A total of 84 clinically stable patients with COPD, referred from a respiratory clinic to a pulmonary rehabilitation program, were recruited. Inclusion criteria included a FEV₁/VC ratio less than 60%, FEV₁ less than 60% predicted, less than 10% reversibility after a bronchodilator, and at least a 10-year smoking pack-year history. Exclusion criteria included a comorbid condition that prevented exercise training and exercise training in the preceding 3 years. The protocol required six visits. The first two visits consisted of familiarization with the exercise tests to minimize learning effects and ensure effort reaching functional maximum. During the remaining four visits, each patient performed four exercise tests (incremental shuttle walk [ISW], endurance shuttle walk [ESW], incremental cycle ergometry [ICE], and endurance cycle ergometry [ECE]) to exhaustion. The order of the tests was randomized with a 48 to 96 hour interval between each test. Standardized pretest instructions to continue exercising until exhaustion were given, but no further encouragement was given during the exercise tests to avoid investigator bias. Before and immediately after exercise, heart rate (HR), transcutaneous oxygen saturation (Sa_O2), and perceived breathlessness and leg effort (using a modified Borg score) were recorded. In addition, patients were asked to name the symptom limiting further exercise: predominantly shortness of breath (SOB), predominantly leg effort (LE), an approximately equal combination of SOB and LE, or other symptoms.

Shuttle walk tests were performed on a flat 10-m course as previously described (10, 11). For ISW, patients were instructed to walk along the course in time with prerecorded signals; initial walking speed was set at 0.50 m/second (m/second) and increased each minute by 0.17 m/second. For ESW, the corresponding walking speed was that producing 80% of predicted peak oxygen consumption (V̇o₂ peak), with predicted V̇o₂ peak calculated using a regression equation from the maximum exercise distance achieved during familiarization ISW (12).

Cycle tests were performed on an electromagnetically braked cycle ergometer. The seat height was adjusted for each individual before each test using a standardized method. Subjects were asked to place the heel of their foot on the pedal, and the seat was adjusted until the leg was almost straight (knee joint 150–170°). For ICE, after 2 minutes of unloaded pedalling, patients cycled at a constant speed of 50 rpm with an initial load of 10 W. This was increased by 10 W every minute to a symptom-limited maximum. For ECE, patients cycled at 50 rpm at a constant load, calculated as 80% of peak workload (Wpeak) achieved during the familiarization ICE.

Twelve stable patients from the original cohort of 84 patients, with previous experience of magnetic stimulation of the femoral nerve, volunteered for this part of the study. On two separate days, each patient performed either ISW or ICE to exhaustion. Twitch quadriceps (TwQ) muscle force was measured as previously described by Polkey and coworkers (9). Unpotentiated and potentiated TwQ (TwQ_u and TwQ_p) and quadriceps compound muscle action potential (CMAP) were measured immediately before and 30 minutes after exercise. Potentiated twitches were measured 5 seconds after a maximum voluntary contraction of 5 seconds duration. Details of the TwQ technique are described in the online supplement.

The McNemar test (paired sample cross-tabulation analysis) was used to ascertain if SOB alone was more commonly cited after walking or cycling tests. Borg scores were compared with the Wilcoxon test, whereas baseline HR and Sa_O2, peak HR, and lowest Sa_O2 were compared using a paired t test. Subgroup comparisons were compared using unpaired t tests. Paired t tests were used to compare before and after exercise TwQ and CMAP and pre- to postexercise differences between walking and cycling. Results are presented as mean ± SD apart from Borg scores where medians are quoted.

Baseline characteristics of the 12 patients who completed this part of the study are shown in Table 4

Figure 2. Mean unpotentiated (A) and potentiated (B) twitch quadriceps tensions before and after exhaustive exercise.

Download Figure | Download Powerpoint

[More] [Minimize]

After cycling there was a significant reduction in maximum voluntary contraction force of the quadriceps (before, 24.8 ± 6.7 kg vs. after, 21.2 ± 4.2 kg [95% CI −6.2, −1.0; p = 0.012]). In contrast, there was no change in maximum voluntary contraction force after walking (before, 23.8 ± 6.8 kg vs. after, 23.7 ± 7.4 kg [95% CI −0.9, 0.6; p = 0.66]). Comparison of precycling to post-cycling differences revealed a significantly greater reduction in maximum voluntary contraction force after cycling (13.0 ± 9.1% vs. 1.5 ± 6.0% [95% CI 4.6, 18.6; p = 0.003).

The within-subject, between-occasion coefficient of variation of TwQ is in the order of 8.0 to 8.5% (7, 9, 13). In this study we defined fatigue as being a greater than 15% reduction in TwQ, as used by previous investigators (7). Only one patient demonstrated a greater than 15% reduction in TwQ_p after walking; in comparison, 8 of the 12 patients (including the patient who fatigued after the walking exercise) demonstrated LFF after cycling. Of these eight patients, four reported leg effort as the most important limiting symptom, three reported an equal contribution from SOB and leg effort, whereas one complained principally of dyspnea.

Acceptable CMAPs, not contaminated by stimulus artifact, were obtained in 11 of the 12 patients. No significant differences were seen between pre- and postexercise CMAP amplitudes. In particular, no differences in pre- and postcycling CMAP amplitudes were seen in those patients demonstrating a greater than 15% reduction in mean TwQ_u and TwQ_p values (unpotentiated: 7.62 and 7.64 mV [95% CI −0.75, 0.69; p = 0.92]; potentiated: 7.90 and 8.00 mV [95% CI −0.48, 0.31; p = 0.61]).

Tradition has resulted in most pulmonary departments using cycle ergometry for exercise testing. It provides a more stable platform for complex physiologic and metabolic measurements, allows workload to be measured directly, and is safer for patients. However, the disadvantage is that it is a less relevant form of exercise for many patients. In comparison, field-walking tests are arguably more realistic at mimicking real life and more familiar to patients, particularly unpaced endurance tests such as the 6-minute walking test. The externally paced ISW is highly reproducible and correlates well with V̇o₂max (12). The recently described ESW test is the only standardized field endurance walking test, shows good reproducibility after only one practice walk, and like the ISW is sensitive enough to detect changes as a result of treatment (11). The ISW and ESW also allow direct comparison with laboratory incremental and endurance cycle tests as a cycling equivalent to the 6-minute walking test does not exist.

A limitation of our study is that no direct metabolic or ventilatory measurements were made. We were primarily interested in the symptoms that occurred at functional maximum and did not wish the presence of mouthpieces, masks, or portable metabolic telemetric equipment to influence symptoms or exercise performance. Previous anecdotal experience has shown that our local patient population often stop prematurely when exercising with a mouthpiece or mask due to symptoms of excessive saliva or claustrophobia.

A further potential limitation of our study is that encouragement was not provided during the data collection exercise tests; this was to avoid any possible investigator bias, and efforts were made during the practice tests to push patients to their functional maximum. On the days when data were collected, performances were as good if not better than those achieved during the encouraged practice tests. In the present study, our patients managed comparable ISW distances and maximal power outputs during ICE to patients in previous studies (8) despite being naive to pulmonary rehabilitation and having more severe disease (lower FEV₁ and diffusing capacity of carbon monoxide).

Fatigue of skeletal muscle is defined as the loss of force-generating capacity resulting from activity under load that is reversible by rest (14). Of particular clinical interest is LFF, which results in loss of force generated in response to low stimulation frequencies (10–20 Hz), the typical motor neurone firing frequencies during human skeletal muscle contractions. LFF may last for more than 24 hours (15). Conventionally the detection of LFF requires the construction of force–frequency curves using tetanic electrical stimulation, but this is often impractical and not tolerable to patients. An acceptable alternative is to measure the pressure or tension elicited from a single supramaximal stimulus (i.e., TwQ) because the basis of the right shift of the force–frequency curve in LFF is the reduced twitch amplitude. We have previously demonstrated that supramaximal stimulation of the femoral nerves can be achieved using magnetic stimulation and that the technique can be used to detect quadriceps muscle fatigue (9), findings confirmed by other investigators (7, 13, 16). Although it is true to say that magnetic stimulation may not be as specific as transcutaneous electrical stimulation, the only motor nerve near the femoral nerve is the obturator, which supplies the adductor group of muscles. No physical evidence of any other movement than knee extension was observed during the stimulations. Magnetic stimulation has some advantages over electrical stimulation: it is well tolerated and less painful to patients, particularly important when repeated stimulations are required. Supramaximal stimulation is easier to achieve than with electrical stimulation, ultimately resulting in fewer stimulations to patients.

It is important to note that a fall in TwQ is only indicative of contractile fatigue, as a consequence of failure of excitation–contraction coupling, if it is accompanied by no change in CMAP amplitude. In contrast, both twitch force and CMAP amplitude decrease with transmission failure due to impairment of neural impulses through nerves or across neuromuscular junctions.

Breathlessness and leg effort are the commonest symptoms cited as limiting exercise. The seminal study from Killian and colleagues (6) reported the unexpected finding that despite severe abnormalities in respiratory mechanics, a good proportion of patients with COPD stop incremental cycling exercise due to the subjective feeling of leg effort rather than breathlessness. In 97 patients with chronic airflow limitation, the rating of dyspnea exceeded leg effort in only 26% of patients, whereas leg effort exceeded dyspnea in 43%. Some investigators criticized this study as the patients had only moderate disease. Although our cohort of 77 patients with COPD was older and had more severe disease, our results for ICE confirm these original findings (see Figure 1A). Subgroup comparisons did not reveal any significant difference in age, FEV₁, transfer factor, or arterial blood gases between the patients complaining primarily of dyspnea and those complaining of leg effort after incremental cycling. We have extended the observations of Killian and coworkers by recording the self-reported symptoms limiting endurance cycling, and incremental and endurance walking in the same cohort of patients. It appears that symptom limitation is exercise-specific with walking much more likely to be limited by breathlessness than leg effort or fatigue.

There are few reports comparing the symptoms limiting walking and cycling. Palange and colleagues demonstrated that the level of breathlessness was greater and level of leg effort smaller after an ISW test compared with an incremental cycling protocol (8). Mathur and coworkers compared incremental walking on a treadmill with incremental cycling and reported that dyspnea was the major symptom limiting both types of exercise (17). Both these studies were of small numbers of patients (nine and eight subjects respectively).

Most studies investigating the physiologic mechanisms of reduced exercise tolerance have analyzed pulmonary gas exchange during cycle ergometry (3, 18). However, cycling is not a familiar or relevant exercise to many patients, and may not reflect the metabolic and hemodynamic events that occur during more familiar exercises such as walking. Recently, Palange and colleagues compared the ventilatory and metabolic requirements during incremental cycling and walking (8), demonstrating a greater ventilatory demand with walking. There was higher V̇e/Vco₂ during walking that was not accounted for by anaerobic metabolism as lactate levels were lower. They demonstrated reduced efficiency of gas exchange with walking as shown by greater arterial hypoxaemia and physiologic dead space: Vt ratio. Another explanation of the differences observed is that walking is more of a whole body exercise involving both upper and lower extremities, compared with cycling where the lower limbs are the primary effector muscles. Hence, during walking, more muscles are at work so that each individual muscle (including the quadriceps) is, despite identical or even higher whole body V̇o₂, working at a lower fraction of its peak performance. Biomechanics data confirm that the locomotor lower limb muscles are more activated during cycling than walking (19, 20). During walking the arms may also be a potential source of increased afferent input to the respiratory centers (21). The finding that dyspnea is the primary symptom limiting walking exercise raises the question of whether field walking tests may be more appropriate or more sensitive to change when assessing the effects of therapeutic interventions on exercise-induced dyspnea.

The physiologic mechanisms underlying the symptoms limiting exercise are not fully understood, and we acknowledge that there is a complex interplay between the respiratory system and locomotor muscles (22). However, as leg effort did not appear to be an important symptom limiting walking exercise, we hypothesized that walking would be terminated before the onset of quadriceps fatigue. In this part of the study, it was demonstrated that a significant fall in TwQ_u and TwQ_p occurs in patients with COPD after incremental cycle exercise to the limits of tolerance, and that there is a significantly greater drop in TwQ tensions after cycling compared with walking. Whereas only 1 of the 12 patients fatigued their quadriceps after an ISW, two thirds of the same patients did so after incremental cycling. This suggests that patients with COPD may terminate walking due to excessive ventilatory demand before the onset of quadriceps LFF. It will be noted that in our cohort of patients there was greater oxygen desaturation after the walking tests compared with the cycling equivalents.

Previous work has also demonstrated the presence of quadriceps fatigue in patients with COPD after cycling exercise (7) with TwQ falling to 75.7% of baseline at 30 minutes postexercise. However, differences exist between that study and the present one. Our protocol was designed to compare quadriceps fatigability after incremental cycling and walking, whereas the study of Mador and colleagues used an endurance cycling protocol and was uncontrolled with no sham exercise run or control group (7). Acceptable CMAPs, uncontaminated from stimulus artifact, were obtained in 11 out of our 12 patients, a higher proportion than previously demonstrated. As discussed earlier, a fall in TwQ is indicative of LFF only when there is no accompanying change in CMAP amplitude.

Although quadriceps dysfunction is well documented in COPD (1–3), this study did not identify leg effort as a major limiting symptom after exhaustive walking. In addition, quadriceps LFF is not a major finding at the termination of walking exercise. It is interesting to note that there is no difference in the proportion of patients with COPD and healthy elderly control individuals reporting leg effort exceeding dyspnea at the end of exercise (6). Similarly, although quadriceps fatigue has been demonstrated in patients with COPD after exhaustive endurance cycling (7), the same authors later demonstrated a greater fall in TwQ in healthy elderly subjects (13). However, the role of quadriceps dysfunction in limiting walking remains unclear. Certainly, as demonstrated in the present study, the quadriceps may have an influence on symptoms during walking exercise in a small proportion of patients with COPD. Several studies have also shown significant associations between muscle strength and both maximal laboratory exercise testing and self-paced walking tests (23, 24), and trials of quadriceps strength training have demonstrated benefits to incremental, endurance, and self-paced walking (4, 25–27). In addition, the improvement in walking distance after pulmonary rehabilitation is now well established (28).

In conclusion, dyspnea is the most important symptom limiting walking exercise in patients with moderate to severe COPD, and LFF of the quadriceps, although present after incremental cycling, is rarely present after incremental walking.