2,max and HR_max were lower (12% and 6%, respectively), though maximal V'_CO2 (V'_CO2,max) was similar. El was similar at SL and HA (table 1⇑). Rl was lower at HA compared to SL (−11%, table 1⇑). The breathing pattern at WR_max was similar at SL and HA. Figure 2a⇓ shows average P_di,sniff values. P_di,sniff was lower at WR_max compared to baseline (SL: average 72%, range 61–80%; HA: average 61%, range 37–80%) in all but one test at SL (94%). P_di,sniff had a positive gastric component (P_di,sniff>P_oes,sniff) before exercise in all instances. By contrast, at WR_max, the gastric component of P_di,sniff became always negative (P_di,sniff< P_oes,sniff) at HA, while at SL it was negative in six subjects and reduced, but still positive in the remaining two. At SL but not at HA P_di,sniff recovered to baseline within 60 min.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.—

Mean±sem of: a) the maximal transdiaphragm pressure (P_di,sniff), and of b) the pressure-time product for the diaphragm expressed as a fraction of the oesophageal pressure-time product (PTP_di/PTP_oes) at baseline (B), maximal exercise (M), and during 60 min of recovery (R₁-R₆₀) measured at sea level (○) and at high altitude (•). During maximal exercise the PTP_di/PTP_oes ratio dropped below unity indicating that the majority of the work of breathing had been shifted from the diaphragm to the accessory inspiratory muscles. *: p<0.05 condition versus B. (sea level versus high altitude)

At SL, P_oes,sniff was higher at WR_max (11.34± 0.49 kPa) than at baseline (9.97±0.39 kPa), whereas at HA it was equal (WR_max: 10.36±0.78 kPa; baseline: 8.99±0.29 kPa). P_{oes,sniff corr} showed similar results at WR_max (SL: 11.24±0.39 kPa; HA: 10.17±0.68 kPa).

At SL, PTP_di averaged 80.74±9.87 kPa·s⁻¹ over 1 min at WR_max, all subjects being on or above the fatigue threshold of 53.76–58.65 kPa·s⁻¹ over 1 min 4. PTP_di at WR_max was lower at HA than at SL (64.13±8.21 kPa·s⁻¹ over 1 min), but five subjects were still above and three lay just below the fatigue threshold (42.03–44.97 kPa·s⁻¹ over 1 min). PTP_di/PTP_oes ratio fell below unity at WR_max both at SL and at HA (but was lower at HA), and recovered to baseline within 5 min in both conditions (fig. 2b⇑).

At SL, lactate increased from 0.9±0.1 mEq·L⁻¹ at rest to 6.2±0.7 mEq·L⁻¹ at the 5th min of recovery, regaining baseline at 60 min (1.0±0.1 mEq·L⁻¹). At HA, lactate increased from baseline (0.9±0.1 mEq·L⁻¹) to the 5th min of recovery (6.7±0.6 mEq·L⁻¹) but, as with P_di,sniff, lactate did not recover to baseline within 60 min (1.3±0.2 mEq·L⁻¹, p<0.05). In contrast to P_di,sniff and lactate, V'_E reached baseline (V'_E at SL: 12.8±1.6 L·min⁻¹; V'_E at HA: 15.3±2.3 L·min⁻¹) within 15 min at SL (15.2±2.4 L·min⁻¹) and within 30 min at HA (14.9±1.5 L·min⁻¹).

Isowork rate

WR_max was lower at HA in three subjects. To data at isowork rate conditions (iso-WR) compared in each subject, WR_max at HA was matched in these three subjects with a similar WR obtained at SL. At iso-WR, P_di,sniff was significantly lower at HA (fig. 3a⇓), P_oes,sniff never being below baseline (fig. 3b⇓). V'_E was higher at HA (117.2±11.4 L·min⁻¹) than at iso-WR SL (91.3± 13.3 L·min⁻¹) because of higher frequency (35.9±3.1 and 30.6±2.1 breaths·min⁻¹, respectively, p<0.05). Similar results were obtained for mean inspiratory flow (WR_max HA: 3.64±0.41 L·s⁻¹; iso-WR SL: 3.03±0.29 L·s⁻¹, p<0.05). Notwithstanding the higher V'_E at HA, PTP_di and W'_di were similar at iso-WR (fig. 4⇓).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.—

Mean±sem of: a) maximal transdiaphragmatic (P_di,sniff) and b) oesophageal (P_oes,sniff) pressures measured at baseline (B) and at isowork rate (iso-WR) at sea level (SL; ○) and at high altitude (HA; •). Iso-WR: workload at SL matched to maximal workload reached at HA. The reduction of P_di,sniff at iso-WR with respect to baseline was greater at HA. By contrast, P_oes,sniff increased slightly both at SL and HA at iso-WR. *: p<0.05 B versus iso-WR (SL versus HA).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.—

Mean±sem of: a) pressure-time product for the diaphragm (PTP_di) and of b) the mechanical work performed by the diaphragm (W'_di). W'_di is the product of PTP_di and mean inspiratory flow (V_T/t_I). These parameters are shown at baseline (B) and at isowork rate (iso-WR) measured at sea level (SL; ○) and at high altitude (HA; •). Iso-WR: workload at SL matched to maximal workload reached at HA. PTP_di was similar at SL and HA at both B and iso-WR. The same finding applied to W'_di which takes into account changes in diaphragm effort induced by flow rate variations. These data indicate that the effort of the diaphragm at iso-WR was similar both at SL and HA. *: p<0.05 B versus iso-WR.

Volitional versus objective measurements

In four subjects retested at SL, both P_di,tw and P_di,sniff decreased significantly 10 min after exercise, and recovered 60 min after exercise (table 3⇓). All subjects showed near complete activation of the diaphragm, as no significant increase of P_di due to CMS was found (fig. 5⇓). P_di,sniff was similar to P_di,max in any condition (fig. 5⇓), showing near complete activation of the diaphragm also with this manoeuvre.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.—

Maximal transdiaphragmatic pressure obtained with a “sniff” manoeuvre a, c, e, g, i, k) and with a Mueller manoeuvre to which a single supramaximal twitch was superimposed (b, d, f, h, j, l) in a representative healthy subject at rest a–d), at 10 min (e–h), and at 60 min (i–l) after the end of an exhaustive incremental exercise on cycle-ergometer performed at sea level. Electromyograms of the right hemidiaphragm in response to supramaximal cervical phrenic nerve stimulation (EME_di) are also provided. The superimposition of supramaximal magnetic stimulation (indicated by the electrical artefact in the EME_di tracing) on maximal quasistatic efforts never evoked a further increase in maximal transdiaphragmatic pressure. Note that P_di obtained with the “sniff” manoeuvre was similar to that obtained during the twitch occlusion test in every condition.

Technical considerations

BPNS is considered to have the best potential as a diagnostic test 21. However, the sniff manoeuvre was primarily used because it can be performed during exercise, at variance with BPNS that requires preparation (5–8 min) after exercise 3.

Poor subject effort might impair the sniff manoeuvre 21. All subjects had previous experience in performing the manoeuvre and were very well motivated. Moreover, sniff intrasubject variability was low (mean coefficient of variation at baseline: 2.3±1.0%, range: 0.1–5.9%). Tachypnoea and general exhaustion could limit P_di,sniff performance 21. However, P_di,sniff recovered to baseline later than the breathing pattern, both at SL and HA. Furthermore, in contrast to P_di,sniff, P_oes,sniff did not decrease at maximal exercise, suggesting maximal inspiratory effort even in these extreme conditions.

P_di,sniff depends on lung volume 22. The end-expiratory P_oes was monitored to ensure that sniff manoeuvres were performed at similar EELV, end-expiratory P_oes being within 0.15 kPa of baseline just before sniffs in any tested condition. Similar results were obtained with end-expiratory P_ga. These data suggest that diaphragmatic length was reasonably constant before P_di,sniff manoeuvres.

Finally, according to Travaline et al. 23, P_di,sniff and P_di,max reliability was tested at SL by CMS in a subset of four subjects on a separate day. The trend of P_di,sniff decay was similar in the two tests, P_di,sniff being within ±5% at each workload tested. Similarly to P_di,sniff, P_di,tw was on average 27% lower than baseline 10 min after exercise (table 3⇑), suggesting that P_di,sniff was accurate in detecting losses of diaphragm FGC, at least in the condition tested. It was extrapolated that similar results could be obtained at HA.

Diaphragm fatigue

A decreased diaphragm FGC was found at maximal exercise both at SL and HA, which recovered to baseline late after exercise (fig. 2⇑a), finding that fits the accepted definition of muscle fatigue 21. These data extend to exhaustive incremental exercise, the evidence that diaphragm fatigue can be generated in normal subjects by endurance exercise 4–, 24.

At end-exercise, P_di,sniff decreased, whereas P_oes,sniff did not or increased compared to baseline (figs. 1 and 3⇑⇑). These data are consistent with data obtained at, for example, the first minute of recovery (fig. 1⇑). It may be argued that this observation cannot be explained by fatigue, since both the diaphragm and the extradiaphragmatic muscles serve to reduce the P_oes. However, P_di,sniff was not only reduced compared to baseline, but it also became lower than P_oes,sniff at end-exercise in almost all the conditions tested (see previously). This suggests diaphragm displacement into the rib cage during maximal manoeuvres, since end-expiratory abdominal pressure (P_ab) was close to baseline values at the beginning of the P_di,sniff manoeuvre (i.e. abdominal muscles phasically inactive). Finally, prolonged forced muscle contractions (similar to those developed by the respiratory muscles during exercise) can cause the phenomenon of potentiation 25. In this condition, a P_oes,sniff increase at end-exercise would be expected, but not a decrease in P_di,sniff (rather it should increase too).

The presented finding contrasts with Levine et al. 26 who reported, in preliminary experiments, no sign of diaphragmatic fatigue at end-incremental exercise. However, they provided neither references nor data to support their statement. Short incremental exercise may appear too light to produce diaphragm fatigue. However, increased diaphragm work generated by sustained hyperpnoea at rest, comparable to that found in the present study, has been demonstrated to cause significant diaphragm fatigue 4. Moreover, end-exercise pH and circulating lactate concentrations similar to those found in the present study have been shown to increase the amount of hyperpnoea-induced diaphragm fatigue 4.

Central, transmission and contractile fatigue have been described 10. In the four subjects in whom the twitch occlusion technique 17 was performed, no signs of central fatigue were detected at least at SL (fig. 5⇑). In the same subjects the M-wave was preserved after exercise, indicating absence of transmission fatigue in the condition tested. Finally, the long lasting duration of P_di,sniff decay after exercise at SL and HA (fig. 2⇑), and the postexercise reduction in twitch pressure observed at SL, suggests that contractile fatigue did occur 3, 4, 11.

Effects of high altitude on diaphragm fatigue

Incremental exhaustive exercise at 3,325 m altitude worsened the FGC compared to iso-WR SL (fig. 3⇑) notwithstanding similar effort of the diaphragm (fig. 4⇑). At HA, significant hypoxia was found. Comparable hypoxaemia has been shown to worsen exercise-induced diaphragm fatigue in normobaric conditions 3. Increased diaphragm work was also found during hypoxic exercise at SL, suggesting that its combination with decreased O₂ transport impairs diaphragm FGC in this condition 3. By contrast, it was found that, although V'_E was higher at HA than at iso-WR SL, PTP_di as well as W'_di (fig. 4⇑) and PTP_oes were equal, indicating that in the experimental condition hypoxia per se hastened exercise-induced diaphragm fatigue.

EELV changes affect inspiratory muscle effort estimation 13. EELV was not directly assessed, but it is generally accepted that in young healthy subjects it does not exceed its baseline value during exercise 27. Conversely, the lower density of the air at HA (resulting in decreased airway flow resistance) could explain the increased inspiratory muscle efficiency in generating V'_E 7. Indeed, similarly to previous studies 7, Rl decreased at HA approaching predicted values 28.

Not only did hypoxia impair independently exercise-induced diaphragm fatigue at HA, but it also influenced respiratory muscle interaction during exercise. Aliverti et al. 29 have shown that rib cage and abdominal muscles are progressively activated during exercise, allowing the diaphragm to act as a flow generator. Similar results were obtained in the present study at SL. Indeed, PTP_di/PTP_oes ratio fell below unity at maximal exercise, indicating a strong activation of extra-diaphragmatic respiratory muscles (fig. 2b⇑). At HA, end-exercise PTP_di/PTP_oes ratio (fig. 2b⇑) as well as absolute values of PTP_di significantly decreased compared to SL, suggesting further recruitment of accessory respiratory muscles to assist impending failure of the diaphragm. These observations are indirectly confirmed by studies on hyperoxic exercise that showed a decrease of accessory muscle activation compared to normoxic exercise 24.

At HA, but not at SL, diaphragm recovery from fatigue took longer than 60 min (fig. 2a⇑). It was associated with lower (three subjects) or equal (five subjects) external work at HA, with significantly reduced end-exercise PTP_di, and with blood lactate higher than baseline 60 min after exercise. As lactate clearance is delayed by hypoxia 30, and the increased level of circulating lactate and other metabolites has been implicated in the fatigue process 31, it might well be that hypobaric hypoxia and not increased workload, represents the major cause of the delay in recovery from exercise-induced diaphragm fatigue at HA.

In conclusion, exercise at high altitude challenges the respiratory muscles as a whole. In fact, not only can high altitude enhance diaphragm fatigue during exhaustive incremental exercise, but it also overloads accessory inspiratory muscles on account of hypobaric hypoxia alone, at least during acute exposure to the altitude tested. Whether this could affect exercise capability, particularly when ventilatory requirements become enormous (e.g. in acclimatized subjects at extreme altitudes), is still questionable. In fact, the effects of hypoxaemia are not confined to the respiratory muscles, since changes in oxygen supply affect locomotor muscles as well 24. Further work is required to assess whether exercise-induced diaphragm dysfunction has any role in determining exercise limitation at high altitude.