Physiologic benefits of exercise training in rehabilitation of patients with severe chronic obstructive pulmonary disease

Physiologic Benefits of Exercise Training in Rehabilitation of Patients with Severe Chronic Obstructive Pulmonary Disease RICHARD CASABURI, JANOS PORSZASZ, MARY R. BURNS, EVE R. CARITHERS, ROBERT S. Y. CHANG, and CHRISTOPHER B. COOPER Division of Respiratory and Critical Care Physiology and Medicine, Harbor-UCLA Medical Center; Pulmonary Rehabilitation Program, Little Companyof Mary Hospital, Torrance, California We determined the effect on exercise tolerance and physiological exercise responses of rigorous rehabilitative exercise training in chronic obstructive pulmonary disease (CaPO). Fifteen men and 10 women (mean age, 68 ± 6 yr; FEV1, 0.93 ± 0.27 L) participated in a rehabilitation program with an exercise component of three per week 45-min sessions of cycle ergometer training for 6 wk with exercise intensity kept near maximal targets. Before and after rehabilitation, patients performed an incremental test and a constant work rate (CWR)test at 80% of the peak work rate in the preprogram incremental test. Ventilation (VE) and gas exchange were measured breath by breath; arterialized venous blood was analyzed for blood gas determinations and lactate. Rehabilitation yielded an average increase in peak work rate in the incremental test of 36% (p < 0.001), and in the duration of the CWR test of 77% (p < 0.001). In the CWRtest, the kinetics of O2 uptake, CO2 output, VE, and heart rate were markedly slower than those of healthy subjects. After training, mean response time decrease averaged 17, 22, 34, and 290/0, respectively (p < 0.02), evidence of a physiologic training effect. Further, for identical CWRtasks, VE was 10% lower (p <0.02) after training, attributable to altered breathing pattern: tidal volume increased by 8% and respiratory rate decreased by 19%, yielding lower Vo /VT (0.46 versus 0.53 p < 0.005). Rigorous exercise training for patients with severe capo yields more efficient exercise breathing pattern and lower VE; this is associated with improved exercise tolerance. Casaburl R, Porszasz J, Bums MR, Carlthen ER, Chang RSY, Cooper CB. Physiologic benefits of exercise training In rehabilitation of patients with severe chronic obstructive pulmonary disease. AM J RESPiRCRIT CARE MED 19117;155:1541-1551. Impaired exercise tolerance is a prominent complaint of patients with chronic obstructive pulmonary disease (COPO). A central goal of pulmonary rehabilitation is to improve exercise tolerance, and a large number of studies have shown substantial increases after a program of rehabilitation (1). However, the question persists as to whether these increases in exercise tolerance are principally due to psychologic benefits of rehabilitation or to improved physiologic ability to perform exercise. Psychologic benefits of rehabilitation include (1) the patient becomes motivated to exert more effort in an exercise task, (2) the patient is taught to "work through" dyspneic sensations, and (3) through mechanisms not well understood, the sensation of dyspnea for a given exercise stimulus decreases (2, 3). These psychologic benefits are most easily observed in effort-dependent measures of exercise capacity, e.g., the 6-min (Received in original form February 12, 1996 and in revised form December 16, 1996) Supported by an Established Investigator Award from the American Lung Association of California. Correspondence and requests for reprints should be addressed to Richard Casaburi, Ph.D., M.D., Diy_on of Respiratory and Critical Care Physiology and Medicine, Harbor-UCLA Medical Center, 1000 W. Carson Street, Torrance, CA90509. Am J Resplr Crlt Care Med Vol. 155. pp. 1541-1551, 1997 or 12-min walk (4). Physiologic benefits that occur through programs of exercise training may also improve exercise tolerance in patients with COPO. In healthy subjects, exercise training produces changes in the exercising muscles that improve O 2 delivery to the site of metabolism, potentially forestalling the onset of lactic acidosis (5). Changes in cardiac structure and function and in body composition are also observed. Whether patients with COPO gain physiologic benefits from exercise programs that are routinely part of pulmonary rehabilitation is not clear (1). We recently demonstrated that patients with COPO who undergo a rigorous program of exercise training show manifestations of physiologic training effect and have substantially improved exercise tolerance in effort-independent tests (6). However, it has been argued that these findings may not be fully applicable to patients who typically undergo pulmonary rehabilitation in the United States (7). First, these patients had predominantly moderate disease severity (FEY!> 56 ± 13% pred) and were a middle-aged group (51 ± 10 yr). Programs of pulmonary rehabilitation in the United States generally cater to an older, more severely ill clientele. Second, an entry requirement for our previous study was that patients be able to elevate blood lactate levels during exercise. We hypothesized a link between training-induced reductions in lactic acidosis, lowered ventilation, and improved exercise tolerance in these patients with ventilatory limitation. Patients with 1542 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL. 155 1997 more severe obstruction may be less likely to achieve a substantial increase in blood lactate levels during exercise if dynamic hyperinflation leads to severe dyspnea at low work rates. Should patients be triaged into programs of rigorous exercise training depending on whether or not they can elevate blood lactate levels? Finally, our previous study involved patients undergoing an 8-wk inpatient rehabilitation program featuring exercise sessions 5 d/wk. Few patients in the United States undergo inpatient rehabilitation, and supervised exercise sessions 5 d/wk are often not practical. We wished to determine whether inclusion of a rigorous program of cycle ergometer exercise training in an outpatient comprehensive pulmonary rehabilitation program elicits a physiologic training effect yielding improved exercise tolerance in patients with COPD. To this end we studied exercise tolerance (and other physiologic variables) before and after rehabilitation in 25 patients with predominantly severe COPD. The results define the alterations in physiologic responses to exercise that occur as a result of rigorous exercise training in severe COPD. As a second objective, we sought to determine whether the response kinetics after the onset of exercise could be used as a noninvasive marker of improved muscle function resulting from exercise training in COPD. It has previously been shown that patients with COPD manifest a substantially slower approach to steady-state responses as compared with healthy subjects (8), presumably a manifestation of severe deconditioning. Breath-by-breath responses after the onset of constant work rate exercise were analyzed to determine the effect of training on ventilatory, gas exchange, and heart rate kinetics. METHODS Study Setting and Experimental Subjects Patients participating in the pulmonary rehabilitation program at the Little Company of Mary Hospital were recruited for this study. This is a well established program (18 yr of operation) that nominally consists of 12 half-day sessions over a 6-wk period. In addition to an exercise program, rehabilitation components included disease education, nutritional counseling, chest physiotherapy, breathing retraining, psychosocial intervention, relaxation therapy, and instruction in the use of medications. Program participants were ambulatory patients referred by their personal physicians. They were invited to participate in this investigation if they had no orthopedic or cardiac contraindications to an exercise program. Patients with pulmonary disease other than COPD were excluded. All participants were nonsmokers at the time of study, although all had a cigarette smoking history. All participants gave institutionally approved informed consent for this study. This study was conducted over an 18-mo period. Exercise Regimen Patients participated in a program of exercise on calibrated cycle ergometers. Exercise sessions were held 3 d/wk for a total of 8 wk. This required adding several "exercise-only" days during the regular 6-wk program and adding 2 wk of "exercise-only" sessions after the formal rehabilitation program had ended. The first 2 wk of exercise sessions were "warm-up" sessions, with no fixed intensity target. Exercise duration was nominally 45 min, though patients were not strongly encouraged to reach this target. The next 6 wk featured specific intensity targets. Exercise sessions were 45 min long, although initially many patients were unable to exercise continuously for 45 min and were allowed to break the session into two or three parts. Initial exercise intensity targets were set at 80% of the peak work rate observed in the pretraining incremental exercise test (see below). Sessions were closely monitored by a rehabilitation therapist who was encouraged to increase the training work rate progressively during the 6-wk period. During exercise sessions, heart rate and arterial 02 saturation, estimated by pulse oximetry, were recorded at approximately 10-min in- tervals. For those patients requiring supplemental oxygen, oxygen was delivered during exercise by nasal cannula at sufficient rate to achieve 02 saturation of 92% or above. Patients were also encouraged to walk an hour a day at home at self-selected rates. Walking diaries were completed by the patients; by the second or third week of the program all patients reported compliance with the walking program. Outcome Measures Patients visited the exercise physiology laboratories at Harbor-UCLA Medical Center for half-day sessions before and within 3 d after the end of their exercise program. Testing sessions were conducted in the morning or the afternoon, but they were conducted at the same time of day for a given patient. Patients first performed resting pulmonary function testing 10 min or more after beta-agonist bronchodilator administration. Spirometry was performed using a 13-L water-seal spirometer (Warren E. Collins, Braintree, MA). From the best of three forced expiratory maneuvers, FEVI and FVC were determined. The diffusing capacity for carbon monoxide (Di to) was determined using a bag-in-box apparatus; the results of three maneuvers were averaged. Results were compared with the normal values derived by Knudsen and colleagues (9) and Cotes (10). Exercise testing was performed on an electromagnetically braked cycle ergometer (Corival; Quinton, Seattle, WA). Patients respired through mouthpieces with noseclips in place. Airflow was directed through a turbine volume transducer (Alpha Technologies, Laguna Hills, CA) with a dead space of 50 ml. Gas was sampled at a rate of 60 ml/min from a point near the mouthpiece and analyzed by a mass spectrometer (Model 1100; Perkin Elmer, Ppmona, CA) for concentrations of 0 2 , CO 2 , and N Z . Heart rite was measured from the R-to-R interval of the electrocardiogram recorded from a three-lead configuration. These signals were relayed to a digital computer that calculated the breath-by-breath time course of a number of ventilatorygas-exchange-, and heart-rate-related variables (11). Ventilation (VE) is expressed as Brns; oxygen uptake (Voz ) and carbon dioxide output (Vco2 ) are expressed as srrn. In this exercise testing system, daily calibration was performed with a 3-L syringe and two precision gas mixtures. The gas analyzer gas transport delay time was measured with a solenoid apparatus. A reciprocating pump device able to simulate known levels of VE, Vo2 , and Vco 2 (12) was also employed. Before exercise testing a 21-gauge butterfly catheter was placed in a vein in the dorsum of the hand or the forearm. Heparinized saline flushed the catheter intermittently. A heat lamp was used to warm the hand and thereby arterialize the venous blood samples. This procedure yields Pco 2 pH, and lactate levels that approximate arterial blood levels (13). In a few patients we were unable to insert or maintain an indwelling venous catheter. In these patients, a single venous sample was drawn from an antecubital vein 2 min after the end of exercise. The same blood sampling site was used in before and after rehabilitation testing in all patients. Patients first performed an incremental exercise test to tolerance. After at least 3 min of sitting at rest on the cycle ergometer, patients performed unloaded pedaling for 3 min with a motor engaged to drive the flywheel so that the patient performed no work to drive the flywheel. Three incremental protocols were employed, depending on an estimate of the patient's exercise tolerance, with the goal that the incremental phase last 8 to 12 min. Work rate was increased continuously at either 5 or 10 W/min or, for the most debilitated patients, a third protocol was used. In these patients, during the 3-min unloaded pedaling period, pedaling rate was set at 20 rpm (rather than at 60 rpm). Pedaling rate was then advanced to 40 rpm and then 60 rpm in subsequent minutes. As the patient continued to pedal at 60 rpm, the flywheel motor was then turned off and, after an additional minute, a 5-W/min increase in work rate was initiated. At rest, at the end of the 3-min unloaded pedaling period, every 2 min thereafter, and 2 min after the end of exercise arterialized venous blood was sampled, blood pressure was measured by sphygmomanometry, and arterial 0 2 saturation was estimated by a pulse oximeter (Model 3700; Ohmeda, Louisville, CO). After a 45-min rest period, patients performed a constant work rate (CWR) test; the work rate was chosen as 80% of peak work rate , Casaburi, Porszasz, Burns, et al.: Exercise Training in Severe COPD achieved in the incremental test. After 3 min of sitting at rest on the cycle ergometer, the work rate was abruptly switched to the chosen level. At rest, every 2 min during exercise, and after 2 min of rest arterialized venous blood samples were obtained. During these tests, patients were verbally encouraged toward a maximal effort. After training, identical exercise tests were conducted; patients were again encouraged toward a maximal effort; however, the maximal duration of the CWR test was set at 15 min. For those patients in whom we were unable to insert indwelling venous catheters, the postrehabilitation CWR test was stopped at the same exercise duration as tolerated in the prerehabilitation test. This assured that the blood sample drawn 2 min after exercise would be in response to an identical exercise stimulus. However, improvements in the time the CWR task could be tolerated was not assessed in these patients. Blood samples were collected anaerobically and immediately iced. Each sample was assayed for lactate level (Model 2300; Yellow Springs Instruments, Yellow Springs, OH), as well as blood gas and pH analysis (Model 1306; Instrumentation Laboratories, Lexington, MA). 1543 1.0 0.8 0.6 á 0 0.4 0.2 0.0 -1 Time (ruin) Analysis of Response Kinetics Those patients whose pretraining CWR tests were at least 4 min in duration and whose assigned work rate was at least 5 W were selected for this analysis. We felt that tests of at least 4-min duration were necessary for accurate determination of response kinetics and that a work rate of at least 5 W was required for an adequate response amplitude. Thirteen of the 25 patients completed a rest-to-exercise transition, meeting these criteria both before and after training. Resting pulmonary function and exercise tolerance (and change in exercise tolerance associated with training) in these patients did not differ significantly from the group as a whole. For determining the response kinetic parameters, the breath-bybreath data from the CWR tests were interpolated at 1-s intervals. The interpolated values of VE, Vco 2 , Vol , and HR were fitted to a three-parameter first-order exponential model y(t) = BL+A * (1-exp [- (t-Q)/Tl ) where BL is the average resting value in the 60 s preceding the exercise onset and nonlinear regression analysis is used to determine the time delay (Q), time constant (T), and the response amplitude (A) (14). The time delay was constrained to assume values greater than or equal to zero. The mean response time (MRT) was then derived as the sum of v and T. This model results in a continuous segmental function in which the response rises above the baseline (BL) after a delay v (see Figure 1). For comparison, response curves were also fit with a single exponential model not including a time delay (a two-parameter model). Superiority of the three-parameter model was tested with an F-test (15). In addition to this kinetic analysis, we determined the magnitude of the initial rapid response component (which was called Phase I) by comparing the average response during the 20 s after exercise onset to the 20 s before exercise onset. The kinetic analyses were done on identical periods of CWR exercise in the pretraining and posttraining studies, even though most subjects increased their tolerated exercise duration after training. Statistical Analysis Comparison between responses before and after the exercise program were made by paired t-tests. Linear regression analysis and Pearson's correlation coefficients were used to test the strength of association between measured variables. Significance was accepted at the p < 0.05 level. Dispersion about mean values is expressed as ± 1 standard deviation, unless otherwise specified. RESULTS Patient Recruitment A total of 43 patients presented to the pulmonary rehabilitation program during the study period; of these, 25 completed the program and were included in this report. A total of 14 patients were excluded from the study because of symptomatic heart disease (four), restrictive lung disease (five), declining to Typical response of oxygen uptake to the transition from rest to constant work rate exercise in a patient with COPD. Superimposed is the best fitting three-parameter model (see text). Exercise starts at time = 0 min. Vertical dotted line is time delay (a). BL = preexercise baseline; T = time constant. Upper horizontal line is asymptotic amplitude (A). Figure 1. participate in a rigorous exercise program (three), scheduling problems (one), postlung transplant rehabilitation (one). Four patients were recruited but failed to complete the study for the following reasons: pulmonary infection (one), nonpulmonary hospitalization (two), noncompliance with exercise program (one). This analysis indicates that 29 of 32 patients (91%) who met inclusion criteria agreed to participate in this study. Further, 25 of 29 patients (86%) who agreed to participate completed the rigorous 8-wk training program. Patient Characteristics The physical characteristics, lung function, and exercise capacities of the patients on entry into the rehabilitative program are presented in Table 1. This was a group of 15 men and 10 women. On average, participants were elderly, with body weights slightly in excess of ideal body weight. FEV 1 was severely reduced on average and FEV 1 /FVC averaged 41%. DL co was moderately reduced, averaging 57% predicted. Resting arterial blood gas analysis revealed moderate hypoxemia on average, with four patients having Pa 02 values less than 55 mm Hg. Acid—base status was consistent with mild, well compensated respiratory acidosis, with two patients having Pa co2 values greater than 50 mm Hg. As determined by the responses to the prerehabilitation incremental exercise test, exercise tolerance was substantially impaired. Peak Vol averaged 54 ± 14% predicted (using prediction equations in Reference 16). Peak VE averaged 33 L. If the maximal voluntary VE is estimated as 40 times FEV 1 (16), breathing reserve averaged 4 L/min, consistent with ventilatory limitation to exercise. Blood lactate levels measured 2 min after exercise were only mildly elevated, suggesting that lactic acidosis was not a substantial stimulus to exercise VE in most patients. Training Program The 25 patients participated in a 6-wk training program, with three sessions of 45 min duration scheduled per week. Three patients missed one and one patient missed two of the 18 sessions. The initial target work rate (80% of the peak work rate AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL. 155 1997 1544 TABLE 1 50 EFFECT OF REHABILITATIVE EXERCISE TRAINING ON PATIENT CHARACTERISTICS, RESTING PULMONARY FUNCTION, AND PEAK EXERCISE RESPONSES* Age,yr Weight, kg Height, cm FEV 1 , L FEV 1 ,%pred FVC, L DLco ml/min/mm Hg Pa 02, mm Hg Pa co2, mm Hg" pHa Peak work rate, W Peak l L/min Peak "E, L/min Breathing reserve, L/min Peak lactate, mMol/L Peak heart rate, beats/minll Vo , Before After 68±6 69 ± 13 169 ± 10 0.93 ± 0.27 36±13 2.29 ± 0.70 13.6 ± 7.6 63 ± 11 44 ± 5 — 69 ± 13 — 1.02 ± 0.33r 7.41 ± 0.03 40 ± 31 0.86 ± 0.29 33 ± 10 4± 7 2.2 ± 0.8 131 ± 18 40 WORK RATE lwatts) 30 20 39±136 2.43t0.84 13.4 t 6.7 — — — 54 ± 32§ 1.00 ± 0.30§ 36 ± 12f 4± 6 3.0 ± 1.2t 1 39 ± 20 10 0 HEART RATE 10 12 14 16 18 120 (Wmini 110 100 90 0 2 4 6 8 10 12 14 16 18 10 12 14 16 18 100 1%1 Changes Induced by Rehabilitation 8 130 S202 Physical characteristics, resting pulmonary function, and peak exercise responses in the incremental exercise test before and after the rehabilitation program are presented in Table 1. Mean body weight did not change significantly. We were surprised to find that on average FEV 1 increased by 9% (p < 0.05). There was a somewhat smaller (approximately 5%) average increase in FVC (NS). We speculate that these modest changes were due to improvements in bronchodilator effectiveness, one of the goals of the rehabilitation program (see DISCUSSION). Resting arterial blood samples were not obtained after rehabilitation. However, arterialized venous samples obtained at rest showed that standard bicarbonate averaged 26.1 ± 1.4 mEq/L before and 26.3 ± 1.4 mEq/L after rehabilitation, suggesting that the training program did not induce appreciable change in chronic CO 2 retention. There was an improvement in exercise tolerance in the incremental test. Peak work rate increased by 35% and peak Vo l increased by 16%. This was accompanied by significantly higher peak VE, blood lactate, and heart rate levels. 6 140 • Values are mean ± 1 SD. Peak exercise responses are those seen during the incremental exercise tests. 1 Significantly different from prerehabilitation response at p < 0.01. Significantly different from prerehabilitation response at p < 0.02. Significantly different from prerehabilitation response at p < 0.001. The distribution of values for Pa0 2 and peak heart rate before training display significant coefficients of skewness and/or kurtosis. Median (range) values for these variables are 42 (38-59) mm Hg and 124 (104-172) beats/min, respectively. achieved in the incremental exercise test) averaged 32 ± 25 W. As can be seen in Figure 2, most patients were initially unable to sustain this work rate for 45 min. However, during subsequent sessions, with the guidance of the rehabilitation therapists, the average reached the target work rate by the eighth session. By the end of the training program, the average work rate sustained for 45 min approximated the peak work rate in the pretraining incremental test. Although the training work rate doubled over the course of the training program, heart rate did not show a similar trend (Figure 2), the average remaining constant at approximately 115 beats/min throughout the program. This is evidence of a physiologic training effect. Further, the increase in training work rate did not engender a decrease in arterial 0 2 saturation estimated by pulse oximetry (Figure 2, bottom panel). 4 2 95 90 EXERCISE SESSION NUMBER Figure 2. Progression of the cycle ergometer training work rate over the 6-wk training program (18 sessions) in 25 patients with COPD. Also shown are the midsession heart rate and arterial oxygen saturation (Sa O2 ) estimated by pulse oximetry. Points are mean ± SEM. Despite progressive increases in work rate, heart rate did not rise appreciably, characteristic of a physiologic training effect. Responses to identical levels of exercise before and after the rehabilitation program, both in the incremental test and in the CWR test, are shown in Tables 2 and 3. In the incremental test, we selected for comparison the highest work rate tolerated by a given patient in both the pretraining and the posttraining test. In the CWR test, we selected for comparison the longest duration tolerated by a given patient in both the pretraining and the posttraining test. Consistent with the abnormalities typical of severe COPD, ventilatory equivalents for CO 2 and 0 2 (VE/Vco 2 ) were high, arterial b2 saturation (estimated by pulse oximetry) was low, and arterialized venous Pco 2 suggested that CO 2 retention occurred during exercise. In the incremental exercise test (Table 2), comparison of responses to identical exercise levels showed that the VE response was lower after rehabilitative training. This occurred without a significant fall in either VCo 2 or Vo2i therefore reductions in VE/Vco2 and VE/Vo 2 were highly significant. The lower VE response could not be accounted for by further CO 2 retention; arterialized venous CO 2 tended to be lower, not higher, after the training program (implying that alveolar ven- 1545 Casaburi, Porszasz, Burns, et al.: Exercise Training in Severe COPD TABLE 2 EFFECT OF REHABILITATIVE EXERCISE TRAINING ON RESPONSES TO IDENTICAL LEVELS OF EXERCISE IN THE INCREMENTAL EXERCISE TEST* Work rate, W VE, L/min Vco 2 , L/min Vo l , L/min Ve/Vco 2 VE/V0 2 HR, beats/min' BP, mm Hg SP o2 ,% Pco 2 , mm Hg pH Lactate, mMol/L Before After 40 ± 31 33 ± 10 0.86 ± 0.34 0.86 ± 0.29 40 ± 31 30 ± 9t 0.86 ± 0.27 0.85 ± 0.24 38 ± 7 35 ± 6* 38 ± 7 131 ± 18 189/89 ± 31/17 88±5 50 ± 6 7.34 ± 0.03 34 ± 7§ 127 ± 16t 178/86 ± 28#/12 88±6 49 ± 6 7.35 ± 0.04 1.8 ± 0.8 1.8 ± 0.6 Definition of abbreviations: Ve = ventilation; Vco 2 = CO 2 output; Vol = 0 2 uptake; HR = heart rate; BP = arterial blood pressure; Sp ot = arterial 0 2 saturation estimated by pulse oximetry. * Values are mean ± 1 SD. In 22 of 25 patients, arterialized venous blood was obtained at identical work rates and analyzed for Pco 2 , pH, and lactate. Significantly different from prerehabilitation response at p < 0.02. Significantly different from prerehabilitation response at p < 0.005. Significantly different from prerehabilitation response at p < 0.001. Significantly different from prerehabilitation response at p < 0.05. The distribution of values for HR and lactate after training display significant coefficients of skewness. Median (range) values for these variables are 125 (106-166) beats/ min and 1.7 (1.0-3.4) mMol/L. tilation was not lower). Nor could reduced stimulation of lactic acidosis to breathing be invoked; average blood lactate levels did not fall. Heart rate and systolic blood pressure were also significantly lower after training. The responses to identical levels of CWR exercise (Table 3) showed a similar pattern of change. VE was significantly lower after training. This was again not accompanied by a reduced metabolic rate; Therefore, VE/Vco 2 and VE/Vo 2 were highly significantly lower after training. As in the incremental test, arterialized venous Pco 2 was not higher and lactate level was not lower. Heart rate was significantly lower after training. The altered physiologic responses to exercise were accom- TABLE 3 EFFECT OF REHABILITATIVE EXERCISE TRAINING ON RESPONSES TO IDENTICAL LEVELS OF EXERCISE IN THE CONSTANT WORK RATE TEST* Workrate,W Duration, min§ Ve, L/min Vco 2 , L/min Vo z , L/min VE/Vco 2 VE/V0 2 HR, beats/min Sp 02 ,3b§ Pco 2 , mm Hg pH Lactate, mMol/L Before After 32±25 5.8 ± 2.9 32 ± 10 0.85 ± 0.28 0.86 ± 0.24 32±25 10.2 ± 4.3 29 ± 8t 0.82 ± 0.22 0.87 ± 0.22 40 ± 8 35 ± 6^ 39 ± 6 132±20 87±6 49 ± 5 7.34 ± 0.03 35 ± 6* 124 ± 16# 88±8 47 ± 5 7.36 ± 0.03 2.0 ± 1.1 2.0 ± 0.8 For definitions of abbreviations, see Table 2. * Values are mean ± 1 SD. t Significantly different from prerehabilitation response at p < 0.02. = Significantly different from prerehabilitation response at p < 0.001. § The distribution of values for duration before training and Sp ot after training display significant coefficients of skewness and/or kurtosis. Median (range) values for these variables are 4.8 (2.1-15) min and 89 (66-98)%. panied by a substantial increase in exercise tolerance. The CWR test was tolerated for an average of 10.2 ± 4.3 min after training (versus 5.8 ± 2.9 min before training), an increase averaging 77%. Further, this increase is likely an underestimate, as after training eight patients exercised for the full 15 min allowed. No patient could exercise this long before training; presumably, these eight could have continued beyond 15 min. The mechanism of the reduced VE requirement for exercise is defined in Figure 3. This figure shows the responses to incremental and CWR exercise of a man with severe COPD (FEV 1 , 17% pred) before and after exercise training. Very low exercise tolerance was apparent before training; peak work rate in the incremental test was 9 W, and 5 W of CWR exercise was tolerated for only 4.5 min. A ventilatory limitation is reached at approximately 30 L/min in both tests. After training, in both tests the VE response was attenuated and exercise tolerance was prolonged. This reduction in VE cannot be explained by a reduction in lactic acidosis; blood lactate did not rise appreciably from resting levels in any test. Apportioning the ventilatory response into the tidal volume and frequency components is instructive. In the pretraining studies, there was a striking tachypnea in the last few minutes of exercise. This tachypnea was appreciably delayed in the post-training study; breathing rate was lower at equivalent exercise durations. The tachypneic pattern was inefficient in terms of gas exchange; a larger fraction of anatomic dead space was rebreathed in each breath. As a result, the dead space to tidal volume ratio (VD/ VT) was lower in the post-training responses, yielding a reduced ventilatory requirement for exercise. The breathing pattern responses are summarized in Table 4. In both the incremental and CWR tests for identical exercise tasks the reduced ventilatory response was composed of an increased tidal volume response and a substantially lower breathing rate. Average VD/VT was therefore lower. We sought to determine patient characteristics that predict the magnitude of the breathing pattern adaptations that accompanied this rigorous training program. The severity of obstructive lung disease, quantitated by percent predicted. FEV I did not correlate significantly with changes in VE, f, or VT at identical levels of exercise in either the ramp or the CWR test. The small changes in FEV I that accompanied rehabilitation were not predictive of the fall in VE or f seen during exercise, but they correlated weakly with the rise in VT in both the ramp and the CWR test (r = 0.55, p < 0.01 and r = 0.54, p < 0.02, respectively). Neither the change in blood lactate nor the change in arterialized venous Pco 2 correlated significantly with the fall in VE during either the ramp or the CWR test. However, changes in Vco 2 were well correlated with changes in VE in both tests (r = 0.66, p < 0.001 and r = 0.83, p < 0.001, respectively), even though Vco 2 was, on average, not significantly lower after training (Tables 2 and 3). Exercise Response Kinetics The average parameters obtained by fitting the single exponential model to the time courses of Vo l , Vco 2 , VE, and HR in the pretraining and post-training CWR studies are presented in Table 5. Before training, end-exercise blood lactate increased only modestly over the resting values (2.4 Y 1.0 mMIL). After the training program the lactate increase associated with the same exercise task was similar (2.3 ± 0.8 mMIL). The time courses of gas exchange, VE, and HR both before and after training during the CWR exercise for a representative subject are shown in Figure 4. The best-fit curves for the three-parameter model are superimposed for each variable. We found that the three-parameter (exponential with delay) model provided a significantly better fit than the two-parame- AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL. 155 1997 1546 30 30 I I 26 „p 25 Ódsf ^^ 20 20 E E 16 10 . > i 6 15 1 1 I 16 I 5 0 _ 1 10 5 0 I 21 10 16 20 5 10 16 20 2 1.5 I I I I I I I I Y^ )_.. I 0 I .; 0 16 10 6 0 E 5 2 0.6 J 0 15 0 60 50 ♦0 4 70 I I 20 I I 30 f E 10 0 0 10 5 I I 20 10 0 I 1 0 16 1.6 1.3 1 i 0.9 I 10 16 20 10 15 20 1 1 I 1.6 5 1.3 7.1 0.7 0.7 a.s 0.5 10 5 0 15 0 5 Time (min) Time (min) Figure 3. Physiologic responses to identical exercise tasks before (closed circles) and after (open circles) a pulmonary rehabilitation program featuring a rigorous exercise component in a 65-yr-old man with severe COPD (FEV, = 17% predicted). (Left panels) Response to an incremental exercise protocol consisting of a 3-min rest followed by unloaded pedaling (3 min at 20 rpm, 1 min at 40 rpm, 1 min at 60 rpm) followed by a continuous 5-W/min increase in work rate. (Right panels) Response to a 3-min rest followed by a constant work rate of 5 W. VE = ventilation; Lactate = arterialized venous blood lactate; f = respiratory frequency; VT = tidal volume. (See text.) ter model (exponential without delay) for most subjects for VE, Vco z and Vo l, but not for HR. Overall, time delays for VE, Vco 2 , and Vo l were less than 10 s in most subjects, and they did not change significantly with training. To facilitate comparisons, we chose to express the relative speed of response of each variable as a mean response time. In Figure 4, the MRT values are shown in each panel. It is visually apparent that these curves describe the data well. After training, the TABLE 4 EFFECTS OF TRAINING ON BREATHING PATTERN AND GAS EXCHANGE EFFICIENCY FOR IDENTICAL LEVELS OF EXERCISE IN PATIENTS UNDERGOING TRAINING* Constant Work Rate Test Incremental Test Before VI:, L/min 33 ± 10 0.99±0.30 VT, L f, breaths/min 33 ± 7 VD /VT90.54 ± 0.07 After % Change Before After % Change 30 ± 9 1.10 ± 0.31 27 ± 6 0.47 ± 0.07 —10%t +11%* —15%II —13% 32 ± 10 1.02 ± 0.28 32 ± 7 0.53 ± 0.07 29 ± 8 1.13 ± 0.30 26 ± 5 0.46 ± 0.07 _10%t +8%5 _19%II Definition of abbreviations: VE = ventilation; Vr = tidal volume; f = respiratory rate; Vo = dead space volume. • Values are mean ± 1 SD. t Significantly different from pretraining response at p < 0.05. Significantly different from pretraining response at p < 0.005. Significantly different from pretraining response at p < 0.01. Significantly different from pretraining response at p < 0.001. 1 Calculated as 1— (k Vco 2 /Pa c02 VE) — (VD,/VT), where k is a constant, Pa CO2 is arterialized venous CO 2 partial pressure, Vco 2 is CO2 output, and V 0 , P is the breathing apparatus dead space Casaburi, Porszasz, Burns, et 01.: Exercise Training in Severe COPD 1547 TABLE 5 TABLE 6 RESPONSE KINETICS TO ONSET OF CONSTANT WORK RATE EXERCISE BEFORE AND AFTER TRAINING PHASE I AMPLITUDE OF RESPONSE AFTER ONSET OF CONSTANT WORK RATE EXERCISE* Before Training Vo 2A, l/min VC02A, L/min VEA, L/min' HR A, beats/min 0.76 0.86 26.3 50.1 87 MRT VC02' 5' MRTVE,5 MRT HR, 5 0.79 ± 0.33 0.81 ± 0.33 23.6 ± 10.2 39.7 ± 16.2t ± 0.29 ± 0.34 ± 10.1 ± ± 136 ± 142 ± 144 ± MRT VOlt 5 After Training 20.7 21 52 66 48 72 ± 15* 106 ± 20§ 94 ± 25 t 101 ± 34 11 Definition of abbreviations: A = amplitude of response; MRT = mean response time; n = 13. For other definitions, seeTable 2. • All values are mean ± 1 SO. Work rate = 45 ± 22 W. By paired t analysis: t Significantly different from pre-training response at p < 0.005. t Significantly different from pre-training response at p < 0.001. § Significantly different from pre-training response at p < 0.02. II Significantly different from pre-training response at p < 0.0001. , Distribution of values for VEamplitude after training and MRT VC02 before training display significant coefficients of skewness and kurtosis. Median (range) values for these variables are 22 (12-49) l/min and 121 (Bl-274) s, respectively. Before Training AVE, L/min* 2.43 0.06 0.08 6.6 AVCOlt l/min AVo2 , l/min AHR, beats/min After Training 4.36 ± 2.69t 0.10 ± 0.07t 0.14 ± 0.08 t 5.9 ± 2.9 ±1.18 ± 0.03 ± 0.04 ± 4.8 For definition of abbreviations, seeTable 2. • Values are mean ± 1 SO. Phase I is defined as the increase in response averaged over the first 20 s of exercise compared with the 20 s preceding exercise; n = 13. t Significantly different from pretraining response at p < 0.05. t Distribution of values for AVE after training display significant coefficients of skewness and kurtosis. Median (range) is 2.8 (1.4-11 .6) l/min. VCOz kinetics were highly correlated both before (r = 0.91, P < 0.001) and after (r = 0.76, P < 0.01) training as previously reported (17). The pretraining HR time constant was prolonged even more than the other variables, a more than three160 r--~--.--~---.-~--r--~--.--~--, .C) responses reached steady state sooner than in the pre training study (i.e., the response kinetics were faster). For each of the four variables, the MRT was shorter after training. As shown in Table 5, the response amplitudes did not change significantly as a result of training, with the exception of a fall in HR amplitude (50 ± 21 versus 40 ± 16 bpm, p < 0.005). Before training, the MRT of these variables were profoundly longer than those previously reported for healthy young subjects. VOz had faster kinetics than did VCOZ. VE and <II en --- 1.2 .C) C\I 124 0.4 0.8 0.8 0.2 45 0.6 0.4 0.0 120 Eo-< 0:: 60 160 73 --- .;> 80 30 0.0 120 ~ 60 15 90 -so 0 so 100 150 200 Time (sec) 250 300 100 0.4 10 100 120 VE 350 r -50 0 110 55 90 M 100 ISO 200 ::a 2SO 300 3SO Time (sec) Figure 4. Effect of training on ventilatory (VE, L/min), gas exchange (Voz and Vcoz, L/min), and heart rate (HR, beats/min) responses to the transition from rest to 50-W constant work rate exercise in a patient with severe COPD. Superimposed curves represent best fits to the three-parameter model (see text). Insert numbers are the mean response time values for each variable. Note that training-ofesults in a considerably faster response for all variables. Note also that the early responses (first 20 s, Phase I) are greater after training for VOz, Vcoz, and VE, but not HR. ±SEM 40 20 100 50 ~ 1---+--+,--+---+----<>---+----+--t-----1 140 ~ en 0.2 86 25 110 .- -t- 40 35 20 M 0.6 40 106 0.2 1.0 r I 100 80 1.2 106 0.6 I---+--+--+---+----<>---+----+--t---I 120 u 0.8 1.2 -t...- - 140 .;> ::a 1.0 :COPD 40 o 77 100 160 1.4 97 • :Healthy 60 en After training • 120 80 <II Before training 140 L-----'~-.l_--'-_ o 2 _L_~_-L_~_..I__~_.l 4 6 8 10 End Exercise Lactate (mM/L) Figure 5. Relation of mean response time (MRT) of oxygen uptake (Voz), COz output (Vcoz), and ventilation (VE) to end-exercise lactate level in healthy subjects, n = 5 (closed circles) and patients with COPD, n = 14 (closed triangles). Bars at each data point are ± 1 SE. Data from healthy subjects are taken from Reference 18 in which subjects performed multiple transitions to seven progressively higher work rates. Arrowheads point from before to after training responses in patients with COPD. Note that MRT is longer (kinetics are slower) in patients with COPD than in healthy subjects for a given level of end-exercise lactate. After training, patients with COPD manifest considerably shorter MRT, but little change in end-exercise lactate. 1548 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL. 155 1997 fold increase from previously reported values for healthy subjects (138 versus 36 to 40 s). After training, the Vol , Vco 2 , Vu, and HR kinetics were clearly speeded, by an average of 17, 22, 34, and 29%, respectively. In a previous study of patients with COPD (8), the amplitudes of Phase I VE, Vo 2 , Vco2 , and HR were found to be smaller than in healthy subjects. We were interested to see whether training of patients with COPD was associated with increased Phase I amplitude. It can be seen in Figure 4 that for Vo l , Vco2 , and VE, after training there was a discernable increase in the response immediately after exercise onset. In contrast, HR had no apparent Phase I component. The average Phase I amplitudes for the 13 patients with COPD are shown in Table 6. VE, Vco2 , and Vo l Phase I amplitudes increased after training by an average of 90, 67, and 75%, respectively. In contrast, the HR amplitude showed a (nonsignificant) trend to decrease as a result of training. The measured MRT were dramatically longer than those previously reported for healthy young subjects performing moderate (i.e., below lactic acidosis threshold) exercise. However, we previously demonstrated that response kinetics are slower as the level of blood lactate that the exercise engenders increases (18). We wished to determine whether the modest elevation in end-exercise blood lactate in these subjects with COPD could explain the slow kinetic responses. End-exercise blood lactate is plotted on the abscissa and MRT on the ordinate for Vo l , Vco 2 , and VE in Figure 5. Average data for the present study, both before and after training, are shown. Superimposed on these plots are data from the study of Casaburi and colleagues (18) in which kinetic responses to seven work rates, ranging in intensity from moderate to very heavy, were assessed in five healthy subjects. It can be seen that the slow response kinetics of the patients with COPD cannot be accounted for by the lactate elevations found in these subjects. Further, training did not return the MRT to values normal for healthy young subjects, with the possible exception of Vu. DISCUSSION We have shown that patients with severe COPD who participated in a rigorous exercise training program as part of pulmonary rehabilitation demonstrated substantial improvements in exercise tolerance. The improvements in exercise tolerance were accompanied by physiologic changes, including an unanticipated alteration in the pattern of breathing during heavy exercise. These data define a new mechanism by which exercise training reduces the ventilatory requirement in patients with a ventilatory limitation to exercise. The participants in this study were fairly typical of patients undergoing outpatient pulmonary rehabilitation in the United States today. The great majority of these patients suffered from COPD. As is common, the patients we studied had a long history of cigarette smoking: 54 ± 32 pack-years. Smoking cessation was a requirement for program entry; patients had stopped smoking a median of 6 yr prior to rehabilitation (range, 1 mo to 27 yr). On average, spirometry showed severe obstruction, resting arterial oxygenation was moderately impaired, and there was mild CO 2 retention. Exercise responses showed a peak Vol averaging 54 ± 14% pred, with a very low breathing reserve indicating a ventilatory limitation to exercise. Exercise Training Strategy The training program design was based on principles found to be effective in healthy subjects. The duration of the rigorous phase of the program (6 wk), the frequency of the sessions (three per week) and the length of the training sessions (45 min) were well within the guidelines for exercise training programs (5). Intensity guidelines are controversial; strategies applicable for healthy subjects based on fixed fractions of maximal HR or Vol ( either fractions of predicted maximum or of the observed maximum for a given patient) are likely to be invalid for patients with COPD. We chose to use the approach advanced by Ries and Archibald (19), which advocates training intensity targets near the maximally tolerated work rate for patients with COPD. Patients with COPD are limited by their ability to breathe, not by limits of muscle metabolism. VE levels mildly below maximally tolerated levels can apparently be sustained for appreciable periods. By encouraging patients toward progressively higher work rates as the training program proceeded, by the end of the study most patients were tolerating work rates higher than their preprogram peak work rate (Figure 2). Our understanding of what constitutes a training intensity adequate to induce a physiologic training effect is evolving, and exercise prescription principles in COPD remain controversial. Our previous study of patients with COPD with predominantly moderate disease (6) postulated that, in order to be effective, a training program must involve exercise intensities associated with lactic acidosis. Indeed, patients who exercised at high intensities associated with lactic acidosis demonstrated greater evidence of a physiologic training effect than did patients whose exercise intensities were not associated with lactic acidosis, though the total work per training session was the same. However, more recentjy (20) we found that healthy subjects were able to attain a substantial training response from an exercise regimen that did not raise blood lactate levels. Apparently, lactic acidosis is not a critical requirement for an effective training intensity. Physiologic Benefits of Training In the present study, appreciable gains in exercise tolerance were observed in patients who were unable to substantially increase blood lactate levels during exercise. What evidence is there that gains were due to physiologic rather than psychologic factors? The improvement in exercise tolerance we observed, though considerable, is not conclusive proof in itself. Although performance in cycle ergometry is not markedly strategy dependent in comparison with treadmill or timed walking tests (21), improvement in motivation might have translated into improved exercise endurance. Although we have no concurrent control group, we think this unlikely based on results obtained from a retrospective comparison (unpublished observations). A group of 26 patients with COPD whose anthropometrics and disease severity did not differ significantly from those in the present study group underwent a rehabilitation program with the same therapeutic team. The program was identical in all respects, except that the rigorous cycle ergometer training program was not present. Instead, a program of "progressive ambulation" was included where walking and treadmill exercise was stressed but no specific exercise intensity targets were utilized. Incremental cycle ergometer exercise testing was performed in a manner identical to the present study. However, after rehabilitation, peak work rate and peak Vo l were not significantly increased, in contrast to the substantial increases seen in the present study. This suggests that the improvements in exercise tolerance we observed were not motivational effects that attended the rehabilitative process. More conclusive evidence of physiologic changes are obtained from responses to identical levels of exercise. In both incremental and constant work rate tests, HR and VE were Casaburi, Porszasz, Burns, et al.: Exercise Training in Severe COPD 1549 lower after rehabilitation. Further, evidence of faster kinetics of Vol and Vco2 strongly supports the presence of intramuscular changes associated with a training response (see below). Moreover, recent evidence has been obtained (22) that an intensive training program increases the levels of the aerobic enzymes of leg muscles of patients with severe COPD—a sure sign of a physiologic training effect. This contradicts the previous impression obtained from an older study (23) in which training was less intense (24) that such patients could not train the exercising muscles. The mechanism of the decreased ventilatory requirement for exercise after undergoing rigorous training merits examination. Alveolar mass balance dictates that: and Vco 2 are linked to an abrupt increase in cardiac output (29), though rapid changes in mixed venous 0 2 and CO2 content after exercise onset have also been demonstrated (30). It has been shown that the amplitude of Phase I is reduced in COPD (8), pulmonary vascular disease (31), cyanotic congenital heart disease (32), and hyperthyroidism (33). In patients with cardiovascular disease (31, 32) the Phase I amplitude showed a positive correlation with exercise tolerance. In hyperthyroidism, there was an increase in Phase I amplitude as the clinical status of the patients improved (33). In the present study, Phase I amplitude for both Vo l and Vco2 increased dramatically after training (Table 6). The increased Phase I gas exchange response is likely related to an accelerated delivery of desaturated blood to the central circulation, possibly because of more efficient action of the muscle pump (29) and a larger abrupt increase in stroke volume. The Phase I VE amplitude also increased substantially as a result of training (by an average of 90%). The mechanism of Phase I VE is controversial. Some authorities have postulated a link to rapid cardiac output increase, though evidence has been presented that appears to refute this linkage (29). A slow exponential phase of Vo l response has been previously observed in patients with COPD (8), pulmonary vascular diseases (31), and chronic heart failure (34), and also in normal subjects with (3-adrenergic blockade (35). Slow Vo l kinetics have also been shown in healthy subjects after a prolonged bedrest (36). More recently, slow Vol kinetics have been demonstrated in the elderly (37). The mechanism of slowed Vo l kinetics in these varied conditions is likely multifactorial. First, slow circulatory kinetics (as suggested by slow heart rate kinetics) (Table 5) presumably leads to slower delivery of desaturated blood from the muscle to the lung. Second, intramuscular oxygen uptake kinetics are likely slower because deconditioning (and myopathic changes in some chronic diseases) leads to inefficient oxygen delivery to the mitochondria. As evidence of this, deconditioned subjects have been shown to have appreciable transient increases in blood lactate levels even during moderate exercise (38). Improvements in both circulatory and intramuscular factors likely contribute to the speeding of Vo l kinetics seen in these patients with COPD after a rigorous training program (Table 5). HR kinetics are substantially more rapid, suggesting more rapid delivery of the products of exercise metabolism to the lung. Further, Martin and colleagues (39) have shown in the elderly that training reduces peripheral vascular resistance and increases maximal calf blood flow during ischemic exercise. Similar effects may have occurred in the patients here, as suggested by a decrease in systolic blood pressure and a trend toward a lower diastolic pressure in response to identical work rates (Table 2). Supporting intramuscular changes, Maltais and colleagues (22) have recently shown increased aerobic enzyme concentrations after a rigorous training program in patients with COPD. Presumably, this and other intramuscular changes associated with training lead to more efficient oxidative metabolism and faster intramuscular oxygen uptake kinetics. However, after training Vol kinetics remain considerably slower in these patients with COPD than in healthy subjects (Figure 5), suggesting that residual circulatory and/or muscle abnormalities remain. Vco2 kinetics are considerably slower than Vo l kinetics, likely because of the large intramuscular and venous CO 2 stores that must be "washed in" when CO2 content in these compartments increases during exercise. The speeding of Vco 2 kinetics that was seen with training (Table 5) can be explained by the more rapid intramuscular kinetics of aerobic metabolism (i.e., more rapid Vol kinetics) as well as more VE = k Vco 2/Pa co2 (1 — VD/VT) where k is a constant. As shown in Tables 2 to 4, the reduction in VE response is predominantly due to a reduced VD/VT (though the significant correlation between changes in Vco2 and VE suggest a small component linked to decreased Vco2 ). In turn, the reduced VD/VT can be traced to a slower, deeper pattern of breathing. As patients with severe COPD exercise, progressive hyperinflation occurs and tidal volume increase is limited. A progressively tachypneic pattern of breathing results (25) that is inefficient in that a large fraction of the breath is composed of anatomic dead space. After training, at a given level of heavy exercise, average respiratory rate is substantially lower and tidal volume is mildly higher. The explanation of the mechanism of the altered breathing pattern seen during heavy exercise must be speculative. Conceivably, patients may be less hyperinflated at a given work rate after training, though we cannot confirm this as we did not assess changes in lung volume during exercise. Dynamic hyperinflation compromises the ability of the inspiratory muscles to generate pressure; thus, functional inspiratory muscle weakness develops. Decreased hyperinflation might be explained by increased respiratory muscle endurance. Inspiratory muscle fatigue is known to cause rapid, shallow breathing (26); tidal volume is less for a given level of VE in the fatigued than in the unfatigued state. If the inspiratory muscles are stronger after training, decreased fatigue might lead to less hyperinflation. However, a previous study (27) did not establish the ability of leg exercise training to increase respiratory muscle endurance, though it might be speculated that the training program involved in this previous study was not sufficiently intense to elicit this effect. Another factor that might tend to slow respiratory rate is breathing retraining techniques that are routinely taught as part of pulmonary rehabilitation, although breathing retraining has not been shown to improve exercise tolerance in most studies (28). Moreover, our retrospective comparison group underwent breathing retraining but did not demonstrate significant reductions in VE or an altered pattern of breathing during exercise (unpublished observation). Finally, the small improvement in spirometry observed after rehabilitation might tend to reduce the tendency for hyperinflation. However, the improvement in FEV I failed to correlate significantly with the fall in VE at identical work rates. Exercise Response Kinetics The speeding of exercise response dynamics we observed after a rigorous training program provides evidence for physiologic improvement in muscle and cardiovascular function. We examined the effect of exercise training on two aspects of the dynamic response to the onset of CWR exercise: Phase I and the subsequent exponential phase. The Phase I increases in Vol 1550 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL. 155 1997 rapid equilibration of the CO 2 stores because of more rapid circulatory kinetics. VE kinetics have been found to be highly correlated with VCO 2 kinetics, suggesting a humoral CO 2 -mediated mechanism of the exponential phase of VE response after exercise onset (17). In the present study, the appreciable speeding of VE kinetics with training is presumably linked to the speeding of Vcoz kinetics; VE, and Vco 2 MRT were highly correlated both before and after training (r = 0.91 and 0.76, respectively). Speeding of response kinetics at a given work rate is a noninvasive effort-independent measure of the physiologic training effect that occurs as a result of a training program in these patients with severe COPD. This is all the more important because many of these patients fail to increase blood lactate levels substantially; a fall in blood lactate level cannot be used as a reliable indicator of a physiologic training effect as it can be in other subject groups. However, we acknowledge that some patients with very severe disease may be unable to generate the data necessary to evaluate response kinetics; 12 of the 25 patients in the current study were unable to generate evaluable data. In summary, we feel that institution of a rigorous training program, with individualized exercise prescription, will be a practical addition for many pulmonary rehabilitation programs. Cardiopulmonary exercise testing on entry to the rehabilitation program helps establish training targets and rules out contraindications to exercise training. We found that a single therapist could supervise as many as six exercising patients at a time and that the group dynamic and staff encouragement yielded excellent compliance. It was our impression that COPD severity or disease characteristics (e.g., CO 2 retention, hyperinflation, low body weight) were not strong predictors of the success of the training program. In fact, except for muscle soreness early in the training program, no adverse effects were detected in any patient. We feel that strategies we used are applicable to a substantial majority of patients with COPD presenting for pulmonary rehabilitation. We conclude that a rigorous program of exercise training as part of a program of pulmonary rehabilitation yields physiologic benefits that improve exercise tolerance in patients with severe as well as moderate COPD; severity of lung disease alone should not be a selection criterion. We hasten to emphasize that we do not believe that cycle ergometer exercise has special advantages over other modes of exercise that yield similar stress to the muscles of ambulation (e.g., stair climbing, treadmill walking). Cycle ergometry was chosen to facilitate similar testing and training modalities. Finally, it deserves mention that the benefits of a rehabilitative program will recede without a continuing care program; the physiologic benefits of training will be lost unless regular exercise is continued (40). Acknowledgment The writers wish to thank Dr. James Daly for help with these studies and Mrs. Jackie Tosolini and Betsy Barnes for supervising the exercise training program. Mrs. Maclovia Wallace contributed expert secretarial support. References 1. Casaburi, R. 1993. Exercise training in chronic obstructive lung disease. In R. Casaburi and T. L. Petty, editors. Principles and Practice of Pul -monaryRehbilt.WBSaundrs,Philep.2042. O'Donnell, D. E., M. McGuire, L. Samis, and K. A. Webb. 1995. The impact of exercise reconditioning on breathlessness in severe chronic airflow limitation. Am. J. Respir. Crit. Care Med. 152:2005-2013. 3. Haas, F., J. Salazar-Schicchi, and R. Axen. 1993. Desensitization to dyspnea in chronic obstructive pulmonary disease. In R. Casaburi and T. L. Petty, editors. Principles and Practice of Pulmonary Rehabilita- tion. WB Saunders, Philadelphia. 241-251. 4. Sinclair, D. J. M., and C. G. Ingram. 1980. Controlled trial of supervised exercise training in chronic bronchitis. B.M.J. 1:519-521. 5. Casaburi, R. 1994. Physiologic responses to training. Clin. Chest Med. 15:215-227. 6. Casaburi, R., A. Patessio, F. Ioli, S. Zanaboni, C. F. Donner, and K. Wasserman. 1991. Reduction in exercise lactic acidosis and ventilation as a result of exercise training in obstructive lung disease. Am. Rev. Respir. Dis. 143:9-18. 7. Belman, M. J., and Z. Mohsenifar. 1991. Reductions in lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease (letter). Am. Rev. Repir. Dis. 144:1220-1221. 8. Nery, L. E., K. Wasserman, J. D. Andrews, D. J. Huntsman, J. E. Hansen, and B. J. Whipp. 1982. Ventilatory and gas exchange kinetics during exercise in chronic airways obstruction. J. Appl. Physiol. 53:15941602. 9. Knudsen, R. J., R. C. Slatin, M. D. Lebowitz, and B. Burrows. 1976. The maximal expiratory flow-volume curve. Normal standards, variability and effects of age. Am. Rev. Respir. Dis.113:587-600. 10. Cotes, J. E. 1975. Lung Function, 3rd ed. Blackwell, Oxford. 11. Beaver, W. L., K. Wasserman, and B. J. Whipp. 1973. On-line computer analysis and breath-by-breath graphical analysis of exercise function tests. J. Appl. Physiol. 34:123-128. 12. Huszczuk, A., B. J. Whipp, and K. Wasserman. 1990. A respiratory gas exchange simulator for routine calibration in metabolic studies. Eur. Respir. J. 3:465-468. 13. Forster, H. V., J. A. Dempsey, J. Thomson, E. Vidruk, and G. A. doPico. 1972. Estimation of arterial Po e , Pco 2 pH and lactate from arterialized venous blood. J. Appl. Physiol. 32:134-137. 14. Lamarra, N., B. J. Whipp, S. A. Ward, and K. Wasserman. 1987. Effect of interbreath fluctuations on characterizing exercise gas exchange kinetics. J. Appi. Physiol. 62:2003-2012. 15. Motulsky, H. J., and L. A. Ran§nas. 1987e Fitting curves to data using nonlinear regression: a practical and nonàipthematical review. FASEB J. 1:365-374. 16. Wasserman, K., J. E. Hansen, D. Y. Sue, B. J. Whipp, and R. Casaburi. 1994. Principles of Exercise Testing and Interpretation. Lea and Febiger, Philadelphia. 17. Casaburi, R., B. J. Whipp, W. L. Beaver, K. Wasserman, and S. N. Koyal. 1977. Ventilatory and gas exchange dynamics in response to sinusoidal work. J. Appl. Physiol. 42:300-311. 18. Casaburi, R., T. Barstow, T. Robinson, and K. Wasserman. 1989. Influence of work rate on ventilatory and gas exchange kinetics. J. Appl. Physiol. 67:547-555. 19. Ries, A. L., and C. J. Archibald. 1987. Endurance exercise training at maximal targets in patients with chronic obstructive pulmonary disease. J. Cardiopulmon. Rehabil. 7:594-601. 20. Casaburi, R., T. W. Storer, C. S. Sullivan, and K. Wasserman. 1995. Evaluation of blood lactate elevation as an intensity criterion for exercise training. Med. Sci. Sports Exerc. 27:852-862. 21. Olopade, C. 0., K. C. Beck, R. W. Viggiano, and B. A. Staats. 1992. Exercise limitation and pulmonary rehabilitation in chronic obstructive pulmonary disease. Mayo Clin. Proc. 67:144-158. 22. Maltais, F., P. LeBlanc, C. Simard, J. Jobin, C. Berube, J. Bruneau, L. Carrier, and R. Belleau. 1996. Skeletal muscle adaptation to endurance training in patients with chronic obstructive pulmonary disease. Am. Respir. Crit. Care Med. 154:442-447. 23. Belman, M. J., and B. A. Kendregan. 1981. Exercise training fails to increase skeletal muscle enzymes in patients with chronic obstructive pulmonary disease. Am. Rev. Respir. Dis. 123:256-261. 24. Haber, P. 1981. Exercise training fails to increase skeletal muscle enzymes in patients with chronic obstructive pulmonary disease (letter) Am. Rev. Respir. Dis. 124:347-348. 25. Marin, J. M., S. N. A. Hussain, W. J. Gibbons, M. Polverino, R. D. Levy, and M. G. Cosio. 1993. Relationship of resting lung mechanics and exercise pattern of breathing in patients with chronic obstructive lung disease. Chest 104:705-711. 26. Gallagher, C. G., V. I. Hof, and M. Younes. 1985. Effect of inspiratory muscle fatigue on breathing pattern. J. Appl. Physiol. 59:1152-1158. 27. Belman, M. J., and B. A. Kendregan. 1982. Physical training fails to improve ventilatory muscle endurance in patients with chronic obstructive pulmonary disease. Chest 81:440-443. 28. Faling, L. J. 1993. Controlled breathing techniques and chest physical therapy in chronic obstructive pulmonary disease and allied conditions. In R. Casaburi and T. L. Petty, editors. Principles and Practice of Pulmonary Rehabilitation. WB Saunders, Philadelphia. 167-182. 29. Casaburi, R., S. Spitzer, R. Haskell, and K. Wasserman. 1989. Effect of , Casaburi, Porszasz, Burns, et al.: Exercise Training in Severe COPD altering heart rate on gas exchange at exercise onset. Chest 95:6-12. 30. Casaburi, R., J. Daly, J. E. Hansen, and R. Effros. 1989. Abrupt changes in mixed venous blood gas composition after the onset of exercise. J. Appl. Physiol. 67:1106-1112. 31. Sietsema, K. E. 1992. Oxygen uptake kinetics in response to exercise in patients with pulmonary vascular disease. Am. Rev. Respir. Dis. 145: 1052-1057. 32. Sietsema, K. E., D. M. Cooper, M. H. Rosove, J. K. Perloff, J. S. Child, M. M. Canoccio, B. J. Whipp, and K. Wasserman. 1986. Dynamics of oxygen uptake during exercise in adults with cyanotic congenital heart disease. Circulation 73:1137-1144. 33. Ben-Dov, I., K. E. Sietsema, and K. Wasserman. 1991. 0 2 uptake in hyperthyroidism during constant work rate and incremental exercise. Eur. J. Appl. Physiol. 62:261-267. 34. Riley, M., J. Pórszász, C. F. Stanford, and D. P. Nicholls. 1994. Gas exchange responses to constant work rate exercise in chronic heart failure. Br. Heart. J. 72:150-155. 35. Petersen, E. S., B. J. Whipp, J. A. Davis, D. J. Huntsman, H. V. Brown, View publication stats 1551 and K. Wasserman. 1983. Effects of (3-adrenergic blockade on ventilation and gas exchange during exercise in humans. J. Appl. Physiol. 54: 1306-1313. 36. Convertino, V. A., D. J. Goldwater, and H. Sandler. 1984. Vo l kinetics of constant-load exercise following bed-rest-induced deconditioning. J. Appl. Physiol. 57:1545-1550. 37. Babcock, M. A., D. H. Paterson, D. A. Cunningham, and J. R. Dickinson. 1994. Exercise on-transient gas exchange kinetics are slowed as a function of age. Med. Sci. Sports Exerc. 26:440-446. 38. Cerretelli, P., D. Pendergast, W. C. Paganelli, and D. W. Rennie. 1979. Effects of specific muscle training on Vo l on-response and early blood lactate. J. Appl. Physiol. 47:761-769. 39. Martin, W. H., W. M. Kohrt, M. T. Malley, E. Korte, and S. Stoltz. 1990. Exercise training enhances leg vasodilatory capacity of 65-yr-old men and women. J. Appl. Physiol. 69:1804-1809. 40. Casaburi, R. 1996. Deconditioning. In A. P. Fishman, editor. Pulmonary Rehabilitation. Lung Biology in Health and Disease Series. Marcel Dekker, New York. 213-230.