Key words: exercise therapy; exercise tolerance; lung diseases; obstructive; oxygen consumption; walking

Exercise Outcomes After Pulmonary Rehabilitation Depend on the Initial Mechanism of Exercise Limitation Among Non-Oxygen-Dependent COPD Patients* John F. Plankeel, MD; Barbara McMullen, RRT; and Neil R. MacIntyre, MD, FCCP Study objectives: Pulmonary rehabilitation (PR) that includes exercise training can improve exercise tolerance and quality of life for patients with COPD. However, the degree of benefit from PR is variable. We hypothesized that the exercise response to PR varies depending on the initial factors that limit exercise. Design, setting, participants, and measurements: We retrospectively analyzed the change in exercise capacity after PR in 290 nonhypoxemic patients with COPD. We classified patients into the following subgroups based on the primary limitation seen on initial exercise testing: (1) ventilatory-limited (VL); (2) cardiovascular-limited (CVL); (3) mixed ventilatory/cardiovascularlimited (VLCVL); and (4) non-cardiopulmonary-limited (NL). We compared outcomes among subgroups. Results: In the entire study population, PR led to increased timed walk distance (30.3%; p < 0.0001) and maximal oxygen consumption (V O2 max) [84.8 ml/min; p < 0.0001]. Stepwise multiple regression selected age, ventilatory reserve at peak exercise, and exercise arterial oxygen pressure as individual predictors of improvement in V O2 max. V O2 max increased in the VL subgroup (30.4 ml/min; p 0.008), the CVL subgroup (109.0 ml/min; p < 0.0001), the mixed VLCVL subgroup (61.3 ml/min; p < 0.0001), and NL subgroups (110.5 L/min; p < 0.0001). The improvement in V O2 max was greater in the CVL subgroup than in the VL subgroup (p < 0.0001). Timed walk distance improved to a similar degree in all subgroups (26 to 36%). Conclusions: Patients with nonventilatory exercise limitations experience the greatest increase in V O2 max after PR. However, even patients with severe ventilatory limitation can improve exercise tolerance with PR. (CHEST 2005; 127:110 116) Key words: exercise therapy; exercise tolerance; lung diseases; obstructive; oxygen consumption; walking Abbreviations: CVL cardiovascular-limited; ExPo 2 Po 2 at peak exercise; HR heart rate; MVV maximum voluntary ventilation; NL non-cardiopulmonary-limited; PR pulmonary rehabilitation; V e minute ventilation; VL ventilatory-limited; VLCVL ventilatory/cardiovascular-limited; V o 2 max maximal oxygen consumption; VR ventilatory reserve Patients with COPD often complain of exercise intolerance. While ventilatory limitation to exercise is often present, other factors are also important. These include cardiovascular deconditioning, skeletal muscle dysfunction, gas exchange abnormalities, right ventricular dysfunction, and psychological factors. 1 *From the Department of Pulmonary and Critical Care Medicine, Duke University Medical Center, Durham, NC. Manuscript received April 2, 2004; revision accepted August 12, 2004. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: permissions@chestnet.org). Correspondence to: Neil MacIntyre, MD, FCCP, Room 7453, Box 3911, Duke University Hospital, Durham NC 27710; e-mail: neil.macintyre@duke.edu Pulmonary rehabilitation (PR) is designed to reverse these exercise-limiting factors through supervised exercise training, respiratory care, and education. Exercise training in COPD patients can lead to improved aerobic fitness with increases in muscle aerobic enzyme content, 2 and reductions in lactate levels, heart rate (HR), and minute ventilation (V e) at isowork rates. 3 Other benefits include improved motivation and exercise technique, desensitization to dyspnea, and an optimized breathing pattern. 4 A prospective, randomized, controlled trial 5 showed that PR improves endurance tests, peak work rate, maximal oxygen consumption (V o 2 max), and quality of life. The American Thoracic Society recommends PR for 110 Clinical Investigations

patients with persistent exercise intolerance despite receiving optimal medical therapy. 6 Despite these overall benefits, the response to PR varies significantly among individuals. This variation may result from individual differences in the factors limiting exercise. For example, Zu Wallack et al 7 showed that ventilatory reserve (VR) is positively correlated with an improvement in 12-min walk distance after PR. Similarly, physiologic training effects may be more pronounced in patients limited primarily by cardiovascular and skeletal muscle deconditioning (poor aerobic fitness). The aim of this study was to describe the relationship between initial exercise limitation and exercise response to PR. We hypothesized that the degree of benefit from PR and the mechanisms of improvement vary depending on the primary factor limiting exercise. Specifically, we expected patients who were limited by poor aerobic fitness to have a greater increase in exercise capacity than ventilatory-limited (VL) patients. To test this hypothesis, we performed a retrospective analysis of 290 COPD patients who had undergone PR. Subjects Materials and Methods We retrospectively reviewed a computerized database, which included all patients entering the Duke Pulmonary Rehabilitation Program between 1985 and 1999. This analysis included patients with a primary diagnosis of COPD or asthmatic bronchitis and a V o 2 max of 80% of predicted. Patients with diagnoses of asthma, interstitial lung disease, or cystic fibrosis, or those who have undergone lung transplantation were excluded. We also excluded patients requiring supplemental oxygen during exercise because the level of oxygen used during exercise testing was not controlled. We thought that a difference in the amount of oxygen used at baseline and after rehabilitation exercise testing could be a significant confounding variable. Finally, patients were routinely excluded from the rehabilitation program for a history of unstable or exertion angina, frequent ventricular extrasystoles, or worsening ECG ST-wave or T-wave abnormalities with exercise. Approval for the use of the clinical data in this study was obtained from our institutional review board on human research. During the period from 1985 to 1999, a total of 636 patients entered the Duke Pulmonary Rehabilitation Program. A total of 450 patients (71%) had a primary diagnosis of COPD or asthmatic bronchitis. Of these, we excluded 150 patients (33%) who required supplemental oxygen during exercise and 4 who lacked documentation of oxygen use. Finally, we excluded six patients due to missing initial exercise test data. This left 290 non-oxygendependent COPD patients (46.5% of all patients undergoing the Duke PR program) for inclusion in this study. Of the 290 patients analyzed, 11 dropped out of the program for the following reasons: inguinal hernia requiring surgery (1 patient); back pain (1 patient); groin pull (1 patient); COPD exacerbation (1 patient); exercise-induced ST-segment changes on ECG (1 patient); hospitalization related to preexisting mitral valve disease (1 patient); palpitations (1 patient); new-onset atrial fibrillation/flutter (2 patients); and stable ventricular tachycardia during exercise (1 patient). One patient died at home 10 days after prematurely discontinuing the program. This death was presumed to have been due to myocardial infarction. In addition, nine patients were missing postprogram exercise test data. Six of these patients successfully completed the program but did not return for a final exercise test. Design The database included the following information: age; sex; primary diagnosis; pulmonary medications; supplemental oxygen use; spirometry; lung volumes; diffusing capacity; resting and exercise arterial blood gas levels; timed walk distance; and data from a maximal incremental exercise test. Spirometry, timed walk distance, and maximal exercise testing were performed before and after PR. Primary outcome measures were the change in timed walk distance, V o 2 max, and oxygen pulse (ie, V o 2 max/peak HR) after completing PR. Secondary outcomes included changes in spirometry, peak V e, and peak HR. Correlation between baseline variables and improvement in V o 2 max also were analyzed. Pulmonary Function and Exercise Testing Pulmonary function testing included spirometry, lung volume determinations, and single-breath diffusing capacity using standard equipment (model 2200 and V max systems; SensorMedics; Yorba Linda, CA). Procedures were carried out according to American Thoracic Society standards. 8 12 The maximum voluntary ventilation (MVV) used in this study was directly measured using a 12-s MVV maneuver. Graded bicycle exercise testing was performed with a system designed to measure oxygen consumption, carbon dioxide production, and V e in 20-s increments using polarographic, infrared, and mass flowmeter devices, respectively (system 2900 and V max systems; SensorMedics). Workload was ramped at 12.5 W/min. Arterial blood gas levels and finger pulse oximetry were recorded at peak exercise. Eight-lead ECG was performed at rest and during all stages of exercise. Testing was usually performed in the week prior to entering the program and during the final week of the program. Timed Walk Testing Tests were performed on an indoor level track. Patients were instructed to walk as far as possible during the test and to give it your best effort. A therapist kept time and counted laps but did not walk alongside the patient. No direct encouragement was offered during the test. Patients were allowed to use assistive devices (eg, walker or cane) for musculoskeletal or balance needs. Vital signs (ie, HR, BP, and oxygen saturation) were monitored at defined intervals during the test. During the 14-year study period, patients were evaluated with either a 12-min or a 15-min walk test, and the same type was used in individual patients before and after PR. Because the type of test may have differed between patients, we only report the percentage improvement in timed walk distance. The supervising therapist was not necessarily the same on pretesting and posttesting. PR PR was performed at Duke University Medical Center (Durham, NC) on an outpatient basis. It consisted of 20 sessions completed over a 4-week period. Sessions were 4hinduration. The program emphasized respiratory care, education, and exercise. The exercise portion of each session consisted of at least 1 h www.chestjournal.org CHEST / 127 / 1/ JANUARY, 2005 111

of leg and arm ergometry and level walking on an indoor track. The target exercise intensity was either an HR of 80% of the predicted maximum (220 beats/min age) or a score of 12 on the modified Borg exertion scale. Stretching, exercise with free weights, and water exercise also were included on an individual basis. Statistical Analysis Paired t tests were used to compare the mean changes in spirometry, exercise testing, and timed walk distance after PR. The correlation between baseline variables and improvement in V o 2 max was performed using Pearson correlation coefficients. The cohort had missing data for some baseline variables. Therefore, the number of available observations is reported with the results. Stepwise multiple regression was used to identify significant individual predictors of changes in V o 2 max after PR. We entered into this model only the variables that were significantly correlated (ie, p 0.05) with improved V o 2 max on univariate regression. In addition, we chose to enter VR (ie, [1 peak V e/mvv] 100) and not baseline MVV into this model due to the obvious interaction among these variables. Significance on multivariate regression was set at a p 0.05. A statistical software package was used for this analysis (SAS; SAS Institute; Cary, NC). Results Overall Changes Induced by PR The preprogram subject characteristics and the changes in spirometry, timed walk distance, and peak exercise performance after the rehabilitation program are summarized in Table 1. As a whole, the study population showed a small but significant improvement in spirometry. In addition, significant improvement in mean peak exercise capacity was seen, as follows: peak work rate improved by 18.2% (8.41 W); V o 2 max improved by 11.0% (84.8 ml/ min); and peak oxygen pulse improved by 10.8% (0.64 ml per beat). Overall, the mean timed walk distance improved 30% over baseline values. Age, FEV 1, MVV, VR (ie, VR [1 peak V e/ MVV]), diffusing capacity, arterial Po 2 at rest, and arterial Po 2 during exercise (ExPo 2 ) were significantly correlated with improvement in V o 2 max (Table 2). However, on stepwise multiple regression, only age, VR, and ExPo 2 remained significant independent predictors of change in V o 2 max. The final regression equation predicting change in V o 2 max after PR was as follows (r 0.45; r 2 0.20; p 0.0001): V o 2 max (in milliliters per minute) 232.5 5.1 (age in years) 180 (VR) 1.8 (ExPo 2 in millimeters of mercury). Response to PR in the Exercise Limitation Subgroups Because this multivariate model suggested that improvement in V o 2 max after PR varies depending on the degree of pre-pr ventilatory and gas exchange function, we chose to further analyze our Table 1 Baseline Subject Characteristics and Changes Induced After PR* Variable No. Mean Baseline Value Mean Change Percentage Change Study population 290 Age, yr 290 66.4 (0.5) Female, % 150 52 Medications, % Bronchodilators 264 91.4 Inhaled steroid 148 51.2 Theophylline 138 47.8 Prednisone 58 20.1 Diffusing capacity, ml/mm Hg/min 254 12.8 (0.4) Po 2 at rest, mm Hg 274 74.4 (0.6) Pco 2 at rest, mm Hg 274 39.5 (0.3) ExPo 2,mmHg 257 80.1 (0.9) Pco 2 during exercise, mm Hg 232 42.3 (0.5) Peak V e/mvv ratio 290 0.77 (0.01) Peak HR/(220 age) 290 0.83 (0.01) FEV 1,L 290 1.22 (0.03) 0.05 (0.02) 6.0 (1.5) FVC, L 288 2.56 (0.05) 0.11 (0.03) 6.3 (1.4) MVV, L/min 290 45.9 (1.3) 3.92 (0.82) 12.6 (2.1) Timed walk distance 268 30.3 (1.8) Peak work rate, W 290 62.5 (1.6) 8.41 (0.74) 18.2 (2.1) V o 2 max, ml/min 290 831.0 (18.3) 84.8 (9.46) 11.0 (1.2) *Values given as mean (SEM), unless otherwise indicated. Significantly different from Pre-PR at p 0.001. Significantly different from Pre-PR at p 0.0001. 112 Clinical Investigations

Table 2 Correlation Between Baseline Variables and Improvement in V O2 max Baseline Variables No. Improvement in V o 2 max R Value p Value Age 269 0.28 0.0001 FEV 1 269 0.18 0.003 MVV 269 0.17 0.005 VR 100 269 0.15 0.01 Diffusing capacity 237 0.32 0.0001 Peak HR/(220 age) 269 0.01 0.83 Baseline V o 2 max 269 0.11 0.08 Baseline O 2 pulse 269 0.05 0.40 Po 2 at rest 255 0.14 0.03 Pco 2 at rest 255 0.07 0.26 ExPo 2 240 0.25 0.0001 Exercise Pco 2 217 0.05 0.43 data by grouping patients according to their mechanism of exercise limitation. To do this, we first divided our population into those with and those without a ventilatory limitation for exercise, as defined by a peak V e/mvv ratio of 0.80. We chose not to further classify our patients by exercise Po 2 response because patients with true gas exchange limitation to exercise (ie, those with hemoglobin desaturation to 90%) were not included in this analysis. We did, however, classify these two groups into those reaching and not reaching cardiovascular limits (ie, 80% of their predicted maximal HR [220 beats/min age]). We reasoned that adding the HR criteria to our patient grouping scheme might help to identify those patients who could reach cardiovascular limits and thus do aerobic training in PR. The following four groups were thus established (Table 3): (1) VL and unable to achieve 80% of the predicted maximal HR; (2) VL and able to achieve 80% of maximal HR (ventilatory/cardiovascular-limited [VLCVL]); (3) not VL and able to achieve 80% of maximal HR (cardiovascular-limited [CVL]); and (4) not VL and unable to achieve 80% of maximal HR (non-cardiopulmonary-limited [NL]). Analysis of variance was used to determine differences in outcome among these four groups and to control for differences in baseline variables between groups. Table 3 Exercise Limitation Subgroup Definitions Primary Exercise Limitation Peak Ventilation/MVV Peak HR Ventilatory 0.8 0.8 Ventilatory and cardiovascular 0.8 0.8 Cardiovascular 0.8 0.8 No limit 0.8 0.8 Adjustment for multiple comparisons was made using the Tukey method (p 0.05). The preprogram demographics, pulmonary function data, and exercise performance for these exercise limitation subgroups are shown in Table 4. The changes in exercise capacity for each group are summarized in Table 5 and in Figure 1. The timed walk distance increased significantly and to a similar degree in all groups. V o 2 max also improved significantly in all groups. However, the VL group had significantly less absolute improvement and percentage of improvement compared to the CVL and NL groups. After controlling for baseline FEV 1, diffusing capacity, and ExPo 2, the exercise limitation group designation remained a significant predictor of improvement for V o 2 max (p 0.005). Mean peak oxygen pulse improved in the CVL group, the mixed VLCVL group, and the NL group, but did not change in the VL group. Postrehabilitation changes in spirometry and the physiologic response to peak exercise in these four groups are shown in Table 5. Both VL groups had a relatively small change in peak ventilation (ie, peak V e) but had a significant increase in FEV 1 and MVV. In contrast, the CVL group had a marked increase in mean peak V e without a change in spirometry. Therefore, in the CVL group, the mean peak V e/ MVV ratio increased, suggesting that many of these patients exercised further, to their ventilatory limits, after completing PR. Finally, the groups reaching cardiovascular limits had a significant reduction in peak HR, whereas the NL group, had an increase in peak HR after PR. Discussion In this study, the two groups reaching cardiovascular limits on initial testing had the greatest increase in V o 2 max. This supports our hypothesis that patients who are limited by poor aerobic fitness experience the greatest increase in exercise capacity after PR. These groups had the highest baseline V o 2 max and peak HR. We suspect that this ability to train at a higher intensity resulted in greater training effects. The evidence for a true improvement in aerobic fitness includes the increase in peak oxygen pulse and the fall in peak exercise HR seen in these groups after PR. In addition, the CVL group had a significant increase in the peak V e/mvv ratio on postprogram testing, implying that many of these patients exercised further, to ventilatory limits as opposed to cardiovascular limits, after completing PR. Prior studies have shown that PR with exercise training can improve cardiovascular and peripheral muscle function in COPD patients. However, be- www.chestjournal.org CHEST / 127 / 1/ JANUARY, 2005 113

Table 4 Characteristics of the Exercise Limitation Subgroups* Variable Ventilatory (n 46) Ventilatory and Cardiovascular (n 75) Cardiovascular (n 87) No Limit (n 82) p Value Peak ventilation/mvv 0.97 (0.02) 1.00 (0.02) 0.61 (0.01) 0.62 (0.01) Peak HR 0.73 (0.01) 0.90 (0.01) 0.92 (0.01) 0.71 (0.01) Female, % 52 49 54 51 Age, yr 67.3 (1.0) 65.9 (1.1) 66.5 (1.1) 66.2 (0.8) Medications, % Bronchodilator 93.5 96.0 87.2 90.2 Inhaled steroid 47.8 54.7 52.3 48.8 Theophylline 58.7 50.7 47.7 39.0 Prednisone 21.7 16.0 22.1 20.7 FEV 1,L 0.97 (0.06) 0.99 (0.04) 1.52 (0.07) 1.26 (0.05) 0.0001 Diffusing capacity, ml/mm Hg/min 10.3 (0.8) 11.9 (0.6) 14.5 (0.7) 13.1 (0.5) 0.0002 Po 2 at rest, mm Hg 71.3 (1.1) 74.3 (1.0) 76.0 (1.2) 74.4 (1.1) 0.05 ExPo 2,mmHg 72.9 (1.8) 79.8 (1.6) 85.5 (1.7) 78.3 (1.8) 0.0001 Pco 2 at rest, mm Hg 40.4 (0.6) 39.2 (0.5) 38.6 (0.6) 40.1 (0.6) 0.14 Exercise Pco 2,mmHg 44.3 (1.1) 42.7 (0.8) 40.0 (0.9) 43.2 (0.9) 0.009 V o 2 max, ml/min 794.1 (49.6) 878.8 (36.4) 898.4 (37.7) 736.3 (23.0) 0.003 Peak oxygen pulse, ml/beat 7.1 (0.4) 6.3 (0.2) 6.3 (0.2) 6.8 (0.2) 0.09 *Values given as mean (SEM), unless otherwise indicated. cause these effects depend on the training intensity, 3 some have thought that patients limited by ventilatory factors cannot train at an intensity that would improve V o 2 max. 13 Our study challenges this notion by showing that VL patients who are unable to exercise to 80% of their HR maximum experience a small improvement in V o 2 max after PR. However, this improvement was significantly less than that of the CVL group (without ventilatory limitations), and it occurred without an increase in peak oxygen pulse. Therefore, factors other than aerobic conditioning are probably important in increasing V o 2 maxinthe VL group. One such factor may be the observed increase in ventilatory capacity after PR. Significant improvement was seen in FEV 1 (10%), MVV (16%), and peak V e (5%) in the VL groups. Casaburi et al 4 reported similar changes among patients with severe ventilatory limitation. These changes may be due to improved bronchodilator use or respiratory muscle strength. A change in breathing pattern may be another mechanism leading to improved exercise capacity. In the study by Casaburi et al, 4 exercise training resulted in a lower respiratory rate, a higher tidal volume, and less dead space ventilation during exercise. Finally, improved motivation, improved Table 5 Mean Changes in Spirometry, Physiologic Exercise Response, and Exercise Capacity After PR in the Exercise Limitation Subgroups* Variable Ventilatory Ventilatory and Cardiovascular Cardiovascular No Limit p Value FEV 1,L 0.08 (0.03) 0.08 (0.03) 0.01 (0.04) 0.06 (0.03) 0.32 FVC, L 0.19 (0.08) 0.17 (0.06) 0.03 (0.06) 0.09 (0.05) 0.20 MVV, L/min 5.3 (1.7) 7.4 (1.5) 1.9 (2.0) 2.1 (1.1) 0.04 Peak ventilation, L/min 2.1 (1.1) 1.1 (0.9) 3.8 (0.9) 4.0 (0.8) 0.07 Peak ventilation/mvv 0.08 (0.02) 0.12 (0.04) 0.10 (0.07) 0.07 (0.02) 0.001 Peak HR, beats/min 4.7 (2.2) 6.4 (1.7) 2.3 (1.6) 5.9 (1.3) 0.0001 Work efficiency 0.005 (0.002) 0.004 (0.002) 0.002 (0.002) 0.002 (0.002) 0.60 Walk distance, % 34.2 (4.3) 26.7 (2.9) 26.4 (3.2) 35.7 (4.1) 0.12 Peak work, W 5.6 (1.5) 7.8 (1.4) 9.6 (1.3) 9.3 (1.6) 0.30 Peak work, % 13.5 (4.6) 13.9 (2.9) 18.2 (3.3) 24.8 (5.4) 0.20 V o 2 max ml/min 30.4 (14.3) 61.3 (18.6) 109.0 (19.9) 110.5 (16.7) 0.01 % 4.8 (1.9) 6.6 (2.1) 13.2 (2.4) 15.9 (2.2) 0.002 Oxygen pulse, ml/beat 0.05 (0.1) 0.75 (0.2) 0.85 (0.1) 0.63 (0.2) 0.003 Oxygen pulse, % 1.3 (2.0) 12.3 (2.3) 14.8 (2.1) 10.5 (2.2) 0.001 *Values given as mean change from baseline (SEM), unless otherwise indicated. p 0.05 (nonsignificant). Work/V o 2 max (in W/mL/min). 114 Clinical Investigations

Figure 1. Mean changes in exercise capacity after PR in the primary exercise limitation subgroups: VL; VLCVL; CVL; and NL. Top left, A: the percentage improvement in timed walk distance was significant for all groups (p 0.0001) and the differences between groups were nonsignificant (NS). Top right, B: group mean changes in V o 2 max (* p 0.008 p 0.0001). Bottom, C: group mean changes in oxygen pulse (* p 0.05; p 0.0001). skeletal muscle function, enhanced work efficiency, and desensitization to dyspnea also may be important. These factors may be the primary mechanisms for the small increase in peak exercise capacity in the VL group. Despite having only a minimal increase in V o 2 max, the VL group did have a marked improvement in timed walk distance. The degree of improvement was similar to the groups without ventilatory limitation. Timed walk testing is a measure of submaximal exercise performance, whereas V o 2 max measures the peak exercise level that can be attained for a short time period. Therefore, an improved physiologic response to submaximal exercise could have contributed to this marked increase in timed walk distance without a marked impact on V o 2 max. In fact, Casaburi et al 4 showed that exercise training can induce faster oxygen consumption and carbon dioxide production kinetics, lower ventilation, and lower HR on submaximal constant-work-rate testing among patients with severe COPD. However, without isowork testing, we cannot prove or disprove such a training effect in our study. This discrepancy between PR-related changes in timed walk distance and V o 2 max in the VL group also could be related to the effects of PR on exercise strategy and effort. Indeed, timed walk distance has been shown to increase up to 5% with practice 14 (a change, however, that is well below our observed 30.3% improvement). More importantly, improvements in timed walk distance may better reflect one s increased ability to perform the activities of daily living than a pure physiologic measure of peak exercise capacity such as V o 2 max. 15 In addition, timed walk distance may have significant prognostic value. One study showed that the 12-min walk distance after the completion of an outpatient PR program was the most important predictor of survival. 16 Therefore, the marked increase in timed walk distance in the VL group may represent an important clinical benefit. This finding supports the continued referral of these VL patients to receive PR. By definition, the NL group did not meet cardiovascular or ventilatory limits on initial testing. This combination of adequate breathing reserve, adequate HR reserve, and a low V o 2 max implies limitation due to poor effort or perhaps musculoskeletal factors. Despite this, the NL group showed the greatest percentage increase in timed walk distance (36%) and V o 2 max (16%). We suspect that this www.chestjournal.org CHEST / 127 / 1/ JANUARY, 2005 115

significant improvement is due in part to the beneficial effects of PR on motivation, exercise strategy, skeletal muscle effects, and desensitization to dyspnea. This is supported by the significant increase in peak exercise HR seen only in this NL group. This study may be limited by the method used to define ventilatory constraint. The peak V e/mvv ratio is used commonly in clinical practice. However, exercise flow-volume loops 17 and the negative expiratory pressure test 18 are other methods that are used to determine the presence of ventilatory limitation. In our study, an inaccurate assessment of ventilatory limitation may have led to an error in assigning the mechanism of exercise limitation. A more precise measure of flow limitation may result in a tighter correlation between the mechanism of exercise limitation and outcome after PR. Further studies are needed to determine the best method for measuring ventilatory limitations. In addition, as mentioned above, this retrospective study is also limited because we did not have measurements of the physiologic response to submaximal, isowork exercise. For this reason, we could not determine with certainty the mechanisms by which PR resulted in an improved timed walk distance and V o 2 max, particularly in the VL group. In summary, we have analyzed the effects of an intensive PR program in patients with non-oxygendependent COPD. We found that the initial mechanism of exercise limitation is an important predictor of response to PR. We showed that CVL patients have a greater mean improvement in V o 2 max than VL patients. However, we also found that marked improvement in timed walk distance occurs regardless of the initial mechanism of exercise limitation. Therefore, this study supports the continued enrollment of all functionally impaired COPD patient groups into PR. References 1 Gallagher CG. Exercise limitation and clinical exercise testing in chronic obstructive pulmonary disease. Clin Chest Med 1994; 15:305 326 2 Maltais F, LeBlanc P, Jobin J, et al. Intensity of training and physiologic adaptation in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1997; 155: 555 561 3 Casaburi R, Patessio A, Ioli F, et al. Reductions in exercise lactic acidosis and ventilation as a result of exercise training in patients with obstructive lung disease. Am Rev Respir Dis 1991; 143:9 18 4 Casaburi R, Porszasz J, Burns MR, et al. Physiologic benefits of exercise training in rehabilitation of patients with severe chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1997; 155:1541 1551 5 Ries AL, Kaplan RM, Limberg TM, et al. Effects of pulmonary rehabilitation on physiologic and psychosocial outcomes in patients with chronic obstructive pulmonary disease. Ann Intern Med 1995; 122:823 832 6 American Thoracic Society. Pulmonary rehabilitation-1999. Am J Respir Crit Care Med 1999; 159:1666 1682 7 Zu Wallack RL, Patel K, Reardon JZ, et al. Predictors of improvement in the 12-minute walking distance following a six-week outpatient pulmonary rehabilitation program. Chest 1991; 99:805 808 8 American Thoracic Society. Standardization of spirometry. Am Rev Respir Dis 1979; 119:831 838 9 American Thoracic Society. Standardization of spirometry: 1987 update. Am Rev Respir Dis 1987; 136:1286 1296 10 American Thoracic Society. Standardization of spirometry: 1994 update. Am J Respir Crit Care Med 1994; 152:1107 1136 11 American Thoracic Society. Single-breath carbon monoxide diffusing capacity (transfer factor): recommendations for a standard technique. Am Rev Respir Dis 1987; 136:1299 1307 12 American Thoracic Society. Single-breath carbon monoxide diffusing capacity (transfer factor): recommendations for a standard technique; 1995 update. Am J Respir Crit Care Med 1995; 152:2185 2198 13 Belman MJ, Kendregan BA. Exercise training fails to increase skeletal muscle enzymes in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1981; 123:256 261 14 Sciurba F, Criner GJ, Lee SM, et al. Six-minute walk distance in chronic obstructive pulmonary disease: reproducibility and effect of walking course layout and length Am J Respir Crit Care Med 2003; 167:1522 1527 15 Carlson DJ. V o 2 max: the gold standard? Chest 1995; 108: 602 603 16 Gerardi DA, Lovett L, Benoit-Connors ML, et al. Variables related to increased mortality following out-patient pulmonary rehabilitation. Eur Respir J 1996; 9:431 435 17 Johnson BD, Beck KC, Zeballos RJ, et al. Advances in pulmonary laboratory testing. Chest 1999; 116:1377 1387 18 Koulouris NG, Dimopoulou I, Valta P, et al. Detection of expiratory flow limitation during exercise in COPD patients. J Appl Physiol 1997; 82:723 731 116 Clinical Investigations