INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It has been suggested that in mechanically ventilated patients, measurement of respiratory mechanics can be important for assessing the status and progress of the disease underlying acute respiratory failure (ARF), and for guiding the ventilator settings (1). Recent studies have provided the first demonstration, to our knowledge, that measurement of respiratory mechanics for setting ventilator patterns can also have a real impact on the clinical outcome of patients with acute respiratory distress syndrome (ARDS) (4, 5). It was shown that the survival of ARDS patients was improved by a new ventilatory strategy based on the ventilation in the linear portion of the sigmoidal total respiratory system pressure-volume (P-Vrs) curve: namely, between a lower and an upper inflection point (LIP and UIP, respectively) (2, 6). The initial poor distensibility at low lung volume is due to the relatively high pressure needed to reopen closed air spaces (2, 7), whereas the upper, flat portion of the P-Vrs curve reflects alveolar overdistension (3, 6). Accordingly, by setting the positive end-expiratory pressure (PEEP) to a value just above the LIP, most recruited zones are kept open, and ventilator-induced lung damage due to continuous stretching from opening and closing is avoided (8, 9). On the other hand, a less than conventional inflation volume maintains alveolar pressure below the UIP, thus preventing undue alveolar overdistension (10).

The interpretation of the LIP as representing alveolar recruitment is based on the assumption that the P-Vrs curve, obtained in mechanically ventilated patients by means of supersyringe lung inflation, mainly reflects the elastic behavior of the lung, with little or no influence from the elastic properties of the chest wall. However, despite the extensive use of the P-V curve in the clinical setting and its recently disclosed clinical implication, this assumption has never actually been tested, although: (1) chest-wall compliance has been found to be decreased in mechanically ventilated patients with ARF (11, 12); and (2) the relatively flat initial portion of the P-V curve could be due to chest-wall elastic characteristics at low lung volume (13).

The purpose of the present study was therefore to partition the total respiratory system P-V relationship into the lung and chest-wall components in order to investigate whether chest-wall mechanics could affect the shape of the P-Vrs curve drawn with the supersyringe method, and could hence influence the choice of ventilator settings.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study was approved by the Ethics Committee of the General and University Hospital of Parma, and informed consent was obtained from all patients or their next of kin.

Patients

Thirteen patients admitted into the Intensive Care Unit (ICU) of the General and University Hospital of Parma (Italy) were recruited for the study. Patients' characteristics and clinical diagnoses are shown in Table 1. Arterial blood gas values and the ventilator settings at the time of inclusion in the study are shown in Table 2. Five patients were intubated orally with endotracheal tubes ranging from 7.5 to 8.5 mm I.D.; the remaining eight patients were tracheotomized (with tracheotomy cannulas ranging from 8 to 9 mm I.D.). All patients were mechanically ventilated with the Drager Evita ventilator (Dragerwerk AG, Lubeck, Germany) in the controlled mode (controlled mechanical ventilation [CMV]). Mechanical ventilation was instituted because of ARF diagnosed according to the following criteria (14): (1) acute onset; (2) Pa_O2/FI_O2 < 300; (3) diffuse bilateral infiltrates on frontal chest radiograph. On average patients were mechanically ventilated for 5 ± 2 d (range: 2 to 10 d) before the study. In general, in our institution, tracheotomy is performed early in the clinical course of ARF, when the clinical condition of the patient clearly indicates that mechanical ventilation will be needed for longer than 10 d.

Patients with one of the following characteristics were excluded: (1) previous history of chronic airway disease (e.g., chronic obstructive pulmonary disease [COPD], asthma); (2) heart failure; (3) shock (suggested by a mean systolic blood pressure of less than 60 mm Hg); and (4) chest-wall abnormalities and pneumothorax.

Measurements

Airflow was measured with a heated Fleisch No. 2 pneumotachograph (Fleisch, Lausanne, Switzerland) connected to a differential pressure transducer (Validyne MP 45 ± 2 cm H₂O; Validyne, Inc., Northridge, CA), which was inserted between the proximal tip of the endotracheal tube and the Y piece to which the supersyringe and the ventilator circuit were connected, as illustrated in Figure 1. During the study period, the humidifier was omitted from the inspiratory line. The instrumental dead space was 120 ml. The pneumotachograph was calibrated with the supersyringe apparatus and the gas mixture being used. Changes in volume were obtained by numerical integration of the flow signal.

View larger version (15K): [in this window] [in a new window]   Figure 1.   Schematic representation of equipment layout. Pes = esophageal pressure; Pao = pressure at airway opening; PT = pneumotachograph.

Pressure at the airway opening (Pao) was sampled through a side port proximal to the pneumotachograph (Figure 1), connected through an air-filled, noncompliant catheter to a disposable pressure transducer (Transpac II; Abbot Critical Care System, Abbott Ireland Ltd., Sligo, Ireland) (15).

Changes in pleural pressure (Ppl) were estimated from changes in esophageal pressure (Pes), using a thin-walled latex balloon (10 cm long, 1 cm in circumference), sealed over one end of a polyethylene catheter (100 cm long, 2 mm ID; Jaeger No 720 199; Jaeger Wurzburg, Germany), that was positioned in the midesophagus and inflated with 0.6 to 0.8 ml of air; the validity of the Pes measurement was assessed with the occlusion test (16), and was satisfactory in all instances. The esophageal balloon was connected to a differential pressure transducer (Validyne MP 45 ± 80 cm H₂O). Pressure transducers were calibrated with a water column.

P-V curves were obtained by means of a supersyringe (volume: 2 L) moved at a low, constant rate (50 ml/s) by an electric engine. The inflation volume was set at 10 to 12 ml/kg. At the final inflation volume, the engine automatically reversed the syringe to the starting pressure. Changes in transpulmonary pressure (PL) were calculated by subtracting the Pes signal from Pao. The total respiratory system P-V curve (P-Vrs), the lung P-V curve (P-VL), and the chest-wall P-V curve (P-Vw) were obtained from the plot of volume against Pao, PL, and Pes, respectively.

All signals were recorded on a personal computer (Macintosh II CI; Apple Computer, Inc., Cupertino, CA) via an analog-to-digital converter (MacLab Analog Digital Instrument, Pty. Ltd., Castle Hill, Australia) at a sample rate of 100 Hz, and were stored on diskettes for subsequent computer analysis.

Arterial blood was sampled from an indwelling catheter inserted into the radial artery for clinical purposes, and pH, Pa_CO2, and Pa_O2 were measured by means of an automated blood gas analyzer (IL BG3; Instrumentation Laboratory, Lexington, MA). Heart rate (HR), pulse oximetry, and systemic blood pressure were monitored throughout the study and a physician not involved in the procedure was always present for patients' care.

Procedure

At the time of the study, the patients were in a clinically and hemodynamically stable condition. Measurements were taken with the patients in the supine position. After insertion of the esophageal balloon through the nose, the occlusion test was performed (16). Patients were then sedated with a continuous infusion of fentanyl (0.02 to 0.03 µg/ kg/min) and diazepam (0.6 µg/kg/min) and paralyzed with vecuronium (0.1 mg/kg followed by 0.05 mg/kg if necessary). Before each set of measurements, the cuff of the endotracheal tube or of the tracheal cannula was inflated with an extra volume of 3 ml of air, and the airways were suctioned free of centrally retained secretions; special care was taken to avoid air leaks in the equipment's connections.

The ventilator settings, established by the responsible physicians and shown in Table 2, were kept constant for each patient throughout the experiment, with the exception of PEEP. Indeed, at 20 to 30 min before the start of the procedure, four different levels of PEEP (i.e., 0, 5, 10, and 15 cm H₂O) were applied in random order. For Patient 8, the FI_O2 had to be increased to 1 to prevent an excessive decrease in Pa_O2 when PEEP was set to zero (ZEEP). In the other patients, FI_O2was kept constant (Table 2). Because the curves had to start at various levels of PEEP, the supersyringe and ventilator circuits were connected to the patient through a Y-piece (Figure 1). Before the P-V maneuver was begun, the syringe and ventilator circuits were put in communication to allow pressure equilibration. Five seconds after the beginning of the expiration, the ventilator circuit was clamped and the syringe flow was started. In that way the patient was not disconnected from the ventilator, and the selected end-expiratory pressure could be maintained at the start of the syringe flow. The entire inflation-deflation maneuver lasted 39 ± 8 s. In no instance did Sa_O2 fall below 90%.

Physiologic measurements, including arterial blood gases, were taken after a minimum of 20 min following each level of PEEP. Thereafter, the P-Vrs and P-Vw curves were obtained simultaneously by means of the supersyringe.

Data Analysis

The P-VL, P-Vw, and P-Vrs curves for each patient at each level of PEEP were drawn from pressure and volume values with the graphic option of the Excel 4 program (Microsoft Corp., Redmond, WA). From the P-VL, P-VW, and P-Vrs curves the following parameters were derived (2, 17):

1. Compliance. Starting compliance (C_start) and inflation compliance (C_inf) were computed according to Gattinoni and colleagues (2). The compliance at the end of the curve (end compliance [C_end]) was computed as the ratio of the last 150 ml of inflation to the corresponding pressure change.
2. Lower Inflection Point (LIP). In nine patients, the LIP was computed on the P-Vrs, P-VL, and P-VW curves as the difference between the pressure at the interception of the lines corresponding to C_start and C_inf and the pressure at the beginning of the supersyringe lung inflation (Figure 2). When PEEP was set by the ventilator, the true value of LIP included PEEP. In four patients (Patients 1, 6, 8, and 11) an intrinsic PEEP (PEEPi) was present at the start of syringe flow (18). In these patients the LIP on the P-Vrs curve was above PEEPi, and was computed relative to the elastic equilibrium volume (which corresponded to atmospheric pressure or PEEP), thus including the value of PEEPi in the LIP value. To take the same phenomenon into account for the lung and chest-wall LIPs, PEEPi was partitioned between the lung and chest wall according to the corresponding C_start. First, we computed the changes in the end-expiratory lung volume (EELV) above the relaxation volume (Vr) according to the formula: EELV = C_start,rs × PEEPi (19, 20). The lung and chest-wall PEEPi were then computed according to the formula: PEEPi (l and w) = EELV/C_start, L and W. The values of PEEPi obtained in this way were added to the values of LIP measured on the corresponding P-VL and P-VW curves as the distance between the beginning of the curves and the C_inf -C_start intersection.
3. Upper Inflection Point (UIP). The UIP was computed on all of the P-V curves as the pressure difference between the pressure at the interception of the lines corresponding to C_inf and C_end and the pressure relative to which LIP was measured (4).

View larger version (17K): [in this window] [in a new window]   Figure 2.   An example of the method used to identify the lower inflection point in a representative patient (Patient 4) with an inflection on the chest-wall P-V curve. Pes = change in esophageal pressure; Pao = pressure at the airway opening; PL = change in transpulmonary pressure.

The values of the lower and upper inflection points were detected by two independent readers, and the mean value was used for subsequent analysis.

Statistical Analysis

Individual respiratory mechanical and arterial blood gas values at the four PEEP levels were compared through a one-way analysis of variance (ANOVA) with adjustment for repeated measurements; when appropriate, and post hoc analysis was performed with the Newman- Keuls test (STATISTICA/Mac 3.0b, StatSoft Inc., Tulsa, OK). A value of p < 0.05 was accepted as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For all 13 patients enrolled in this study, the P-Vrs curve was not linear throughout the volume span of the syringe inflation. An LIP separating the initial, less compliant portion of the curve from the subsequent linear portion was observed in every instance on ZEEP, as shown in Table 3. The LIPrs ranged from 2.5 to 17.7 cm H₂O, averaging 7.5 ± 3.9 cm H₂O at a volume of about 0.15 L above the relaxation volume. In four patients, the value of LIPrs included PEEPi, which amounted to 3.4, 4, 9, and 3.2 cm H₂O in Patients 1, 6, 8, and 11, respectively. As illustrated in Figures 3 and 4, the LIPrs was not always due to the LIPL. In two patients the LIPrs was due exclusively to LIPL (Figure 3), whereas in seven patients it was due to the characteristics of the P-VW curve (Figure 4) rather than to the P-VL curve. In the other four patients an inflection point was simultaneously present on the P-VL and P-VW curves. Values of LIPL and LIPW are also reported in Table 3. In the six patients who exhibited an inflection point on the P- VL curve, the LIPL averaged 8.6 ± 2.6 cm H₂O, ranging from 6.7 to 13.6 cm H₂O. In the 11 patients with an inflection point on the P-VW curve, the LIPW averaged 2.9 ± 1.1 cm H₂O, ranging from 1.3 to 4.8 cm H₂O.

View larger version (28K): [in this window] [in a new window]   Figure 3.   P-V curves of the lung, chest wall, and respiratory system in a representative patient (Patient 3) with a lower inflection point on the P-V curve of the lung; V = volume; Pao = pressure at airway opening; Pes = change in esophageal pressure; PL = change in transpulmonary pressure.

View larger version (29K): [in this window] [in a new window]   Figure 4.   P-V curves of the lung, chest wall, and respiratory system in a representative patient (Patient 4) with a lower inflection point on the P-V curve of the chest wall; V = volume; Pao = pressure at airway opening; Pes = change in esophageal pressure; PL = change in transpulmonary pressure.

The P-VL, P-Vw, and P-Vrs curves at ZEEP were fitted to a power equation of the type: P = a* · V^b*, where a* represents elastance of the relevant respiratory component at a V of 1 L and b* is a dimensionless number that indicates the variation of elastance with inflating volume (21). For coefficient b* < 1, elastance decreases with inflating volume, whereas it increases for values of coefficient b* > 1. A value of b* = 1 indicates a linear P-V relationship. The correlation coefficients ranged between 0.94 and 0.99. The mean values of a* for the lung, chest wall, and respiratory system were 14.6 ± 6.3, 6.7 ± 2.4, and 20.6 ± 6.1 cm H₂O/L, respectively. The mean values of b* for the lung in patients with and without an LIP on the P-VL curve were 0.73 ± 0.05 and 0.98 ± 0.05, respectively. Likewise, the values of b* for the chest wall were 0.75 ± 0.08 and 0.92 ± 0.01 in patients with and without an LIP, respectively. The mean value of coefficient b* for the respiratory system was 0.69 ± 0.1.

From the P-VL and P-Vw curves with ZEEP, we computed the lung volumes corresponding to LIP, which amounted to 0.17 ± 0.05 L and 0.18 ± 0.04 L, respectively. In two patients exhibiting an inflection point on both the P-VL and P-Vw curves, the LIPw exceeded the LIPL by 52 ml and 160 ml, respectively, whereas in two other patients LIPL was above LIPw.

Application of PEEP reduced the inflection point on all the P-V curves, as also illustrated by the examples in Figures 3 and 4. At a PEEP of 10 cm H₂O, only two patients still exhibited an LIP, which amounted to 3.6 cm H₂O on the P-VL and P-Vrs curves for Patient 1, whereas Patient 8 had a total LIP of 7.4 cm H₂O, corresponding to an LIP of 5.9 and 2.7 cm H₂O on the P-VL and P-Vw curves, respectively. After application of 15 cm H₂O of PEEP, no LIP was detectable in any instance, whereas the UIP was disclosed on the P-Vrs curve, as shown in Table 3, and in a representative patient as shown in Figure 5. The UIP was detected only on the lung curve; no UIP was observed on the P-Vw curve. Values of UIP, which are reported in Table 3, were detected at a volume of 0.59 ± 0.09 L above PEEP.

View larger version (14K): [in this window] [in a new window]   Figure 5.   P-V curves of the lung, chest wall, and respiratory system of a representative patient (Patient 11) with an upper inflection point during ventilation at a PEEP value of 15 cm H2O. Note the absence of the upper inflection on the chest-wall P-V curve. UIP = upper inflection point; V = volume; Pao = pressure at airway opening; Pes = change in esophageal pressure; PL = change in transpulmonary pressure.

Table 4 shows the mean values of C_start, C_inf, and C_end for the lung, chest-wall, and total respiratory system at different levels of PEEP. At 0 and 5 cm H₂O of PEEP, C_inf and C_end are significantly greater than C_start in all instances, in keeping with the presence of an LIP. At higher levels of PEEP, and in particular at 15 cm H₂O, C_end is significantly lower than C_inf and C_start for the P-VL and P-Vrs curves, whereas there is no difference in chest-wall compliance. This result is in keeping with the presence of a UIP on the P-VL curve at high PEEP. Application of PEEP did not significantly change C_start,L and C_inf,L, whereas C_end,L decreased significantly. In contrast, C_start,w increased significantly with PEEP, whereas both C_inf,w and C_end,w did not change. On the P-Vrs curve, C_start,rs increased significantly with PEEP, paralleling the change in C_start,w , whereas C_end,rs decreased significantly with PEEP, paralleling the change in C_end,l.

Additionally, as shown in Table 5, Pa_O2 increased progressively with PEEP when averaged over the entire group of patients. However, when we considered the six patients with an LIP on the P-VL curve apart from the seven patients with an LIP on the P-Vw curve, Pa_O2 increased significantly only in the former group and not in the latter, suggesting that a significant improvement in pulmonary gas exchange was obtained only when PEEP abolished the lung LIP (Figure 6).

View larger version (10K): [in this window] [in a new window]   Figure 6.   Changes in PaO2 with PEEP (mean and one SD). Closed circles and solid lines: patients with a lower inflection point on the P-V curve of the lung. Open squares and dashed lines: patients without a lower inflection point on the P-V curve of the lung but with an inflection on the P-V curve of the chest wall. *p < 0.05 versus corresponding value at PEEP of 0 cm H2O; p < 0.05 versus corresponding value at PEEP of 5 cm H2O.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this study show that in mechanically ventilated patients with ARF: (1) the LIP on the P-Vrs curve can be due to the shape of the P-Vw curve rather than to the P-VL curve; (2) PEEP may fail to improve Pa_O2 in patients with an LIP on the P-Vrs curve if the LIP is due to the chest-wall mechanics.

In this study, as in previous reports by other authors (2, 22), the inflation limb of the P-Vrs curve obtained by means of the supersyringe method in mechanically ventilated patients with ARF exhibited an "inflection point" separating the initial, less compliant portion of the curve from the subsequent, steeper relationship. The most widely accepted interpretation of this feature is that initially a relatively high pressure is needed to reopen closed air spaces, such that LIP represents the pressure at which most air spaces are recruited (2, 4). Accordingly, it has been suggested that PEEP should be set at a value slightly greater than the LIP to ventilate the lung in the linear portion of the P-V relationship (2). However, we show here that the P-Vrs curve may not reflect the mechanical properties of the lung, but in some patients can be significantly influenced by chest-wall mechanics such that the LIPrs is due to the LIPw and not to the LIPL. We also show that the shape of the P-Vw curve is modified by PEEP (i.e., the LIPw disappears with increasing PEEP), suggesting that a volume-dependent mechanism determines the LIPw in the P-Vw curve.

It is worth considering that the P-Vw curves were obtained by means of the esophageal balloon technique with the assumption that the average stress on the chest wall (i.e., Ppl) may be approximated by a pressure measurement at a single location in the esophagus. The correspondence between the absolute value of Ppl and Pes has been validated in healthy humans in the erect position (23). In the supine position, owing to the weight of the mediastinal structures on the esophagus, the agreement between Ppl and Pes is less satisfactory (24). However, in this case, it has been demonstrated (16) that the changes in Pes can accurately reflect the changes in Ppl, provided that proper balloon positioning has been accomplished (16, 25, 26). In the ICU setting, with the patient in the supine position and with inhomogeneous lung, the assumption of correspondence between Ppl and Pes has been questioned. However, several authors (6, 11, 12, 21) have used the esophageal balloon technique in mechanically ventilated patients to partition lung from chest-wall mechanics by scrupulously following a standardized technique (16, 25). Although some limitations are inherent with this approach, it represents the only means for measuring lung and chest-wall mechanics.

To our knowledge, ours is the first partitioning of the supersyringe P-Vrs curve into the P-VL and P-Vw curves. We found that in 11 of our 13 patients, the slope of the chest-wall P-V relationship increased with the progression of inflation after the initial, less compliant portion of the curve, exhibiting an LIP. In four of these 11 patients, an inflection point was also present on the P-VL curve, whereas in the remaining seven patients the P-VL curve was linear, such that the LIP detected on the P-Vrs curve was determined exclusively by the chest-wall mechanics. A major clinical implication of this finding is that caution is required in the use of the P-Vrs curve to set the best PEEP in mechanically ventilated patients (2, 5, 27). Indeed, when the inflection point is due to chest-wall mechanics, application of PEEP does not improve Pa_O2 (Figure 6), whereas the ventilating lung units may be unduly overdistended, as suggested by the appearance of a UIP (Table 3), hence exposing the patients to an enhanced risk of barotrauma (4, 28).

In accordance with the analysis of the P-V curve suggested by Gattinoni and colleagues (2), we computed C_start and C_inf for the total respiratory system, the lung, and the chest wall. The low lung compliance in patients with ARF is determined by alveolar flooding rather than by changes in the elastic properties of the lung (2, 29). For the lung, the improvement from C_start,L, to C_inf,L is explained by a reopening of previously collapsed air spaces (2, 9). The interpretation of the low C_start,w needs further analysis.

First, positive fluid balance, abdominal distension, edema of the soft tissues, and pleural effusion (12, 30) may alter the intrinsic mechanical properties of the chest wall. Second, it is known that in the supine position, gravity has a marked expiratory effect on the abdomen-diaphragm (13, 33). In the normal human in the upright position, the gravitational effect of the abdominal contents generates, just beneath the dome of the diaphragm, a subatmospheric pressure of 3 or 4 cm H₂O that equals the inward recoil force of the lung facing the diaphragmatic dome. In contrast, in the supine position, due to gravitational forces, the abdominal content exerts a pressure of about 9 cm H₂O on the abdominal surface of the diaphragm; as a consequence, the diaphragm is displaced upward and resting lung volume decreases. The passive tension developed across the diaphragm in this way is responsible for the increase in the stiffness of the chest wall (30) and for the decrease in chest-wall compliance observed at low lung volume (13).

The application of PEEP can reduce the upward displacement of the diaphragm-abdominal compartment into the chest wall and move the chest wall into the linear portion of the P-V curve. This effect of PEEP on the P-Vw curve supports the hypothesis that in our patients, a volume-related mechanism determined the low C_start,w. Indeed, C_start,w increased with PEEP, approaching the value of C_inf,w, and the LIPw disappeared at PEEP > 10 cm H₂O. Pelosi and associates (12) failed to find any increase in chest-wall compliance, computed with the end-expiratory occlusion technique, with PEEP of up to 15 cm H₂O in patients with moderate and severe ARF. Accordingly, a decrease in FRC as a possible mechanism for changes in chest-wall mechanics was ruled out (12). In contrast, we found that C_start,w increased significantly with increasing PEEP, approaching the mean value of C_inf,w at PEEP > 5 cm H₂O. On average, changes in C_start,rs paralleled the behavior of C_start,w and not of C_start,L, which did not improve with PEEP, in the overall group of patients.

Recruitment of previously unventilated or poorly ventilated air spaces has been invoked as a major mechanism for the improvement in Pa_O2 following application of PEEP (9, 35, 36). In fact, we found that Pa_O2 improved in patients with an LIP on the P-VL curve, which was abolished by PEEP. In contrast, Pa_O2 did not increase significantly in patients in whom the LIP on the P-Vrs curve was due to an LIP on the P-Vw curve. It has been observed for a long time that in some mechanically ventilated patients with ARF, PEEP may not improve Pa_O2 (9). The traditional explanation for this finding is the lack of alveolar recruitment in such patients, due to P-V characteristics (2, 35, 37) or to the advanced stage of the disease, when pulmonary fibrosis is likely to have replaced edema (22, 38, 39). Our patients were examined mainly in the first week of mechanical ventilation, and hence in what is usually defined as an early stage of ARF (40), when florid pulmonary edema is the characteristic of the disease and PEEP is likely to be effective by inducing lung recruitment (40). We offer here an alternative explanation for why PEEP may not improve Pa_O2 in some patients with early ARF. In those patients, the lungs are already ventilated on the linear portion of the P-V curve, despite the fact that an LIP was present on the P-Vrs curve and was abolished by PEEP. In fact, in those patients the improvement in the P-Vrs curve reflects modifications such as the abolition of an LIP determined by the effect of PEEP on chest-wall mechanics, with little recruitment of poorly or nonaerated alveoli (41). In our patients, C_start,rs decreased at 15 cm H₂O PEEP, the curve being apparently displaced toward the upper, flat portion and so describing an upper inflection. This result is consistent with the recent work by Roupie and colleagues (6), who reported an inflection in the P-Vrs curve in its upper portion at a pressure of 26 ± 6 cm H₂O, corresponding to a volume above PEEP of 610 ± 234 ml, which was very similar to our findings of 27 ± 5 cm H₂O and 590 ± 88 ml, respectively. A similar observation has been reported by Brunet and colleagues (41). Because the slope of the P-Vw curve never decreased with increasing PEEP, this phenomenon was due exclusively to the lung mechanics and was thought to reflect an overdistension of normally aerated alveoli (41). UIP was present in six patients at a PEEP of 10 cm H₂O, and in 10 patients at a PEEP of 15 cm H₂O, with a greater risk for barovolutrauma and ventilator-induced injury (28).

Therefore, the P-Vrs curve can be safely used to prevent excessive lung hyperinflation by avoiding the appearance of an upper, flat portion that would reflect lung mechanics. Hence, if our work questions on one hand the use of P-Vrs curves to set the best PEEP in some patients, on the other hand it supports the use of the P-Vrs curve to prevent excessive lung overdistension and eventual pneumothorax.

In conclusion, our results demonstrate that the LIP on the P-Vrs curve can be determined by chest-wall rather than lung mechanics. This can provide a possible explanation, among others, for cases in which Pa_O2 does not improve with increasing PEEP, even when the P-V curve shows an LIP (Figure 4). In such cases clinicians may overtitrate PEEP (6) with an increasing risk of barotrauma and without advantages on the side of oxygenation. Furthermore, our work extends previous observations (12, 32) that chest-wall mechanics, and not only lung mechanics, can be abnormal in mechanically ventilated patients with ARF, suggesting the need for further investigation to better clarify the mechanism(s) underlying these abnormalities as well as their clinical implications.