INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In patients with chronic obstructive pulmonary disease (COPD), the inspiratory muscles are faced with an increased load. Airflow obstruction increases resistive work; hyperinflation places the patient on the upper flattened portion of the pressure- volume curve, increasing elastic work; and the development of auto-PEEP (positive end-expiratory pressure) creates a threshold load that needs to be overcome with each breath. Hyperinflation shortens the inspiratory muscles, particularly the diaphragm, placing them at a mechanical disadvantage. During exercise, ventilatory requirements increase, further exacerbating the potential imbalance between inspiratory muscle load and capacity (1, 2). For these reasons, patients with COPD may be particularly predisposed to the development of inspiratory muscle fatigue during exercise. However, data evaluating this hypothesis are limited. Prior studies have demonstrated a change in the diaphragmatic electromyogram (EMG) power spectrum (a fall in the high/low ratio) during exercise in some patients with COPD (3), and a slowing of inspiratory muscle relaxation immediately postexercise consistent with a potentially fatiguing pattern of contraction (4). However, evidence of overt contractile fatigue of the diaphragm is lacking (5).

Measurement of twitch transdiaphragmatic pressure (Pdi) after bilateral supramaximal phrenic nerve stimulation appears to be a sensitive effort-independent index for detecting contractile fatigue of the diaphragm (6). This technique has allowed diaphragmatic fatigue to be demonstrated after inspiratory muscle loading (7), isocapnic hyperpnea (8), and heavy endurance exercise (9) in normal subjects. In this study, we measured twitch Pdi before and after high-intensity constant-load cycle exercise to the limits of tolerance in patients with moderately severe COPD to determine whether contractile fatigue of the diaphragm occurs in such patients after exercise.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twelve male patients with COPD, 61.4 ± 3.0 (SE) yr of age, volunteered for this study. Their mean height and weight were 1.76 ± 0.02 m and 83.9 ± 3.1 kg, respectively. The diagnosis of COPD was based on a prior history of smoking and pulmonary function tests demonstrating moderate to severe irreversible airflow obstruction (FEV₁ and FEV₁/FVC ratio both < 65%). All patients were clinically stable at the time of study and were free from overt cardiovascular, musculoskeletal, or neurologic disease that could affect exercise performance. All patients were taking inhaled bronchodilators. None of the patients were receiving oral steroids or had received oral steroids either continuously or as a short burst (for exacerbations) in the year before the study.

Pulmonary Function Testing

Spirometry was performed in accordance with ATS recommendations (10). The maximum voluntary ventilation of each patient was measured over 12 s. Lung volumes were measured by body plethysmography (P. K. Morgan; Chatham, Kent, UK). Single-breath diffusing capacity was also measured (P. K. Morgan). Predicted normal values were those of Crapo and coworkers (11) and Kory and colleagues (14). Maximal inspiratory pressure (PI_max) was measured with a differential pressure transducer (model MP-45, ± 350 cm H₂O; Validyne, Northridge, CA) while performing a maximal inspiratory effort against an occluded airway near residual volume. Visual feedback of the pressure signal was provided. PI_max maneuvers were repeated until at least three reproducible measurements that could be sustained for at least 1 s were recorded. Age-specific predicted values were those of McElvaney and coworkers (15). Pulmonary function measurements are shown in Table 1.

Exercise

Several days before the actual experimental studies a maximal symptom-limited exercise test was performed on an electronically braked cycle ergometer (CPE-2000; MedGraphics, St. Paul, MN) to determine the maximal work capacity (Wmax). After 2 min of unloaded cycling, the workload was increased in a ramp fashion by 10-15 W every minute until the patient was unable to continue. The last workload for which the subject was able to complete the full minute of cycling was designated Wmax.

Constant workload exercise was performed at 60-70% of Wmax until the patient had to stop because of intolerable symptoms (dyspnea, leg fatigue, or both). The patients were allowed 3 min to acclimatize to the breathing circuit. They then exercised for 1 min at 0 W and 2 min at 10 W (warm-up period) before initiating exercise at 60- 70% of Wmax.

The patients breathed through a two-way nonrebreathing valve of low resistance and dead space (model 2700; Hans Rudolph, Kansas City, MO). Inspiratory flow was measured with a pneumotachograph (model 3813; Hans Rudolph) and a ± 5 cm H₂O differential pressure transducer (MP-45; Validyne). Tidal volume (VT) was obtained by integration of the digitized flow signal. Expired gas was sampled breath-to-breath and analyzed with a zirconia O₂ analyzer and infrared CO₂ analyzer, respectively (MedGraphics). Oxygen consumption was measured breath-to-breath and then averaged over 30-s intervals. The heart rate was determined from a 12-lead electrocardiograph. Esophageal (Pes) and gastric (Pga) pressures were measured with balloon catheters by standard techniques. All signals were digitized and stored on disk, using Windaq software (Dataq Instruments, Akron, OH). Peak and mean inspiratory pressure swings were calculated breath-to-breath from the digitized pressure tracings. The beginning and end of inspiration were determined from the inspiratory flow signal and mean pressures were calculated for this time period. Baseline Pes and Pdi were taken as the immediately preceding end-expiratory pressure. The esophageal pressure-time integral was calculated as the product of mean pressure, inspiratory time, and respiratory rate.

Phrenic Nerve Stimulation

Phrenic nerve stimulation was achieved by cervical magnetic stimulation (Magstim 200; Magstim, Whitland, Dyfed, Wales) using a circular 90-mm coil (16). Stimulation was performed with the patient in the seated position, with the neck flexed, and breathing through a mouthpiece while wearing noseclips. Before nerve stimulation, the mouthpiece was occluded via a mouth shutter (4200C; Hans Rudolph) to prevent any change in volume during nerve stimulation. The coil was placed over the cervical spine and moved up and down between the fifth and seventh spinous processes until the site with the largest Pdi response was identified. This site was marked and all subsequent twitches were obtained from this coil position.

Transcutaneous electrical phrenic nerve stimulation was also performed in five of the patients. The technical details of this procedure have been described previously (8, 17). M waves were recorded with surface electrodes placed in the seventh and eighth intercostal spaces, 2 to 3 cm from the costal margin. The EMG signals were amplified, bandpass filtered (bandwidth, 20 Hz-2 kHz), and then passed through a 12-bit analog-to-digital converter and displayed on a computer by Windaq software (Dataq Instruments). The sampling rate was 1,000 Hz. Stimulus current was progressively increased to each nerve separately until no further increase in the amplitude of the ipsilateral M wave was seen. To ensure that the stimulus intensity remained supramaximal, the current was increased by an additional 20 to 50% during all experimental studies.

Twitch Pdi was defined as the peak amplitude above the immediately preceding Pdi baseline. Approximately 8-10 twitches were obtained before and 10, 30, and 60 min after exercise. Individual twitches were rejected from analysis if there was (1) failure to initiate the twitch near the resting end-expiratory lung volume as determined by end-expiratory esophageal pressure, (2) esophageal peristalsis during or just before initiation of the twitch, and (3) lack of diaphragmatic relaxation as demonstrated by diaphragmatic EMG activity and/or Pga in excess of baseline (8, 17) before twitch onset.

Arterialized Venous Blood Determinations

A Teflon catheter (18 or 20 gauge) was inserted into a vein in the dorsum of the hand. Before sampling, the dorsum of the hand was heated until the skin temperature exceeded 43° C for at least 2 min. Blood was then withdrawn for measurement of lactate levels. Blood samples were obtained before and 5, 15, and 30 min after exercise.

Data Analysis

Changes in variables over time were analyzed by repeated measures analysis of variance. If the analysis of variance was significant, measurements postexercise were compared with baseline by paired t test with Bonferroni correction (21). For each patient, a fall in twitch Pdi postexercise of  10% at 10 and 30 min was considered potentially indicative of diaphragmatic fatigue. Data are expressed as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cardiopulmonary Parameters during Exercise

The patients exercised for an average of 10.2 ± 2.0 min at a workload of 59.9 ± 4.3 W (65.3 ± 2.2% of Wmax). Peak O₂ (oxygen consumption) during exercise was 1.46 ± 0.10 L/min or 17.7 ± 0.9 ml/kg/min, which was 62.4 ± 4.6% of the predicted maximum. The peak O₂ during constant-load exercise represented 108.1 ± 2.8% of the peak O₂ obtained during the maximal incremental exercise test. The maximal heart rate was 139 ± 4 beats/min (83.7 ± 2.8% of the predicted maximum). Peak exercise ventilation (VE) was 55.6 ± 4.1 L/min, which represented 78.8 ± 4.5% of the 12-s maximum voluntary ventilation. In no patient did oxygen saturation fall below 88% during or after exercise.

Bilateral Phrenic Nerve Stimulation

Twitch Pdi was not significantly different from baseline at any time postexercise (Figure 1). Of the individual patients, four had a greater than 10% fall in twitch Pdi 10 min postexercise, potentially indicative of diaphragmatic fatigue (Figure 2). However, in two of these patients, twitch Pdi was within 10% of the baseline value at 30 min postexercise. Twitch Pdi reflects low-frequency fatigue, which recovers slowly (7) and should not have recovered by 30 min postexercise. Thus, it is unlikely that diaphragmatic fatigue truly occurred in these two patients. In contrast, in the other two patients, twitch Pdi remained depressed at 30 and 60 min. The gastric and esophageal components of the twitch were also not significantly different from baseline at any time postexercise (Figure 3). In the two patients who had unequivocal evidence of diaphragmatic fatigue postexercise, twitch Pes fell by 19.8 ± 5.7% and twitch Pga fell by 22.2 ± 10.2%. The twitch Pes/twitch Pga ratio was 2.11 ± 0.34 at baseline and 2.18 ± 0.27 at 10 min postexercise. Five patients also had twitch Pdi measured during transcutaneous electrical stimulation (Figure 2). Twitch Pdi results obtained with cervical magnetic stimulation and transcutaneous electrical stimulation were congruent. Three patients had a  10% fall in twitch Pdi 10 min postexercise by both techniques, whereas the other two patients had no change in twitch Pdi postexercise by both techniques. M-wave amplitude measured during transcutaneous electrical stimulation was not significantly different from baseline at any time postexercise.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Twitch transdiaphragmatic pressure (twitch Pdi) before and 10, 30, and 60 min postexercise. Data are expressed as means ± SE, n = 12. Twitch Pdi was not significantly different from baseline at any time postexercise.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Twitch transdiaphragmatic pressure (twitch Pdi) before and 10 min after exercise in each patient. Patients who displayed at least a 10% reduction in twitch Pdi postexercise (that persisted at 30 min postexercise), potentially indicative of contractile fatigue of the diaphragm, are represented by closed circles; patients in whom twitch Pdi was unchanged postexercise are represented by open circles; while patients who displayed a transient reduction in twitch Pdi are represented by closed squares. Five of the patients also had twitch Pdi measured during transcutaneous electrical stimulation and these results are represented by the closed triangles ( 10% fall in twitch Pdi 10 min postexercise) and open triangles. Twitch Pdi was persistently decreased postexercise in only two of the 12 patients. In the five patients who had twitch Pdi measured by both techniques, the results were congruent.

View larger version (17K): [in this window] [in a new window]   Figure 3.   Twitch esophageal pressure (twitch Pes) and twitch gastric pressure (twitch Pga) before and 10, 30, and 60 min postexercise. Twitch Pes and twitch Pga were not significantly different from baseline at any time postexercise.

Breathing Pattern during Exercise

Mean inspiratory Pdi, Pes, and VE are shown in Figure 4. As exercise progressed, mean inspiratory Pes and VE continued to rise while mean inspiratory Pdi plateaued early in the exercise period. As exercise continued and patients recruited their expiratory muscles, peak expiratory Pga averaged 11.9 ± 2.9 cm H₂O (range, 4-36 cm H₂O) at end exercise. The pressure- time product (PTPes) was 519 ± 42 cm H₂O · s · min ¹ at end exercise. We used end-expiratory Pes as our baseline to calculate PTPes (18). As exercise progresses, end-expiratory Pes increases owing to auto-PEEP (which truly increases inspiratory muscle work) and abdominal muscle contraction during expiration. It is not possible to quantify reliably the relative contribution of auto-PEEP and abdominal muscle activity during intense exercise. However, all of our patients recruited their abdominal muscles as exercise progressed. Thus, our estimate of PTPes must be an overestimate of the true value. If we chose the baseline to be zero (22) (assuming that the increase in end-expiratory Pes is entirely due to abdominal muscle activity, which is unlikely to be true), PTPes at end exercise would be 367 ± 30 cm H₂O · s · min¹ or 70.8 ± 3.0% of our original estimate. The true PTPes must lie somewhere between these two values. Peak inspiratory flow was 2.96 ± 0.16 L/min. No significant differences were observed in breathing pattern or respiratory pressures between the two patients who definitely displayed evidence of contractile fatigue of the diaphragm and the eight patients who did not.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Mean inspiratory transdiaphragmatic (Pdi) and esophageal (Pes) pressure and minute ventilation at baseline and during exercise. As exercise progressed, mean inspiratory Pes and VE continued to rise while mean inspiratory Pdi plateaued early in the exercise period.

Lactate Measurements

Lactate levels before and after exercise are shown in Figure 5. Peak lactate levels postexercise were 4.8 ± 0.5 mmol/L, indicating that our subjects, despite their respiratory disease, were capable of generating a significant lactate acidosis. Elderly healthy men performing a similar exercise protocol generated peak lactate levels of 6.8 ± 0.6 mmol/L (23).

View larger version (14K): [in this window] [in a new window]   Figure 5.   Lactate levels before and 5, 15, and 30 min postexercise. Lactate levels increased significantly postexercise. The asterisk (*) indicates a significant difference from the baseline value.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The major finding of this study was that overt contractile fatigue of the diaphragm occurred uncommonly after high-intensity endurance exercise to the limits of tolerance in patients with moderately severe COPD.

Critique of Methods

Cervical magnetic stimulation was used in this study to detect overt contractile fatigue of the diaphragm. In our laboratory, cervical magnetic stimulation is highly reproducible (8, 9, 17). The within-subject between-trial (within a single day) coefficient of variation is 6.4 ± 0.5%. Cervical magnetic stimulation has been successfully employed to detect diaphragmatic fatigue in a number of prior studies (7, 24, 25). Cervical magnetic stimulation is technically much easier to apply than transcutaneous stimulation, which we thought would be highly advantageous for this study, in which measurements had to be performed repeatedly at precise time intervals in elderly patients with respiratory disease. To check further on the validity of our measurements, we performed both transcutaneous electrical and cervical magnetic stimulation in five of our more stoical patients who had a favorable body habitus (absence of a short, thick neck, which makes transcutaneous stimulation more difficult). Twitch Pdi results with the two techniques were congruent. Three of the patients demonstrated a reduction in twitch Pdi of > 10% at 10 min postexercise by both techniques while the other two patients had no change in twitch Pdi postexercise by both techniques. However, the percentage fall in twitch Pdi was not identical by the two techniques (r = 0.81). The M-wave amplitude during transcutaneous stimulation was unchanged postexercise, indicating that the fall in twitch Pdi, when it occurred, was not due to transmission failure.

Twitch Pdi is larger during cervical magnetic stimulation compared with transcutaneous electrical stimulation (8, 17, 25). Some of the observed difference may be caused by unappreciated submaximal transcutaneous electrical stimulation due to difficult patient anatomy. However, even when transcutaneous electrical stimulation is clearly maximal, this difference persists and is entirely due to increases in twitch Pes. During cervical magnetic stimulation, the sternomastoids, trapezii, deltoids, and rhomboids are activated in addition of the diaphragm (16, 25). It is currently believed that contraction of these muscles stiffens the upper thorax, leading to increased twitch Pes during diaphragmatic contraction. In the five patients in whom both transcutaneous electrical stimulation and cervical magnetic stimulation were obtained, twitch Pdi in the fresh state was 18.9 ± 3.0 cm H₂O during transcutaneous stimulation and 19.5 ± 1.7 cm H₂O (p = NS) during cervical magnetic stimulation. The difference in twitch Pes between the two techniques narrows as end-expiratory lung volume increases (25) and all of our patients were hyperinflated.

A number of factors can influence twitch Pdi. Changes in lung volume can alter twitch Pdi considerably (26). In patients with COPD, end-expiratory lung volume increases during exercise (27) but returns to baseline levels within the first few minutes after exercise is terminated (5), well before our initial postexercise twitch measurements. Twitch potentiation (augmentation of the twitch after a vigorous voluntary contraction) can markedly increase twitch Pdi (28). We waited 10 min after exercise ended to allow potentiation from the exercise itself to resolve before we repeated the twitch measurements. We were also careful to avoid any maneuvers during the study that could potentially cause twitch potentiation.

Diaphragmatic Fatigue and Exercise

In this study, we found no change in twitch Pdi after exhausting endurance exercise in a group of patients with moderately severe COPD (Figure 1). Examination of individual patient data revealed that two of the 12 patients displayed a persistent reduction in twitch Pdi postexercise that could be indicative of diaphragmatic fatigue (Figure 2). The within-subject between-trial coefficient of variation for twitch Pdi during cervical magnetic stimulation in our laboratory is 6.4% (8, 9, 17, 20). A 10% reduction in twitch Pdi as potentially indicative of diaphragmatic fatigue may be too low a cutoff point, given this degree of variability. However, even if we chose 15% (which is greater than twice the average variability), the two patients with a persistent fall in twitch Pdi would still meet this criteria (average fall in twitch Pdi, 21.5 ± 0.5%). Furthermore, similar reductions in twitch Pdi postexercise were observed with transcutaneous electrical stimulation (Figure 2). Thus, high-intensity constant-load cycle exercise can induce diaphragmatic fatigue in some patients with COPD, but in the majority of patients contractile fatigue of the diaphragm was not observed. There were no obvious differences between the two patients who developed diaphragmatic fatigue postexercise and the rest of the group in terms of baseline pulmonary function or exercise parameters. One variable we did not measure during exercise that could be important is the degree of exercise induced hyperinflation. Perhaps the patients who developed diaphragmatic fatigue had a greater degree of dynamic hyperinflation during exercise.

The exercise performed in this study was intense. The peak O₂ reached during constant-load exercise was 108.1 ± 2.8% of the peak O₂ obtained during a preliminary maximum incremental symptom-limited exercise test. It is unlikely that our patients were capable of exercising more intensely. Despite the high intensity of exercise performed, the majority of patients did not display evidence of diaphragmatic fatigue postexercise. During activities of daily living, in which the level of exertion is much less, diaphragmatic fatigue should occur even less frequently if at all.

In a prior study, Polkey and coworkers measured twitch Pdi in a small group of patients with severe COPD before and after walking on a treadmill to the limits of tolerance (5). Twitch Pdi did not fall in any subject (n = 6) postexercise. Patients with severe COPD have usually been disabled for long periods of time and are often severely deconditioned. Quadriceps strength and cross-sectional area are positively correlated with FEV₁, indicating that peripheral muscle strength declines as COPD worsens (29). Furthermore, quadriceps strength correlates with maximal exercise capacity in patients with COPD (29), suggesting that this peripheral muscle weakness can be functionally significant. We reasoned that it might be easier to push less severely affected patients who have better peripheral muscle function to exercise more intensely and, thus, diaphragmatic fatigue might be more easily elicited. While we succeeded in maximally exercising our patients, diaphragmatic fatigue occurred uncommonly. Taking the two studies together, it appears that contractile fatigue of the diaphragm is an uncommon event after intense exercise in patients with COPD. Thus, it is highly likely that contractile fatigue of the diaphragm does not occur to any clinically important extent when patients with COPD perform their normal activities of daily living.

Ribcage Muscle Fatigue and Exercise

Similowski and colleagues have shown that twitch Pdi during cervical magnetic stimulation can be reduced by both diaphragmatic and ribcage muscle fatigue (30). Diaphragmatic fatigue leads to reductions in both twitch Pga and twitch Pes, while ribcage fatigue leads to a reduction primarily in twitch Pes (30). In the two patients in this study who had a persistent reduction in twitch Pdi, twitch Pes fell by 19.8 ± 5.7%, twitch Pga fell by 22.2 ± 10.2%, while the twitch Pes/twitch Pga ratio was basically unchanged, suggestive of primarily diaphragmatic fatigue (this interpretation is supported by our transcutaneous stimulation results, in which twitch Pes and twitch Pga fell in equal proportions). Thus, we also found no evidence of ribcage muscle fatigue postexercise. The ability of cervical magnetic stimulation to detect either diaphragmatic and/or ribcage muscle fatigue is a major advantage for this particular study since the ribcage muscles are primarily responsible for increasing inspiratory pleural pressure during exercise in patients with COPD (31) and may, therefore, also be predisposed to develop fatigue postexercise.

Comparison with Unaffected Subjects

We have previously shown that healthy young subjects can develop diaphragmatic fatigue after high-intensity exercise to exhaustion (18, 19). It is somewhat surprising that young healthy subjects with normal lungs commonly develop diaphragmatic fatigue postexercise while patients with COPD rarely do so. Of course, our young subjects were able to exercise at much greater exercise intensity than our patients with COPD. We studied a group of elderly subjects (n = 9) who were healthy but leading sedentary lifestyles (23). For the group as a whole, twitch Pdi was not significantly different from baseline at any time postexercise. Only one subject developed diaphragmatic fatigue postexercise according to the criteria employed in this study. Thus, elderly sedentary subjects were not able to exercise at a sufficient intensity to fatigue their diaphragm. Our elderly subjects exercised at a higher work intensity than our patients with COPD (75.0 ± 7.4 versus 59.9 ± 4.3 W). However, owing to their derangements in respiratory mechanics, our patients with COPD will perform more inspiratory muscle work at any given workload.

We have measured the pressure-time product (PTPes) and the tension-time index (TTes = mean Pes/Pes_max · respiratory duty cycle [TI/Ttot]) from the esophageal pressure curve during exercise to obtain an index of global inspiratory work. PTPes and TTes were significantly higher in our patients with COPD compared with the healthy elderly (519 ± 42 versus 316 ± 29 cm H₂O · s · min¹, p < 0.001; 0.092 ± 0.008 versus 0.057 ± 0.005, p < 0.002). In contrast, PTPes and TTes in our patients with COPD were similar to that observed in a group of healthy sedentary young subjects who displayed diaphragmatic fatigue postexercise (18) (519 ± 42 versus 585 ± 52 cm H₂O · s · min¹, p = NS; 0.092 ± 0.008 versus 0.083 ± 0.009, p = NS). It would appear, therefore, that our patients should have developed diaphragmatic fatigue postexercise. The fact that they did not suggests that the diaphragm in patients with COPD may have increased resistance to fatigue.

Several studies have obtained biopsy specimens from patients with and without COPD who were undergoing thoracic surgery (32). These studies have shown that the diaphragm adapts to chronic loading and that the degree of adaptation correlates with the severity of COPD. These adaptations include an increased proportion of type I (slow twitch fatigue-resistant) fibers (32, 33), an increase in mitochondrial content (34), and a reduction in sarcomere length (34) (which will allow the diaphragm to generate pressure better at high lung volumes). All of these adaptations would make the diaphragm more fatigue resistant and offer an explanation for our findings. Analogous to our findings, maximum voluntary ventilation for 2 min produced diaphragmatic fatigue in normal subjects (24) but not in patients with COPD (35), further suggesting that the diaphragm is fatigue resistant in patients with COPD.

In conclusion, the majority of patients with COPD did not develop diaphragmatic fatigue after high-intensity cycle exercise to the limits of tolerance. These results support the current view that the diaphragm is fatigue resistant in patients with COPD.