INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The elastic pressure-volume (Pel-V) curve of the respiratory system has been used for many years to describe the mechanical characteristics of the lungs in the acute respiratory distress syndrome (ARDS) (1). Moreover, several parameters measured directly or indirectly from this curve have been proposed as a help to adjust the ventilatory settings (3, 5). Special attention has been paid to the measurement of compliance, both as a marker of the severity of the disease, and as an indicator of the effect of positive pressure ventilation in terms of alveolar recruitment (3, 4).

Unfortunately, the use of compliance, a parameter easy to measure at the bedside, has not been convincingly demonstrated to be a reliable indicator of recruitment (8), despite early encouraging findings (5). The shape of the Pel-V curve in ARDS is often sigmoidal, or described as having three main segments. A lower segment with a low compliance indicates that some lung compartments do not receive insufflated gas because airway closure and/or alveolar collapse prevent their inflation. Above a certain portion of the curve, frequently referred to as the lower inflection point, the slope of the curve, i.e., compliance, increases over a nearly linear segment. This often sudden change has been interpreted as a reopening of alveoli and as a marker of recruitment.

Dynamic aspects of the collapse-reopening phenomena (11), and their influence on the shape of the curve, have been given limited attention in previous studies, which may explain discrepant findings in the literature. Indeed, specific pressure- volume (P-V) curve gives insight into the state of the system only at the time when the maneuver is performed. After an increase of positive end-expiratory pressure (PEEP), a progressive time-dependent recruitment may affect the slope of the P-V curve (8, 12). Because events immediately preceding the recording may influence the state of the lung when the P-V recording is performed, using a single insufflation to trace the P-V curve may be important. For this purpose a low-flow inflation technique that allows recording of P-V curves during a single insufflation following an adjustable expiration has been designed.

The P-V curve has the potential to be used to adjust ventilatory settings and it is important to well understand how the curve and the measurement of compliance are affected by alveolar recruitment. This has been investigated by several studies in the past (8, 12, 16); however, this problem was only partially solved since (1) the main goal of most of these studies was to look at the predictive value of the curve performed on zero end-expiratory pressure (ZEEP) (8, 16); (2) all previous studies used the "effective" compliance using "relatively" high tidal volumes (VT), which may markedly differ from the chord compliance both because of the inflection of the P-V curve at low lung volume and because of its flattening at end-inspiration (18). Because the net result on "effective" compliance will depend, on the one hand, upon the relative magnitude of these inflections and, on the other hand, upon the true changes in chord compliance, the result may be totally misleading for the clinician and will depend on the size of the VT; (3) a single inflation technique as the continuous flow technique was not used for tracing P-V curves, whereas the use of multiple VT at random may by itself influence volume history; and (4) alveolar recruitment may also happen during the inflation maneuver used to trace the P-V curve and should be discussed separately from the progressive time-dependent recruitment observed with PEEP (12).

The present study evaluated the amount of derecruitment that occurs in acute lung injury (ALI) when, in a steady-state situation, PEEP is eliminated during a single prolonged expiration to the elastic equilibrium volume. This allowed estimation of the amount of recruitment that had been previously gained with PEEP. Thus, the effect of this recruitment/derecruitment on the shape of the P-V curve and on the measurement of compliance was studied.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

During the period of the study, patients requiring mechanical ventilation with a fraction of inspired oxygen (FI_O2) equal to or greater than 0.5 for more than 24 h and fulfilling the criteria for ALI as defined by a Pa_O2/FI_O2 < 300 mm Hg, bilateral opacities on the chest X-ray, and no elevated pulmonary artery wedge pressure when available or no history suggestive of elevated left atrial pressure, were regarded as potential candidates. Exclusion criteria were a documented history of chronic obstructive pulmonary disease (COPD), presence of a chest tube with persistent leak, or contra-indication for sedation and paralysis. Ten patients were studied on 11 occasions, one patient being studied twice for two different episodes. In nine cases, patients met the criteria for ARDS (PO₂/FI_O2 < 200 mm Hg). The lung injury score (LIS) was computed as described by Murray and coworkers (18), using the number of quadrants involved on the chest X-ray, the Pa_O2/FI_O2 ratio, the level of PEEP, and the quasi-static compliance (measured as tidal volume divided by plateau pressure minus total end-expiratory pressure), each scored from 0 to 4; the mean of these scores gave the LIS. The general characteristics of the patients are reported in Table 1. The protocol was approved by the Henri Mondor Ethics Committee and informed consent was obtained from patients' next of kin.

All patients were intubated and ventilated in the volume-controlled mode (Servo Ventilator 900C; Siemens-Elema AB, Sweden). The VT was 8 to 10 ml · kg¹, rate about 18 min¹, an inspiratory/total time ratio 0.33. The level of PEEP was chosen according to clinical requirements to improve oxygenation without causing hemodynamic compromise. Patients were supine, sedated, and paralyzed by a continuous infusion of midazolam-fentanyl and vecuronium bromide. Routine care such as aspiration was performed some minutes before the study. Treatment according to the clinical routine was maintained stable during the period of data collection. The absence of leaks in the circuit was ensured by establishment of a nearly stable airway pressure toward the end of a 12-s postinspiratory pause as estimated from the computer screen display.

Equipment

A system comprising a computer-controlled Servo Ventilator 900C, a ventilator/computer interface, and an IBM-compatible computer was used (Figure 1). The flow and expiratory pressure transducers of the ventilator were regularly calibrated. Volumes were measured at BTPS. The signals were fed to the computer and A/D converted at 50 Hz.

View larger version (10K): [in this window] [in a new window]   Figure 1.   The equipment comprises a ServoVentilatory 900C, a ventilator/computer interface (VCI), and a computer. Signals from D/A converters control the ventilator. Signals from the flow and pressure transducers of the ventilator are read by A/D converters.

Application of analog signals to the external control socket of the ventilator allowed the computer to control respiration rate, level of PEEP, and minute volume. The external control signal has an immediate effect. Accordingly, if, for example, the respiration rate is lowered during a specific expiration, this expiration will be prolonged. If the external signal for minute ventilation is oscillating during a specific inspiration, this will lead to a modulated oscillating inspiratory flow.

Recording and Analysis of the Pel-V Curve

After an expiration prolonged to 6 s during which the pressure was either maintained at PEEP or brought to ZEEP, a predefined target volume was insufflated during a 6-s-long inspiratory phase. The target volume was chosen so that one would expect a pressure of about 40 cm H₂O. Should the pressure reach 50 cm H₂O before the target volume was reached, the insufflation was stopped. During the insufflation the flow was modulated in a sinusoidal way at 1 Hz (Figure 2). The variations in flow rate thus accomplished make it possible to calculate inspiratory resistance of the respiratory system as shown subsequently. The following expiration was prolonged in order to allow expiration of the large insufflated volume.

View larger version (15K): [in this window] [in a new window]   Figure 2.   Superimposed original recordings of inspiratory Pel-V curves from ZEEP (heavy line) and from PEEP (thin line) from one patient. Flow and pressure tracings versus time for the two recordings are similar until the prolonged expiration (6 s) which occurs at ZEEP and PEEP, respectively. During the insufflation at modulated flow the pattern is similar for the two recordings until maximal pressure is reached. The two volume tracings are aligned at A, which for both curves represents the volume after an ordinary expiration at PEEP. Reference for volume is then for both curves the elastic equilibrium volume attained after the expiration at ZEEP.

The recorded data for flow () and pressure from the insufflation period were analyzed in order to construct the Pel-V curve. The data were transferred to a spreadsheet (EXCEL 5.0; Microsoft) in which the analysis was automatically performed. Flow rate was corrected for gas compression in ventilator tubes. To obtain the elastic pressure (Pel) from the measured total airway pressure (Ptot), two steps are used as depicted in Figure 3. First, in order to calculate the tracheal pressure (Ptr), the resistive pressure of the connections from the site of pressure measurement (Y-piece) up to the tracheal end of the endotracheal tube (Pres_tube), was subtracted from the measured total pressure, Ptot.

(1)

View larger version (18K): [in this window] [in a new window]   Figure 3.   Steps in the analyses of the inspiratory Pel-V curve. From the recorded total pressure (Ptot) the resistive pressure from the Y-piece to the end of the endotracheal tube is subtracted to derive tracheal pressure (Ptr). With a numerical technique Ptr is analyzed to derive resistance of the respiratory system. Pel is then calculated by subtraction of the resistive pressure of the respiratory system from Ptr. The Pel-V curve is shown both as a thinner line corresponding to the measured data and as a heavier line which shows the mathematical description according to Equation 3. The linear part of the Pel-V curve is extrapolated (interrupted line). The lower and upper inflection points (LIP and UIP) are determined as described in text.

K₁ and K₂ were determined in vitro for each type of tube and connections. This gives an estimate of Ptr, but the in vivo characteristics of the tube may correspond to higher resistance. The calculation of the total resistive pressure drop will not be affected, however, because the calculation of the resistive pressure due to the lower airways will take into account this difference; the distribution of resistances between the tube and the airways may be slightly inaccurate in this case. The calculated Ptr comprises the resistive pressures of the respiratory system, Pres_RS, and the dynamic elastic recoil pressure, Pel.

During the insufflation Ptr varies with time (t) because its components Pres_RS and Pel vary with flow rate, , and volume, V, respectively.

(2)

where R_RS is the inspiratory resistance of the respiratory system.

Modeling of the curve was necessary to get quantified information. The aim was to mathematically define the main part of the Pel-V curve in order to make possible unbiased statistical comparisons between conditions and patients. We decided to analyze each curve as comprising three segments, separated by a lower and an upper inflection point (LIP and UIP, respectively). We considered the segment between these two points as linear and the upper segment as curvilinear. The following mathematical model of the Pel-V relationship was thus applied. No mathematical description of the lowest segment was used. The model therefore only describes the Pel-V curve above the LIP.

(3)

For the linear segment between LIP and UIP the relationship is described by the coefficients Pel₀ and C_lin. Pel₀ is the intercept of the extrapolated linear segment with the pressure axis and C_lin is compliance over the linear segment. Accordingly, the UIP was defined as the point where the statistical analysis indicated the beginning deviation of the Pel-V curve from a straight line. Additional volume increments lead to increments in pressure from Pel_UIP according to the second term in the lower part of Equation 3. This implies that compliance falls linearly with additional volume from its value C_lin, until it reaches zero at maximum distension of the lungs, i.e., at V_max. The equation for the segment above UIP complies with the equation of Salazar and Knowles (19) and has been applied in recent studies (20, 21).

R_RS and the coefficients that define the Pel-V curve from LIP (i.e., Pel₀, C_lin, V_uip, V_max) were estimated from Ptr/ data with a numerical technique, in accordance with the principle of least sum of squared deviations (Solver, Excel 5.0). This implies that coefficients were sought which together yielded a minimum sum of squared differences between measured values of tracheal pressure and values calculated according to Equations 2 and 3.

The two Pel-V curves recorded from ZEEP and PEEP in each patient were aligned at the end-expiratory volume of the ordinary breath preceding the measurement (Figure 2). Accordingly, the volume scale of both curves refers to the elastic equilibrium volume reached after the prolonged expiration at ZEEP. The pressure at the lower inflexion point, P_LIP, was defined from the point of the Pel-V curve at which the measured pressure deviated from the mathematical model with more than 0.5 cm H₂O (Figure 3). Resistance of the respiratory system was also measured as effective resistance for the complete respiratory cycle, R_EFF,RS, obtained during passive tidal ventilation in accordance with Varène and Jacquemin (22).

Recording of the Static Pel-V Curve

To ensure that the method used in this study gave comparable results to more conventional static measurements, the lower inflexion point and linear compliance were also determined with the multiple occlusion technique described previously (7, 8, 15). The recording of the static Pel-V curves was performed after similar prolonged expirations at ZEEP and at PEEP as were applied for the dynamic recording. The mathematical analysis of the static Pel-V curve was made following Equation 3.

Protocol

During the whole study the patient was ventilated with PEEP. The first Pel-V curve was recorded after a prolonged expiration during which PEEP was maintained. Because any maneuver by itself may change the state of the system, a period of 10 to 15 min after this first curve then allowed baseline conditions to be reestablished. The second Pel-V curve was then recorded after a single prolonged expiration during which PEEP was eliminated. This Pel-V curve was accordingly recorded from the elastic equilibrium volume reached at ZEEP. The volume exhaled from end inspiration at PEEP to end expiration with ZEEP was measured. After new periods of 15 min the static Pel-V curves from PEEP and from ZEEP were recorded with the multiple occlusion technique.

Statistics

Values are given as means ± SD. Differences between curves recorded from ZEEP and from PEEP were compared with Wilcoxon test for paired samples. Significant differences are considered for p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In all patients and over a wide range of Pel, the volume for a given pressure was lower in recordings starting from ZEEP compared with those starting from PEEP (Figure 4). Over the range of 15 to 30 cm H₂O data were obtained both in ZEEP and in PEEP for all subjects. The significant volume deficit of the curve recorded from ZEEP was 205 ± 100 ml at 15 cm H₂O (p < 0.01) and decreased to 78 ± 93 ml at 30 cm H₂O (p < 0.01) (Figure 5).

View larger version (11K): [in this window] [in a new window]   Figure 4.   Inspiratory Pel-V curves recorded from ZEEP and from PEEP in a representative patient.

View larger version (11K): [in this window] [in a new window]   Figure 5.   Average volume against Pel in the range within which data were obtained in all patients at both PEEP and ZEEP, i.e., 15 to 30 cm H2O). **p < 0.01.

Over the range 15 to 30 cm H₂O, the Pel-V curves recorded from ZEEP had a significantly higher "chord" compliance, i.e., a higher slope, compared with those recorded from PEEP (Figure 6). Thus in the presence of significant recruitment, for a given pressure, the compliance was always lower on the curve from PEEP than from the curve from ZEEP. The average difference in compliance was 10.0 ± 8.7 ml/cm H₂O at a Pel of 15 cm H₂O (p < 0.01), and 5.4 ± 5.5 ml/cm H₂O at a Pel of 30 cm H₂O (p < 0.01).

View larger version (13K): [in this window] [in a new window]   Figure 6.   Average compliance plotted against Pel. Statistical analysis was only made over the range in which data were obtained in all subjects at both PEEP and ZEEP, i.e., 15 to 30 cm H2O. **p < 0.01.

To examine the relationship between the individual values of the LIP and the volume difference found between Pel-V curves recorded from PEEP and ZEEP, P_LIP observed at ZEEP was correlated to the volume difference between the two curves at 15 cm H₂O (V_15cmH2O). A significant correlation was found: P_LIP = 2.26 + 0.024 ·V_15cmH2O (with P_LIP in cm H₂O and V in L; r = 0.67, p < 0.01).

Inspiratory resistance, R_RS, determined according to the iterative technique (Equations 2 and 3) was on average 4.3 ± 2.6 cm H₂O/(L/s). Effective resistance for a full respiratory cycle was 5.3 ± 2.8 cm H₂O/(L/s). The difference was significant (p = 0.005), while the correlation coefficient between the two measures was 0.9.

C_lin measured with the dynamic method at ZEEP and at PEEP correlated closely to C_lin measured with the multiple occlusion technique (r = 0.95). With the dynamic method C_lin was 2.6 ± 4.5 ml/cm H₂O lower than the values obtained with the static method (p = 0.01). Similarly, P_LIP, identified only at ZEEP, measured with the dynamic method correlated closely to P_LIP measured with the static method (r = 0.95) although it was 0.9 ± 1.1 cm H₂O lower (p = 0.01).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The first finding of this study was that, in all patients, the Pel-V curves obtained from the end-expiratory lung volume resulting from a PEEP setting were shifted, compared with the curves traced from the elastic equilibrium volume of the respiratory system, toward upper values of lung volume, in a pressure range far above the LIP observed on ZEEP. This finding, suggestive of alveolar recruitment, was accompanied by a decrease in the chord compliance with PEEP, for any given pressure.

The methods for recording and analysis of static and dynamic Pel-V curves have been presented earlier (20, 23, 24). Both methods allow strict standardization of the procedure in a setup which offers the possibility for the operator to select the breaths preceding the inflation during which P-V curves are recorded. These features, together with the automated and unbiased derivation of results, imply that methodological errors are low (20). A prerequisite for the present study was that the method allowed alignment of the Pel-V curves recorded from ZEEP and from PEEP along a common volume scale as described by Ranieri and coworkers (8). This was achieved with the computer control of the measurement sequence, illustrated in Figure 2. The dynamic inflation technique is an attractive alternative to the time-consuming recording and analysis of the static Pel-V curve measured with the multiple occlusion technique. The comparisons between the static and the dynamic methods confirm previous findings that similar results are obtained, particularly with respect to compliance and the position of the LIP (23). A lower compliance and a more evident UIP observed with the dynamic method is expected as the buildup of viscoelastic pressure during the dynamic insufflation (and/or alterations in distribution of ventilation causing pendelluft) reduces the effective compliance. The viscoelastic properties are nonlinear at high distending pressures in normal and in sick lungs (21, 25). Such nonlinearity may significantly contribute to higher recoil pressures during insufflation with high pressures as in the present study. Although this study was not primarily designed to study viscoelastic properties, the lower values of dynamic compliance than the static compliance suggest that viscoelastic pressures can be substantial in ALI/ARDS. This notion supports the idea of using dynamic rather than static recordings of Pel-V diagrams, as the former give more realistic information about the pressures to which peripheral lung units are exposed during insufflation.

The inspiratory resistance during the insufflation could in this study be measured because of the flow modulation. The oscillating flow therefore offers the possibility to more accurately determine the elastic recoil pressures (Pel). The inspiratory resistance determined during insufflation was significantly lower than effective resistance which represents the full respiratory cycle. As inspiratory resistance is lower than expiratory resistance, this lower value was expected. The difference between inspiratory R_RS and effective resistance R_EFF,RS, was modest, reflecting that the subjects had no significant obstructive disease. Furthermore, R_EFF,RS was measured at PEEP, and low lung volumes at which expiratory resistance tends to be particularly high, were avoided (25).

The pressure-volume curve of the respiratory system is influenced by chest wall elastic properties. Several studies in medical patients, however, have shown that chest wall compliance did not influence the shape of the curve (7, 17, 26). Ranieri and coworkers have shown, however, that abnormally high value of the elastance of the chest wall could be observed in patients with postabdominal surgery ALI, and may lead to underestimation of the volume reaching the UIP of the total P-V curve, with regard to the effects on the lungs (26). Mergoni and coworkers also recently showed that small values of the LIP, in the range 2 to 4 cm H₂O, could be entirely explained by the chest wall (27). This is certainly an additional limitation for the study of the P-V curve of the total respiratory system. We do not think, however, that the absence of esophageal pressure measurement in our study is an important limitation. First, highly abnormal values of chest wall elastance substantially influencing airway pressure measurements have been described in patients with trauma or complicated abdominal surgery, i.e., situations known to alter chest wall properties (26, 28). It is unlikely that, in these medical patients, the chest wall played a substantial role in the compliance values, as already shown (7, 17, 26). Second, the chest wall elastance does not influence the measurement of recruited volume, as estimated in our study. Lastly, validation of the esophageal balloon technique may be debated in such disease where heterogeneity in lung parenchyma induces heterogeneity in regional displacement of the lung adjacent to the esophagus.

The difference between two Pel-V curves recorded from ZEEP and from PEEP showed that, in all patients, the volume at a given pressure was lower when the lung was inflated from the elastic relaxation volume reached at ZEEP than when it was inflated from the higher end-expiratory volume at PEEP. Such a difference can be attributed to a derecruitment of lung units that occurs during a single expiration down to the elastic equilibrium volume attained during a 6-s-long expiration at ZEEP. Such findings are in agreement with the findings of Ranieri and coworkers (8) and Valta and coworkers (16). The loss of lung volume observed during insufflation after such an expiration suggests airway and alveolar collapse. When the lung was inflated from ZEEP, a longer linear segment with a higher compliance was observed than when the lung was inflated from PEEP. A segment with high compliance should, according to commonly held views, represent a well-recruited lung. At first sight, this could seem to contradict the finding of a lower volume observed during the insufflation from ZEEP. This warrants discussion of physical and biological significance of compliance.

By definition, compliance is volume change over pressure change. During insufflation, lung volume change may represent two physical phenomena. One is distension of previously open lung units. A second phenomenon is recruitment of previously collapsed or closed units. At a certain pressure, when a lung unit "pops" open, this unit has an "infinitely high" compliance. Accordingly, as long as lung units of a partially collapsed lung are recruited during an insufflation, these units contribute to a higher compliance than that observed at the same airway pressure in the same lung when it is fully recruited. Accordingly, after an expiration to ZEEP, the finding of a long linear segment of the Pel-V curve with a high compliance can be explained by a continuous recruitment of collapsed lung units occurring during the insufflation. Once this recruitment has been established, the curve presents a steeper slope, because no recruitment is occurring during the maneuver. At a pressure of about 30 cm H₂O the two curves recorded from ZEEP and from PEEP tend to merge. Only above that pressure a full, or nearly full, recovery of the volume deficit caused by the deep expiration is regained.

Regarding the uniform sequence in which the PEEP and ZEEP P-V curves were performed, we used this approach for two reasons: first, the time constant for progressive recruitment could be longer than for derecruitment and collapse; second, this protocol allowed ventilation of the patient without PEEP only for the 6-s-long expiration, which precluded any clinical problems with oxygenation.

The LIP is an established sign of recruitment of lung units with increasing airway pressure. Our observations are in agreement with the concept that at the LIP, or within a narrow range of airway pressure around LIP, a sufficiently large number of lung units open up and thereby contribute to compliance so that a rather distinct inflection is often observed. The lung is, however, nonhomogeneous, and particularly so in ALI/ARDS. Furthermore, at a certain airway pressure the distending pressure of different lung units varies because of the gravitational pleural pressure gradient. It appears inconceivable that all collapsed lung units should open up at nearly the same airway pressure. Indeed, Gattinoni and coworkers have shown that more and more lung zones get re-aerated at successively higher airway pressure (29). In this study, during insufflation from ZEEP, the deficit in lung volume compared with PEEP progressively decreased to become nonsignificant at a pressure above 30 cm H₂O. In other words, a continuous recruitment occurred during insufflation from ZEEP up to 30 cm H₂O. That pressures up to and above 30 cm H₂O are needed before a complete recruitment is achieved is in agreement with the data of Gattinoni and coworkers (29).

Although a continuous recruitment of more and more lung units offers a reasonable explanation for the observations after a single expiration to ZEEP, other mechanisms may be considered. If, above the LIP, the same number of lung units would be open during insufflation from ZEEP as from PEEP, the findings would imply that lung units at a given pressure are smaller when inflated from ZEEP than when inflated from PEEP. Or, in other terms, at a given volume, lung units would exert a higher Pel when inflated from ZEEP. Such a higher Pel could be caused by a higher surface tension (30). The surfactant alveolar film might, when compressed at low lung volumes, loose active molecules at the gas-liquid interface. During a subsequent expansion, the surface tension would be increased. Such phenomena, which are related to surfactant-dependent pressure-volume hysteresis of the lung, might contribute to apparent derecruitment of lung units during insufflation from ZEEP. Beydon and coworkers have, however, not been able to demonstrate a significant Pel-V hysteresis in ALI/ARDS at ventilation at ZEEP nor at ventilation with PEEP, at least in the VT range (25). Numerous morphological studies have also demonstrated that reopening of collapsed tissue is an important feature in ventilation of surfactant-deficient lungs as in ALI/ARDS (4, 29, 31, 32). Collapse of airways and/or alveoli is thus considered to be the predominant explanation for our findings although we cannot exclude a contribution from surfactant film hysteresis.

Measurement of compliance has been recommended as an aid to adjust ventilatory settings and, particularly, for titration of PEEP (5). When, in some studies, measurements were made in the lowest or in the linear part of the Pel-V curve the value of compliance was found to correlate to recruitment achieved with PEEP (12, 33). Other investigators, however, did not find significant changes in compliance with PEEP despite the demonstration of recruitment, or no correlation (8, 12, 16). At the bedside, measurements of compliance are thus often disappointing (8, 12). Several researchers pointed out that one factor introducing confusion was the size of the VT in relation to the upper, curvilinear part of the P-V curve (17, 34). Our findings indicate that another drawback exists. When a single curve from ZEEP is considered, the considerable difference of slope of the initial part of the P-V curve with that of the linear part, is a valid indication of alveolar recruitment (3, 13) although it does not quantify this process. Bedside measurements of the compliance at different end-expiratory pressures, however, may be totally misleading, and this is understood on the basis of the results of this study. The change in effective compliance within tidal ventilation will indeed depend on opposite factors: whereas the ongoing recruitment occurring during one inflation maneuver, as observed within a single P-V curve, will tend to increase compliance, the effect of the established, time-dependent, recruitment obtained with PEEP will tend to decrease its value, as shown in this study. Measurement of "effective" compliance over the full VT is not a useful tool in setting PEEP.

Recently, Gattinoni and coworkers described different patterns of recruitment and of lung mechanics when comparing ARDS of pulmonary origin to extrapulmonary ARDS (28). In particular, they found no recruitment with 15 cm H₂O of PEEP in patients with pulmonary ARDS. We believe that these latter results, which contradict previous results from the same group using computed tomographic (CT) scan (4), and our present results as well, are partly explained by the methodology used based on "effective" compliance. Because calculating this compliance assumes a linear relationship between pressure and volume, it can markedly overestimate the lung volume at ZEEP over a large range of pressure in case of pronounced inflection at the curve. This will lead to underestimation of the recruitment with PEEP.

In addition to showing evidence of recruitment on the P-V curve, all the patients in this study improved their oxygenation with PEEP. Although this was not studied systematically in conjunction with the P-V curves, all our patients had either a PEEP trial (from ZEEP to PEEP) or an increase in the PEEP level in the 48 h surrounding this study. For a mean increase of 5.6 ± 2.8 cm H₂O of PEEP, Pa_O2 increased in all patients, by 73 ± 73 mm Hg (10-228) at an FI_O2 of 0.93 ± 0.14.

In conclusion, in ALI and ARDS, a single prolonged expiration to ZEEP leads to a derecruitment of lung units. During insufflation, when at certain pressure a significant recruitment of lung units begins, a rather distinct LIP may be observed in the Pel-V curve. A straight segment of the Pel-V curve with a high compliance is compatible with an ongoing continuing recruitment of lung units after a single expiration to ZEEP.