American Journal of Respiratory and Critical Care Medicine

Physiologic adaptations after an 8-wk endurance training program were examined in 13 patients with chronic obstructive pulmonary disease (COPD) (age, 64 ± 4 [SD] yr; FEV1, 43 ± 9% pred; PaO2 , 72  ± 8 mm Hg; and PaCO2 , 36 ± 2 mm Hg) and in eight healthy sedentary control subjects (61 ± 4 yr). Both pre- and post-training studies included: (1) whole-body oxygen consumption (V˙ o 2) and one-leg O 2 uptake (V˙ o 2leg) during exercise; and (2) intracellular pH (pHi) and inorganic phosphate to phosphocreatine ratio ([Pi]/[PCr]) during exercise; and half-time of [PCr] recovery. After training, the two groups increased peak V˙ o 2 (p < 0.05 each) and showed a similar fall in submaximal femoral venous lactate levels (p < 0.05 each). However, control subjects increased peak V˙ e (p < 0.01) and raised peak O 2 delivery (p = 0.05), not shown in patients with COPD. Both groups increased post-training O 2 extraction ratio (p < 0.05). The most consistent finding, however, was in patients with COPD, who had a substantial improvement in cellular bioenergetics: (1) half-time of [PCr] recovery fell from 50 ± 8 to 34 ± 7 s (p = 0.02); and (2) at a given submaximal work rate, [Pi]/[PCr] ratio decreased and pHi increased (p < 0.05 each). We conclude that beneficial effects of training in patients with COPD essentially occurred at muscle level during submaximal exercise.

Respiratory rehabilitation, including lower limb exercise training, is recommended as part of the management for patients with chronic obstructive pulmonary disease (COPD), because it has been consistently shown that this relieves dyspnea and improves health-related quality of life (HRQL) (1). It is of note, however, that the physiologic mechanisms underlying the beneficial consequences of training are poorly understood (1-5). In 1991, Casaburi and colleagues (6) demonstrated that a relatively high-intensity training program was required to improve exercise endurance in patients with moderate COPD. More recently, Maltais and coworkers (7, 8) reported that patients with COPD concurrently showed early rise of blood lactate levels during exercise, together with low concentration of skeletal muscle oxidative enzymes, as compared with normal subjects. The two findings were at least partially corrected after a controlled endurance training program (8). These two groups of investigators (6, 8) hypothesized that reduction of blood lactate levels after training decreased ventilatory requirements during exercise, which in turn would relieve dyspnea and improve exercise tolerance. More recently, Casaburi and colleagues (9) reported that rigorous physical training in patients with severe COPD, in whom blood lactate levels did not increase during exercise, yielded a more efficient breathing pattern and reduced minute ventilation at a given submaximal work rate, hence improving exercise tolerance.

An alternative hypothesis is that improvement of HRQL after endurance training is directly and mainly related to enhancement of skeletal muscle bioenergetics during submaximal exercise, rather than to changes in ventilation (V˙e). This notion may help to resolve the apparent discrepancies between the consistently beneficial effects of physical training on HRQL (1-4) and the variability among studies regarding the impact of rehabilitation on peak O2 uptake (V˙o 2 peak) (1). The present study examined the relationships among oxygen consumption (V˙o 2), muscle O2 transport, and cellular bioenergetics after an 8-wk endurance training program in 13 patients with COPD and in eight healthy sedentary subjects used as control subjects.

Study Group

Thirteen clinically stable male, steroid-free patients with COPD (mean age, 64 ± 4 [mean ± SD] yr; height, 167 ± 7 cm; and weight, 73 ± 8 kg) without recent history (6 mo) of an acute episode of exacerbation, and eight healthy sedentary men (age, 61 ± 4 yr; height, 170 ± 9 cm; and weight, 73 ± 7 kg) were enrolled in the study. The patients with COPD were selected a priori because of their moderate-to-severe ventilatory impairment, but preserved single-breath carbon monoxide transfer capacity (Dl CO). All of them showed only moderate hypoxemia at rest without exercise-induced oxyhemoglobin desaturation. Age, anthropometric variables, lung function measurements, therapy, and smoking habits of the two groups are listed for each individual in Tables 1 and 2. The control group of sedentary subjects was carefully selected on the basis of no previous history of regular or even occasional physical exercise above that required for average daily activities. In each subject, the same measurements described below were carried out before and after the training period. All were informed of any risks and discomfort associated with the experiment, and written informed consent was obtained in accordance with the Committee on Investigations Involving Human Subjects at the Hospital Clı́nic, Universitat de Barcelona, which approved the study.


Age (yr)Weight (kg)Height (cm)BMI (kg · m 2 )Tobacco ( pack-years)Medication ( puffs per day)[Hb] ( g · dl 1 )
Patients with COPD
  166791722730(SB) + 12 IB + 4 S + 4 B13.0
  267721682690(SB) + 4 S15.5
  3636316423508SB + 4 S + 4 B16.5
  4618317328 5(SB) + 2 IB15.8
  561791752650(SB) + 16 IB + 2 S + 4 B15.0
  865651632480(SB) + 12 IB + 6 S + 6 B15.8
  966641672365(SB) + (IB) + (B)15.7
 11677916629306 SB + 6 IB14.6
 1369781772540(SB) + 4 S + 4 IB + 4 B13.7
 SD 4 8  7 221 1.0
Healthy Sedentary Subjects
 SD 4 7  9 2 0.90

Definition of abbreviations: B = budesonide, 200 μg/puff; BMI = body mass index; COPD = chronic obstructive pulmonary disease; Hb = hemoglobin concentration; IB = ipratropium bromide, 18 μg/puff; ( ) = bronchodilators used only occasionally are indicated within parentheses; S = salmeterol, 25 μg/puff; SB = salbutamol, 100 μg/puff. Only two of eight control subjects were smokers. Results expressed as individual data and mean ± standard deviation (SD).


FEV1 FEV1/FVC (%)TLC (% pred )RV/TLC (%)Dl CO sb(% pred )PaO2 (mm Hg)Wpeak(W )o 2peak (L · min−1 )
(% pred )(L)
Patients with COPD
  1 461.54010740 73 82 801.2
  2 391.24211665 86 60 501.3
  3 401.24310655 54 81 501.1
  4 391.44310257108 72 501.7
  5 371.35111167103 75 951.4
  6 501.157 8956 87 821001.1
  7 471.553 9555 74 681001.4
  8 300.938 9359 65 59 601.1
  9 541.75410456 65 73 801.5
 10 331.13310656 57 70 801.2
 11 541.64611347 98 761201.6
 12 571.85710951 86 621402.2
 13 311.33511952 72 79 701.2
 Mean 431.44610555 79 72 831.4
 SD  90.3 8  9 7 17  8 280.3
Healthy Sedentary Subjects
  1 831.962 931001.3
  2 873.166102431071171402.1
  31013.791 7838 721021202.1
  4 963.7651042102.8
  5 983.578 96391171061602.1
  61113.66411232 94 921001.4
  7 953.174 9535 98 971402.0
  8 782.57610053 80 95 801.2
 Mean 943.172 9740 951011311.9
 SD 100.710 11 7 17  8 410.5

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; Dl CO = single-breath diffusing capacity for carbon monoxide expressed as percent predicted; RV/TLC = residual volume to TLC ratio expressed as actual percentage; TLC = total lung capacity, expressed as percent predicted; V˙ o 2peak = whole-body O2 uptake at peak exercise, in L · min−1; W = peak work rate, in watts. Results expressed as individual data and mean ± standard deviation (SD).

31P-Magnetic Resonance Spectroscopy

In eight of the 13 patients with COPD and in all eight healthy sedentary subjects, 31P-magnetic resonance spectroscopy (31P-MRS) measurements in the quadriceps of the left thigh were carried out less than 1 wk before the study with catheters as described below. Exercise tests during MRS measurements were performed using an ergometer made of nonmagnetic materials conceived to fit into a standard whole-body magnet. The system allowed flexion and extension of both legs by alternatively pressing two foot pedals connected to a hydraulic-controlled resistance system while the subject lay supine on the ergometer table (10). A piezoelectric force transducer (9251A model; Kistler Instruments AG, Winterthur, Switzerland) placed inside one of the pedals was used to provide an “on-line” measure of both strength and rate of pedaling during the exercise test.

Each subject began with an incremental protocol that consisted of a maximum of five periods of 2-min of exercise each, pedaling at 60 rpm. The two first periods were set at 3 W · kg−1 muscle mass, with subsequent increases at 7, 9, and 11 W · kg−1 muscle mass. After a full 1-h recovery at rest, the subject performed a constant work rate exercise for 2 min. This second protocol was designed to allow evaluation of [PCr] recovery from data collected for 8 min after exercise. The constant work rate used in each subject was chosen to produce a rise in [Pi]/[PCr] ratio of approximately 1 ± 0.5 while keeping pHi higher than 6.90 units, as measured in the first spectrum of the recovery period. The purpose was to prevent a deleterious effect of low intracellular pH on the rate of oxidative phosphorylation (11-13).

31P-MRS measurement details. Studies were carried out using a 1.5-T Signa System (Signa Advantage; General Electric Medical Systems, Milwaukee, WI) operating at a frequency of 63.65 MHz and 25.86 MHz for the hydrogen-1 and the phosphorus-31 nucleus, respectively, connected to a SPARCStation 20 Workstation (Sun Microsystems, Mountain View, CA). Subjects were placed in the magnet in a supine position, head first, and the elliptical distributed capacitance surface coil (14.5 cm × 6.5 cm), pretuned at the phosphorous resonant frequency, was positioned and fixed over the vastus medialis muscle of the left leg. Then the position of the surface coil was checked with spin-echo T1-weighted images (10). Phosphorous-31 spectra were obtained using 180° pulses as measured at the center of the coil. A total of 1,024 data points were accumulated for the measurements at rest, 12 scans for each work rate of the incremental exercise, and blocks of eight scans were continuously recorded during the 8-min recovery period form the constant work rate exercise.

Free induction decays were analyzed in the time domain with the magnetic resonance user interface software (14). All resonances were fitted to single lorentzian functions with the nonlinear least-squares variable projection method (15, 16) and the estimated amplitudes of the time domain signals, which correspond to the area under each resonance in the frequency domain, and their chemical shifts were directly used for calculations. The individual half-time of phosphocreatine ([PCr]) recovery was calculated by fitting the time domain amplitudes to a monoexponential function (Figure P; Biosoft, Cambridge, UK). The intracellular pH (pHi) was calculated using the formula: pHi = 6.75 + log [(d − 3.27)/(5.69 − d)], where di is the chemical shift distance in ppm between the Pi and the PCr resonances (17).

Whole-body and One-leg O2 Uptake Measurements

Subject preparation, safety precautions, and technical aspects of the central measurements (arterial and femoral venous blood gases and femoral venous blood flow) have been described in detail elsewhere (18-21). Briefly, one catheter was placed in the radial artery of the nondominant arm to measure Po 2, Pco 2, pH, SaO2 , lactate, and hemoglobin in arterial blood. In the femoral vein of the left leg, a 7-F catheter was advanced 7 cm into the vessel with the tip oriented distally and a 2.5-F thermistor was advanced 5 cm proximally into the same vessel. Each subject performed an incremental cycle exercise test (10-watt [W] increments every 2 min in patients with COPD and 20-W increments every 2 min in healthy sedentary control subjects, without an initial period cycling at zero watts) breathing room air (Fi O2 = 0.21) until exhaustion. The exercise protocol was done using an electromagnetically-braked cyclo ergometer (CardiO2 cycle; Medical Graphics Corporation, St. Paul, MN) with a mechanical assist to overcome the internal frictional resistances.

On-line calculations of whole-body V˙o 2, CO2 output (V˙co 2), minute ventilation (V˙e), respiratory exchange ratio (RER), heart rate (HR), and respiratory rate (RR) were averaged sequentially over 15-s intervals and displayed on a screen monitor to observe the progress of the tests. In each subject, simultaneous arterial and femoral venous blood samples were collected at rest and during the second minute of each incremental work rate. Femoral venous blood flow measurements were made by short-term steady-state thermodilution using iced saline (18, 19) immediately after femoral venous blood sampling. In each instance, the following measurements were made: (1) Po 2, Pco 2, pH (IL model 1302, pH/blood gas analyzer and tonometer model 237; Instrumentation Laboratories, Milan, Italy), oxyhemoglobin saturation, hemoglobin concentration (Hb) (IL 482 co-oximeter), and whole-blood lactate concentrations (YSI 23L blood lactate analyzer; Yellow Springs Instruments, Yellow Springs, OH) from simultaneous arterial and femoral venous blood samples; and (2) femoral venous blood flow (Q˙leg) and arterial pressure. As indicated above, V˙e, fraction of expired oxygen (Fe O2 ), fraction of expired carbon dioxide (Fe CO2 ), and HR were continuously monitored. Technical aspects of these measurements have been previously provided in detail (18-21). In the pre-training studies of five of the 13 patients with COPD, femoral venous flow was measured only at approximately 30, 60, 80, and 100% of peak work rate (as assessed in a preliminary incremental exercise protocol done in all subjects before inclusion in the study). Measurements of leg blood flow at each work rate were done in all the remaining pre- and post-training studies of the two groups of subjects (patients with COPD and control subjects).

In the present study, blood O2 content was calculated as follows: [(1.39 · Hb · measured oxyhemoglobin saturation) + (0.003 · Po 2)]. This was done for arterial (CaO2 ) as well as venous (CfvO2 ) blood. The O2 delivery to the exercising leg (Q˙o 2leg) was calculated as the product of arterial O2 content and leg blood flow [Q˙o 2leg = CaO2 · Q˙leg]. Leg O2 uptake (V˙o 2leg) was obtained as the product of Q˙leg and the arterial − femoral venous difference of O2 content [V˙o 2leg = Q˙leg · (CaO2 − CfvO2 )]. Leg O2 extraction ratio (O2ER) was calculated as the ratio of the arterial to femoral venous O2 content difference and the arterial O2 content [O2ER = 100 · (CaO2 − CfvO2 )/CaO2 ]. In each subject, measured O2 saturation and the corresponding Po 2 from all samples were used to estimate the oxygen half-saturation pressure (P50) of hemoglobin.

Training Program

The two groups of subjects (patients with COPD and control subjects) exercised on a cycle ergometer (Monark model 810; Monark, Sweden) 5 d per week for 8 wk. The training sessions were directly supervised by one of the members of the team (J.N.). The compliance of the subjects was excellent, with no withdrawals during the training period. During the cycling sessions, HR was continuously monitored (SportTester PE 3000 System; Polar Electro, Kemple, Finland). The training sessions, approximately 60-min duration, consisted of (1) 5 min of cycling at the low work rate, 40% of the peak work rate (40% Wpeak) achieved in the previous control; (2) 20 min cycling at the high work rate, 70–90% Wpeak; (3) 5 min cycling at 40% Wpeak; (4) 20 min cycling at 70–90% Wpeak; and (5) 5 min at 40% Wpeak. The rate of pedaling during the sessions was kept at 60 rpm. The progress of work rate during the training period was decided on an individual basis to maximize the training effect. During the first 3 wk of the program, cycling at the high work rate was set at least 70% Wpeak. Thereafter, it was increased at least by 5% every week up to a maximum of 90% of the actual Wpeak during the last 2 wk of the training program.

Data Analysis

Results are expressed as mean ± SD. Comparisons of pre- and post-training incremental exercise within each group (13 patients versus eight control subjects) were done, unless otherwise stated, using Student's paired t test to examine the slopes (and intercepts) derived from individual least-squares regression lines of each variable versus work rate and V˙o 2 (both whole-body V˙o 2 and V˙o 2leg). Comparisons of peak exercise values between pre- and post-training were analyzed similarly. The effects of training on femoral venous lactate concentrations ([La]fv) compared post-training to pre-training values by regression analysis. Comparisons of pre-training data between patients with COPD and control subjects were done using Student's unpaired t test to examine the slopes (and intercepts) derived from individual results. Comparison of the profiles of the O2 extraction ratio (O2ER) response to exercise was carried out using a repeated measurements analysis of variance (MANOVA). Since we found that the O2ER profile was substantially different between groups (p = 0.03), the analysis of the O2ER response during exercise was done for each group separately. Statistical significance was set up at p value equal or lower than 0.05.

Lung Function and Exercise Tolerance before Training

The characteristics of the patients with COPD and the healthy sedentary control subjects are depicted in Tables 1 and 2. The 13 patients with COPD presented moderate to severe airflow obstruction (FEV1, 1.4 ± 0.3 L, 43 ± 9% predicted) (22-24), slight single-breath Dl CO impairment (79 ± 17% predicted) (25), and only moderate hypoxemia at rest (72 ± 8 mm Hg). Arterial Po 2 did not show significant changes from rest to peak exercise (+8.4 ± 9.5 versus +10.2 ± 8.4 mm Hg, patients with COPD and control subjects, respectively). While arterial Pco 2 did not change (from 35.7 ± 2.4 to 34.1 ± 2.8 mm Hg) during exercise in control subjects, PaCO2 significantly increased at peak exercise in patients with COPD (from 35.5 ± 2.9 to 40.8 ± 4.4 mm Hg) (p = 0.0001). Arterial O2 content (CaO2 ) was similar in the two groups both at rest (19.6 ± 1.1 versus 19.5 ± 1.3 ml O2 · 100 ml−1 blood, patients with COPD and control subjects, respectively) and at peak exercise (19.7 ± 1 versus 19.6 ± 1.2 ml O2 · 100 ml−1 blood).

As expected, the patients with COPD showed exercise intolerance as assessed by both lower peak work rate (85 ± 26 versus 131 ± 41 W) (p = 0.005) and peak whole-body O2 uptake (1.4 ± 0.3 versus 1.9 ± 0.6 L · min−1) (p = 0.008) than control subjects. Peak O2 uptake expressed as percent of predicted values was 69 ± 16% and 85 ± 16% (p = 0.04), respectively. The patients with COPD in the present study showed, at a given submaximal work rate (40 W), significantly higher whole-body V˙o 2 (p < 0.009) and V˙o 2leg (p < 0.02) than the healthy sedentary control subjects with similar slopes of O2 uptake versus work rate relationships in the two groups for both whole-body O2 uptake (12 ± 3 versus 12 ± 1 ml · min−1 · W−1, respectively), as shown in Figure 1.

The ventilatory and heart rate responses to exercise in patients with COPD were similar to those reported in the literature (30). Peak ventilation was lower in patients with COPD than in control subjects (45.6 ± 7.1 versus 70.9 ± 22.2 L · min−1, respectively) (p = 0.001), but peak respiratory rate was similar (34.9 ± 8.4 versus 33.3 ± 10 min−1). At a given submaximal work rate (40 W), V˙e was higher in patients with COPD than in control subjects (28.8 ± 7.4 versus 21.5 ± 4.4 L · min−1) (p = 0.02). The response of Vt to exercise was also different between groups; while in the control group mean Vt rose progressively throughout exercise, the patients with COPD showed a well-defined plateau in Vt above approximately 60 W. Mean peak heart rates were not different between groups (133 ± 20 versus 145 ± 15 min−1, patients with COPD and control subjects, respectively), but HR at a given submaximal work rate (40 W) was significantly higher in patients with COPD than in control subjects (107 ± 17 versus 93 ± 10 min−1) (p = 0.05). The RER at peak exercise was also significantly lower in patients with COPD than in control subjects (1.02 ± 0.08 versus 1.16 ± 0.09) (p = 0.002).

Training Response

Muscle O2 transport and O2 uptake. Peak V˙o 2leg (Figure 1, upper panel ) significantly increased after training in the control group (by 0.18 ± 0.11 L · min−1) (p = 0.002). The difference in peak V˙o 2leg of the patients with COPD (0.07 ± 0.2 L · min−1) was not significant (p = 0.2). In contrast, peak whole-body V˙o 2 (Figure 1, bottom panel ) rose in both groups (by 0.15 ± 0.25 and by 0.27 ± 0.20 L · min−1, patients with COPD and control subjects) (p = 0.05 and p = 0.005, respectively). Likewise, peak work rate significantly increased with training in both groups (by 15 ± 12 W and by 33 ± 10 W, patients with COPD and control subjects, respectively) (p < 0.01 each). RER at peak exercise did not change after training in patients with COPD (from 1.02 ± 0.08 to 1.00 ± 0.1), but it increased in the control group (from 1.16 ± 0.09 to 1.21 ± 0.09) (p = 0.04). As depicted in Figure 1, the slope of post-training V˙o 2 (both whole-body V˙o 2 and V˙o 2leg) versus work rate was not different between patients with COPD and control subjects, and in the two groups it essentially overlaid the pre-training data.

For both subject groups, the plots of both post-training one-leg blood flow (Q˙leg) and post-training one-leg O2 delivery (Q˙o 2leg) versus work rate (or O2 uptake) essentially overlaid the pre-training data, as shown for Q˙o 2leg in Figure 2 (upper panel ). The normal subjects showed a significant increase in these two variables at peak exercise: peak Q˙leg increased by 1.15 ± 1.19 L · min−1 (p = 0.03) and peak Q˙o 2leg rose by 0.19 ± 0.23 L · min−1 (p = 0.05). However, no changes were observed in the patients with COPD after training (Table 3). Both leg blood flow and leg O2 delivery at a given submaximal work rate were not significantly different between patients with COPD and control subjects in either of the studies (pre- and post-training). The arterial minus femoral venous blood O2 content gradient at peak exercise did not show significant changes with training in patients with COPD (from 13.7 ± 1.7 to 14.2 ± 1.2 ml O2 · 100 ml−1 blood) nor in the control group (from 13.2 ± 1.4 to 13.8 ± 1.5 ml O2 · 100 ml−1 blood). Peak oxygen extraction ratio (O2ER, %) increased in both the patients with COPD (from 69.6 ± 8.3 to 74 ± 5.4%) (p = 0.02) and the control group (from 67.4 ± 6.0 to 73.7 ± 6.0%) (p = 0.05), but did not differ between groups. O2ER during submaximal exercise was significantly higher in patients with COPD than in control subjects (p = 0.03) both in the pre- and post-training (Figure 2, bottom).


Training Response (mean post- minus pre-training difference)Healthy SubjectsPatients with COPD
Difference%p ValueDifference%p Value
WR peak, W33 ± 10 20< 0.0115 ± 12 15< 0.01
Whole-body V˙ o 2 peak, L · min−1 0.3 ± 0.2 13< 0.010.2 ± 0.3 10  0.05
e peak, L · min−1 18 ± 12 23< 0.013.6 ± 8  7
e submax, L · min−1 −3.3 ± 3.2−15  0.01−1.9 ± 4.4 −7
HR peak, beats · min−1 14 ± 14  9  0.030.4 ± 10  0
HR submax, beats · min−1 −4.3 ± 10 −5−9.5 ± 12 −8  0.01
leg peak, L · min−1 1.2 ± 1.2 24  0.030.3 ± 1.5 11
o 2 peak, L · min−1 0.2 ± 0.2 20  0.050.04 ± 0.3  6
Peak (a-fv) O2 diff, mlO2 · 100 ml−1 0.6 ± 1.8  50.5 ± 1.6  4
O2 ER peak, %6.2 ± 7.3  9  0.054.8 ± 6.5  6  0.02
[La]fv peak, mM · L−1 1.6 ± 2.2 19−0.5 ± 2 −9
[La]fv submax, mM · L−1 −0.3 ± 0.4−26  0.05−0.6 ± 0.9−32  0.03
[Pi]/[PCr] peak−2.2 ± 3.8−42−1.7 ± 3.5−41
[Pi]/[PCr] submax−0.01 ± 0.6 −1−0.8 ± 1−51  0.05
pHi peak0.07 ± 0.2  10.1 ± 0.2  2
pHi submax−0.02 ± 0.08  00.1 ± 0.1  2  0.02
T1/2 of [PCr] recovery, s−6.2 ± 10.3−18−17.6 ± 11.3−35  0.02

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; HR = heart rate; [La]fv = femoral venous lactate concentration; [PCr] = phosphocreatine concentration; Peak (a-fv) O2 diff = difference between arterial and femoral venous blood oxygen content; pHi = intracellular pH; [Pi] = inorganic phosphorus concentration; Q˙ leg = leg blood flow; Q˙ o 2leg = leg oxygen delivery; T1/2 of [PCr] recovery = half-time of phosphocreatine concentration recovery. Training response expressed as absolute post- minus pre-training differences and as percentage of change after training; p value is Student's paired t test between pre- and post-training absolute values. Results at submaximal work rate were calculated at 40 W for the following variables measured during cycling exercise: V˙ e, HR, Q˙ leg, Q˙ o 2leg, and [La]fv. Results for [Pi]/[PCr] and pHi correspond to pooled data obtained during the incremental protocol done within the magnet.

Figure 3 (upper panel ) shows that while the normal subjects moderately decreased V˙e at a given submaximal work rate (40 W) from 22 ± 4.4 to 18 ± 3.5 L · min−1 (p = 0.02) and substantially increased peak V˙e from 71 ± 22.2 to 89 ± 27.2 L · min−1 (p = 0.004), no significant effects of training on submaximal V˙e nor in peak V˙e were observed in patients with COPD (Table 3). It is of note, however, that submaximal V˙e after training showed a trend to fall, by −1.9 L · min−1 (p = 0.07, one tail). Control subjects significantly increased post-training peak HR (from 145 ± 15 to 158 ± 13 cycles · min−1) (p = 0.03), a change not seen in patients with COPD (133 ± 19 compared with 133 ± 20 cycles · min−1). The COPD group did, however, show a significant reduction of HR at a given submaximal work rate (40 W), by 10 ± 12 min−1 (p = 0.01). Despite the substantial differences in FEV1 between patients with COPD and control subjects, physical training decreased femoral venous lactate concentrations ([La]fv) at any work rate in a similar magnitude in the two groups, as indicated in Figure 3 (lower panel ) (p < 0.05, each). After training, however, while differences in base excess between patients with COPD and control subjects disappeared, [La]fv levels were still higher in patients with COPD than in control subjects (at 40 W, 1.9 ± 0.36 versus 1.2 ± 0.28 mM · L−1, respectively) (p = 0.0001).

Skeletal muscle bioenergetics. Pre-training half-time of [PCr] recovery was significantly longer in the patients with COPD (50 ± 8 s) than in control subjects (35 ± 7 s) (p = 0.02). After training, half-time of [PCr] recovery substantially fell to normal levels in patients with COPD (34 ± 7 s) (p = 0.02), but this variable did not show significant changes in the healthy sedentary subjects (29 ± 8 s). The analysis of changes in half-time recovery was done in only six control subjects, since two of them (subjects no. 1 and 2) had to be excluded because pHi fell below 6.9 either in the pre- or post-training study. Endurance training improved the skeletal muscle bioenergetic status to normal levels in patients with COPD during incremental exercise (Figure 4), as shown by the lesser increase in the [Pi] to [PCr] ratio and the lesser reduction in pHi at a given submaximal exercise. For example, at 27 W the [Pi]/[PCr] ratio fell from 2.8 to 1.3 (p = 0.05) and the pHi increased from 6.83 to 6.98 (p = 0.04). In contrast, the normal sedentary subjects did not show significant changes in the [Pi]/[PCr] ratio nor in pHi at this level of exercise. It is of note, however, that the control group showed a trend toward improvement of the bioenergetic status after training at the last step of the exercise protocol (Figure 4).

Nature of the Training Response

The present study shows that patients with COPD and healthy sedentary control subjects presented clear physiologic training effects both at peak work rate and during submaximal exercise. After training, peak whole-body V˙o 2 and peak work rate significantly increased in the COPD group (by 10% and 15%, respectively) (p < 0.01 each) and in the control group (by 13% and 20%) (p = 0.05 and p < 0.01). Likewise, post-training [La]fv during submaximal exercise (at 40 W) fell by −32% (p = 0.03) in the patients with COPD and by −26% (p = 0.05) in the healthy sedentary control group. The two groups, however, showed quite different physiologic adaptations to endurance training (Table 3). Whereas the healthy subjects markedly increased the central factors governing convective O2 transport at peak exercise: (1) femoral venous blood flow and heart rate; (2) ventilation; and (3) O2 delivery, no changes in any of these variables at peak exercise were observed in the patients with COPD (Figures 2 and 3, upper panels ). In fact, similarities of pre- and post-training ventilatory responses to exercise (as well as the observed plateau in tidal volume above 60% peak work rate) in the patients with COPD support the notion that ventilatory capacity was one of the factors (albeit not the only one) limiting peak exercise in these patients. The lack of a significant reduction of post-training V˙e during submaximal exercise (by −1.9 L · min−1) should be attributed to a type II error. It is of note that post-training heart rate during submaximal exercise (at 40 W) fell by 8% (p = 0.01) in the COPD group.

The most consistent effects of the training program in patients with COPD were: (1) a significant increase in peak O2 extraction ratio (also seen in the control group); (2) reduced half-time of [PCr] recovery; and (3) improved cellular bioenergetics during submaximal exercise ([Pi]/[PCr] ratio and pHi) (Figure 4). Overall, the post-training findings in patients with COPD indicate that the physiologic changes provoked by endurance training essentially took place at the level of the skeletal muscle during submaximal exercise. The increase in peak whole-body V˙o 2 after training in the patients can be interpreted as the end-result of the skeletal muscle changes alluded to above.

The O2 extraction ratio (V˙o 2/Q˙o 2), at any given work rate, can be analyzed as the ratio of peripheral to central components of O2 transport (18, 31). While the denominator (Q˙o 2) includes only central components of O2 transport (leg blood flow and arterial O2 concentration), the numerator (V˙o 2) can be expressed as a reflection of peripheral O2 transfer. According to the laws of diffusion, V˙o 2 is the product of muscle capillary O2 conductance and the Po 2 difference between muscle capillaries and mitochondria. In the present study, since peak Q˙o 2 did not change (patients with COPD) or even increased (healthy sedentary control subjects) (Figure 2, bottom panel), the rise in peak %O2 ER after training should be interpreted as an improvement of the muscle O2 transfer. There are three potential mechanisms to explain such an increase in peak %O2 ER with training: (1) higher capillary-to-muscle fiber ratio due to training-induced angiogenesis, which would effectively enlarge the area for O2 transfer in the muscle microcirculation; (2) improved mitochondrial oxidative capacity, which might potentially increase the O2 transfer gradient between capillary and mitochondria; and (3) the combined effects of these two phenomena. Unfortunately, the analysis of structural changes provoked by training was beyond the scope of the present study. In the patients with COPD, however, since the half-time of [PCr] recovery fell to normal levels after training, it can be hypothesized that physical training significantly improved mitochondrial oxidative capacity, as suggested by previous studies (8). The half-time of [PCr] recovery is an overall marker of the dynamics of muscle bioenergetics. It is well accepted, however, that when this variable is measured after a low-intensity constant work rate exercise preventing a marked fall in pHi would otherwise negatively interfere oxidative phosphorylation, the half-time of [PCr] recovery reflects skeletal muscle mitochondrial oxidative capacity (11– 13, 32).

A fundamental outcome of the post-training study in patients with COPD is the improvement of cellular bioenergetics ([Pi]/[PCr] ratio and pHi) during submaximal exercise up to normal levels, despite the lack of improvement in convective O2 transport (Q˙o 2) (Figure 2, upper panel ). The results of the present study suggest that physical deconditioning plays a key role in the impairment of skeletal muscle bioenergetics in patients with COPD. The prominence of the training effects on cellular bioenergetics in the COPD group may not be representative of the average training response usually seen in the clinical setting. Both the intensity of the controlled training program and the inclusion criteria of the patients can be important factors in explaining our results.

It is well accepted that healthy sedentary subjects show clear training effects on both cellular metabolism and muscle O2 transport (33). The lack of improvement in the cellular bioenergetic status ([Pi]/[PCr] ratio and pHi) during submaximal incremental exercise in our control group can be largely attributed to the design of the 31P-MRS study. The use of an identical incremental protocol for the two groups of subjects had some advantages, but only allowed exploration of very low work rates relative to the expected peak exercise for the control group. Post-training improvement of cellular bioenergetics in the healthy sedentary subjects likely would have been shown by exploring further steps in the incremental protocol within the magnet. This was not, however, a central aim of the present study. Moreover, the absence of a significant fall in half-time of [PCr] recovery after training (from 35 to 28 s) in normal subjects can be most likely attributed to a type II error since only six subjects could be included in the final analysis.

O2 Uptake at Early Submaximal Exercise

It is well accepted that patients with COPD show relationships between O2 uptake and work rate within the normal range such that O2 uptake for a given submaximal work rate in these patients is similar to that seen in healthy sedentary subjects (30). It is of note, however, that in the present study the patients with COPD showed, at a given submaximal work rate, significantly higher whole-body V˙o 2 (p < 0.009) and V˙o 2leg (p < 0.02) than the healthy sedentary control subjects (Figure 1) with no differences in the slopes of O2 uptake (whole body and leg) versus work rate between groups. We can reasonably assume the reliability of our results because whole-body V˙o 2 and V˙o 2leg are independent measurements, and similar differences between the two groups were obtained in the pre- and post-training studies. Since the problem appears in the transitional zone from rest to slight exercise without measurements obtained during unloaded (zero watts) cycling, it is difficult to provide a clean explanation for such a difference in O2 uptake at early submaximal exercise. The design of the incremental exercise protocol without the recommended 3-min unloaded exercise (34, 35) could be considered a weakness of the study, but to our understanding it does not constrain the interpretation of the beneficial effects of physical training in patients with COPD, which, in fact, is the main outcome of the present investigation.

In summary, the present study suggests that the sedentary lifestyle of these patients with COPD due to psychological and/or physiologic factors provokes functional changes in the skeletal muscle that can be partly or completely reversed by physical rehabilitation, as shown by the improvement of cellular bioenergetics in skeletal muscles to normal levels. The results of the investigation stress the importance of physical deconditioning in the skeletal muscle dysfunction observed in patients with COPD. It can be hypothesized that such an enhancement of the cellular bioenergetic status after physical rehabilitation may contribute to the underlying physiologic basis of the improved HRQL observed in most of the studies (1). Physiologic adaptations to physical training essentially take place at muscle level, and they do not necessarily have an appreciable impact on the ventilatory response to incremental exercise. Our results support the notion that improvement of the bioenergetic status of the skeletal muscle during submaximal exercise should constitute one of the crucial components of the expected goals of physical training. The findings of the present study may be of help in better identifying outcome variables for physical rehabilitation programs.

The authors are grateful to Felip Burgos, Jaume Cardús, and all the technical staff of the Lung Function Laboratory for their skillful support during the study; to Vicky Cabestany, Lola Núñez, and Conxi Gistau for their outstanding support in the training program; to Narcis Gusı́ for his advice in conducting the training program; and to Mirjam Hillenius for typing the manuscript.

Supported by grants FIS 94-1106 and 97-0794 from the Fondo de Investigaciones Sanitarias; ALFA ETIR 2.42 (8) from the European Union (DG XII); Comissionat per a Universitats i Recerca de la Generalitat de Catalunya (1997 SGR-0086); and HL-17731 from the National Heart, Lung, and Blood Institute.

1. Lacasse Y., Wong E., Guyatt G. H., King D., Cook D. J., Goldstein R. S.Meta-analysis of respiratory rehabilitation in chronic obstructive pulmonary disease. Lancet348199611151119
2. Celli B.Is pulmonary rehabilitation an effective treatment for chronic obstructive pulmonary disease? Yes. Am. J. Respir. Crit. Care Med.1551997781783
3. Albert R.Is pulmonary rehabilitation an effective treatment for chronic obstructive pulmonary disease? No. Am. J. Respir. Crit. Care Med.1551997784785
4. Celli B. R.Pulmonary rehabilitation in patients with COPD. Am. J. Respir. Crit. Care Med.1521995861864
5. Wijkstra P., Van Der Mark T., Kraan J., Van Altena R., Köeter G., Postma D.Long-term effects of home rehabilitation on physical performance in chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med.153199612341241
6. Casaburi R., Patessio A., Ioli F., Zanaboni S., Donner C. F., Wasserman K.Reductions in exercise lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease. Am. Rev. Respir. Dis.1431991918
7. Maltais F., Simard A. A., Simard C., Jobin J., Desgagnes P., Leblanc P.Oxidative capacity of the skeletal muscle and lactic acid kinetics during exercise in normal subjects and in patients with COPD. Am. J. Respir. Crit. Care Med.1531996288293
8. Maltais F., Leblanc P., Simard C., Jobin J., Bérubé C., Bruneau J., Carrier L., Belleau R.Skeletal muscle adaptation to endurance training in patients with chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med.1541996442447
9. Casaburi R., Porszasz J., Burns M. R., Chang R. S. Y., Cooper C. B.Physiologic benefits of exercise training in rehabilitation of patients with severe chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med.155199715411551
10. Gonzalez de Suso J. M., Bernús G., Alonso A., Alay A., Capdevila A., Gili J., Prat J. A., Arús C.Development and characterization of an ergometer to study the bioenergetics of the human quadriceps muscle by 31P-NMR spectroscopy inside a standard NMR scanner. Magn. Reson. Med.291993575581
11. Kemp G. J., Thompson C. H., Taylor D. J., Hands L. J., Rajagopalan B., Radda G. K.Quantitative analysis of defects of mitochondrial oxidation during recovery from exercise. Biochem. Soc. Trans.211993176S
12. Thompson C. H., Kemp G. J., Sanderson A. L., Radda G. K.Skeletal muscle mitochondrial function studied by kinetic analysis of postexercise phosphocreatine resynthesis. J. Appl. Physiol.78199521312139
13. Thompson C. H., Davies R. J. O., Kemp D., Taylor D. J., Radda G. K., Rajagopalan B.Skeletal muscle metabolism during exercise and recovery in patients with respiratory failure. Thorax481993486490
14. Van den Boogaart, A. 1995. The Use of Signal Processing Algorithms to Obtain Biochemically Relevant Parameters from Magnetic Resonance Data Sets (doctoral thesis). University of London, London.
15. Golub G. H., Pereyra V.The differentiation of pseudo- inverses and nonlinear least squares problems whose variables separate. SIAM J. Number Anal.101973413432
16. Van der Veen J. W. C., de Beer R., Luyten R., van Ormondt D.Accurate quantification of in vivo 31P-NMR signals using the variable projection method and prior knowledge. Magn. Reson. Med.619889298
17. Taylor D. J., Bore P. J., Stylen P., Gadian D. G., Radda G. K.Bioenergetics of intact human muscle: a 31P-nuclear magnetic resonance study. Mol. Biol. Med.119837794
18. Roca J., Agustı́ A. G., Alonso A., Barberà J. A., Rodriguez-Roisin R., Wagner P. D.Effects of training on muscle O2 transport at Vo2max. J. Appl. Physiol.73199210671076
19. Agustı́ A. G., Roca J., Barberà J. A., Casademont J., Rodriguez-Roisin R., Wagner P. D.Effect of sampling site on femoral venous blood gas values. J. Appl. Physiol.77199420182022
20. Marrades R. M., Roca J., Campistol J. M., Diaz O., Barberà J. A., Torregrosa J. V., Masclans J. R., Cobos A., Rodriguez-Roisin R., Wagner P. D.Effects of erythropoietin on muscle O2 transport during exercise in patients with chronic renal failure. J. Clin. Invest.97199620922100
21. Marrades R. M., Alonso J., Roca J., González de Suso J. M., Campistol J. M., Barberà J. A., Diaz O., Torregrosa J. V., Masclans J. R., Rodriguez-Roisin R., Wagner P. D.Cellular bioenergetics after erythropoietin therapy in chronic renal failure. J. Clin. Invest.97199621012110
22. Roca J., Sanchis J., Agustı́-Vidal A., Cobo E., Casan P., Segarra J., Rodriguez-Roisin R.Spirometric reference values for a Mediterranean population. Bull. Eur. Physiopathol. Respir.221986217224
23. Roca J., Burgos F., Barberà J. A., Sunyer J., Rodriguez-Roisin R., Castellsague J., Sanchis J., Antó J. M., Casan P., Clausen J. L.Prediction equations for plethysmographic lung volumes. Respir. Med.921998454460
24. Roca J., Burgos F., Sunyer J., Saez M., Chinn S., Antó J. M., Rodriguez-Roisin R., Quanjer P. H., Nowak D., Burney P.Reference values for forced spirometry. Eur. Respir. J.11199813541362
25. Roca J., Rodriguez-Roisin R., Cobo E., Burgos F., Perez J., Clausen J. L.Single-breath carbon monoxide diffusing capacity (DCLO) prediction equations for a Mediterranean population. Am. Rev. Respir. Dis.141199010261032
26. Wenger H. A., Bell G.The interactions of intensity, frequency, and duration of exercise training in altering cardio-respiratory fitness. Sports Med.31986346356
27. Kilbon A., Hartley L., Saltin B., Bjure J., Astrand G.Physical training in sedentary middle-aged and older men. Scand. J. Clin. Lab. Invest.241969315322
28. Wilmore J., Royce J., Girandola R., Katch F., Katch V.Physiological alternatives resulting from a 10-week jogging program. Med. Sci. Sports91970714
29. Hickson R., Hagberg J., Ehsani A., Holloszy J.Time course of the adaptive responses of aerobic power and heart rate to training. Med. Sci. Sports Exerc.1319811720
30. Gallagher C. G.Exercise limitation and clinical exercise testing in chronic obstructive pulmonary disease. Clin. Chest Med.151994305326
31. Wagner, P. D. 1988. An integrated view of the determinants of maximum oxygen uptake. In N. Gonzalez and M. R. Fedde, editors. Oxygen Transfer from Atmosphere to Tissues. Plenum Press, New York. 245–256.
32. McCully K., Vanderborne K., Posner J. D., Leigh J. S.Muscle metabolism in track athletes, using 31P magnetic resonance spectroscopy. Can. J. Physiol. Pharmacol.70199213531359
33. Astrand, P. O., and K. Rodahl. 1986. Physical training. In P. O. Astrand and K. Rodahl, editors. Textbook of Work Physiology: Physiological Bases of Exercise, 3rd ed. McGraw-Hill, New York. 412–476.
34. ERS Task ForceClinical exercise testing with reference to lung diseases: indications, standardization and interpretation strategies. Eur. Respir. J.10199726622689
35. Wasserman, K., J. E. Hansen, D. Y. Sue, B. J. Whipp, and R. Casaburi. 1994. Protocols for exercise testing. In K. Wasserman, editor. Principles of Exercise Testing and Interpretation, 2nd ed. Lea & Febiger, Malvern, PA. 97–111.
Correspondence and requests for reprints should be addressed to Josep Roca, M.D., Servei de Pneumologia, Hospital Clı́nic, Villarroel 170, Barcelona 08036, Spain. E-mail:

Ernest Sala, M.D., was a Research Fellow supported by the Hospital Clı́nic (1997).


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