Mask Proportional Assist vs Pressure Support Ventilation in Patients in Clinically Stable Condition With Chronic Ventilatory Failure*

Mask Proportional Assist vs Pressure Support Ventilation in Patients in Clinically Stable Condition With Chronic Ventilatory Failure* Roberto Porta, MD; Lorenzo Appendini, MD; Michele Vitacca, MD; Luca Bianchi, MD; Claudio F. Donner, MD, FCCP; Roberta Poggi, MD; and Nicolino Ambrosino, MD, FCCP Objective: To compare the short-term physiologic effects of mask pressure support ventilation (PSV) and proportional assist ventilation (PAV) in patients in clinically stable condition with chronic ventilatory failure (CVF). Design: Randomized, controlled physiologic study. Setting: Lung function units of two pulmonary rehabilitation centers. Patients: Eighteen patients with CVF caused by COPD (11 patients) and restrictive chest wall diseases (RCWDs) [7 patients]. Methods: Assessment of breathing pattern and minute ventilation (V E), respiratory muscles and lung mechanics, and patient/ventilator interaction during both unassisted and assisted ventilation. After baseline assessment during spontaneous breathing (SB), mask PSV and PAV were randomly applied at the patient s comfort, with the addition of the same level of continuous positive airway pressure (2 cm H 2 Oor4cmH 2 O in all patients), for 30 min each, with a 20-min interval of SB between periods of assisted ventilation. Results: A longer time was spent to set PAV than PSV (663 179 s and 246 58 s, respectively; p < 0.001). Mean airway opening pressure (Pao) computed over a period of 1 min, but not peak Pao, was significantly lower with PAV than with PSV (151 45 cm H 2 O/s/min and 207 73 cm H 2 O/s/min, respectively; p < 0.002). Tidal volume (VT) exhibited a greater variability with PAV than with PSV (variation coefficient, 16.3% 10.5% vs 11.6% 7.7%, respectively; p < 0.05). Compared with SB, both modalities resulted in a significant increase in VT (by 40% and 36% with PAV and PSV, respectively, on average) and V E (by 37% and 35%) with unchanged breathing frequency and duty cycle. Both modalities significantly reduced esophageal (by 39% and 51%) and diaphragmatic (by 42% and 63%) pressure-time products, respectively. Ineffective efforts were observed with neither modes of assistance in any patient. Conclusions: In resting, awake patients in clinically stable condition with CVF caused by either COPD or RCWD, noninvasive application of PAV, set at the patient s comfort, was not superior to PSV either in increasing VT and V E or in unloading the inspiratory muscles. We failed to find any difference in patient/ventilator interaction between ventilatory modes. (CHEST 2002; 122:479 488) Key words: breathing pattern; COPD; hypercapnia; noninvasive mechanical ventilation; respiratory failure; respiratory muscles; restrictive chest wall disease Abbreviations: CPAP continuous positive airway pressure; CVF chronic ventilatory failure; El,dyn dynamic elastance; El,dyn residual lung elastance; FA flow assist; IE ineffective efforts; IPS inspiratory pressure support; NPPV noninvasive positive pressure ventilation; Pao airway opening pressure; PAV proportional assist ventilation; Pdi transdiaphragmatic pressure; Pes esophageal pressure; PEEP positive end-expiratory pressure; PEEPi,dyn dynamic intrinsic positive end-expiratory pressure; Pga gastric pressure; PSV pressure support ventilation; PTPao mean airway opening pressure computed over a period of 1 min; PTPdi pressure-time product of the diaphragm calculated over a period of 1 min; PTPes pressure-time product of the inspiratory muscles calculated over a period of 1 min; RCWD restrictive chest wall disease; Rl pulmonary resistance at mid inspiration; Rl residual pulmonary resistance; SB spontaneous breathing; Ti/Ttot duty cycle; VA volume assist; V e minute ventilation; Vt tidal volume www.chestjournal.org CHEST / 122 / 2/ AUGUST, 2002 479

Long-term noninvasive positive pressure ventilation (NPPV) is used widely in the management of chronic ventilatory failure (CVF) that results from restrictive chest wall disease (RCWD) and from COPD 1 ; however, particularly in the latter, a strong evidence of significant clinical benefit is lacking. 2,3 Pressure support ventilation (PSV) is the most common mode of providing ventilatory assistance in the long-term setting, with and without some level of positive end-expiratory pressure (PEEP). 4,5 Studies have shown that proportional assist ventilation (PAV) 6 provides some physiologic benefit in patients with CVF by improving the breathing pattern and arterial blood gases, 7 decreasing breathlessness during exercise, 8,9 and unloading the respiratory muscles. 10 The theoretic background suggests that these effects of PAV should be associated with a better patient/ventilator interaction because PAV shifts the control of ventilatory assistance from the caregiver to the patient. In fact, the pattern of mechanical breath is proportional to the amplitude and timing of the patient s ventilatory demand. 6 However, this theoretic advantage was never tested in the clinical practice of NPPV. Among others, one reason might be that a direct comparison between PAV and PSV was never performed to understand whether the former may improve the patient s acceptability of long-term NPPV. This study was undertaken to investigate whether PAV will provide a better patient/ventilator interaction than PSV, and thus a better acceptability in patients requiring long-term use of noninvasive ventilation. Materials and Methods The investigative protocol was approved by the Ethics Committee of the S. Maugeri Foundation IRCCS. The study was conducted according to the declaration of Helsinki. Patients gave their informed consent to participate in the study. Patients The study was conducted in the Lung Function Units of Gussago and Veruno, S. Maugeri Foundation, from March 1999 to March 2000. Eighteen patients (11 patients with COPD and 7 patients with RCWD caused by kyphoscoliosis) with CVF were recruited for this study. Diagnosis of COPD was made according From the Fondazione Salvatore Maugeri IRCCS, Pulmonary Departments, Scientific Institutes of Gussago (Drs. Porta, Vitacca, Bianchi, and Ambrosino) and Veruno (Drs. Appendini and Donner), Ospedale Maggiore di Borgo Trento, Azienda Ospedaliera di Verona (Dr. Poggi), Italy. This study was partially supported by Respironics Inc., Murrysville, PA. Manuscript received July 12, 2001; revision accepted February 20, 2002. Correspondence to: Nicolino Ambrosino, MD, FCCP, Fondazione S. Maugeri, Lung Function Unit, Istituto Scientifico di Gussago, I-25064 Gussago (BS), Italy; e-mail: nambrosino@fsm.it to the American Thoracic Society. 11 The diagnosis of CVF was based on the clinical records showing values of Paco 2 persistently 45 mm Hg during room air spontaneous breathing (SB) in the months if not years preceding the study. All patients were in stable clinical condition, as assessed by an arterial ph 7.35 and were free from exacerbations in the preceding 4 weeks. Patients with other chronic organ failure, cancer, or inability to cooperate were also excluded from the study. All of the patients were receiving drug treatment according to the prescriptions of their primary physicians. In particular, patients with COPD were receiving regular treatment with inhaled bronchodilators, avoiding either systemic or inhaled steroids apart from exacerbations. At the time of the study, 16 of the 18 patients were receiving long-term oxygen therapy. Six patients with COPD and two patients with RCWD had been receiving long-term home NPPV by nasal mask for 16 to 36 months, with volume-cycled ventilators in assisted mode for two patients, and with pressure-cycled ventilators in PSV mode for six patients. Indications for long-term NPPV for these patients were chronic hypercapnia and nocturnal hypoventilation for all patients. The mean use of NPPV was approximately 7 h per night. Six of 18 patients were admitted to the hospital for indication of domiciliary NPPV. Four of these six patients were discharged with home NPPV. The other patients underwent respiratory rehabilitation programs. The patients characteristics, according to their diagnoses, are shown in Table 1. Measurements Lung function and arterial blood gas levels were assessed 1 to 3 days before the study. Routine static and dynamic lung volumes were measured with constant volume body plethysmographs (CAD-NET system 1085; Medical Graphics; St. Paul, MN, and V max Series/6200 Autobox DL; SensorMedics; Yorba Linda, CA) with the patients in the seated position according to standard procedure. 12 The predicted values of Quanjer 13 were used. Arterial blood was sampled at the radial artery with the patients in a semirecumbent position and breathing room air. Pao 2, Paco 2, and ph were measured with automated analyzers (model 840; Ciba Corning; Medfield, MA, and ABL 620; Radiometer; Copenhagen, Denmark). For the experimental procedure of this study, flow was measured using a heated pneumotachograph (Fleisch n 1; Fleisch; Lausanne, Switzerland) connected to a flow transducer (Hewlett- Packard 47304A; Hewlett-Packard; Cuppertino, CA) inserted between the nasal mask and the plateau valve of the ventilator Table 1 Demographic, Anthropometric, and Functional Characteristics of Patients* Characteristics COPD (n 11) RCWD (n 7) Age, yr 66 10 55 15 Male/female sex, No. 9/2 6/1 Weight, kg 76 17 58 11 FEV 1, % predicted 26 6 31 9 FVC, % predicted 47 13 35 8 FEV 1 /FVC, % 45 11 75 15 Residual volume, % predicted 189 43 82 17 Inspiratory capacity, % predicted 42 22 35 9 Total lung capacity, % predicted 104 15 53 11 ph 7.40 0.04 7.38 0.02 Pao 2,mmHg 54 11 53 10 Paco 2,mmHg 58 8 57 5 *Data are presented as mean SD unless otherwise indicated. 480 Clinical Investigations

circuit. 14 Volume was obtained by numerical integration of the flow signal. Airway opening pressure (Pao) was measured with a differential pressure transducer (Honeywell 143PCO3D; Honeywell; Freeport, IL) connected to one port of the nasal mask. Changes in pleural and abdominal pressures were estimated from changes in esophageal pressure (Pes) and gastric pressure (Pga), respectively (Transducer Motorola X2010 100 cm H 2 O; Colligo, Elekton; Agliano Terme, Italy) using the balloon-catheter technique, with esophageal and gastric balloon catheters, as previously described. 10 Transpulmonary pressure and transdiaphragmatic pressure (Pdi) were obtained by subtraction of Pes from Pao and Pga, respectively. Data Analysis All signals were digitized by an analog-to-digital converter with 12-bit resolution, connected to a personal computer, at a sampling frequency of 100 Hz. Subsequent analyses of breathing pattern and pulmonary mechanics were performed using an edition of the software package (ANADAT 5.2; RHT-Infodat; Montreal, PQ, Canada) interfaced with the respiratory monitoring system used in the present study. Using the Abreath facility of ANADAT, the mean value of each physiologic variable was computed and used subsequently for statistical analysis. Tidal volume (Vt), respiratory frequency, minute ventilation (V e), total cycle duration, inspiratory time, expiratory time, and duty cycle (Ti/Ttot) were calculated from the flow signal as average values from 5 min of continuous recording of flow and volume. Breathing pattern variability was also computed on flow tracing and expressed as coefficient of variation (SD/mean 100). Dynamic intrinsic PEEP (PEEPi,dyn) was measured as the negative deflection in Pes from the onset of the inspiratory effort to the start of the inspiratory flow. In the presence of expiratory muscle activity, the value of PEEPi,dyn was reduced by the decrease in Pga measured in the same time interval. 15 Changes in the magnitude of the inspiratory muscles and diaphragm effort were estimated from changes in Pes and Pdi swings, respectively, as well as from changes in the pressure-time product of the inspiratory muscles calculated over a period of 1 min (PTPes) and pressure-time product of the diaphragm calculated over a period of 1 min (PTPdi). 15 The latter measurements were expressed also as pressure developed per liter of ventilation (PTPes/V e and PTPdi/V e). Transpulmonary pressure was used to calculate dynamic elastance (El,dyn) according to the Mead and Whittenberger technique, 16 and pulmonary resistance at midinspiration (Rl) according to the Neergaard-Wirtz elastic subtraction technique. 16 The maximal positive level of Pao signal (peak Pao) was calculated as average values from 10 consecutive respiratory acts in which breathing pattern and mechanics were calculated. In addition, the average area subtended by Pao of the same tidal swings, from the onset of inspiratory effort to the inspiratory flow tracing inversion (from inspiration to expiration) was computed and calculated over a period of 1 min, multiplying it for respiratory rate (mean Pao computed over a period of 1 min [PTPao]) [Fig 1]. Patients inspiratory efforts that were unable to trigger a new ventilator cycle despite a negative deflection in Pes were termed ineffective efforts (IE). 17 The mean number of IE per minute, recorded over 5 min, was expressed as percentage of the respiratory rate (IE per minute/respiratory frequency [beats per minute] 100). Figure 1. Representative tracing of Pao delivered by PAV (upper panel) and PSV (lower panel). The maximal positive level of the Pao signal (peak Pao) was calculated as average values from 10 consecutive respiratory acts in which breathing pattern and mechanics were also calculated. The average area subtended by Pao of the same 10 tidal swings (magnified dashed areas in squares) was computed and calculated over a period of 1 min multiplying it for respiratory rate (PTPao). www.chestjournal.org CHEST / 122 / 2/ AUGUST, 2002 481

Ventilatory Settings A commercial nasal mask of adequate size for each patient s nose was used (Respironics; Murrysville, PA). In each patient, both modalities were delivered using the same portable ventilator able to compensate for leaks and to operate in the continuous positive airway pressure (CPAP) mode, PSV mode, and PAV mode (BiPAP Vision; Respironics). The ventilator circuit was equipped with the Sanders NRV-2 plateau valve (Respironics) to prevent CO 2 rebreathing. 14 Before starting the protocol, PAV and PSV settings were separately assessed as follows. PAV: The ventilator delivers pressure according to the motion equation, generating a pressure that is proportional to patient s spontaneous effort. 6 A portion of the total mechanical workload, ie, elastance and resistance, is taken over according to a level of assistance, which has been decided by the caregiver and can specifically unload the resistive burden (flow assist [FA]) and the elastic burden (volume assist [VA]). VA and FA were set separately; VA and FA were set initially at the minimum value of 2 cm H 2 O/L/s and 1 cm H 2 O/L/s, respectively, in all patients. Then, leaving FA unchanged, VA was increased slowly by steps of 2 cm H 2 O/L until the patient indicated that breathing was uncomfortable. Each step lasted the time necessary for the ventilatory setting to be described as comfortable or uncomfortable. Then, that level of assist was decreased by 2 cm H 2 O/L. This level was considered as the maximum tolerated. To set FA, a similar stepwise approach was used, by keeping VA at 2 cm H 2 O/L and slowly increasing FA from 1 cm H 2 O/L/s by small steps of 1 cm H 2 O/L/s until the patient noted being uncomfortable with that level of assistance. Then, that level of assistance was decreased by 1cmH 2 O/L/s. This level was considered to be the maximum tolerated. The setting of PAV that was applied to patients corresponded to 80% of the maximal individual tolerated values of VA and FA. 10 Those values of FA and VA were used to evaluate the residual pulmonary resistance ( Rl) and residual lung elastance ( El,dyn) faced by respiratory muscle contraction to generate inspiratory flow and volume, being subtracted from Rl ( Rl Rl FA) and El,dyn ( El,dyn El,dyn VA), respectively. No data were available about the residual resistance and elastance of the chest wall. PSV: The level of inspiratory pressure support (IPS) was increased slowly by steps of 2 cm H 2 O, starting from 2 cm H 2 O, until the patients indicated that breathing was uncomfortable. Hence, that level of IPS was decreased by 2 cm H 2 O, and the resultant level was applied. A default level of 2 cm H 2 O of CPAP was added to both modalities in each individual patient. Experimental Procedure The whole procedure was performed under continuous monitoring of arterial oxygen saturation by pulse oximetry (Oxicap Monitor; Ohmeda; Louisville, CO). The patients were evaluated in the morning (patients with COPD were evaluated at least 2 h after inhalation of their bronchodilating medications) and were free to choose the most comfortable position. All patients adopted a semirecumbent position. The procedure to evaluate respiratory muscles and mechanics has been extensively detailed elsewhere. 4,5,10,15 Briefly, in all patients, after the application of topical anesthesia (xylocaine spray 10%), two balloon-tipped catheters were consecutively inserted through the nose into the middle third of the esophagus and into the stomach and thereafter inflated to 0.5 ml and 1 ml for the esophageal and gastric catheters, respectively. Then, the nasal mask was applied and connected to the pneumotachograph. The occlusion test 18 was finally performed to check the proper functioning of the esophageal balloon. A pneumatic shutter was inserted in line and proximally to the pneumotachograph only to perform this maneuver, and then removed. The occlusion test was satisfactory in every instance. Initially, all patients were not connected to the ventilator and breathed room air through the nose mask for approximately 20 min. All of the patients were instructed to breathe through the nose mask and to keep their mouths closed during the experimental procedure to prevent leaks. Subsequently, PAV and PSV were applied in random order for 30 min, and the trials were separated by 20-min periods of SB through the mask. All of the measurements were obtained at an inspiratory oxygen fraction of 0.21. All of the physiologic signals were recorded in the last 5 min of each unsupported and supported breathing period. In two patients with RCWD, the signal from the gastric balloon was unsatisfactory. Hence, complete data of respiratory muscles and mechanics were available in 16 of 18 patients. At the end of each period of assistance, 10 patients (5 patients with each diagnosis) of the 18 patients scored their comfort sensation under NPPV by means of an arbitrary scale ranging from 0 (worst comfort) to 10 (best comfort). At the end of the two ventilation sessions, the patients indicated their preferred mode of ventilation. The time spent by the operator to set the two modalities was recorded by an attending nurse. Statistical Analysis Results are shown as mean 1 SD. Differences between treatments and within treatment were evaluated by analysis of variance for repeated measures. Because of the lack of differences between values assessed during the two SB periods, baseline values were considered as the mean levels of the two measurements. Differences between paired groups of data were evaluated with post hoc paired t test with Bonferroni adjustment and were applied as requested by analysis of variance interaction. A p value 0.05 was considered significant. Results All patients accepted NPPV well throughout the procedure. No patients reported side effects or refused the procedure with either modality. Because no significant difference between PAV and PSV was found according to the diagnosis, when not specified, data from patients with different diagnoses are reported together. Setting of the Ventilator The individual and mean levels of assistance with the two modalities and patients elastance and resistance are shown in Table 2. With PAV, no runaway phenomenon 6 was observed in any patient. The run-away phenomenon was defined by the continuation of positive pressure after the end of the patients inspiratory effort into the patients neural expiration. In this condition, the pressure delivered by the ventilator exceeds the patients elastic recoil opposing force, generating inspiratory flow and volume well beyond the termination of the inspiratory muscle effort, until total respiratory elastance increases above VA value at total lung capacity. 6 482 Clinical Investigations

Table 2 Individual and Mean Levels of Assistance and Lung Mechanics During Spontaneous and Supported Ventilation PSV PAV Patient No. Group IPS cm H 2 O VA cm H 2 O/L FA cm H 2 O/L/s El,dyn cm H 2 O/L Rl cm H 2 O/L/s El,dyn cm H 2 O/L Rl cm H 2 O/L/s 1 COPD* 16 16 4 10.5 14.6 5.5 10.6 2 COPD 6 10.4 4 29.1 19.9 18.7 15.9 3 COPD 14 4.8 2.4 4.5 12 0.3 9.6 4 COPD 13 10.4 5.6 62 31 51.6 25.4 5 COPD* 8 7.2 4.8 3.1 2.7 4.1 2.1 6 COPD* 10 3 9 7.2 15.3 4.2 6.3 7 COPD* 14 8 10 10.1 16.7 2.1 6.7 8 COPD 10 10 4 17.2 23 7.2 19 9 COPD* 16 11 3 13.2 16.3 2.2 13.3 10 COPD* 17 16 4 16.7 12.4 0.7 8.4 11 COPD* 10 6 1 6.9 4.8 0.9 3.8 Average SD COPD 12.2 3.6 12.2 5.6 4.7 2.7 16.4 16.8 15.3 7.9 7.1 16.1 10.6 7.6 12 RCWD* 11 24.8 4 7.1 2.8 17.7 1.2 13 RCWD 11 29.6 8 8.8 6.4 20.8 1.6 14 RCWD* 10 12 4 22.2 6.4 10.3 2.4 15 RCWD 11 22.4 8.8 17.7 5.3 4.7 3.5 16 RCWD* 14 22.4 11.2 6.9 8.5 15.5 2.7 17 RCWD* 10 22 7 22.7 13.5 0.7 6.5 18 RCWD* 16 28 5 41.7 13.6 13.7 8.6 Average SD RCWD 11.9 2.3 27.8 8 6.9 2.7 18.2 12.4 8.1 4.1 4.9 13.8 1.2 4.7 Average SD Total 12 3 14.7 8 5.5 2.8 17.1 14.9 12.5 7.5 2.4 16 7 8 *Receiving home ventilation. p 0.001 between COPD and RCWD patients. Table 2 also shows patients discharged with home mechanical ventilation. There was no significant difference in levels of VA and FA between patients who required home ventilatory assistance and patients who did not. Patients with RCWD tolerated higher levels of VA under PAV than patients with COPD. Peak Pao was not significantly different between ventilator modes (11 4cmH 2 O and 12 3cmH 2 O with PAV and PSV, respectively). By contrast, PTPao was lower with PAV (151 45 cm H 2 O/s/min) than PSV (207 73 cm H 2 O/s/min; p 0.002). During PAV, Rl and El,dyn workloads showed a different behavior in patients with COPD and patients with RCWD (Table 2). In fact, in patients with RCWD, VA was above the El,dyn, whereas Rl fell within normal limits. 19 By contrast, in patients with COPD, El,dyn fell within normal limits, and, on average, Rl was mildly increased (Table 2). The caregivers spent more time to set PAV than PSV (663 179 s and 246 58 s, respectively; p 0.001). Breathing Pattern and Lung Mechanics Figure 2 shows a polygraphic tracing from a representative patient during the different conditions of the study protocol. Ineffective efforts were not observed with either modes of assistance in any patient. In the breathing pattern, Vt showed a greater variability under PAV than under PSV (coefficient of variation, 16.3% 10.5% and 11.6% 7.7%, respectively; p 0.01). Mean values of breathing pattern and lung mechanics during SB and assisted ventilation are shown in Table 3. On average, both PAV and PSV improved Vt (by 40% and 36%, respectively), V e (by 37% and 35%, respectively), and mean inspiratory flow (by 33% and 51%, respectively) without changes in respiratory frequency and Ti/Ttot. Neither mode modified El,dyn or Rl compared with baseline. Inspiratory Muscles Table 4 shows that on average, the magnitude of the patients inspiratory muscle effort was significantly reduced by both PAV and PSV, which resulted in lower Pes ( 33% and 51%), Pdi ( 34% and 52%), PTPes ( 39% and 51%), PTPdi ( 42% and 63%), PTPes/V e ( 53% and 63%), and PTPdi/ V e ( 56% and 72%, respectively; Fig 3). Although there was no significant difference, PSV unloaded the inspiratory muscles slightly more than PAV. Comfort Comfort score was available in 10 patients (5 patients with COPD and 5 patients with RCWD). www.chestjournal.org CHEST / 122 / 2/ AUGUST, 2002 483

Figure 2. Polygraphic recording during different periods of protocol study. From top to bottom: flow, Pao, Pes, Pdi, and Vt. No patients rated low scores of comfort under either modality of ventilation, indicating good adaptation to NPPV (7.0 1.3 and 7.4 1.6 for PAV and PSV, respectively). Moreover, no significant difference was found between scores during PAV and PSV. There was no clear preference about a specific ventilatory mode because six patients preferred PSV and four patients preferred PAV. Discussion The results of this study show that in resting, awake patients in clinically stable condition with Table 3 Breathing Pattern and Lung Mechanics During Spontaneous and Supported Ventilation* Variables SB PAV PSV Respiratory frequency, 20.8 5.7 20.8 5 19.6 5.4 beats/min Vt, ml 452 168 579 190 621 182 Ve, L/min 8.7 1.6 11.5 2.9 11.5 2.4 Ti/Ttot 0.36 0.05 0.34 0.04 0.32 0.05 Vt/Ti, L/s 0.40 0.1 0.52 0.17 0.59 0.17 PEEPi,dyn, cm H 2 O 1.6 1.2 1.5 1.1 1.3 1.3 El,dyn, cm H 2 O/L 16.7 12.3 12.2 7.4 12.3 6.2 Rl, cmh 2 O/L/s 12.6 6.6 12.1 5.7 11.7 4.5 *Data are presented as mean SD. Vt/Ti mean inspiratory flow p 0.001 vs SB. p 0.02 vs SB. CVF caused by either COPD or RCWD, noninvasive application of PAV, set at the patient s comfort, was not superior to PSV either in increasing Vt and V e and or in unloading the inspiratory muscles. The only differences between PSV and PAV were that the latter achieved those physiologic benefits at a lower level of PTPao but required more time to set the ventilator. We failed to find any difference in patient/ventilator interaction between ventilatory modes. This study confirms and extends previous observations, 4,5,10 but it provides the first comparison in the NPPV setting between PAV, a recently proposed mode of mechanical ventilation, and PSV, which still remains the most widely used mode of ventilatory assistance for both short-term and long-term patients. 1 In patients with COPD and in patients with Table 4 Inspiratory Effort During Spontaneous and Supported Ventilation* Variables SB PAV PSV Pes, cm H 2 O 12.6 4.7 8.3 4.5 6.8 4.4 Pdi, cm H 2 O 14.3 4.9 9.7 4.9 7.6 5.5 PTPes, cm H 2 O/min 218 83 126 81 106 82 PTPdi, cm H 2 O/min 249 106 143 75 97 76 PTPes, cm H 2 O/s/L V e 27 14 12 8 10 8 PTPdi/V e, cmh 2 O/s/L 30 15 13 7 8 8 *Data are presented as mean SD. p 0.001 vs SB. p 0.001 vs SB. 484 Clinical Investigations

Figure 3. Individual and mean changes in PTPes/V e (upper panel) and PTPdi/V e (lower panel) corrected per liter of ventilation during SB and assisted ventilation with the two modes studied. Values are given according to diagnosis (full circles and continuous line COPD; open squares and dashed lines RCWD). Individual and mean values of SB represent the actual baseline measurement before each assisted period. RCWD, the magnitude of improvement in V e and Vt (Table 3), as well as the amount of the reduction of the patients inspiratory effort (Table 4), were not significantly different between PAV and PSV, although PSV determined a slightly greater reduction of patients effort, on average. Similar changes in breathing pattern 7 and respiratory mechanics 10 were obtained in our previous studies with PAV. In this connection, it should be mentioned that all of the patients in this study and in the previous studies 7,10 were receiving low levels of CPAP, which contributed to the reduction of the patients inspiratory effort. 4,15 A direct comparison between PAV and PSV was www.chestjournal.org CHEST / 122 / 2/ AUGUST, 2002 485

performed in intubated patients with acute respiratory failure. 20 22 The settings (ie, ICUs), the patients (ie, in acute condition), and the method of delivering mechanical ventilation (ie, through the endotracheal tubes) are too different from the conditions of our study to make any comparison. However, it should be mentioned that in those studies, 20 22 PAV allowed a greater variability of Vt than PSV in the face of an increased ventilatory demand. Also in our study, patients receiving PAV showed a greater variability of Vt than those receiving PSV. However, we failed to find any significant clinical benefit from this difference in our long-term patients. In fact, none of the two modes (ie, PAV or PSV) prevailed in the patients choice of a greater comfort. This might be caused by the fact that, among others, both modalities were set at the patients comfort since the beginning of the procedure, whereas in the acute settings, PAV was set after measuring the patients respiratory mechanics using the occlusion technique. 20 22 This procedure, which might allow a better physiologic tailoring of the level of assistance following the theoretic background of PAV, can be applied without discomfort in ventilator-dependent patients, but it is difficult to use in awake patients reacting to the airway occlusion. Furthermore, NPPV needs patients cooperation, which cannot be obtained in condition of discomfort. Therefore, regardless the mode of mechanical assistance, the ventilator must be set at the patient s comfort to ensure cooperation. In comparing different modalities of mechanical ventilation, a major problem is to avoid the risk to match up to apples and oranges. 23 In our study, this risk was reduced by a few issues. First, the same ventilator was used to deliver the mechanical assistance. Second, the two modes were applied according to a random order with a similar period (30) of spontaneous, unsupported breathing before each mode. Third, the patients were in stable clinical condition; therefore, it is unlikely that changes in the patient s state might affect the results of the physiologic measurements. Finally, both PSV and PAV were set with the same procedure, namely at the patient s comfort. Having obtained similar physiologic results, PAV was associated with a lower mean Pao, as assessed by PTPao, than PSV (151 45 cm H 2 O/s/min vs 207 73 cm H 2 O/s/min). This may be relevant in view of studies 24 showing that NPPV can significantly reduce cardiac output in patients with COPD. It could be extrapolated that the lower mean Pao with PAV was likely to determine less changes in the patients hemodynamics than PSV. However the lower mean Pao with PAV was associated with a trend to less respiratory muscle unloading than with PSV. PAV required a longer time to be set. There are two possible explanations for this finding. First, PAV is not common in the clinical practice, whereas PSV is the most widely used mode of NPPV. Conceivably, the caregivers are far more familiar with PSV than with PAV. In these circumstances, the time to set PAV will be remarkably reduced by practice. Second, with PSV only Pao must be regulated whereas with PAV two independent variables (ie, VA and FA) must be coordinated. Although the manufacturers are trying to simplify this issue, there is little doubt that at present the setting of PAV requires more time and awareness than PSV. Time might not be so relevant for long-term patients in stable condition, but it may be crucial for patients needing mechanical ventilation to treat acute respiratory failure. In line with previous studies, 25 both PAV and PSV unloaded the patients respiratory muscles without causing changes in lung mechanics (Tables 3, 4). The low levels of PEEPi,dyn in our patients with COPD are not surprising in view of their stable conditions. The most interesting aspect of Table 2 is the comparison between the levels of assistance (VA and FA) and patients lung mechanics (lung resistance and elastance). Apparently, FA underassisted the resistive component in patients with COPD, whereas VA overassisted the elastic component in patients with RCWD. The first of these results is in line with our previous study. 10 Similar to those patients, most of the patients with COPD in this study also did not tolerate levels of FA 5cmH 2 O/L/s. The present study was not tailored to address directly this finding, but it can be hypothesized that patients with COPD, by means of dyspnea perception, chose a low level of FA to avoid excessive increase in Vt, thus protecting themselves from the development of dangerous levels of dynamic hyperinflation and intrinsic PEEP. In fact, dynamic hyperinflation is known to increase dyspnea perception in patients with COPD. 26 Moreover, the fact that FA algorithm is linear, and Rl during inspiration is not, 6 may add to the above hypothesis. It may be that early during inspiration, Rl was underassisted because of its high value; on the contrary, with increasing the inspiratory volume and airway caliber, FA might have been better matched to a decreasing Rl. In a recent study, Younes and colleagues 27 suggested a new method for measuring resistance to set PAV in intubated patients with acute respiratory failure. Whether that method may allow a better tailoring of FA also with NPPV remains to be established in the clinical setting. Finally, the present study does not provide experimental data explaining the apparent VA overassistance of the patients with RCWD. In fact we could only measure El,dyn, whereas in patients with 486 Clinical Investigations

RCWD the major mechanical abnormality is well known to be related to stiff chest wall. Chest wall elastance could not be measured in this study, but it can be assumed to be greater than the level of VA overassistance indicated by El,dyn. This hypothesis is indirectly supported by the fact that the run-away phenomenon was not observed when VA overwhelmed lung elastance. All together, the above data on chest wall and lung mechanics suggest that the measurement of lung mechanics using the esophageal catheter system can be not only uncomfortable for the patients, but also of little help to set PAV for clinical purposes when chest wall mechanics abnormalities are present. The regulation at patient s comfort still seems the most acceptable way to set any mode of NPPV in the common clinical situation. 5 Limitations of the Study PAV has been designed to assist the respiratory muscles in coping with changes of the ventilatory demand. 6 This particular feature of PAV should distinguish it from other modes of ventilatory assistance. This study failed to show any advantage of PAV over PSV to claim as relevant for clinical purposes, apart from the lower mean Pao. However we documented that PAV behaved as well as PSV under resting conditions in very sick patients in stable clinical condition. Our study was performed in awake patients, whereas home NPPV is usually prescribed at night. Therefore, the correct ventilator setting theoretically should be tested during a formal sleep study. Nevertheless, we think that the lack of studies of comparison of PAV and PSV would warrant a daytime investigation, in particular when one takes into account the invasive techniques needed to measure patient s respiratory muscle function. Furthermore, daytime mechanical ventilation in awake patients was reported to be as equally effective in reversing chronic hypercapnia as nocturnal mechanical ventilation. 28 In conclusion, in resting, awake patients in clinically stable condition with CVF, noninvasive application of PAV, set at patient s comfort, was not superior to PSV either in increasing Vt and V e, in unloading the inspiratory muscles, or in patient/ ventilator interaction. ACKNOWLEDGMENT: We want to thank Dr. Andrea Rossi for useful talks, comments, and suggestions. References 1 Mehta S, Hill NS. Noninvasive ventilation. Am J Respir Crit Care Med 2001; 163:540 577 2 Rossi A, Hill NS. Pro-con debate: noninvasive ventilation has been shown to be effective/ineffective in stable COPD. Am J Respir Crit Care Med 2000; 161:688 691 3 Pauwels RA, Buist AS, Calverley PM, et al. Global strategy for the diagnosis, management and prevention of chronic obstructive lung disease: NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary. Am J Respir Crit Care Med 2001; 163:1256 1276 4 Nava S, Ambrosino N, Rubini F, et al. Effect of nasal pressure support ventilation and external PEEP on diaphragmatic activity in patients with severe stable COPD. Chest 1993; 103:143 150 5 Vitacca M, Nava S, Confalonieri M, et al. The appropriate setting of noninvasive pressure support ventilation in stable COPD patients. Chest 2000; 118:1286 1293 6 Younes M. Proportional assist ventilation. In: Tobin MJ, ed. Principles and practice of mechanical ventilation. New York, NY: McGraw-Hill, 1994; 349 369 7 Ambrosino N, Vitacca M, Polese G, et al. Short-term effects of nasal proportional assist ventilation in patients with chronic hypercapnic respiratory insufficiency. Eur Respir J 1997; 10:2829 2834 8 Dolmage TE, Goldstein RS. Proportional assist ventilation and exercise tolerance in subjects with COPD. Chest 1997; 111:948 954 9 Bianchi L, Foglio K, Pagani M, et al. Effects of nasal proportional assist ventilation (PAV) on exercise performance and breathlessness of COPD patients with chronic hypercapnia. Eur Respir J 1998; 11:422 427 10 Polese G, Vitacca M, Bianchi L, et al. Nasal proportional assist ventilation unloads the inspiratory muscles of stable patients with hypercapnia due to COPD. Eur Respir J 2000; 16:491 498 11 Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease: American Thoracic Society. Am J Respir Crit Care Med 1995; 152:S77 S121 12 Quanjer PH, Tammeling GJ, Cotes JE, et al. Lung volumes and forced ventilatory flows. Report working party standardization of lung function tests, European community for steel and coal. Official Statement of the European Respiratory Society. Eur Respir J 1993; 6(suppl 16): 5 40 13 Quanjer PH, ed. Standardized lung function testing. Report, ECCS, Luxemborg. Bull Eur Physiopathol Respir 1983; 19(suppl):1 95 14 Ferguson T, Gilmartin M. CO 2 rebreathing during BiPAP ventilatory assistance. Am J Respir Crit Care Med 1995; 151:1126 1135 15 Appendini L, Patessio A, Zanaboni S, et al. Physiological effects of positive end expiratory pressure and mask pressure support during exacerbations of chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1994; 149:1069 1076 16 Tobin MJ, Van de Graaff WB. Monitoring of lung mechanics and work of breathing. In: Tobin MJ, ed. Principles and practice of mechanical ventilation. New York, NY: McGraw- Hill, 1994; 967 1003 17 Nava S, Bruschi C, Rubini F, et al. Respiratory response and inspiratory effort during pressure support ventilation in COPD patients. Intensive Care Med 1995; 21:871 879 18 Baydur A, Behrakis K, Zin WA, et al. A simple method for assessing the validity of the esophageal balloon technique. Am Rev Respir Dis 1982; 126:788 791 19 Frank NR, Mead J, Ferris BG. The mechanical behaviour of the lungs in healthy elderly persons. J Clin Invest 1957; 36:1680 1686 www.chestjournal.org CHEST / 122 / 2/ AUGUST, 2002 487

20 Ranieri VM, Giuliani R, Mascia L, et al. Patient-ventilator interaction during acute hypercapnia: pressure support vs proportional assist ventilation. J Appl Physiol 1996; 81:426 436 21 Wrigge H, Golisch W, Zinserling J, et al. Proportional assist versus pressure support ventilation: effects on breathing pattern and respiratory work of patients with chronic obstructive pulmonary disease. Intensive Care Med 1999; 25: 790 798 22 Grasso S, Puntillo F, Mascia L, et al. Compensation for increase in respiratory workload during mechanical ventilation: pressure-support versus proportional-assist ventilation. Am J Respir Crit Care Med 2000; 161:819 826 23 Ambrosino N. Apples and oranges: comparing different modalities of mechanical ventilation. Intensive Care Med 1999; 25:775 777 24 Ambrosino N, Nava S, Torbicki A, et al. Haemodynamic effects of pressure support and PEEP ventilation by nasal route in patients with stable chronic obstructive pulmonary disease. Thorax 1993; 48:523 528 25 Appendini L, Purro A, Gudjonsdottir M, et al. Physiologic response of ventilator-dependent patients with chronic obstructive pulmonary disease to proportional assist ventilation and continuous positive airway pressure. Am J Respir Crit Care Med 1999; 159:1510 1517 26 O Donnell DE, Revill SM, Webb KA. Dynamic hyperinflation and exercise intolerance in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001; 164:770 777 27 Younes M, Kun J, Masiowski B, et al. A method for noninvasive determination of inspiratory resistance during proportional assist ventilation. Am J Respir Crit Care Med 2001; 163:829 839 28 Schoenhofer B, Geibel M, Sonneborn M, et al. Daytime mechanical ventilation in chronic respiratory insufficiency. Eur Respir J 1997; 10:2840 2846 488 Clinical Investigations