Jadranka Spahija, PhD; Michel de Marchie, MD; and Alejandro Grassino, MD

Effects of Imposed Pursed-Lips Breathing on Respiratory Mechanics and Dyspnea at Rest and During Exercise in COPD* Jadranka Spahija, PhD; Michel de Marchie, MD; and Alejandro Grassino, MD Study objectives: To investigate the effect of volitional pursed-lips breathing (PLB) on breathing pattern, respiratory mechanics, operational lung volumes, and dyspnea in patients with COPD. Subjects: Eight COPD patients (6 male and 2 female) with a mean ( SD) age of 58 11 years and a mean FEV 1 of 1.34 0.44 L (50 21% predicted). Methods: Wearing a tight-fitting transparent facemask, patients breathed for 8 min each, with and without PLB at rest and during constant-work-rate bicycle exercise (60% of maximum). Results: PLB promoted a slower and deeper breathing pattern both at rest and during exercise. Whereas patients had no dyspnea with or without PLB at rest, during exercise dyspnea was variably affected by PLB across patients. Changes in the individual dyspnea scores with PLB during exercise were significantly correlated with changes in the end-expiratory lung volume (EELV) values estimated from inspiratory capacity maneuvers (as a percentage of total lung capacity; r 2 0.82, p 0.002) and with changes in the mean inspiratory ratio of pleural pressure to the maximal static inspiratory pressure-generating capacity (PcapI) [r 2 0.84; p 0.001], measured using an esophageal balloon, where PcapI was determined over the range of inspiratory lung volumes and adjusted for flow. Conclusion: PLB can have a variable effect on dyspnea when performed volitionally during exercise by patients with COPD. The effect of PLB on dyspnea is related to the combined change that it promotes in the tidal volume and EELV and their impact on the available capacity of the respiratory muscles to meet the demands placed on them in terms of pressure generation. (CHEST 2005; 128:640 650) Key words: breathing pattern; breathlessness; end-expiratory lung volume; exercise; respiratory mechanics Abbreviations: EELV end-expiratory lung volume; EILV end-inspiratory lung volume; fb breathing frequency; FRC functional residual capacity; Pcapi maximum static inspiratory pressure-generating capacity; Pdi transdiaphragmatic pressure; Pdimax maximum inspiratory transdiaphragmatic pressure; PEEP positive end-expiratory pressure; Pes esophageal pressure; Pga gastric pressure; Pimax maximum inspiratory esophageal pressure; PLB pursed-lips breathing; Ppl pleural pressure; Pplmax maximum static inspiratory pleural pressure; Te expiratory time; Ti inspiratory time; TLC total lung capacity; TTdi tension-time index of the diaphragm; Ttot total breathing cycle time; V airflow; VAS visual analog scale; VC vital capacity; V e minute ventilation; Vl lung volume; Vt tidal volume; Weres expiratory resistive work of breathing; Wires inspiratory resistive work of breathing; W(i e)res total resistive work of breathing; Ẇlmax maximum exercise workload Pursed-lips breathing (PLB) is a technique whereby exhalation is performed through a resistance created by constriction of the lips. Although the breathing maneuver is often spontaneously *From the School of Physical and Occupational Therapy (Dr. Spahija), McGill University, Montreal, QC, Canada; the Department of Adult Critical Care (Dr. de Marchie), Sir Mortimer B. Davis Jewish General Hospital, Montreal, QC, Canada; and Centre Hospitalier de l Université de Montréal (Dr. Grassino), Notre-Dame Pavillon, Université de Montréal, Montreal, QC, Canada. Dr. Spahija was the recipient of a research fellowship award from the Fonds de la Recherche en Santé du Québec. This study was supported by the Medical Research Council of Canada. adopted by COPD patients, it is also routinely taught as a breathing-retraining exercise in pulmonary rehabilitation programs because it is thought to alleviate dyspnea. It appears, however, that not all patients obtain symptom benefits from PLB. 1 Despite this, Manuscript received August 31, 2004; revision accepted January 25, 2005. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Jadranka Spahija, PhD, Hôpital du Sacré- Coeur de Montréal, L Axe de Recherche en Pneumologie, 5400 Blvd Gouin Ouest, Montréal, QC, Canada H4J 1C5; e-mail: spahija@crhsc.umontreal.ca 640 Clinical Investigations

few studies have tried to determine the actual degree of dyspnea relief provided, and none, to our knowledge, have examined the effect of PLB on dyspnea when it is performed volitionally during exercise. Previous studies of PLB performed volitionally during resting breathing, have found that it improves arterial oxygenation 1 and saturation 2,3 and reduces arterial carbon dioxide levels 4 by promoting a slower and deeper breathing pattern. 1 4 Despite similar changes in breathing pattern with PLB performed during exercise, under such conditions no improvements in the arterial blood gas levels have been documented. 1 Exercise-induced dyspnea has been associated with the increased intensity and duration of respiratory muscle force generation as well as with an increased amplitude and velocity of muscle shortening. 5,6 More recently, the ratio relating inspiratory effort (ie, esophageal pressure [Pes], expressed as a fraction of the maximum Pes at isovolume [maximal inspiratory pressure (Pimax)], to tidal volume [Vt], expressed as a fraction of the vital capacity [VC]) was reported to be the best predictor of the inspiratory difficulty experienced by patients with COPD during incremental bicycle exercise. 7 Moreover, the (Pes/ Pimax)/(Vt/VC) ratio was shown to be significantly related to the degree of hyperinflation developed during exercise. 7 Dynamic hyperinflation or an increase in the end-expiratory lung volume (EELV), as typically occurs during exercise in patients with COPD, 8 11 can reduce the pressure-generating capacity of the respiratory muscles and has been associated with increased breathing effort and dyspnea. 12,13 Although EELV has been reported to be unaffected 4 or reduced 14 when PLB was performed at rest by patients with COPD, the extent to which PLB might alter EELV during exercise and might subsequently affect dyspnea remains unknown. The purpose of the present study was therefore to examine the effect of volitionally performed PLB on respiratory mechanics, EELV, and dyspnea in a group of patients with COPD at rest and during constant-work-rate bicycle exercise. We hypothesized that any possible effect of PLB on dyspnea would be related to the change that it produced in ventilatory muscle force-generating capacity. Subjects Materials and Methods Eight patients (2 women and 6 men) with stable mild-to-severe COPD 15,16 participated in the study (Table 1). Subjects with known cardiovascular disease, neurologic or psychiatric illness, or impaired lower extremity function or those requiring supplemental oxygen were excluded from the study. All patients received their regular treatment of inhaled bronchodilators, and none received oral steroid therapy. Four of the patients were receiving inhaled steroid therapy, and one patient was receiving oral theophylline therapy. No change in the medications was made for the purpose of the study. Patients were asked to abstain from smoking on the day of the study and to avoid eating for at least 2 h prior to undergoing testing. The study was approved by the ethics committee of the hospital, and all subjects gave written informed consent. Protocols and Instrumentation One week prior to the actual study, subjects underwent pulmonary function testing (system 1085; Medical Graphics Table 1 Anthropometric and Lung Function Data on Patients Studied* Subject/Sex/Age, yr Ht, m Wt, kg FEV 1,L FVC, L FEV 1 /FVC, % RV, L TLC, L FRC, L 1/M/62 1.64 54.4 1.22 2.81 43 4.13 7.56 5.31 (45) (74) (208) (128) (176) 2/M/68 1.66 70.0 1.37 3.27 42 5.02 8.95 6.68 (53) (87) (233) (147) (212) 3/M/60 1.75 99.0 1.24 3.13 40 4.00 7.49 4.93 (39) (70) (183) (111) (140) 4/F/55 1.55 56.7 1.56 2.21 71 2.62 5.06 3.15 (74) (78) (151) (97) (123) 5/F/46 1.60 63.0 2.12 3.29 64 2.55 5.97 3.05 (84) (100) (154) (122) (112) 6/M/40 1.68 63.0 1.04 2.11 49 4.35 7.24 5.15 (29) (46) (270) (118) (172) 7/M/57 1.83 54.5 0.73 2.32 31 7.48 9.85 8.38 (21) (46) (326) (133) (217) 8/M/72 1.68 72.2 1.32 2.91 45 3.04 6.29 4.32 (53) (77) (133) (101) (131) Mean SD 1.67 0.09 66.6 14.7 1.34 0.44 2.77 0.48 49 14 4.15 1.61 7.30 1.56 5.12 1.77 58 11 (50 21) (72 19) (207 66) (120 17) (160 40) *Values in parentheses are percent predicted. RV residual volume. www.chestjournal.org CHEST / 128 / 2/ AUGUST, 2005 641

Corp; St. Paul, MN) and performed an incremental exercise test on a bicycle ergometer (model 400L; Medical Fitness Equipment; Maarn, the Netherlands) to determine their maximum exercise workload (Ẇlmax). Subjects maintained a mean ( SD) pedaling rate of 60 5 revolutions per minute, and workload was increased by 10 W every minute until subjects could no longer continue. During this initial session, a physiotherapist instructed subjects on how to perform the PLB technique (ie, nasal inspiration followed by expiratory blowing against partially closed lips avoiding forceful expiration 17 ), and subjects also practiced the technique while pedaling and wearing a tight-fitting facemask in order to familiarize them with the study protocol. None of the subjects had difficulty in learning the breathing technique. On the day of the study, while seated on the bicycle ergometer, each subject first performed maximal inspiratory maneuvers (ie, the Mueller maneuver) and combined maneuvers consisting of a Mueller maneuver and abdominal expulsive maneuvers with the glottis open 18 to determine their maximum static inspiratory pleural pressure (Pplmax) and maximum inspiratory transdiaphragmatic pressure (Pdimax), respectively. Pdimax was measured at functional residual capacity (FRC), and the highest value obtained from three or more attempts was used in the subsequent analysis. For Pplmax, subjects performed the maximal inspiratory maneuvers at several lung volumes ranging from FRC to total lung capacity (TLC). While still seated on the bicycle, patients then breathed for 8 min using PLB and 8 min without using PLB (ie, control breathing). This was followed by 8-min periods of control breathing and PLB during constant-work-rate exercise at 60% of Ẇlmax. The order of control breathing and PLB were alternated among subjects, whereas exercise always followed the resting condition. Subjects were allowed to rest for at least 10 to 15 min between the two exercise runs. The breathing circuit consisted of a tight-fitting facemask (dead space, 90 ml) that was connected to a heated pneumotachograph (model No. 3; Fleisch; Lausanne, Switzerland), which permitted the measurement of inspiratory and expiratory airflow (V ). Vt was obtained by integrating the flow signal. The facemask was transparent, enabling investigators to verify that subjects were performing the PLB maneuver appropriately and when requested. Pleural pressure (Ppl) and gastric pressure (Pga) were measured using two balloon-tipped catheters that were passed transnasally, and transdiaphragmatic pressure (Pdi) was obtained by subtracting Ppl from the Pga. EELV was estimated by having subjects perform inspiratory capacity maneuvers every minute during the last 4 min of each experimental condition. EELV was obtained by subtracting the inspiratory capacity values from measures of TLC that had been previously obtained with whole-body plethysmography. 19 Subjects were asked to rate the sensation of breathlessness that they perceived every minute during each of the conditions studied using a visual analog scale (VAS). The VAS was displayed on a 10-cm oscilloscope screen with the verbal anchors no breathlessness and maximal breathlessness corresponding to the numerical values of 0 and 10, respectively, positioned at the bottom and top of the scale. Subjects were able to control the position of the line representing their breathlessness by means of a variable potentiometer attached to the bicycle handlebar. Data Analysis All signals were acquired online at a sampling rate of 100 Hz. Offline breath-by-breath analysis was performed on the last 4 min of each 8-min data segment. Timing parameters including inspiratory time (Ti), expiratory time (Te), total breathing cycle time (Ttot), and duty cycle were determined from the flow signal. Mean pressure swings were calculated between points of end-expiratory and end-inspiratory zero flow. The tension-time index of the diaphragm (TTdi) was calculated as the product of the ratio of the delta mean inspiratory Pdi to the Pdimax and the inspiratory duty cycle. Resistive work of breathing was determined by measuring the area enclosed by plots of Ppl vs Vt (see A panels in Fig 2) and was partitioned into inspiratory resistive work of breathing (Wires) and expiratory resistive work of breathing (Weres) components. 20 The individual average loops presented for each subject were obtained by combining ensemble averaged inspiratory and expiratory data sections, with each normalized to mean Ti and Te values, respectively. The maximum static inspiratory pressure-generating capacity (Pcapi) was determined first by plotting the Pplmax values along with their corresponding lung volumes, normalized to TLC, on a pressure-volume diagram and fitting a polynomial curve through these data points. For each subject, individual Ppl vs lung volume (Vl) loops were then positioned on this pressure-volume diagram, using the EELV values obtained under each of the experimental conditions and normalized to TLC. For every data point comprising the inspiratory portion of the individual Ppl-Vl loops, the corresponding Pplmax values were then adjusted for flow by applying a correction that reduced Pplmax by 5% for every liter per second increase in inspiratory V according to the following equation 21 : Pcapi Pplmax [Pplmax 0.05 V (L/s)]. For each data point in the inspiratory Ppl-Vl loop, Ppl was then expressed as a percentage of Pcapi, and an average value for inspiration was calculated. The ratio of Ppl to Pcapi (percent) was also determined at the point at which Ppl was at the peak value during inspiration. Statistical analysis for the comparison of variables between control breathing and PLB during rest and exercise was performed using the Student t test for paired data. Associations between the changes in the dyspnea scores and the changes promoted to breathing pattern variables, operational lung volumes, FEV 1 (percent predicted), and the ratio of the mean Ppl to Pcapi, were assessed by computing the Pearson product moment correlation coefficient. Significance for all tests was considered to be p 0.05. Results Effect of PLB on Dyspnea Whereas none of our subjects claimed to have dyspnea at rest, during control exercise at 60% of Ẇlmax, dyspnea scores ranged between 3.5 and 9. With PLB, breathlessness increased in four of the eight patients, was relatively unaltered in two patients, and decreased in two patients (Fig 1). Effect of PLB on Breathing Pattern Individual and average values for breathing pattern variables at rest and during exercise are shown in Tables 2 and 3, respectively. PLB prolonged total breath duration, slowed breathing frequency (fb), and increased Vt both at rest and during exercise. 642 Clinical Investigations

Figure 1. Average breathlessness scores (VAS) from the last 4 min of constant-work-rate exercise in each of the eight COPD patients during control breathing (E) and PLB (F). Lines of asterisks represent the average dyspnea scores for the group. Effect on End-Expiratory and End-Inspiratory Lung Volumes Whereas EELV increased on average by approximately 350 ml with exercise alone, it was variably altered by PLB at rest and during exercise across patients (Tables 2 and 3). On average, the increased Vt and unaltered EELV caused the end-inspiratory lung volume (EILV), which was 81% of TLC at rest without PLB, to increase to 86% of TLC with PLB (p 0.05). Furthermore, due to a combined dynamic hyperinflation and increased Vt values, EILV during exercise increased to 91% of TLC without PLB and was 92% of TLC with PLB. Changes in the dyspnea scores promoted by PLB were significantly Table 2 Ventilatory Parameters in Eight Patients With COPD During Resting Breathing Control Breathing PLB f B, breaths/ min V e, L/min V t /T i, L/s EELV, L f B, breaths/ min V e, L/min V t /T i, L/s EELV, L T i,s T tot,s V t,l T i,s T tot,s V t,l Subject 1 1.16 2.98 0.63 20.13 12.68 0.54 5.85 2.56 9.02 1.41 6.65 9.38 0.55 5.45 2 1.39 4.02 0.86 14.93 12.84 0.60 6.36 1.93 5.84 1.33 10.27 13.66 0.70 6.34 3 4.13 9.74 1.21 6.16 7.45 0.30 5.01 4.28 11.55 1.49 5.19 7.74 0.35 5.35 4 1.02 2.57 0.38 23.35 8.87 0.38 3.51 0.93 2.09 0.86 28.71 24.69 0.94 3.76 5 0.97 2.56 0.84 23.44 19.69 0.86 3.35 0.89 2.48 0.99 24.19 23.95 1.12 3.19 6 3.30 7.07 1.24 8.49 10.52 0.39 4.38 3.30 9.77 2.00 6.14 12.28 0.62 4.30 7 1.12 2.76 0.69 21.74 15.00 0.61 7.98 1.98 4.97 1.27 12.07 15.33 0.64 7.64 8 1.22 3.12 0.76 19.23 14.16 0.63 4.49 1.09 3.37 0.80 17.80 14.24 0.74 4.42 Mean SD 1.79 1.22 4.35 2.64 0.83 0.29 17.18 6.69 12.71 3.87 0.54 0.18 5.12 1.56 2.12 1.21 6.14 3.58* 1.27 0.39* 13.88 8.83 15.16 6.19 0.71 0.24 5.06 1.45 *p 0.05 vs rest control breathing. www.chestjournal.org CHEST / 128 / 2/ AUGUST, 2005 643

Table 3 Ventilatory Parameters in Eight Patients With COPD During Exercise Control Breathing PLB f B, breaths/ min V e, L/min V t /T i,l/s EELV, L T i,s T tot,s V t,l f B, breaths/min V e, L/min V t /T i,l/s EELV, L T i,s T tot,s V t,l Subject 1 0.78 2.15 0.89 27.91 24.84 1.14 5.92 1.06 3.76 1.30 15.96 20.74 1.24 6.05 2 0.86 2.01 0.98 29.85 29.25 1.14 7.56 0.94 2.72 1.15 22.06 25.37 1.23 7.48 3 1.06 2.81 1.24 21.35 26.47 1.18 5.32 1.57 4.34 1.72 13.82 23.78 1.10 5.50 4 0.65 1.35 0.84 44.44 37.33 1.29 3.57 0.72 1.52 0.89 39.47 35.13 1.25 3.68 5 0.57 1.31 1.40 45.80 64.12 2.48 3.31 0.58 1.42 1.67 42.25 70.56 2.89 3.24 6 0.89 2.37 1.71 25.32 43.29 1.92 5.17 0.93 2.59 1.87 23.17 43.32 2.05 5.05 7 0.75 2.10 1.11 28.57 31.71 1.50 8.31 0.97 2.68 1.32 22.39 29.55 1.36 7.98 8 0.81 2.18 1.18 27.52 32.48 1.47 4.76 0.80 2.23 1.17 26.91 31.48 1.47 4.72 Mean SD 0.80 0.15 2.04 0.50 1.17 0.29 31.35 8.89 36.19 12.75 1.52 0.47 5.49 1.75 0.95 0.30 2.66 1.00* 1.39 0.34 25.75 10.22 34.99 16.03 1.57 0.61 5.46 1.67 *p 0.05 vs exercise control breathing. p 0.01 vs exercise control breathing. correlated with changes in EELV (as a percentage of TLC; r 2 0.82; p 0.002). Effect of PLB on Respiratory Mechanics As shown in Table 4 and also illustrated by the respective horizontally lined and hatched areas of the average plots of Vt vs Ppl shown for two representative subjects in Figure 2 (panel A in upper and lower panels) because PLB promoted larger Vt values, Wires and Weres were significantly increased with PLB at rest and during exercise. However, the total resistive work of breathing [W(i e)res] per minute was not significantly altered by using PLB during exercise. As evidenced by the higher expiratory Pga values observed in the average Vt-Pga loops presented in Figure 2 (panel B in upper and lower panels), PLB performed at rest promoted an increased abdominal expiratory muscle recruitment compared to control breathing. This was a consistent finding in all subjects. As presented also in Figure 2 (panel B in upper and lower panels), exercise without PLB promoted greater clockwise looping in all patients compared to control breathing at rest, indicating the presence of more expiratory abdominal muscle activity. However, PLB during exercise further increased abdominal muscle recruitment in only three of the patients. In general, neither mean Pdi nor TTdi were altered by PLB at rest or during exercise (Table 4). However, as shown in the individual exercise Vt-Pdi loops of Figure 2 (panel C in upper and lower panels), end-expiratory Pdi was increased as a result of Pdi increasing prior to the onset of inspiratory flow, which was a consistent finding in all COPD patients breathing with and without PLB during exercise, and was observed in four of the eight patients during control breathing at rest. The mean end-expiratory Pdi, which was 2.5 1.6 cm H 2 O during control breathing at rest and increased to 8.0 3.6 cm H 2 O with exercise, was not significantly altered by PLB under either condition. Effect of PLB on Inspiratory Muscle Capacity The mean Pcapi at FRC was 67.5 18.4 cm H 2 O, where the mean FRC for the group was 69.1 8.2% of TLC, while at TLC the mean Pcapi was 24.2 9.5 cm H 2 O. Individual and average group values of the mean inspiratory ratio of Ppl/ Pcapi (in percent) are presented in Table 4. The ratio was increased significantly with PLB at rest but was inconsistently altered by PLB during exercise. Changes occurring in the Ppl/Pcapi ratio with PLB during exercise were significantly correlated with changes in EELV (as a percentage of TLC; 644 Clinical Investigations

Table 4 Respiratory Mechanics in Eight Patients With COPD During Exercise Control Breathing PLB W i e res, cm H 2 O L min W e res, cm H 2 O L W i res, cm H 2 O L Mean Pdi, cm H 2 O TTdi Mean Ppl, cm H 2 O Mean Ppl/ Pcapi, % W i e res cm H 2 O/L/min W e res, cm H 2 O/L W i res, cm H 2 O/L Mean Pdi, cm H 2 O TTdi Mean Ppl, cm H 2 O Mean Ppl/ Pcapi, % Subject 1 36.83 16.11 14.58 0.08 7.53 2.49 279.66 43.89 16.58 10.64 0.04 10.29 3.77 224.4 2 41.01 11.74 13.26 0.04 7.01 4.80 352.53 49.36 13.31 15.20 0.02 7.16 11.76 417.38 3 29.38 14.30 19.80 0.07 12.83 17.81 654.16 43.29 15.92 8.26 0.07 12.29 17.99 418.47 4 25.9 13.89 17.86 0.04 4.85 6.50 504.39 37.54 18.77 26.23 0.05 10.05 15.37 1003.33 5 23.76 19.00 10.64 0.03 18.64 14.54 1519.64 31.61 23.59 14.84 0.04 24.36 27.7 2199.54 6 36.19 18.26 14.22 0.05 25.22 21.05 1171.56 40.49 19.43 12.16 0.05 32.82 25.43 1349.65 7 60.64 23.56 26.97 0.03 14.80 27.25 1201.37 54.92 22.89 26.04 0.02 16.49 26.13 954.26 8 41.35 15.06 19.48 0.04 9.74 10.89 567.74 40.35 15.33 19.75 0.04 10.24 14.02 652.84 Mean SD 36.76 11.61 16.48 3.69 17.10 5.10 0.05 0.02 12.58 6.82 13.17 8.60 781.38 454.79 42.68 7.12 18.23 3.64* 16.67 6.82 0.04 0.02 15.46 8.81* 17.77 8.27* 902.48 642.19 *p 0.05 vs control breathing. r 2 0.65, p 0.016) and EILV (as a percentage of TLC; r 2 0.60; p 0.024). Figure 3 illustrates the relationship between the concurrent changes occurring in the dyspnea scores and the ratio of mean inspiratory Ppl to Pcapi with PLB during exercise. Subjects who experienced less breathlessness when using PLB exhibited decreases in the mean inspiratory ratio of Ppl to Pcapi (r 2 0.84; p 0.001). Discussion The present study shows that PLB can have a variable effect on dyspnea when performed volitionally by patients with COPD during exercise, and that the effect of PLB is dependent on changes promoted to the Vt, the EELV, and, ultimately, to respiratory muscle function. Patients who experience increases in breathlessness with PLB have the greatest combined increases in EELV and/or Vt, forcing the inspiratory muscles to operate close to the limits of their pressure-generating capacity, whereas patients who experienced reductions in dyspnea had reductions in EELV and greater reserves in inspiratory muscle pressure-generating capacity. Effect of PLB on Dyspnea, Breathing Pattern, and Respiratory Mechanics Patients reported having no dyspnea at rest, with or without PLB, whereas during exercise PLB had a variable effect on the dyspnea perceived by individual patients, with half of patients experiencing more dyspnea when using PLB. This is the first study that objectively corroborated previous anecdotal evidence suggesting that some patients with COPD obtain relief of dyspnea with PLB, whereas others do not. 1 During both breathing at rest and exercise, PLB promoted a slower and deeper breathing pattern, findings that concur with those of previous studies of volitional PLB. 1 4,14 Mueller et al 1 observed that COPD patients who obtained dyspnea relief with PLB exhibited larger increases in Vt and greater reductions in fb than did subjects who failed to experience any symptom benefit. These investigators proposed that the relief of breathlessness provided by PLB was related to its ability to promote a slower and deeper breathing pattern. In our study, however, such breathing pattern changes were not found to correlate with breathlessness. Certain patients who exhibited large increases in Vt combined with reductions in the respiratory rate experienced increases in breathlessness with PLB (eg, subjects 1 and 2). Likewise, there was no relation between the changes in breathing pattern promoted and the www.chestjournal.org CHEST / 128 / 2/ AUGUST, 2005 645

Figure 2. Average loops of Ppl (A panels), Pga (B panels), and Pdi (C panels) plotted against Vt and obtained with breath-by-breath analysis during the last 4 min of resting breathing (upper panels) and exercise (lower panels) for two representative COPD patients. Dashed lines control breathing; solid lines PLB; points of end-expiration; F points of end-inspiration. The loops in the A panels move in the clockwise direction, the direction of the loops in the B panels are indicated by the arrows, and the loops in the C panels move in the counterclockwise direction. Error bars represent the SD for Vt and the respective pressures. Horizontally lined and hatched areas in the A panels represent the increased Wires and Weres values that occurred in the two subjects with PLB. 646 Clinical Investigations

Figure 3. Relationship between changes in dyspnea and the concurrent changes in the ratio of mean inspiratory Ppl to Pcapi (as a percentage) occurring with PLB during steady-state exercise. The points represent the values for each individual patient with corresponding patient number. existing degree of expiratory airflow obstruction (as FEV 1 percent predicted) and/or static hyperinflation (ie, the ratio of residual volume to TLC [as a percentage]). Indeed, healthy individuals performing volitional PLB have also previously been found to exhibit significant increases in Vt during resting breathing and exercise, suggesting that the ability of PLB to promote changes in breathing patterns does not depend on the presence of expiratory flow obstruction. 22 In our patients, both at rest and during exercise, PLB had no effect on minute ventilation (V e). In addition, PLB also significantly increased both the Wires and Weres per breath, the result of joint increases in Vt and the expiratory flow resistance (Table 4). Thus, despite a reduction in the respiratory rate, the increased work of breathing was sufficient to offset the former, precluding significant change in the W(i e)res per minute. Although the use of PLB by patients in our study promoted larger mean inspiratory Ppl swings, mean inspiratory Pdi was not altered, suggesting that PLB may have the propensity to increase the recruitment of the rib cage and accessory muscles, as has been proposed in previous studies. 3,23 However, it is not certain to what extent such changes in respiratory muscle recruitment ultimately affect dyspnea, given that dyspnea may not depend on the activation of any one specific respiratory muscle group but, rather, may involve the integration of neural afferent information from any number of respiratory muscles. This premise is supported by previous findings 24 of the existence of a close relationship between respiratory effort sensation and Ppl swings, irrespective of whether the rib cage/accessory muscles or the diaphragm are used to generate those pressures. PLB performed at rest in the present study led to an increase in the recruitment of the abdominal muscles, whereas during exercise it had a less consistent effect. The expiratory muscles were already significantly recruited by exercise alone. While it is presumed 25,26 that abdominal muscle recruitment during expiration can lengthen the diaphragm, improving its tension-generating capacity during inspiration, it has also been postulated 9 that abdominal muscle recruitment can store elastic and gravitational energy within the diaphragm/abdomen, which, when released during the initial part of inspiration, enhances inspiratory pressure generation and lung inflation. Because our patients exhibited substantial expiratory muscle recruitment during exercise alone, PLB may have been unable to further improve on the existing pattern, thus having little impact in altering dyspnea. While in some studies, 10,27 expiratory muscle recruitment has been associated with a worsening of dyspnea, the extent to which the expiratory muscles contribute to the genesis of dyspnea in COPD patients is uncertain, given that hyperinflation improves the mechanical advantage of the expiratory muscles while requiring the inspiratory muscles to work at a higher fraction of their force-generating capacity. Although it was previously suggested 3 that PLB may act to unload the diaphragm and consequently may help to protect it against the development of fatigue, in our study PLB was found to have no effect on TTdi at rest or during exercise. However, using TTdi during dynamic exercise may be problematic due to the static nature of the Pdimax measurement. Because it can vary with changing lung volumes as well as with the rate and extent of muscle shortening, its application under conditions of exercise may be limited. Effect of PLB on EELV Among our subjects, PLB was found to have an inconsistent effect on EELV both at rest and during exercise (Tables 2 and 3, respectively). This concurs with the results obtained by Thoman et al 4 for PLB performed at rest, whereas, to the best of our knowledge, no other study has examined the effect of PLB on EELV in patients with COPD during exercise. However, the fact that patients are able to variably modulate the pressure generated with PLB www.chestjournal.org CHEST / 128 / 2/ AUGUST, 2005 647

from breath to breath and within a breath 22 and the wide range of expiratory flow obstruction exhibited by patients in our study may explain the variable effect that PLB was observed to have on EELV. Studies of positive end-expiratory pressure (PEEP) have demonstrated that the effect of PEEP on EELV varies depending on the presence or absence of dynamic airway collapse. In subjects who are not limited, expiratory muscle recruitment has been shown to effectively reduce the normal increase in EELV that occurs with the addition of PEEP, 28 whereas in the presence of expiratory flow limitation the resulting effect appears to be dependent on the degree of existing airway collapse, the amount of expiratory pressure reflected upstream from the mouth toward the airways, the level of V e, and the extent to which the expiratory muscles are recruited. 29 Several studies have elucidated the impact of dynamic hyperinflation on respiratory muscle function and dyspnea in COPD patients performing exercise. 7,11 O Donnell and Webb, 11 found dynamic EILV (as a percentage of TLC) to be the strongest independent predictor of exertional dyspnea, while combined changes in the EILV, EELV, Vt, and fb accounted for 61% of the variance in the change in dyspnea. While such lung volume components may contribute to the absolute dyspnea experienced by patients, in the present study 82% of the variance in the change in dyspnea promoted by PLB was accounted for by the alteration produced in EELV (as a percentage of TLC). Relationship Between Dyspnea and Inspiratory Muscle Force-Generating Reserve With PLB The mean Pplmax at FRC achieved in our patients at rest was 67.5 18.4 cm H 2 O, and it declined by about 1.4 cm H 2 O for every 1% increase in Vl (as a percentage of TLC). While the Pplmax was below normal, which is consistent with previous values reported in patients with COPD, 18,21,30 the rate of decline in the pressure-generating capacity with increasing lung volume was similar to that observed in healthy subjects. 21,31 The capacity to generate pressure also declines with increasing inspiratory flow, 32,33 and increased inspiratory flows during exercise in healthy subjects can result in substantial reductions in Pcapi. 21 In contrast, our patients exhibited significantly smaller peak inspiratory flows during exercise, and therefore had more modest reductions in Pcapi. Although patients with COPD typically demonstrate reduced exercise capacities and achieve below normal levels of maximum ventilation and V, the smaller inspiratory flows seen in our patients could also be attributed to the fact that they were exercising at 60% of their Ẇlmax. In our patients, the mean inspiratory Ppl/Pcapi ratio accounted for 84% of the variance observed in the change in dyspnea produced by PLB (p 0.001) [Fig 3]. Those patients who experienced more breathlessness with PLB demonstrated an increase in the ratio of mean Ppl to Pcapi, whereas subjects who reported improvements in dyspnea, exhibited a reduction in the ratio. Both the changes in EELV and EILV (as percentages of TLC) were correlated with the changes promoted in the Ppl/Pcapi ratio by PLB. Changes in both the EELV and Vt contributed to altering the EILV and, consequently, had an impact on the ratio of Ppl to Pcapi (Fig 4). Thus, PLB was able to modify dyspnea by promoting changes in the operational lung volumes, which consequently altered respiratory muscle capacity and performance. In patients with COPD, mechanical derangements of the ventilatory system and an increased flow resistance require that such patients generate larger inspiratory forces to achieve a given V e. Because hyperinflation promotes the shortening of the inspiratory muscles, a higher than normal neural motor output, or central neural drive, is needed. 34,35 Although conscious awareness of the outgoing motor command resulting from corollary discharges to the cerebral cortex has been suggested to mediate the sense of effort, 12,36,37 it has been proposed 38,39 that the discrepancy between the perceived effort and the actual ventilatory output (ie, neuroventilatory dissociation) contributes to the manifestation of dyspnea during exercise in such patients. The findings of the current study relate specifically to volitionally performed PLB and cannot be generalized to spontaneous PLB. Volitional PLB may involve greater focusing of the individual s attention on breathing pattern, which could have affected the results. It should also be noted that we did not ask whether our patients had used PLB prior to enrollment in the current study. Although none of the patients had previously participated in any formal pulmonary rehabilitation program, it is not known how many patients may have discovered the breathing technique on their own. Future studies of patients who naturally incorporate PLB into their breathing pattern may provide added information regarding this breathing technique. In summary, the present study shows that volitionally performed PLB by patients with COPD promotes a slower and deeper breathing pattern both at rest and during exercise, while prolonging expiratory and total breath durations, particularly at rest. Although PLB during exercise is capable of relieving dyspnea by decreasing EELV in some patients, it can 648 Clinical Investigations

Figure 4. Average loops of Ppl plotted against Vl expressed as a percentage of the TLC (%TLC) in two representative patients during exercise. Dashed lines control breathing; solid lines PLB; points of end-expiration; F points of end-inspiration. Loops move in the clockwise direction. The dotted line to the left of the loops indicates the Pplmax values that could be generated at a given Vl, and the corresponding curves represent the available inspiratory pressure generating capacity of the inspiratory muscles once adjusted for V (Pcapi). likewise be ineffective or even detrimental to dyspnea in others when Vt values increase. The effect that PLB has on dyspnea seems to be related to the combined changes that it promotes in EELV and Vt, and the impact that this has on the available capacity of the respiratory muscles to meet the demands that are placed on them in terms of pressure generation to achieve a given ventilation. ACKNOWLEDGMENT: The authors thank Dr. Heberto Ghezzo for his helpful advice on the statistical analysis. References 1 Mueller R, Petty T, Filley G. Ventilation and arterial blood gas changes induced by pursed-lips breathing. J Appl Physiol 1970; 28:784 789 2 Tiep BL, Burns M, Kao D, et al. Pursed lips breathing training using ear oximetry. Chest 1986; 90:218 221 3 Breslin E. The pattern of respiratory muscle recruitment during pursed-lip breathing. Chest 1992; 101:75 78 4 Thoman R, Stoker G, Ross J. The efficacy of pursed lips breathing in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1966; 93:100 106 5 Leblanc P, Bowie DM, Summers E, et al. Breathlessness and exercise in patients with cardiorespiratory disease. Am Rev Respir Dis 1986; 133:21 25 6 El-Manshawi A, Killian KJ, Summers E, et al. Breathlessness during exercise with and without resistive loading. J Appl Physiol 1986; 61:896 905 7 O Donnell DE, Bertley JC, Chau LK, et al. Qualitative aspects of exertional breathlessness in chronic airflow limitation. Am J Respir Crit Care Med 1997; 155:109 115 8 Stubbing DG, Pengally LD, Morse JLC, et al. Pulmonary mechanics during exercise in subjects with chronic airflow obstruction. J Appl Physiol 1980; 49:511 515 9 Dodd DS, Brancatisano T, Engel LA. Chest wall mechanics during exercise in patients with severe chronic air-flow obstruction. Am Rev Respir Dis 1984; 129:33 38 10 O Donnell DE, Sanii R, Younes M. Improvement in exercise endurance in patients with chronic airflow limitation using continuous positive airway pressure. Am Rev Respir Dis 1988; 138:1510 1514 11 O Donnell DE, Webb KA. Exertional breathlessness in patients with chronic airflow limitation. Am Rev Respir Dis 1993; 148:1351 1357 12 Killian KJ, Gandevia SC, Summers E, et al. Effect of increased lung volume on perception of breathlessness, effort, and tension. J Appl Physiol 1984; 57:686 691 13 Killian KJ, Jones NL. Mechanisms of exertional dyspnea. Clin Chest Med 1994; 2:247 257 14 Bianchi R, Gigliotti F, Romagnoli I, et al. Chest wall kinematics and breathlessness during pursed-lip breathing in patients with COPD. Chest 2004; 125:459 465 15 American Thoracic Society. Lung function testing: selection of reference values and interpretive strategies. Am Rev Respir Dis 1991; 144:1202 1218 16 American Thoracic Society. Standards for the diagnosis and www.chestjournal.org CHEST / 128 / 2/ AUGUST, 2005 649

care of patients with chronic obstructive pulmonary disease (COPD) and asthma. Am Rev Respir Dis 1987; 136:225 241 17 American Thoracic Society. Pulmonary Rehabilitation-1999. Am J Respir Crit Care Med 1999; 159:1666 1682 18 Laporta D, Grassino A. Assessment of transdiaphragmatic pressure in humans. J Appl Physiol 1985; 58:1469 1476 19 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 20 Roussos C, Campbell EJM. Respiratory muscle energetics. In: Macklem P, Mead J, eds. Mechanics of breathing: part II. Bethesda, MD: American Physiologic Society, 1986; 481 509 21 Leblanc P, Summers E, Inman MD, et al. Inspiratory muscles during exercise: a problem of supply and demand. J Appl Physiol 1988; 64:2482 2489 22 Spahija JA, Grassino A. Effects of pursed-lips breathing and expiratory resistive loading in healthy subjects. J Appl Physiol 1996; 80:1772 1784 23 Roa J, Epstein S, Breslin E, et al. Work of breathing and ventilatory muscle recruitment during pursed lip breathing in patients with chronic airway obstruction [abstract]. Am Rev Respir Dis 1991; 143:A77 24 Fitting JW, Chartrand DA, Bradley TD, et al. Effect of thoracoabdominal breathing patterns on inspiratory effort sensation. J Appl Physiol 1987; 62:1665 1670 25 Grimby G, Elgefors B, Oxhoj H. Ventilatory levels and chest wall mechanics during exercise in obstructive lung disease. Scand J Respir Dis 1973; 54:45 52 26 Sharp JT, Drutz WS, Moisan T et al. Postural relief of dyspnea in severe chronic obstructive pulmonary disease. Am Rev Respir Dis 1980; 122:201 211 27 Petrof BJ, Calderini E, Gottfried SB. Effect of CPAP on respiratory effort and dyspnea during exercise in severe COPD. J Appl Physiol 1990; 69:179 188 28 Chandra A, Coggeshall JW, Ravenscraft SA, et al. Hyperpnea limits the volume recruited by positive end-expiratory pressure. Am J Respir Crit Care Med 1994; 150:911 917 29 Marini JJ. Should PEEP be used in airflow obstruction? Am Rev Respir Dis 1989; 140:1 3 30 Gibson GJ, Pride NB. Static transdiaphragmatic pressures in normal subjects and in patients with chronic hyperinflation. Am Rev Respir Dis 1981; 124:685 689 31 Agostoni E, Rahn H. Abdominal and thoracic pressures at different lung volumes. J Appl Physiol 1960; 15:1087 1092 32 Hyatt RE. Expiratory flow limitation. J Appl Physiol 1983; 55:1 7 33 Agostoni E, Fenn WO. Velocity of muscle shortening as a limiting factor in respiratory air flow. J Appl Physiol 1960; 16:349 353 34 Grassino A, Goldman MD, Mead J, et al. Mechanics of the human diaphragm during voluntary contraction. J Appl Physiol 1978; 44:829 839 35 Sinderby C, Spahija J, Beck J, et al. Diaphragm activation during exercise in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001; 163:1637 1641 36 McCloskey DI. Corollary discharges: motor commands and perception. In: Brookhart JM, Mountcastle VB, eds. The nervous system: handbook of physiology. Bethesda, MD: American Physiologic Society, 1981; 1415 1447 37 Altose MD, Leitner J, Cherniak NS. Effects of age and respiratory efforts on the perception of resistive ventilatory loads. J Gerontol 1985; 40:147 153 38 Killian KJ, Campbell EJ. Historical aspects of dyspnea. In: Adams L and Guz A, eds. Respiratory sensation. New York, NY: Marcel Dekker, 1996; 1 17 39 O Donnell DE. Breathlessness in patients with chronic airflow limitation: mechanism and management. Chest 1994; 106:904 912 650 Clinical Investigations