American Journal of Respiratory and Critical Care Medicine

Although it has been postulated that central inhibition of respiratory drive may prevent development of diaphragm fatigue in patients with chronic obstructive pulmonary disease (COPD) during exercise, this premise has not been validated. We evaluated diaphragm electrical activation (EAdi) relative to maximum in 10 patients with moderately severe COPD at rest and during incremental exhaustive bicycle exercise. Flow was measured with a pneumotachograph and volume by integration of flow. EAdi and transdiaphragmatic pressures (Pdi) were measured using an esophageal catheter. End-expiratory lung volume (EELV) was assessed by inspiratory capacity (IC) maneuvers, and maximal voluntary EAdi was obtained during these maneuvers. Minute ventilation (V˙ e) was 12.2 ± 1.9 L/min (mean ± SD) at rest, and increased progressively (p < 0.001) to 31.0 ± 7.8 L/min at end-exercise. EELV increased during exercise (p < 0.001) causing end-inspiratory lung volume to attain 97 ± 3% of TLC at end-exercise. Pdi at rest was 9.4 ± 3.2 cm H2O and increased during the first two thirds of exercise (p < 0.001) to plateau at about 13 cm H2O. EAdi was 24 ± 6% of voluntary maximal at rest and increased progressively during exercise (p < 0.001) to reach 81 ± 7% at end-exercise. In conclusion, dynamic hyperinflation during exhaustive exercise in patients with COPD reduces diaphragm pressure-generating capacity, promoting high levels of diaphragm activation.

Several studies (1, 2) have demonstrated that exercise to exhaustion in patients with chronic obstructive pulmonary disease (COPD) is not typically accompanied by the development of diaphragm contractile fatigue. These and others (3, 4) have found that exercise in such patients is accompanied by only modest increases in transdiaphragmatic pressure (Pdi). Mador and colleagues (2) demonstrated that despite a progressive increase in minute ventilation (V˙e) during constant workrate exercise, patients with COPD exhibited a small rise in mean inspiratory Pdi, which then reached a plateau after 40% of total exercise time. One explanation for the modest increases in Pdi observed is that despite an increased activation, progressive dynamic hyperinflation, as typically occurs in patients with COPD during exercise (5, 6), may reduce the diaphragm's ability to generate pressure. Another possibility is that central inhibition of the respiratory drive could attenuate the rise in Pdi (7). Central inhibition and decreases in central motor neuron firing rate have also been proposed as possible explanations for the lack of contractile fatigue found during exercise in patients with COPD (1, 5). The present study was designed to determine whether the small increases in Pdi, which occur during exhaustive exercise in patients with COPD, are due to a central inhibition of respiratory drive, or to an inability of the diaphragm to develop pressure. We hypothesized that with muscle failure and no central inhibition, diaphragm activation would increase progressively to a level close to maximum, whereas with central inhibition, diaphragm activation would likely plateau at or decrease to some submaximal level.


Ten clinically stable male patients with moderately severe COPD (FEV1 < 50% predicted) participated in the study after giving their informed consent. The study was approved by the ethics board of the institution. Anthropometric data and maximum sniff Pdi values obtained at the time of the study, as well as pulmonary function and blood-gas values obtained within 1 mo of the study, are presented in Table 1. At the time of the study, patients were alert and not prescribed any sedative medication. Patients with known cardiovascular disease and other systemic diseases or conditions likely to influence exercise were excluded from the study.


VariableAbsolutePercentage of Predicted Value
Age, yr61 ± 8
Height, cm171.2 ± 7.3
Weight, kg 75.1 ± 31.2
FEV1, L1.04 ± 0.2432.9 ± 5.7
VC, L3.12 ± 0.66 75.7 ± 15.3
FEV1/VC, %34 ± 6
TLC, L7.77 ± 1.41117.4 ± 22.5
FRC, L5.63 ± 1.29160.7 ± 34.6
RV, L4.66 ± 1.05199.6 ± 46.4
Pdimax sniff, cm H2O82.1 ± 20.7
PaO2 , mm Hg47.8 ± 7.2
PaCO2 , mm Hg49.1 ± 4.1

Experimental protocol

While seated on a bicycle ergometer each patient performed: (1) maximum sniff inhalations from FRC with one nostril occluded, and (2) inspirations to TLC, to determine peak Pdi (8) and maximum voluntary diaphragm activation (9), respectively. The maneuvers were repeated until three reproducible values were obtained. At least 15 s of rest was allowed between attempts for a given maneuver and 10 to 15 min between the two types of maneuvers.

Subsequently, while still seated on the bicycle, each patient breathed at rest for 5 min through a mouthpiece with a noseclip on. Measurements were made during the last 2 min of this period. Patients then performed incremental bicycle exercise, with exercise initiated at a workload of 10 W and increased by 10 W every minute until subjective exhaustion. Patients were asked to rate their perceived breathlessness, using a visual analog scale, during each minute of exercise. While still pedaling, patients performed an inspiratory capacity (IC) maneuver at the end of each exercise workload, in order to estimate end-expiratory lung volume (EELV) (10).


The electrical activity of the crural diaphragm (EAdi) was obtained via an esophageal catheter with a multiple-array electrode and two balloons for measurement of esophageal (Pes) and gastric (Pga) pressures. Mouth pressure (Pmo) and flow were also measured. Detailed description of electrode positioning, equipment, and data acquisition for EAdi, Pes, Pga, and flow can be found in the report of Sinderby and colleagues (9). Transcutaneous O2 saturation (Stc O2 ) was measured with a pulse oximeter (Pulsox TS-7; Minolta, Ramsey, NJ).

EAdi Signal Processing

Automatic on-line digital processing of EAdi segments of 16 ms duration consisted of an initial filtering for optimization of the signal-to-disturbance ratio (9, 11, 12). The position of the center of the electrically active region of the activated diaphragm (EARdi) was determined by a cross-correlation method (13, 14). The root mean square (RMS) was calculated using the double subtraction technique (13), with the modification that the signal obtained from the center electrode pair was also added. Signal segments with residual disturbances caused by cardiac electric activity or common mode signals were evaluated via specific detectors and replaced by a predicted value, e.g., the previously accepted value (9, 11). To ensure that the RMS value was not influenced by power spectral shifts, the power spectrum center frequency was computed according to the methods described in Sinderby and colleagues (15) in order to implement corrections described by Beck and colleagues (16). Our standard methodology for power spectrum analysis and center frequency calculation of diaphragm electrical activity has been described in detail and validated in the following works (13-18). Throughout the report, EAdi is presented as RMS expressed as a percentage of the voluntary maximum RMS.

Analysis of Respiratory Mechanics

Pdi was calculated as the difference between Pga and Pes. In order to include the component of Pdi generated to overcome intrinsic PEEP, mean Pdi swings were calculated between the onset of EAdi and the end of inspiratory flow. Timing parameters, including inspiratory duration (Ti), total breath duration (Tt), and duty cycle (Ti/Ttot) were determined from the flow signal. Volume was obtained by integrating flow. EELV was obtained by subtracting IC values from measures of TLC, previously obtained with whole-body plethysmography. End-inspiratory lung volumes (EILV) were calculated by adding the tidal volume (Vt) to the EELV. EELV and EILV are presented as percentages of TLC.

Statistical Analysis

Comparison of variables over time was performed using one-way repeated measures ANOVA. Post-hoc comparisons were made with the Student-Newman-Keuls test.

All patients exercised to subjective exhaustion. Heart rate was 87.2 ± 21.4 beats/min at rest and 117.6 ± 18.6 beats/min at end-exercise. At the end of exercise, the average workload was 60.0 ± 21.1 W and the mean dyspnea score was 7.9 ± 2.1. Stc O2 was 92.7 ± 3.1% at rest and 87.5 ± 4.4% at end-exercise. At rest, maximal EAdi values were higher during sniff inhalation in eight patients and during the inspiration to TLC in two patients. In four of these patients, maximal EAdi values were further increased with TLC maneuvers performed during exercise. Maximal EAdi values obtained at rest (sniff inhalation or TLC maneuvers) and during different levels of exercise (TLC maneuvers), are presented in Table 2. Also, the peak EAdi during tidal breathing, normalized to these maximum values, are presented for end-exercise. Values of peak EAdi, expressed as percent of maximum during tidal breathing, presented in the following text, are those normalized to the maximum EAdi ever obtained in each patient.


Maximal EAdi (a.u.)Peak EAdi at End-exercise(% of each presented maximum)p Values versus Rest
Rest210.0 ± 76.2 88.1 ± 12.1
Exercise 0–33%165.4 ± 62.0116.6 ± 23.2< 0.05
Exercise 33–66%191.3 ± 69.4 99.5 ± 13.7NS
Exercise 66–100%223.9 ± 67.4 83.5 ± 11.0NS
Exercise (last TLC) 215.5 ± 69.3 87.5 ± 14.7NS
Exercise (anytime)§ 225.7 ± 66.6 82.7 ± 10.3NS
Rest/Exercise (anytime)II 231.5 ± 64.980.0 ± 7.1

Definition of abbreviations: EAdi = electrical activity of the crural diaphragm; a.u. = arbitrary units.

*Values are means ± SD.

  Peak EAdi values for tidal breathing at end-exercise are expressed as percentages of each corresponding maximum EAdi value obtained at rest and at different levels of exercise.

Exercise (last TLC) indicates the value obtained from the last TLC maneuver performed during the exercise period.

§Exercise (anytime) refers to a maximum EAdi value obtained at anytime during exercise.

  Rest/Exercise (anytime) refers to a maximum value obtained at anytime either at rest or during exercise.

The group mean values for EAdi and Pdi plotted versus V˙e, during breathing at rest and during the early, middle, and late periods of exercise as well as the last five breaths at end-exercise are shown in Figure 1. The peak EAdi increased progressively (p < 0.001) from a resting value of 24 ± 6% of maximum to an end-exercise value of 81 ± 7%. Two patients with Stc O2 below 90% at rest received supplemental oxygen during the exercise. Peak EAdi at end-exercise was 83 ± 4% in patients with Stc O2 below 90% (n = 4) and 77 ± 5% in patients with Stc O2 above 90% (n = 6, p = 0.089). The power spectrum center frequency did not change significantly in any of the subjects at the end of exercise, and therefore correction of the EAdi amplitudes was not necessary. Mean Pdi increased (p < 0.001) from a value of 9.4 ± 3.2 cm H2O at rest, to attain a plateau of about 13 cm H2O two-thirds into the exercise run. V˙e increased progressively (p < 0.001) from a resting value of 12.2 ± 1.9 L/min to an end-exercise value of 31.0 ± 7.7 L/min. As presented in Table 3, the increases in V˙e during the initial phases of exercise were caused by both increments in Vt and breathing frequency, whereas during the later stages of the exercise, breathing frequency was the sole contributor to the increasing V˙e. No changes were observed in breathing duty cycle, and consequently the Vt/Ti increased progressively from rest to end of exercise (Table 3).


VariablesRestExercise 0–33%Exercise 33–66%Exercise 66–99%Exercise 100%p Value ANOVA
Vt, L0.69 ± 0.160.90 ± 0.261.01 ± 0.261.13 ± 0.291.14 ± 0.32< 0.001
RR, bpm18.2 ± 4.821.5 ± 4.623.2 ± 6.026.5 ± 6.528.1 ± 5.8< 0.001
Vt/Ti, L/s0.55 ± 0.110.80 ± 0.120.99 ± 0.191.20 ± 0.241.36 ± 0.33< 0.001
Ti/Ttot0.37 ± 0.040.38 ± 0.010.38 ± 0.010.38 ± 0.030.38 ± 0.04NS

Definition of abbreviations: RR = respiratory rate (breaths per minute); Ti = inspiratory duration; Ti/Ttot = duty cycle; Tt = total breath duration; Vt = tidal volume.

The operational lung volumes during quiet breathing and progressive exercise are presented in Figure 2. EELV was 77 ± 5% of TLC at rest and increased (p < 0.001) to a plateau of about 82 ± 6% of TLC during exercise. The absolute increase in EELV from rest to end of exercise was 0.37 ± 0.30 L (p = 0.004). The EILV was 86 ± 4% of TLC at rest and increased progressively (p < 0.001) by 0.81 ± 0.29 L with exercise to attain 97 ± 3% of TLC at end-exercise (Figure 2).

This study has demonstrated that diaphragm activation increases progressively during incremental symptom-limited maximum bicycle exercise in patients with COPD and reaches high values at the end of exercise. In contrast, Pdi increases only modestly and reaches a plateau shortly after onset of exercise. These results indicate that the lack of increase in Pdi during incremental exercise cannot be attributed to a central inhibition of the diaphragm but rather to the diaphragm's inability to generate pressure.

This is the first study that has measured and evaluated EAdi during exercise using a validated and standardized methodology (9, 11, 13-20), which enables normalization of EAdi to its maximum value (9). Appropriate electrode positioning is crucial for accurate determination of EAdi amplitudes (13, 14, 17). This becomes especially important, considering the progressive diaphragm-shortening anticipated with dynamic hyperinflation during exercise in patients with COPD. In the current study, the multilead configuration of the esophageal electrode as well as the methodology of signal acquisition and processing ensured minimum influence of electrode-to-diaphragm filtering (13, 14) and reduced signal contamination (9, 15). With this methodology, the diaphragm EMG signal strength and frequency content are also unaffected by changes in chest wall configuration (16-18). We are therefore confident that the EAdi obtained in the present study represents the electrophysiological events occurring at the level of the diaphragm sarcolemma, reflecting neural drive to the diaphragm.

It has previously been suggested that central inhibition of respiratory drive may act to protect the diaphragm against the development of contractile fatigue (7). In our study, EAdi increased progressively with incremental exercise and was high at end-exercise, not supporting the existence of such a mechanism in patients with COPD. The fact that maximum EAdi during the TLC maneuvers at end-exercise was similar to those obtained with TLC/sniff maneuvers at rest, is additional evidence that central inhibition played a minor role at end- exercise. Although an average EAdi of 81% of maximum might suggest a 19% reserve of diaphragm activation at end-exercise, it is necessary to point out that the maximum EAdi value used for normalization can vary depending on when and how it is obtained. In the present study, EAdi during tidal breathing was normalized to the maximum EAdi value observed at any time; however, if normalized to maximum values obtained at rest (TLC/sniff maneuvers), the resulting EAdi would be 88% of maximum at end-exercise. Determination of maximum EAdi has been validated only for resting conditions (9); however, we felt it was more appropriate to use the highest maximum EAdi ever obtained. In the present study, values obtained at rest and during exercise were not significantly different. However, comparing across levels of exercise revealed that maximum EAdi values were lower at low submaximum levels of exercise when compared with values obtained at rest and at end-exercise. This indicates that TLC maneuvers at low exercise levels may not be reliable for determining maximum voluntary EAdi. The reason for the lower maximum EAdi values during TLC maneuvers at low submaximal levels is not clear. However, factors such as motivation, greater relative contribution of the rib-cage muscles, and a learning effect for performance of TLC maneuvers during exercise may have influenced the outcome. We have previously reported that maximum values of EAdi are not achieved at all times when patients with COPD perform a TLC maneuver at rest (9). In that study, the coefficient of variation for the EAdi during TLC maneuver was 8% (9). In the present study coefficient of variation for TLC maneuvers performed at rest was also 8%, suggesting a similar reproducibility of the maneuver. However, different from our previous study, where the majority of patients obtained maximum EAdi values during the TLC maneuver, in the present study, the maximal EAdi at rest was obtained during the sniff maneuver. This supports our previous suggestion that it is important to combine sniff and TLC maneuvers in order to increase the probability of obtaining a EAdi as near to maximum as possible (9).

On average, mean Pdi increased by only ∼ 4 cm H2O early into the exercise and remained unaltered during the remainder of the run. These findings suggest that, although reaching 81% of maximum activation, the diaphragm is unable to generate sufficient tension. Like other skeletal muscles, the force the diaphragm can generate is largely dependent on its length (16, 21). By decreasing diaphragm muscle length, hyperinflation can reduce the pressure-generating capacity of the diaphragm. In patients with COPD, dynamic hyperinflation, which occurs when minute ventilation increases (5, 6), can worsen the situation. It is commonly assumed that hyperinflation impairs the diaphragm's force-generating capacity by affecting the length-tension relationship and/or increasing the radius of curvature of the diaphragm. However, recent evidence obtained in both patients with COPD and healthy subjects, indicates that the length-tension relationship of the diaphragm plays a more significant role than does the radius of curvature (22, 23). In the present study, patients showed evidence of static hyperinflation at rest, as indicated from their FRC values, as well as a further dynamic hyperinflation during the exercise, as evidenced by a reduction in the IC. The combined increases in Vt and EELV resulted in the EILV to approach TLC, and thus the diaphragm to be markedly shortened, at end-inspiration. It is therefore not surprising that the increase in Pdi during exercise in the present study was modest. From the results, one might conclude that an inspiratory reserve volume of 3% at end-exercise is not compatible with a 19% reserve in EAdi. However, EAdi is zero at EELV and maximum at TLC. If the inspiratory reserve volume is expressed as a percentage of inspiratory capacity, it becomes 19% at end-exercise.

Another factor to promote increasing EAdi in absence of increasing Pdi would be progressive weakness caused by diaphragm fatigue. Our finding of no change in the EAdi power spectrum center frequency indicates that diaphragm fatigue did not develop, which reduces the likelihood that peripheral fatigue caused Pdi to plateau with increasing EAdi. It should, however, be noted that the interaction between force/Pdi, length/volume, and activation/EAdi and their influence on fatigue/center frequency is complex (24, 25).

The capacity of the respiratory system to generate pressure can decline with increasing inspiratory flow (26, 27) and increased inspiratory flows during exercise can result in substantial reductions of inspiratory pressure-generating capacity (28). It has been estimated that in healthy subjects, the velocity of diaphragm shortening at maximal exercise may increase by several times when compared with that during resting breathing (29). We do not know to what extent the velocity of diaphragmatic shortening increased during exercise in our patients, but we assume it to be relatively small, as mean inspiratory flow of our patients was only 1.36 L/s at the end of exercise. In healthy subjects, increases in mean inspiratory flow as great as 1.4 L/s did not increase the diaphragm activation for any given Pdi at any given lung volume (12). This suggests that diaphragmatic contractions under the current conditions probably occurred at the lower velocity range of the diaphragm force-velocity curve. Accordingly, the diaphragm velocity of shortening likely played a less significant role than the length-tension relationship with respect to the effect on the diaphragm pressure-generating capacity for a given level of activation.

In conclusion, this study shows that Pdi increases only modestly during incremental symptom-limited maximum exercise, whereas EAdi increases progressively to high values at end-exercise. The lack of increase in Pdi is likely due to static and dynamic hyperinflation, which impairs the diaphragm's pressure-generating capacity, rather than to a central inhibition of respiratory drive to the diaphragm.

Supported by grants from the Swedish Association for Traffic and Polio Disabled, the Swedish Association for Neurologically Disabled, and the Bureau for Foreign Assistance and European Integration of Poland.

1. Polkey MI, Kyroussis D, Keilty SEJ, Hamnegard CH, Mills GH, Green M, Moxham JExhaustive treadmill exercise does not reduce twitch transdiaphragmatic pressure in patients with COPD. Am J Respir Crit Care Med1521995959964
2. Mador MJ, Kufel TJ, Pineda LA, Sharma GKDiaphragmatic fatigue and high-intensity exercise in patients with chronic obstructive pulmonary disease. Am J Crit Care Med1612000118123
3. Yan S, Kaminski D, Sliwinski PInspiratory muscle mechanics of patients with chronic obstructive pulmonary disease during incremental exercise. Am J Respir Crit Care Med1561997807813
4. Bye PT, Esau SA, Levi RD, Shiner RJ, Macklem PT, Martin JG, Hardy RLVentilatory muscle function during exercise in air and oxygen inpatients with chronic airflow limitation. Am Rev Respir Dis1321985236240
5. Dodds DS, Brancatisano T, Engel LAChest wall mechanics during exercise in patients with severe chronic air-flow obstruction. Am Rev Respir Dis12919843338
6. O'Donnell DE, Webb KAExertional breathlessness in patients with chronic airflow limitation. Am Rev Respir Dis148199313511357
7. Polkey MI, Kyroussis D, Hamnegard CH, Mills GH, Hughes PD, Green M, Moxham JDiaphragm performance during maximal voluntary ventilation in chronic obstructive pulmonary disease. Am J Respir Crit Care Med1551997642648
8. Miller JM, Moxham J, Green MThe maximal sniff in the assessment of diaphragm function in man. Clin Sci (Colch)6919859196
9. Sinderby C, Beck J, Weinberg J, Spahija J, Grassino AVoluntary activation of the human diaphragm in health and disease. J Appl Physiol85199821462158
10. Yan S, Kaminski D, Sliwinski PReliability of inspiratory capacity for estimating end-expiratory lung volume changes during exercise in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med15619975559
11. Sinderby C, Navalesi P, Beck J, Skrobik Y, Comtois N, Friberg S, Gottfried SB, Lindström LNeural control of mechanical ventilation. Nat Med5199914331436
12. Beck J, Sinderby C, Lindström L, Grassino ACrural diaphragm activation during dynamic contractions at various inspiratory flow rates. J Appl Physiol51998451458
13. Sinderby C, Beck JC, Lindström LH, Grassino AEEnhancement of signal quality in esophageal recordings of diaphragm EMG. J Appl Physiol82199713701377
14. Beck J, Sinderby C, Lindström L, Grassino AInfluence of bipolar electrode positioning on measurements of human crural diaphragm EMG. J Appl Physiol81199614341449
15. Sinderby C, Lindström L, Grassino AEAutomatic assessment of electromyogram quality. J Appl Physiol79199518031815
16. Beck J, Sinderby C, Lindström L, Grassino AEffects of lung volume on diaphragm EMG signal strength during voluntary contractions. J Appl Physiol85199811231134
17. Beck J, Sinderby C, Weinberg J, Grassino AEffects of muscle-to-electrode distance on the human diaphragm electromyogram. J Appl Physiol791995975985
18. Beck J, Sinderby C, Lindström L, Grassino ADiaphragm interference pattern EMG and compound muscle action potentials: effects of chest wall configuration. J Appl Physiol821997520530
19. Sinderby C, Lindström L, Comtois N, Grassino AEEffects of diaphragm shortening on the mean action potential conduction velocity. J Physiol (Lond)4901996207214
20. Weinberg J, Sinderby C, Sullivan L, Grassino A, Lindström LEvaluation of diaphragm electromyogram contamination during progressive inspiratory maneuvers in humans. Electromyogr Clin Neurophysiol371997143153
21. McCulley KK, Faulkner JALength-tension relationship of mammalian diaphragm muscles. J Appl Physiol54198316811686
22. Gauthier AP, Verbanck S, Estenne M, Segebarth C, Macklem PT, Paiva MThree-dimensional shape at different lung volumes. J Appl Physiol761994495506
23. Cassart M, Pettiaux N, Gevenois PA, Paiva M, Estenne MEffect of chronic hyperinflation on diaphragm length and surface area. Am J Respir Crit Care Med1561997504508
24. Sinderby C, Spahija J, Beck CChanges in respiratory effort sensation over time are linked to the frequency content of diaphragm electrical activity. Am J Respir Crit Care Med1632001905910
25. Sacco P, McIntyre DB, Jones DAEffects of length and stimulation frequency on fatigue of the human tibialis anterior muscle. J Appl Physiol77199411481154
26. Agostoni E, Fenn WOVelocity of muscle shortening as a limiting factor in respiratory airflow. J Appl Physiol161960349353
27. Hyatt RE, Flatt RERelationship of airflow to pressure during maximum respiratory effort in man. J Appl Physiol211966477482
28. Leblanc P, Summers E, Inman MD, Jones NL, Campbell EJM, Killian KJInspiratory muscles during exercise: a problem of supply and demand. J Appl Physiol64198824822489
29. Aliverti A, Cala SJ, Duranti R, Ferrigno G, Kenyon CM, Pedotti A, Scano G, Sliwinski P, Macklem PT, Yan SHuman respiratory muscle actions and control during exercise. J Appl Physiol83199712561269
Correspondence and requests for reprints should be addressed to Christer Sinderby, Guy-Bernier Research Center, Maisonneuve-Rosemont Hospital, University of Montreal, 5415 Boulevard de l'Assomption, Montreal, H1T 2M4 Canada. E-mail:

Christer Sinderby and Jennifer Beck are recipients of fellowships from Fonds de la Recherche en Santé du Quebec (FRSQ).


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record