Dynamic Hyperinflation and Exercise Intolerance in Chronic Obstructive Pulmonary Disease DENIS E. O’DONNELL, SUSAN M. REVILL, and KATHERINE A. WEBB Respiratory Investigation Unit, Department of Medicine, Queen’s University, Kingston, Ontario, Canada The role of dynamic hyperinflation (DH) in exercise limitation in chronic obstructive pulmonary disease (COPD) remains to be defined. We examined DH during exercise in 105 patients with COPD (FEV1  37  13% predicted; mean  SD) and studied the relationships between resting lung volumes, DH during exercise, and · peak oxygen consumption ( V O2). Patients completed pulmonary function tests and incremental cycle exercise tests. We measured the change in inspiratory capacity (∆IC) during exercise to reflect changes in DH. During exercise, 80% of patients showed significant DH above resting values. IC decreased 0.37  0.39 L or 14  15% predicted during exercise (p  0.0005), but with large variation in range. ∆IC correlated best with resting IC, both expressed · %predicted (r  0.50, p  0.0005). Peak V O2 (%predicted maximum) correlated best with the peak tidal volume attained (VT standardized as % of predicted vital capacity) (r  0.68, p  0.0005), which, in turn, correlated strongly with IC at peak exercise (r  0.79, p  0.0005) or at rest (r  0.75, p  0.0005). The extent of DH during exercise in COPD correlated best with resting IC. DH curtailed the VT response to exercise. This inability to expand VT in response to increasing metabolic demand contributed importantly to exercise intolerance in COPD. Keywords: COPD; exercise; inspiratory capacity; dynamic lung hyperinflation; emphysema; dyspnea Chronic obstructive pulmonary disease (COPD) is a heterogeneous disorder characterized by dysfunction of the small and large airways, as well as destruction of the lung parenchyma and its vasculature in highly variable combinations. The pathophysiological hallmark of COPD is expiratory flow limitation, which, in more advanced disease, occurs even during resting quiet breathing. As a consequence, resting lung volume (functional residual capacity [FRC]) is dynamically, and not statically, determined. During exercise, as ventilatory demands increase in flow-limited patients, progressive air trapping and further dynamic lung hyperinflation (DH) above already increased resting values is inevitable (1, 2). Recent studies have shown that DH during exercise contributes to perceived respiratory discomfort (3, 4). Indirect evidence of the importance of DH comes from studies that have demonstrated that alleviation of dyspnea following bronchodilator therapy and lung volume reduction surgery (LVRS) was explained, in part, by reduced operating lung volumes (5, 6). However, it is not clear from previous studies to what extent the behavior of operating lung volumes during exercise influences peak exercise capacity in COPD. Moreover, earlier studies have shown wide variability in the extent of DH with (Received in original form December 26, 2000 and in revised form May 8, 2001) Supported by the Ontario Thoracic Society and the Ontario Ministry of Health. Correspondence and requests for reprints should be addressed to Dr. Denis O’Donnell, Richardson House, 102 Stuart Street, c/o Kingston General Hospital, Kingston, ON, K7L 2V7 Canada. E-mail: odonnell@post.queensu.ca This article has an online data supplement, which is accessible from this issue’s table of contents online at www.atsjournals.org Am J Respir Crit Care Med Vol 164. pp 770–777, 2001 Internet address: www.atsjournals.org exercise and the factors that determine this variability have not been elucidated (3–7). We hypothesized that DH and the consequent restrictive constraints on volume expansion during exercise would contribute importantly to reduced exercise performance in COPD. Although volume constraints are, by no means, an exclusive source of ventilatory limitation in COPD, they are likely to be important. They contribute to exertional dyspnea and influence breathing pattern responses during exercise. Furthermore, the operating lung volumes determine, in part, the magnitude of fractional inspiratory muscle force generation (relative to maximum). High inflation volumes may also affect cardiac performance and, thus, peripheral muscle function during exercise in COPD. Therefore, the objectives of this study were (1) to determine the range and pattern of change in the various operating lung volume components during incremental exercise in a large COPD population; (2) to examine factors contributing to the intersubject variability in DH during exercise; (3) to examine the relationship between resting hyperinflation, further · DH during exercise, and symptom limited peak V O2; and (4) to compare operating lung volumes and exercise performance in a subgroup of patients with a more “emphysematous” clinical profile with patients who were matched for FEV1 but with a better preserved diffusion capacity. We conducted incremental cardiopulmonary cycle exercise testing in 105 clinically stable patients with COPD and 25 healthy age-matched control subjects. We measured and compared ventilation, breathing pattern, operating lung volumes, metabolic factors, and exertional symptoms. We evaluated dynamic changes in end-expiratory lung volume (EELV) from resting FRC by collecting serial inspiratory capacity (IC) measurements throughout exercise, having established the reliability of this measurement in a previous study (7). METHODS Subjects We studied 105 clinically stable patients with COPD (FEV1  70% predicted, FEV1/FVC  70%). Exclusion criteria included a history of asthma, atopy, or nasal polyps; other active lung disease; significant disease that could contribute to dyspnea or exercise limitation; and oxygen desaturation to  75% during exercise on room air. Twentyfive age-matched ( 50 yr), healthy subjects were also studied. Study Design COPD subjects included patients who had performed pulmonary function tests (PFTs) and an incremental cycle exercise test during assessment before pulmonary rehabilitation or as part of screening prior to entering various clinical research studies. Healthy normal subjects were recruited from the local community to perform spirometry and incremental cycle exercise tests for comparison of the behavior of operational lung volumes during exercise. All subjects signed written informed consent at the time of their first assessments and were aware that their test data might be used in future analyses. Subjects were familiarized with all procedures prior to collection of the test results evaluated in this study. 771 O’Donnell, Revill, and Webb: Exercise and Hyperinflation in COPD Procedures Spirometry, body plethysmography, single-breath diffusing capacity (DLCO), and maximal inspiratory mouth occlusion pressure (MIP) were performed as previously described (8) (see online data supplement). Chronic dyspnea was assessed using the modified Baseline Dyspnea Index (9). Symptom-limited incremental cycle exercise tests were conducted as previously described (8) (see online data supplement). Subjects breathed through a mouthpiece with noseclips in place. In the majority of subjects (COPD n  74, normal subjects n  20), flow signals were sampled at 100 Hz using computer-based data acquisition software (CODAS; Dataq Instruments Inc., Akron, OH), from which breath-by-breath measurements of volume, flow, and timing were calculated. In these subjects, expired air channelled through a 10-L mixing chamber was analyzed for fraction of O2 (S-3A Oyxgen Analyzer; Applied Electrochemistry, Pasadena, CA) and CO2 (LB-2 Gas Analyzer; SensorMedics, Anaheim, CA). In all remaining subjects, breathby-breath measurements were collected using a Vmax229d Cardiopulmonary Exercise Testing Instrument (SensorMedics, Yorba Linda, CA). Electrocardiography and pulse oximetry were monitored continuously. Blood pressure was auscultated at rest, each stage of exercise, peak exercise, and recovery. The modified Borg Scale (10) was used to rate intensity of dyspnea (i.e., “breathing discomfort”) and leg discomfort at rest, each stage of exercise, and peak exercise. Subjects also specified why they stopped exercise. Operational lung volumes: (see online data supplement). Changes in operational lung volumes were derived from measurements of dynamic inspiratory capacity (IC), assuming that total lung capacity (TLC) remained constant during exercise (11, 12). This has been found to be a reliable method of tracking acute changes in lung volume (7, 11–13). Techniques for performing and accepting IC measurements have been previously reported (2, 3). IC was measured at the end of a steady-state resting baseline, at 2 min intervals during exercise, and at end exercise. Statistical Analysis Results are means  SD. COPD subgroup comparisons using unpaired Student’s t tests included (1) patients who had an emphysematous profile with DLCO  50% predicted and FRC  130% predicted (Group A), versus patients with DLCO  50% predicted and FRC  130% predicted (Group B); (2) patients stopping primarily due to breathing discomfort versus leg discomfort; and (3) patients with different patterns of DH with exercise. Relationships between exercise capacity, exertional dyspnea, and operational lung volumes in COPD were evaluated using Pearson’s correlations. Stepwise multiple regression analysis established the best · predictive equations for peak V O2, Borg ratings of dyspnea, and DH (dependent variables). Independent variables included standardized · exercise measurements of minute ventilation (V E), breathing pattern · · (F, VT, TI, TE, TI/Ttot), gas exchange ( V CO2/ V O2, SpO2), volume constraints (IC, IRV, VT/IC, EELV, EILV), and DH (change in IC), as well as resting pulmonary function and lung volume measurements (expressed as % of predicted normal). RESULTS Subject Characteristics Subject characteristics are summarized in Table 1. In the COPD group as a whole, there was a wide range of airflow obstruction (FEV1 from 12 to 68% predicted), lung hyperinflation (plethysmographic FRC from 94 to 307% predicted), diffusing capacity (DLCO from 16 to 121% predicted), and chronic activity-related dyspnea (modified Baseline Dyspnea Index focal scores from 2 “very severe” to 9 “mild”). The healthy control subjects had normal spirometry and were well matched for age, sex, and body mass index. The majority of patients with COPD (80%) stopped exercise due to severe breathing discomfort, either alone or in combination· with leg discomfort, at a low peak oxygen consumption ( V O2) (Table 2, Figure 1). In contrast, the majority of normal subjects (76%) stopped exercise primarily because of leg discomfort. Compared with normal subjects during exercise, patients with COPD had significantly greater ventila· · tory slopes ( V E/ V CO2) and reduced ·ventilatory reserve at end exercise. In this latter regard, peak V E expressed as a percentage of maximal ventilatory capacity (MVC estimated as 40 FEV1) was 92  31% versus 64  22% in patients with COPD and normal subjects, respectively (p  0.0005). Also of note, the exercise breathing pattern was significantly more rapid and shallow in patients with COPD than in normal subjects. The patients with COPD who stopped exercise primarily due to breathing discomfort (n  64) had significantly greater resting airflow limitation (i.e., decreased FEV1) and thoracic hyperinflation (i.e., increased FRC with reduced IC) than those who stopped primarily due to leg discomfort (n  19) (Table 3). Those limited by dyspnea also had greater impairment in dynamic mechanics during exercise, that is, significantly reduced IC and IRV, increased EILV/TLC and VT/IC, and less VT expansion during exercise (Table 3). Measurements of Operational Lung Volumes There are no current equations for predicting normal spirometric IC values, therefore, a predicted normal value for IC was calculated as predicted TLC minus predicted FRC. In our normal sample, mean resting IC was 3.11  1.13 L or 110  32% predicted; the latter value indicates that this method of calculating a predicted normal value for IC was reasonable, or possibly an underestimation, in this older population. In the COPD sample, mean resting IC was significantly reduced at 1.89  0.72 L or 69  23% predicted, with measurements as TABLE 1. SUBJECT CHARACTERISTICS* COPD (n  105) Normal Subjects (n  25) p Value† Male %: female % Age, yr Height, cm Weight, kg Body mass index, kg/m2 Baseline dyspnea index 64:36 66  8 167  91 72.6  18.6 25.8  5.7 5.0  1.5 60:40 63  7 70  10 76.3  14.7 26.4  4.0 11.9  0.6 NS NS NS NS NS  0.0005 Pulmonary function FEV1, L %predicted FVC, L %predicted FEV1/FVC, % %predicted PEFR, L/s %predicted FEF50%, L/s %predicted MIP, cm H2O %predicted TLC, L %predicted FRC, L %predicted RV, L %predicted DLCO, ml/min/mm Hg %predicted 0.94  0.40 37  13 2.18  0.75 60  17 43  10 62  14 2.98  1.04 43  12 0.43  0.27 11  6 54  31 65  35 7.21  1.67 122  20 5.53  1.55 174  43 4.71  1.49 219  71 11.4  5.3 57  21 2.85  0.85 106  16 3.81  1.15 100  16 75  6 106  9 6.96  2.46 94  21 3.55  1.49 85  33 79  40 94  38 ND  0.0005  0.0005  0.0005  0.0005  0.0005  0.0005 ND ND ND Definition of abbreviations: COPD  chronic obstructive pulmonary disease; DLCO  diffusing capacity of the lung for CO; FEF  forced expiratory flow; FRC  functional residual capacity; MIP  maximum inspiratory pressure; ND  not done in the normal group; NS  not significantly different; PEFR  peak expiratory flow rate; RV  residual volume; TLC  total lung capacity. * Values are mean  SD. Normal predicted values for spirometry, maximal expiratory flow rates, lung volumes, diffusing capacity, and inspiratory muscle strength are from references 14 through 18, respectively. † Comparisons were made for demographics and %predicted values only. 772 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 164 2001 TABLE 2. SYMPTOM-LIMITED INCREMENTAL CYCLE EXERCISE Reason for stopping exercise (% of Group) Breathing discomfort Leg discomfort Both breathing and legs Other§ · V O2, ml/kg/min Heart rate, beats/min SaO2, % · V E, L/min · · V E/ V CO2, % F, breaths/min TI/Ttot VT, L VT, %predicted VC VT/IC, % IC, L IC, %predicted ∆IC from rest, L ∆IC from rest, %predicted IRV, L IRV, %predicted TLC EILV/TLC, % EILV, %predicted TLC COPD Peak Exercise (n  105) Normal Subjects Peak Exercise (n  25) Normals at a VE Similar to Peak COPD (n  25) 61 18 19 2 12.6  5.0 69  10 91  6 33.1  14.6 41.2  9.6 30.3  6.5 0.36  0.06 1.10  0.44 31  10 74  14 1.52  0.06 55  2 0.37  0.39 14  15 0.42  0.33 76 94  5 115  22 8* 76* 12 4 31.3  13.7* 83  11 96  2* 73.9  35.8* 32.7  3.8* 30.7  6.0 0.48  0.04* 2.41  1.04* 63  19* 74  15 3.26  1.30* 114  34* 0.17  0.46* 4  14* 0.85  0.59* 14  9* ND ND 15.9  3.6† 63  10 97  2* 32.4  4.9 34.3  7.4† 22.8  5.4* 0.45  0.06* 1.48  0.33* 40  9* 52  4* 3.23  1.29* 112  34* 0.13  0.51* 3  16* 1.75  1.16* 28  17* ND ND Definition of abbreviations: COPD  chronic obstructive pulmonary disease; EILV  end-inspiratory lung volume; F  respiratory frequency; IC  inspiratory capacity; IRV  inspiratory reserve volume; ND  not done; SaO2  arterial oxygen saturation; TI/Ttot  respira· · tory duty cycle; TLC  total lung capacity; V E  minute ventilation; VT  tidal volume; V O2  oxygen consumption. * p  0.0005, †p  0.01. Significant difference from COPD. § Other in COPD was due to general fatigue (n  1) and discomfort from noseclips (n  1); other in normal subjects was discomfort on the bicycle seat (n  1).  Values are mean  SD. low as 0.74 L or 23% predicted. The 95% CI for resting IC measurements was  0.14 L or  4.5% predicted within the COPD group, indicating that a reproducibility criteria of within 150 ml, or approximately 10%, may be appropriate for testing IC in this population. The 95% CI for peak IC was similar at  0.12 L or  3.9% predicted. During exercise in COPD, IC decreased significantly by 0.37  0.39 L (p  0.0005), with the change (∆) in IC ranging between 1.42 and 0.77 L (Figures 1 and 2): this corresponds to a mean ∆IC of 18  19% or 14  15% predicted. On average, the reduction· in IC occurred progressively throughout exercise, with ∆IC/∆·V E during the first 2–3 min of exercise matching the ∆IC/∆ V E in the later stages of exercise. In contrast to COPD, there was no significant change in IC from rest to peak exercise in the normal group (∆IC  0.17  0.46 L or 4  14% predicted), although ∆IC ranged between 0.56 and 1.14 L (Figures 1 and 2). Whereas 80% of patients with COPD significantly decreased IC during exercise (i.e., outside the 95% confidence limits or  4.5% predicted), the majority of our older normal subjects either increased (40% of subjects) or did not change (40% of subjects) IC during exercise. In COPD, the VT response to exercise was limited from both above (i.e., the TLC envelope) and below (i.e., due to a reduced IC, which decreased even further as ventilation increased) (Figure 3). At the peak of symptom-limited exercise, patients breathed with a tidal end-inspiratory lung volume (EILV) that approached, but never quite reached, their TLC. We defined this upper volume boundary as the “minimal IRV” that could be achieved during exercise, and set its level at the lower 95% confidence limit for peak IRV in this COPD group (i.e., 0.35 L or 5.9% of the predicted TLC). Over half of our patients with COPD reached a “minimal IRV”  5.9% predicted TLC (n  56), with 15 of these patients still having apparent ventilatory reserve by traditional estimates (i.e., · peak V E/MVC  75%). Volume constraints on VT expansion were significantly less in normal subjects than in patients with COPD at a standard· ized V E during exercise (Table 2). Even at the end of exercise Figure 1. Inspiratory capacity (IC) measurements expressed as liters or as a % of predicted normal in 105 patients with COPD and a healthy age-matched normal control group (n  25) for a given ventilation during exercise. Predicted normal IC was calculated as predicted normal TLC minus predicted normal FRC. Values shown are means (solid lines)  95% confidence interval (dotted lines). 773 O’Donnell, Revill, and Webb: Exercise and Hyperinflation in COPD TABLE 3. COPD SUBGROUPS: (1) GROUP A VERSUS GROUP B, (2) DYSPNEA VERSUS LEG LIMITED Group B (n  24) Limited by Dyspnea (n  64) Limited by Leg Discomfort (n  19) 50:50 65  7§ 26.4  4.2 4.8  1.0 71:29 66  9 28.1  4.9 5.4  1.6 67:33 66  8 24.9  5.4 4.7  1.3 53:47 66  8 26.3  4.8 5.3  1.7 38  11 65  17 58  9 94 126  15* 182  38* 56  15* 38  8† 38  12 60  16 63  10 11  4 115  14 157  31 77  34 73  4 35  12* 59  17 59  14 10  6 124  21 184  43* 63  37 54  19 42  12 64  18 66  8 11  4 116  20 157  38 68  24 61  25 46 29 21 4 5.2  1.6 4.8  2.0 13.7  5.3 62  24 71  12 91  5 33.7  12.4 36.6  10.4 30.1  6.0 0.38  0.06 31  9 71  15 60  21 0.32  0.28 11  9 0.52  0.39 97 92  8 106  18 100† 0† 0 0 5.5  1.6‡ 3.5  1.8† 12.4  5.3 56  23 69  9 89  6* 32.0  14.7 42.9  8.4 31.1  7.0 0.35  0.05‡ 29  9 75  12 50  17‡ 0.35  0.42 13  16 0.36  0.26* 6  5* 95  4‡ 119  22* 0 100 0 0 4.3  1.5 6.6  1.7 12.6  3.4 57  20 70  11 93  4 32.6  8.4 38.8  7.8 28.4  4.3 0.39  0.04 33  9 70  16 64  19 0.45  0.29 17  13 0.55  0.42 98 91  9 107  24 Group A (n  24) Male %:Female % Age, yr Body mass index, kg/m2 Baseline dyspnea index Pulmonary function FEV1, %predicted FVC, %predicted FEV1/FVC, %predicted FEF50, %predicted TLC, %predicted FRC, %predicted MIP, %predicted DLCO, %predicted Peak exercise Reason for stopping exercise (% of group) Breathing discomfort Leg discomfort Both breathing and legs Other Dyspnea, Borg Leg discomfort, Borg · V O2, ml/kg/min · V O2, %predicted max Heart rate, %predicted max SaO2, % · V E, L/min · · V E/ V CO2, % F, breaths/min TI/Ttot VT, %predicted VC VT/IC, % IC, %predicted ∆IC rest-to-peak, L ∆IC rest-to-peak, %predicted IRV, L IRV, %predicted TLC EILV/TLC, % EILV, %predicted TLC 75* 13 8 4 5.1  1.6 3.4  2.3 10.6  3.3* 51  18 66  9 90  4 32.0  13.8 42.0  6.5* 30.4  7.0 0.36  0.06 31  9 77  10 54  19 0.43  0.36 17  13 p0.07 0.34  0.24 64 95  3 119  17‡ Definition of abbreviations: COPD  chronic obstructive pulmonary disease; DLCO  diffusing capacity of the lung for CO; EILV  end-inspiratory lung volume; F  respiratory frequency; FRC  functional residual capacity; IC  inspiratory capacity; IRV  inspiratory reserve volume; MIP  maximum inspiratory pressure; SaO2  arterial oxygen saturation; TI/Ttot  respiratory duty cycle; TLC  total lung capac· · ity; V E  minute ventilation; V O2  oxygen consumption; VT  tidal volume. * p  0.05, †p  0.0005, ‡p  0.01. Significant difference between subgroups (A versus B, dyspnea versus leg discomfort). § Values are mean  SD.  Other in Group A was discomfort from noseclips (n  1); Other in Group B was general fatigue (n  1). in normals, IRV did not reach the same minimal level as it did in COPD (Table 2 and Figure 3). Increased versus No Change in DH during Exercise in COPD Of the 84 patients who decreased IC during exercise outside the 95% CI at rest, 62 decreased IC by at least 10% predicted. This latter subgroup (DH subgroup) was compared with the subgroup of 14 patients who did not change IC during exercise (∆IC within  4.5% predicted). These subgroups had similar mean baseline FEV1 %predicted, FRC %predicted, and DLCO %predicted. Although both subgroups reached a similar peak IC %predicted (and VT/%predicted VC), patients who did not change IC during exercise tended to have greater volume constraints at rest, that is, smaller IC (p  0.06) and IRV (p  0.08). Reduced DLCO (Group A) versus Preserved DLCO (Group B) As selected, Group A (n  24) had significantly greater baseline lung hyperinflation and a greater reduction in diffusing capacity than Group B (n  24) (Table 3). These subgroups were well matched for age, sex, height, and body mass index, but Group A had greater exercise impairment due to exer- tional dyspnea than Group B (Figure 4). Although the overall extent of change in IC during exercise was similar in both subgroups, Group A had a significantly greater rate of DH, which occurred in the early stages of exercise, than Group B. Therefore, Group A had an earlier attainment of a limiting mechanical restriction (i.e., minimal IRV) resulting in a reduced peak · V O2 (Table 3 and Figure 4). Correlates of Dynamic Hyperinflation in COPD The total extent of change in IC (%predicted) during exercise was determined primarily by resting volume constraints, that is, IC expressed as %predicted (r  0.503, p  0.0005) and IRV expressed as % of predicted TLC (r  0.497, p  0.0005). By stepwise multiple regression analysis, the FEF50% and DLCO (both expressed as %predicted) added an additional 8% to the variance in ∆IC %predicted (p  0.05 each). The rate of· change in IC during exercise (slope of IC %predicted over V O2 %predicted maximum) correlated best with DL/VA %predicted (r  0.412, p  0.005). Comparison of subgroups best illustrates this model: the subgroup with a reduced DLCO had a significantly faster rate of DH, occurring early in 774 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 164 2001 mined primarily by the peak IC (r  0.791, p  0.0005) (Figure 5) or the resting IC (r  0.745, p  0.0005), both expressed as %predicted. As seen in Figure 5, the relationship between peak VT and peak IC was strong in the 85 patients with an IC  70% predicted (r  0.866, p  0.0005), but was not significant within the 20 patients with a preserved IC (r  0.273, p  0.244). Finally, an index of the mechanical constraints on tidal volume expansion (VT/IC) as exercise progressed was the best correlate of · concurrent estimates of the level of ventilatory limitation ( V E/MVC), and accounted for 43% of its variance after accounting for repeated measurements within patients (p  0.0005). Correlates of Exertional Dyspnea in COPD Figure 2. The distribution of the extent of change (∆) in IC during exercise is shown in patients with COPD (n  105) and in age-matched normal subjects (n  25). A negative ∆IC reflects dynamic hyperinflation (DH) during exercise; each bar width corresponds to a ∆IC range of 0.10 L. In contrast to normal subjects, the majority of patients with COPD experienced significant DH during exercise despite reaching a much lower peak ventilation, that is, 33 versus 64 L/min in patients with COPD and normal subjects, respectively. exercise, than those with a preserved DLCO (Group A versus B, p  0.05) (Figure 4). Correlates of Exercise Capacity in COPD · In COPD, the best physiological correlate of peak V O2 (expressed %predicted maximum) was the peak VT (standardized as %predicted VC) (r  0.682, p  0.0005) (Figure 5 and · Table 4). By stepwise multiple regression analysis, peak V O2 %predicted was best described by the combination of peak · · VT/%predicted VC, peak F, and the slope of V E/ V O2 %predicted (r2  0.816, p  0.0005). Within each of the COPD subgroups (see above), peak VT continued to be the best correlate · of peak V O2 (p  0.0005 each). In turn, peak VT was deter- The strongest correlate of exertional dyspnea intensity was an index of the concurrent constraints on tidal volume: for all points during exercise, after accounting for repeated measurements within patients, the VT/IC ratio accounted for 32% (p  0.0005) of the variance in concurrent Borg dyspnea· ratings. Less important contributing variables included V E/MVC, breathing frequency, and IRV/predicted TLC, each accounting for 25% of the variance in Borg dyspnea ratings (p  0.0005). DISCUSSION The novel findings of this study are as follows. (1) Although the pattern and magnitude of DH was variable among COPD patients during exercise, the majority (80%) demonstrated significant dynamic increases in lung volumes above resting values. (2) The extent of DH during exercise varied inversely with the level of resting hyperinflation. (3) For a given level of airway obstruction, patients with a more emphysematous clinical profile (low DLCO) had faster rates of DH, greater constraints on tidal volume expansion during exercise, greater dyspnea, · and a lower peak V O2. (4) Finally, there was a clear statistical association between the level of resting and dynamic hyperinflation, the degree of tidal volume restriction (i.e., peak VT) during exercise, and peak exercise performance. Operational Lung Volumes during Exercise in COPD Serial IC measurements have been used to track dynamic EELV during exercise for more than 30 yr (2, 3, 11, 21–23). This approach is based on the assumption that TLC does not change appreciably during exercise in COPD, and that reduc- Figure 3. Changes in operational lung volumes are shown as ventilation increases with exercise in patients with COPD and in normal subjects. “Restrictive” constraints on tidal volume (VT, solid area) expansion during exercise are significantly greater in the COPD group from both below (reduced IC) and above (minimal IRV, open area). 775 O’Donnell, Revill, and Webb: Exercise and Hyperinflation in COPD Figure 4. Ventilatory responses to exercise are shown in COPD (n  105) and its subgroups: A with a low DLCO  50% predicted (n  24), and B with a better preserved DLCO  50% predicted (n  24). Group A had significantly (p  0.05) greater exertional dyspnea, greater levels of lung hyperinflation, and earlier attainment of a limiting mechanical restriction (i.e., minimal IRV, shaded area) than Group B. tions in dynamic IC must therefore reflect increases in dynamic EELV (or FRC) (11). However, regardless of any possible changes in TLC with exercise, progressive reduction of an already diminished resting IC means that VT becomes positioned closer to the actual TLC and the upper alinear extreme of the respiratory system’s pressure–volume relationship, where there is increased elastic loading of the inspiratory muscles (1). The reduction of IC as exercise progresses is likely a true reflection of shifts in EELV, rather than simply the inability to generate maximum effort because of dyspnea or functional muscle weakness. In fact, several studies have established that dyspneic patients, even at the end of exhaustive exercise, are capable of generating maximal inspiratory efforts as assessed by peak inspiratory esophageal pressures (4, 13). Moreover, we have recently shown that exercise IC measurements are both highly reproducible and responsive in patients with severe COPD, provided due attention is taken with their measurement (7). The extent of DH from rest to the peak of exercise was variable between patients (Figure 2). The change in IC correlated best with the resting IC (or IRV): those with the greatest IC (or IRV) reduction at rest tended to have smaller changes in both EELV and EILV with exercise (Figures 1 and 2). After accounting for the resting IC, the maximal mid expiratory flow rate also contributed to the variance in DH: those with higher expiratory flows available over the tidal volume operating ranges tended to have less DH. Not surprisingly, this was a weak correlation because the measurement of forced expiratory flow rates, which is prone to measurement artifact (gas and airway compression effects), is a very crude index of the extent of expiratory flow limitation, which is likely the crucial determinant of DH during exercise in COPD. Pattern and Magnitude of DH in COPD Independent Variables · Slope of dyspnea/ V O2, Borg/%predicted max Peak exercise measurements Peak VT, %predicted VC Peak IC, %predicted Peak VT/IC, % · Peak V E, L/min · Peak V E, %MVC Peak F, breaths/min Peak heart rate, %predicted Peak SpO2, % Resting measurements IC, %predicted IRV, %predicted TLC FEV1, %predicted FEV1/FVC, %predicted VC, %predicted FRC, %predicted TLC, %predicted DLCO, %predicted DLCO/VA, %predicted MIP, %predicted This is the first large study to document dynamic volume components during incremental exercise in COPD. Although, in absolute terms, a mean reduction in IC of 0.37 L seems small, this represents a significant further reduction of an already diminished baseline value. In the COPD group, the mean change in IC with exercise was well beyond the within-group 95% confidence interval for the resting IC measurement (i.e.,  0.14 L or  4.5% of the predicted normal value) (Figure 2). Figure 5. In· COPD (n  105), the best correlate of peak oxygen consumption ( V O2) was the peak tidal volume attained (VT standardized as %predicted vital capacity). In turn, the strongest correlate of peak VT was the peak inspiratory capacity (IC). TABLE 4. CORRELATIONS WITH SYMPTOM-LIMITED PEAK · V O2 (%PREDICTED MAXIMUM) IN COPD (n  105) Pearson’s Correlation Coefficient (r) p Value 0.628  0.0005 0.682 0.446 0.286 0.427 0.314 0.057 0.364 0.255  0.0005  0.0005 0.004  0.0005 0.001 0.576  0.0005 0.010 0.451 0.325 0.453 0.147 0.399 0.271 0.104 0.437 0.264 0.196  0.0005 0.001  0.0005 0.145  0.0005 0.006 0.304  0.0005 0.017 0.068 Definition of abbreviations: COPD  chronic obstructive pulmonary disease; DLCO  diffusing capacity of the lung for CO; F  respiratory frequency; FRC  functional residual capacity; IC  inspiratory capacity; IRV  inspiratory reserve volume; MIP  maximum inspiratory pressure; MVC  estimated maximum ventilatory capacity, i.e., 40 FEV1; SpO2  pulmonary oxygen saturation; TLC  total lung capacity; VA  alve· · olar volume; VC  vital capacity; V E  minute ventilation; V O2  oxygen consumption; VT  tidal volume. 776 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE Of interest, the rate of change in DH correlated inversely with resting diffusion capacity (DLCO/VA). Patients with a lower DLCO would be expected to have a greater propensity to expiratory flow limitation because of reduced lung recoil and airway tethering effects. We have previously reported that patients with COPD and a lower DLCO (an average of 32% predicted) had greater chronic activity-related dyspnea and poorer exercise performance than patients with a similar FEV1 but with an average DLCO of 65% predicted (24). In this study we further explored the mechanistic link between low DLCO and poor exercise performance. Thus, patients with a lower DLCO (Group A) had greater resting hyperinflation, greater rates of DH at lower exercise levels, greater exertional dyspnea, earlier attainment of critical volume constraints, accelerated · · breathing frequency, and a lower peak V E and peak V O2 than patients with a better preserved DLCO (Group B) (Figure 3). Operational Lung Volumes and Exercise Intolerance Several recent studies have confirmed that exercise intolerance in COPD is multifactorial and ultimately reflects integrated abnormalities of the ventilatory, cardiovascular, peripheral muscle, and metabolic systems in variable combinations (18, 25–32). In our study patients in whom dyspnea was the main symptom limiting exercise (61% of patients), ventilatory factors played a predominant role in exercise curtailment. Those who stopped primarily because of leg discomfort had significantly less ventilatory constraints at peak exercise. A recent study by Diaz and coworkers (25) has shown that the resting· IC %predicted correlated well with symptom-limited peak V O2. The present study extends these findings to highlight the importance of ventilatory restriction during exercise in flow-limited patients. Compared with age-matched healthy control subjects at a similar low level of ventilation, IC and IRV were markedly diminished in the COPD group. At this point of comparison, VT/IC ratios in COPD and in health were 74% and 52%, respectively (Table 2). These patients with COPD, therefore, had a very limited ability to further expand VT in the face of the increasing metabolic demand of continued exercise. The resting IC (not the VC) and, in particular, the dynamic IC with exercise, represent the true operating limits for VT expansion in any given patient. When the VT during exercise approximated the peak IC, or the dynamic EILV encroached on the TLC envelope, further volume expansion was impossible, even if it were possible to further increase inspiratory muscle effort. In a· multiple regression analysis with symptom-limited peak V O2 as the dependent variable, and several relevant physiological measurements as independent variables, includ· ing FEV1, FEV1/FVC ratio, and V E/MVC, peak VT emerged as the strongest contributory variable, explaining 47% of the variance. Peak VT, in turn, correlated strongly with both the resting and peak dynamic IC. It is noteworthy that this correlation was particularly strong (r  0.9) in approximately 80% of the sample who had a diminished peak IC (i.e.,  70% predicted). Similarly, the intensity of breathlessness throughout exercise correlated better with concurrent measurements of VT/IC (p  0.0005) than any other ventilatory variable. We can conclude, therefore, that volume constraints contribute importantly to both exercise intolerance and dyspnea in patients with COPD. Previous small studies provide a basis for a possible mechanistic link between lung hyperinflation, volume restriction, and exercise intolerance in COPD (2–7, 21–23). The greater the increases in dynamic lung volume components during exercise, the greater the elastic and threshold loads on inspiratory muscles already burdened with increased resistive work. VOL 164 2001 With progressive DH, inspiratory muscles become functionally weakened. This combination of increased loading and reduced strength means that the inspiratory muscles are operating at a high fraction of their maximal force-generating capacity during tidal breathing. The constrained VT response means a greater reliance on tachypnea to increase ventilation, but this rebounds to further aggravate DH in a vicious cycle. Progressive volume restriction in the face of increasing inspiratory effort during exercise ultimately reflects neuromechanical uncoupling of the respiratory pump, which, in turn, may contribute to the quality and intensity of exertional dyspnea that these patients experience (4). The reduced peak VT, as a result of a reduced IC, has similarly been shown to correlate strongly with poor exercise performance in patients with ventilatory restriction due to interstitial lung disease (33), as well as in normal healthy subjects where chest wall restriction was imposed (34). The contention that restrictive mechanics, secondary to lung hyperinflation, contribute to exercise intolerance in severe COPD is bolstered by recent interventive studies (i.e., bronchodilators and oxygen therapy) that show that reduction of resting and/or exercise lung volumes improves exercise endurance in severe COPD (5, 7, 35). · Traditionally, assessment of breathing reserve (i.e., 1  V E/ MVC ratio) has been used to assess ventilatory limitation to exercise in COPD. This study shows that additional measurements of dynamic lung volumes during exercise provide insights into the nature of the critical ventilatory constraints on exercise performance. There was a strong correlation between · the V E/MVC and the VT/IC ratio during exercise (p  0.0005). However, a full 14% of patients with apparent ventilatory re· serve at peak exercise (i.e., V E/MVC  75%) had coexisting limiting restrictive ventilatory constraints as indicated by an EILV of 96% of TLC (i.e., a significantly reduced peak IRV) at the same time point. In summary, the inability to further expand VT in response to the increased respiratory drive of exercise contributes importantly to exercise intolerance in patients with moderate to severe COPD. The main clinical implication of our findings is that exercise performance and dyspnea should be improved by therapeutic interventions that reduce operational lung volumes at rest and during exercise in severe COPD. Measurement of IC and its derived volume components during exercise complement the traditional assessments of ventilatory constraints and can provide additional insight into the impairment–disability interface in patients with COPD. References 1. Pride NB, Macklem PT. Lung mechanics in disease. In: AP Fishman, editor. Handbook of physiology, Section 3, Volume III, Part 2: The respiratory system. Bethesda MD: American Physiological Society; 1986. p. 659–692. 2. Hyatt RE. Expiratory flow limitation. J Appl Physiol 1983;55:1–8. 3. O’Donnell DE, Webb KA. Exertional breathlessness in patients with chronic airflow limitation: the role of lung hyperinflation. Am Rev Respir Dis 1993;148:1351–1357. 4. O’Donnell DE, Chau LKL, Bertley JC, Webb KA. Qualitative aspects of exertional breathlessness in chronic airflow limitation: pathophysiologic mechanisms. Am J Respir Crit Care Med 1997;155:109–115. 5. Belman MJ, Botnick WC, Shin JW. Inhaled bronchodilators reduce dynamic hyperinflation during exercise in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1996;153:967–975. 6. Martinez FJ, Montes de Oca M, Whyte RI, Stetz J, Gay SE, Celli BR. Lungvolume reduction improves dyspnea, dynamic hyperinflation and respiratory muscle function. Am J Respir Crit Care Med 1997;155:1984–1990. 7. O’Donnell DE, Lam M, Webb KA. Measurement of symptoms, lung hyperinflation and endurance during exercise in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998;158:1557–1565. O’Donnell, Revill, and Webb: Exercise and Hyperinflation in COPD 777 8. O’Donnell DE, D’Arsigny C, Raj S, Abdollah H, Webb KA. Ventilatory assistance improves exercise endurance in stable congestive heart failure. Am J Respir Crit Care Med 1999;60:1804–1811. 9. Stoller JK, Ferranti R, Feinstein AR. Further specification and evaluation of a new clinical index for dyspnea. Am Rev Respir Dis 1986; 135: 1129–1134. 10. Borg GAV. Psychophysical basis of perceived exertion. Med Sci Sports Exerc 1982;14:377–381. 11. Stubbing DG, Pengelly LD, Morse JLC, Jones NL. Pulmonary mechanics during exercise in subjects with chronic airflow obstruction. J Appl Physiol 1980;49:511–515. 12. Stubbing DG, Pengelly LD, Morse JLC, Jones NL. Pulmonary mechanics during exercise in normal males. J Appl Physiol 1980;49:506–510. 13. Yan S, Kaminski D, Sliwinski P. Reliability of inspiratory capacity for estimating end-expiratory lung volume changes during exercise in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1997;156:55–59. 14. Morris JF, Koski A, Temple WP, Claremont A, Thomas DR. Fifteen year interval spirometric evaluation of the Oregon predictive equations. Chest 1988;93:123–127. 15. Knudson RJ, Lebowitz MD, Holberg CJ, Burrows B. Changes in the normal maximal expiratory flow-volume curve with growth and aging. Am Rev Respir Dis 1983;127:725–734. 16. Crapo RA, Morris AH, Clayton PD, Nixon CR. Lung volumes in healthy nonsmoking adults. Bull Eur Physiopathol Respir 1982;18:419–425. 17. Burrows B, Kasik JE, Niden AH, Barclay WR. Clinical usefulness of the single-breath pulmonary diffusing capacity test. Am Rev Respir Dis 1961;84:789–806. 18. Hamilton AL, Killian KJ, Summers E, Jones NL. Muscle strength, symptom intensity, and exercise capacity in patients with cardiorespiratory disorders. Am J Respir Crit Care Med 1995;152:2021–2031. 19. Jones NL. Clinical exercise testing, 3rd ed. Philadelphia: W.B. Saunders Company; 1988. 20. O’Donnell DE, Lam M, Webb KA. Spirometric correlates of improvement in exercise performance after anticholinergic therapy in COPD. Am J Respir Crit Care Med 1999;160:542–549. 21. Grimby G, Bunn J, Mead J. Relative contribution of rib cage and abdomen to ventilation during exercise. J Appl Physiol 1968;24:159–166. 22. Potter WA, Olafson S, Hyatt RE. Ventilatory mechanics and expiratory flow limitation during exercise in patients with obstructive lung disease. J Clin Invest 1971;50:910–919. 23. Dodd DS, Brancatisano T, Engel LA. Chest wall mechanics during exercise in patients with severe chronic airway obstruction. Am Rev Respir Dis 1984;129:33–38. 24. O’Donnell DE, Webb K. Breathlessness in patients with severe chronic airflow limitation: physiological correlates. Chest 1992;102:824–831. 25. Diaz O, Villafranco C, Ghezzo H, Borzone G, Leiva A, Milic-Emil J, Lisboa C. Exercise tolerance in COPD patients with and without tidal expiratory flow limitation at rest. Eur Respir J 2000;16:269–275. 26. Grassino A, Gross D, Macklem PT, Roussos C, Zagelbaum G. Inspiratory muscle fatigue as a factor limiting exercise. Bull Eur Pathophysiol Respir 1979;15:105–111. 27. Montes de Oca M, Rassulo J, Celli BR. Respiratory muscle and cardiopulmonary function during exercise in very severe COPD. Am Rev Respir Dis 1996;154:1284–1289. 28. Bauerle, O, Chrusch CA, Younes M. Mechanisms by which COPD affects exercise tolerance. Am J Respir Crit Care Med 1998;157:57–68. 29. Neder JA, Jones PW, Nery LE, Whipp BJ. Determinants of the exercise endurance capacity in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2000;162:497–504. 30. Gosselink R, Troosters T, Decramer M. Peripheral muscle weakness contributes to exercise limitation in COPD. Am J Resp Crit Care Med 1996;153:976–980. 31. Carlson DJ, Ries AL, Kaplan RM. Predictors of maximum exercise tolerance in patients with COPD. Chest 1991;100:307–311. 32. Murariu C, Ghezzo H, Milic-Emili J, Gautier H. Exercise limitation in obstructive lung disease. Chest 1998;114:965–968. 33. O’Donnell DE, Chau LKL, Webb KA. Qualitative aspects of exertional dyspnea in interstitial lung disease. J Appl Physiol 2998;84: 2000–2009. 34. O’Donnell DE, Hong HH, Webb KA. Effects of chest wall restriction and deadspace loading on dyspnea and exercise tolerance in healthy normals. J Appl Physiol 2000;88:1859–1869. 35. O’Donnell DE, D’Arsigny C, Webb KA. Effects of hyperoxia on ventilatory limitation during exercise in advanced chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;163:892–898.