People with moderate to severe COPD trap air in their lungs. This lessens their ability to exercise. The investigators developed a computerized breathing retraining program to help patients empty their lungs during exercise. The investigators believe that learning to control breathing will help patients be more active. The purpose of this study is to test the effects of using breathing retraining (intervention group) versus no breathing retraining (control group) as part of a 12-week exercise program. The investigators are examining the long term effects of this training on exercise capacity and the ability to empty ones lungs during exercise. Patients will participate in a 12-week exercise program, with or without breathing retraining. They will then participate in a 42 week adherence program to maintain physical activity. Patient testing will be competed at baseline, 12 weeks, 6 months and 1 year. 250 patients with moderate to severe COPD will be enrolled.

Rationale: Development of dynamic hyperinflation is a primary limiting factor of exercise tolerance in chronic obstructive pulmonary disease (COPD). To lessen the development of dynamic hyperinflation, and to improve exercise tolerance in COPD patients, the investigators have developed a breathing-retraining technique to be used during pulmonary rehabilitation. This breathing-retraining technique is designed to decrease respiratory rate and prolong exhalation. The investigators demonstrated that, in the short-term, hyperinflation and exercise duration improved more with breathing-retraining plus exercise-training than with exercise-training alone. Because a crucial aspect of pulmonary rehabilitation is the maintenance its short-term benefits over the long-term, the investigators now propose to test whether short-term benefits of breathing-retraining plus exercise followed by an adherence-intervention program are sustainable over the long-term Primary Hypothesis: (H1) In COPD patients, improvements in exercise duration on a constant- load treadmill test will be greater after 12 weeks of breathing-retraining plus exercise-training followed by a 42-week adherence-intervention program (1 yr. total) than after 12 weeks of exercise-training alone followed by a 42-week adherence-intervention program (1 yr. total). Secondary Hypotheses: One year after randomization, exercise-induced dynamic hyperinflation will be less during a constant-load treadmill test (H2), 6-minute walking distance will be longer (H3), and mastery over breathing will be greater (H4) in the breathing-retraining plus exercise-training group than in the exercise-alone group. Lastly (exploratory objective), the investigators will assess the effect of the patient's physiologic, psychologic and clinical phenotype on short- term and long-term responsiveness to pulmonary rehabilitation.

Methods: The proposed study is a randomized, controlled clinical trial in which 250 patients with moderate-to-severe COPD will be enrolled. One hundred forty of these patients are expected to qualify for randomization into the breathing-retraining plus exercise-training group or the exercise-training alone group. Patients will receive 12-weeks of supervised training according to group assignment (three times weekly) followed by an adherence-intervention program (weekly phone motivational interviews, home-exercise program, and monthly laboratory booster sessions). Follow-up testing will be completed at 12 weeks, and at 6 and 12 months. Testing will include pulmonary function test, incremental-load and constant-load treadmill tests, 6-minute walk test, measurements of dyspnea, assessment of respiratory and quadriceps muscle strength and endurance and quadriceps ultrasonography. Analysis: In the principal analysis of the primary outcome measure (exercise duration; H1) the investigators will compare changes in exercise duration (constant work-rate treadmill test) from baseline to end of study using 2-sample t-test (two-tailed =.05). The primary analysis will be based on intention-to-treat principles. Multiple imputation will be used for study subjects missing the 12-month measurement. This imputation model will be based on baseline characteristics of study participants. Several secondary analyses of the primary outcome measure will be performed (H2-H4). Linear regression will be used to determine whether the observed treatment benefits persist after adjustment for baseline covariates and measures of adherence to treatment. Since several measurements will be taken on each patient, mixed-models analysis will be used to compare the changes on the constant-load treadmill test over time between the two groups.

Inclusion Criteria:

• >/= 40 yrs
• FEV1 </= 70% predicted
• FEV1/FVC <70% predicted
• RV/TLC >/=120%
• SpO2 >/= 90% at peak exercise (with or without exercise)
• inspiratory capacity decline >.15L from rest to peak exercise

Exclusion Criteria:

• respiratory infection/exacerbation within the previous 4 weeks
• exercise limiting heart disease
• primary asthma
• congestive heart failure
• exercise limiting peripheral arterial disease
• stops exercise due to arthritic pain in the knee or hips
• inability to walk on the treadmill
• pregnancy
• methadone use
• any unforeseen illness or disability that would preclude exercise testing or training
• participation in a formal exercise program within the previous 12 weeks

Adverse event data were collected over the course of study participation (1 year or the time the patients remained in the study).