INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is generally believed that avoidance of both overdistension and end-expiratory collapse of lung units will reduce the mechanical stress of artificial ventilation (1). Analyzing respiratory mechanics assesses this effect. It has been found that between the lower and the upper inflection point (LIP and UIP, respectively) of the inspiratory limb of a quasistatic pressure- volume (PV) loop, the PV relationship is steep and linear (2). It is widely assumed that this segment of the PV curve, with its constantly ascending slope, indicates the range in which all potentially recruitable lung units are recruited but not overdistended. By translating the PV curve into a compliance curve (with compliance corresponding to the first derivative of the PV relationship), this segment between the LIP and UIP, with its maximally ascending slope, can also be described as a segment of maximal and linear compliance in the form of a flat horizontal line. Accordingly, by setting the ventilator in terms of respiratory mechanics, the end-inspiratory volume is delivered between the boundaries of the LIP and UIP (i.e., where compliance is both maximal and constant), and is seen as a flat horizontal line over the tidal volume (VT). It is unclear, however, whether and how information about elastic properties obtained during the highly artificial quasistatic PV loop maneuver applies to ongoing ventilation in which flow, pressure, and volume differ from quasistatic conditions (3). More importantly, recent studies (4) have indicated that recruitment also takes place along the steep, constantly ascending segment of the PV curve. Obscured by this segment of the PV relationship, the reopening of lung units during inspiration may therefore go undetected. With a modified occlusion method that "freezes" the images of mechanical events at consecutive increments of VT, we attempted to visualize these nonlinearities of compliance within the VT under study. The elastic properties of the respiratory system were assessed by determining the compliance of small increments in VT (up to full VT), and this procedure was repeated at steps of positive end-expiratory pressure (PEEP) from 3 cm H₂O to 24 cm H₂O. We then mapped the configuration of the resulting compliance curves over VT, and compared this map with the results of conventional analysis of respiratory mechanics and gas exchange. The principal aims of the study were to determine whether: (1) a PEEP level at which compliance was maximum and linear over the full VT could be identified in individual piglets; and (2) whether this PEEP level was also associated with maximum gas exchange.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The investigations were conducted at the laboratories of the Section of Anesthesiology and Intensive Care Medicine of the Department of Surgical Sciences of the University Hospital Uppsala, in conformity with the Helsinki convention for the use and care of animals. The local medical ethics committee for animal experimentation reviewed and approved the study protocol. The study used 14 healthy male and female piglets of Swedish landrace breed (weight: 25.5 ± 1.8 [mean ± SD] kg). At the end of the experimental phase of the study the anesthetized animals were killed with an overdose of potassium chloride.

Study Design

The animals were subjected to bronchoalveolar lavage (BAL) under general anesthesia (as described subsequently), and after a 20-min stabilization period a quasistatic PV curve (as subsequently described in detail) was obtained at zero PEEP (ZEEP), with a further curve being recorded 4 h later, at the end of the experiment. The lungs of each animal were ventilated at PEEP settings of 3, 6, 9, 12, 15, 18, 21, and 24 cm H₂O. Each PEEP level was maintained for a period of 20 min before measurements (including assessments with multiple inert gas elimination) were performed. A 2-min recruitment procedure (as described subsequently) was conducted before ventilation at each PEEP setting. To compensate for time-related effects, the sequence of the PEEP levels had been determined beforehand, using a computer-generated random number list and a randomized complete block design.

Respiratory Mechanics

The anesthetized/paralyzed animals were studied in the prone position. Their lungs were ventilated through an endotracheal tube (I.D. = 7.5 mm; Mallinckrodt, Athlone, Ireland), connected via a 60-cm rigid tubing system to a Servo 300 ventilator (Siemens-Elema, Solna, Sweden). Airway pressure and flow were measured with a pressure-flow transducer (CP-100 Pulmonary Monitor; Bicore Monitoring Systems, Irvine, CA) placed between the tracheal tube and the ventilator circuit. The ventilatory frequency was 22 breaths/min, the inspired oxygen fraction (FI_O2) was 1.0, the inspiration-to-expiration (I:E) ratio was 1:1, and the inspiratory flow was kept constant in the volume-controlled mode during all study conditions. VT was set to reach an arterial carbon dioxide tension (Pa_CO2) of 42 ± 5 mm Hg at the end of the 20-min stabilization period. Under this condition, the observed VT (12 ± 0.9 ml/kg on average) remained unchanged during the subsequent study conditions.

BAL

BAL was performed as described previously (7), and 11 lavages (each with 1.2 to 1.5 L normal saline) were done in one session. The lavage effects were assessed in terms of gas exchange, respiratory mechanics, and extravascular lung water through double-indicator dilution [8, 9]), with comparison of volume-controlled ventilation at ZEEP under healthy conditions with that under the surfactant-deficient conditions immediately after lavage. To assess the stability of the induced lung injury over time, this assessment was repeated at the end of the experiment.

Recruitment

The lungs were recruited immediately after lavage, before the establishment of each PEEP setting used in the study, through pressure-controlled ventilation for 2 min with a frequency of 22 breaths/min, on I:E ratio of 1:1, an FI_O2 of 1.0, a PEEP of 25 cm H₂O, and a pressure above PEEP set such as to produce a peak inspiratory airway pressure of 50 cm H₂O and a VT of approximately 600 ml.

LIP and UIP

After BAL and recruitment, the LIP and UIP of the quasistatic PV curve (2) were determined with a modified multiple-occlusion method (10) as follows: PEEP was set to 15 cm H₂O, VT to 12 ml/kg, and ventilatory frequency to 22 breaths/min. According to previous findings (11), these settings would ensure that alveolar recruitment was maintained in most of the animals. After seven breaths, the insufflated volume was reduced to 50 ml and PEEP was set to zero, and seven further breaths were then applied at these settings. At the end of the seventh breath, the end-inspiratory occlusion pressure was measured and the ventilator was switched back to 15 cm H₂O PEEP with a VT of 12 ml/kg for seven additional breaths. In the following run, 100 ml of air was insufflated again at ZEEP, and this procedure was repeated with 50-ml increments until an insufflated volume of 600 ml was reached. The resulting end-inspiratory pressures were used to construct the PV curve from which the LIP and UIP were determined by tracing a straight line with the best visual fit on the linear part of the curve. The first volume step was excluded from analysis, since the pressure reading at 50 ml is not accurate enough to rule out artifacts. This is also the reason why the PV curves at ZEEP in Figure 1 do not start at the 0,0 coordinate on the PV curve. From another viewpoint, the procedure just described may be regarded as a sequence of monotonous ventilations at 15 cm H₂O PEEP and a VT of 12 ml/kg, periodically interrupted by VT values of 50, 100, and 150 ml and so on up to 600 ml and at ZEEP. Measurement of the exhaled volume when PEEP at 15 cm H₂O was stopped allowed assessment of the volume recruited at that particular PEEP level (i.e., the difference in end-expiratory lung volumes at 15 cm H₂O PEEP and ZEEP) (12). Measurement of the exhaled volumes when switching from a VT of 12 ml/kg to the VT values above that level (300 ml to 600 ml) and back allowed the recruiting effects of VT values > 12 ml/kg at 15 H₂O PEEP to be estimated.

View larger version (15K): [in this window] [in a new window]   Figure 1.   Quasistatic PV curve after BAL (bold line) and 4 h later (thin line) in individual animals. x-axis: end-inspiratory (hold) pressure (Paw). y-axis: insufflated volume. An LIP immediately after lavage, determined by tracing a straight line on the linear part of the PV relationship as the best visual fit, is marked with an open circle. Note that neither a distinct LIP nor UIP is a constant finding in these surfactant-deficient lungs. The animal number (#) is indicated near the y-axis, and the animals are arranged according to the severity of lung injury as indicated by the impairment of gas exchange immediately after lavage, with the least severe injury placed at the top of the left column and the most severe injury at the bottom of the right column.

It should be noted that the study protocol did not explicitly include a setting at which PEEP was equal to the LIP that was obtained. However, the small PEEP increments permitted study of a PEEP level very close to this point.

Two-Point Compliance of the Respiratory System

The conventional two-point quasistatic compliance of the respiratory system (C_2P) was calculated at each studied PEEP level at the end of the 20-min application of the PEEP setting under study, according to the formula: VT/(end-inspiratory pressure  end-expiratory pressure). To obtain the end-inspiratory and end-expiratory pressures, the hold functions of the ventilator were used for 5 s.

Sequential Chord Compliance of the Respiratory System

To determine the nonlinearities in respiratory system compliance, we divided VT into 10 fractions of equal magnitude (0.1 VT; 0.2 VT ... 1.0 VT; Figure 2). With a VT of 308 ± 12 (mean ± SD) ml, the average volume increment was thus close to 30 ml per step. The setting under study with full VT and the PEEP level in question was applied for seven breaths, and the insufflated volume was then reduced to 0.1 VT, with PEEP and all other ventilator settings unchanged. Thereafter, 0.1 VT was applied for seven breaths, and at end-inspiration of the seventh breath, after a 5-s hold, the airway pressure at end-inspiratory occlusion was measured. The insufflated volume was then reset to 1.0 VT for seven further breaths, after which it was again reduced, in this case to 0.2 VT, and the end-inspiratory airway pressure was again measured at the end of the seventh breath. This procedure was repeated with further increases in the insufflated volume until 1.0 VT was reached. This measurement cycle was repeated at each PEEP level. The insufflated volumes and the resulting pressures were used to calculate the sequential chord compliance of the respiratory system (C_chord) according to the formula: insufflated volume/(end-inspiratory hold pressure  end-expiratory pressure), and C_chord was plotted over VT (Figure 3). The first volume step was excluded from the analysis, since the pressure reading at 30 ml is not sufficiently accurate to rule out artifacts. For analyzing the configuration of C_chord over the VT curve, the segments were classified as horizontal (representing constant compliance), ascending (representing recruitment), and descending (representing overdistension). The reproducibility of the method used for this had been evaluated in four animals in a pilot study, in which the end-inspiratory pressures at 0.1 VT through 1.0 VT were determined with PEEP ranging from 3 cm H₂O to 24 cm H₂O. The coefficient of variation (CV) for three consecutive measurements was 1.6 ± 0.2% over all PEEP levels, with a tendency toward higher CVs at low PEEP (2.5 ± 0.3% at a PEEP of 3 cm H₂O) than at high PEEP (0.6 ± 0.03% at a PEEP of 24 cm H₂O). The results for a representative animal are shown in Figure 2 (bottom).

View larger version (37K): [in this window] [in a new window]   Figure 2.   Top: Use of multiple occlusion method for determining end-inspiratory pressures at VT increments: VT is divided into 10 fractions of equal magnitude (0.1 VT, 0.2 VT ... 1.0 VT). Full VT (300 ml) was applied for seven breaths at the PEEP level under study, and the insufflated volume was then reduced to 0.1 VT (30 ml), with PEEP and all other ventilator settings unchanged. Thereafter, 0.1 VT was applied seven times, and the end-inspiratory occlusion airway pressure of the seventh breath was measured. The insufflated volume was then reset to 1.0 VT for seven breaths, after which it was again reduced, this time to 0.2 VT, and the end-inspiratory airway pressure was again measured at the end of the seventh breath. This procedure was repeated until 1.0 VT was reached. Cchord was calculated from the insufflated volumes and the resulting pressures. Bottom: Reproducibility of determination of the end-inspiratory pause pressure over the entire range of VT in a representative animal. Three consecutive determinations of end-inspiratory pause pressure were made at each of the PEEP levels of 3, 9, 15, and 24 cm H2O. x-axis: fraction of VT.

View larger version (26K): [in this window] [in a new window]   Figure 3.   Construction of the Cchord over VT curve. (Top panel ) PV curve for a representative animal with end-inspiratory pressures at 0.1 to 1.0 VT plotted over VT, with 9 cm H2O PEEP and volume increments of 30 ml. (Bottom panel ) Resulting sequential chord compliance of the respiratory system. Cchord was calculated and is plotted against VT. Note that nonlinearities over VT are more pronounced and easily discernible in the Cchord curve.

Hemodynamics and Gas Exchange

Intravascular catheters were placed surgically for collection of blood for arterial and mixed venous blood gas analyses (ABL5 and Hemoximeter OSM3, respectively; Radiometer, Copenhagen, Denmark) and measurement of central venous, pulmonary arterial (via the external jugular vein), and aortic pressures (via the carotid artery). The position of the catheters was confirmed by pressure tracing. Cardiac output (CO) was determined from arterial thermodilution curves (13) (Pulsion Medical Systems, Munich, Germany).

Anesthesia and Fluid Management

After premedication (3 mg/kg tiletamine, 3 mg/kg zolazepam, 2.2 mg/kg xylazine, and 0.04 mg/kg atropine, intramuscularly), anesthesia was induced with 500 mg ketamine, 0.5 mg atropine, and 20 mg morphine given intravenously. Anesthesia was maintained with a ketamine infusion (20 mg/kg/h) and morphine (0.5 mg/kg/h), and muscle relaxation was maintained by continuous infusion of pancuronium bromide (0.25 mg/kg/h). To ensure normovolemia, the animals were given a solution of 4.5 g/L NaCl with 25 g/L glucose (Rehydrex; Pharmacia Infusion AB, Uppsala, Sweden) at 10 mg/kg/h, and a 10 ml/kg bolus of dextran-70 (Macrodex 70; Pharmacia Infusion AB).

Multiple Inert Gas Elimination Technique

Ventilation-perfusion ratios (A/) were assessed in seven animals with the multiple inert gas elimination technique (MIGET) (14). In brief, isotonic saline containing a mixture of six inert gases (sulfur hexafluoride, ethane, cyclopropane, enflurane, diethyl ether, and acetone) was infused continuously into a peripheral vein. After an equilibration period of 40 min, arterial, pulmonary arterial, and mixed expired gas samples were collected for subsequent analysis by gas chromatography (Models 5880A and 5890; Hewlett-Packard, Palo Alto, CA). Through mathematical analysis of the inert gas data, each A/ distribution was recovered in a 50-compartment model, and the result with the best fit of data (smallest remaining sum of squares [RSS]) of each sample was used for further statistical analysis. Intrapulmonary shunt (S/T) was defined as the fraction of total blood flow perfusing lung units with a A/ < 0.005. Low A/ was defined as the fraction of total blood perfusion supplying lung units with 0.005 < A/ < 0.1; high A/ was defined as ventilation of regions with ratios of 10 < A/ < 100; and dead space was defined as A/ > 100. The dispersion of ratios was expressed as the logarithmic SD of perfusion distribution (log SDQ). The mean RSS for all measurements was 2.1, with no RSS exceeding 4, indicating a good fit of the measured retention/ excretion data to the derived distributions.

Data Presentation

Individual animals at particular PEEP levels are indicated as animal number/PEEP level (thus "3/9" means animal no. 3 at a PEEP level of 9 cm H₂O). The data are presented as mean ± SD or as median (interquartile range [IQR]). Differences were assessed through a nonparametric analysis of variance (Friedman's test) (15). The significance of any difference was evaluated with the paired-sign test. To evaluate a potential association between variables, Spearman's rank correlation was calculated and the rank correlation coefficient, , was indicated. The level of statistical significance adopted was p  0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of Lavage

With lavage, extravascular lung water increased from 8 ± 1.3 ml/kg (with healthy lungs) to 20 ± 4 ml/kg. It remained at this level during the 4-h period to the end of the experiment. With ZEEP, C_2P was 28 ± 6 ml/cm H₂O before and 17 ± 4 (mean ± SD) ml/cm H₂O immediately after lavage, and 15 ± 4 ml/cm H₂O 4 h later. Mean pulmonary artery pressure () increased from 22 ± 6 mm Hg to 30 ± 7 mm Hg, and calculated venous admixture increased from 13 ± 5% before lavage (ZEEP, FI_O2 = 1.0) to 39 ± 13% immediately after lavage, and was 39 ± 14% with ZEEP at the end of the experiment. The ratio of arterial oxygen tension to FI_O2 (Pa_O2/FI_O2) decreased from 540 ± 105 mm Hg before to 150 ± 105 mm Hg after lavage, and was 165 ± 120 mm Hg at the end of the experiment (for all differences between pre- and postlavage values, p  0.001). In Figures 1 and 4, the animals are arranged according to the severity of the induced lung injury as indicated by the impairment of gas exchange immediately after lavage.

View larger version (37K): [in this window] [in a new window]   Figure 4.   Individual mechanical behavior of the respiratory system in 14 animals with surfactant deficiency at PEEP of 3 to 24 cm H2O and constant VT. Sequential Cchord (smoothed lines) over VT and C2P (gray shaded bars) are shown in individual animals at 3 to 24 cm H2O PEEP (indicated at the bottom). Each row represents one animal (animal number [#] is indicated to the left of the panel ) at increasing PEEP. The animals are arranged according to the severity of their induced lung injury, with the most severely injured at the bottom of the panel. At each PEEP level, 0.1 to 1.0 VT is plotted on the x-axis, with compliance on the y-axis. Note that the points for Cchord do not constitute a true continuum; for ease of visual identification, a smooth line connects them.

LIP and UIP

Immediately after lavage, a lower inflection point was observed in seven animals at a median airway pressure level of 21 cm H₂O (minimal: 16 cm H₂O; maximal: 25 cm H₂O; IQR: 3 cm H₂O), whereas in seven other animals the PV curve was concave. In eight animals (Animals 14, 7, 2, 3, 1, 8, 11, and 13), the PV curve recorded at the end of the experiment showed essentially the same shape as that obtained immediately after lavage, whereas in six animals (Animals 12, 5, 9, 6, 10, and 4) it was shifted to the right (i.e., toward higher pressures) (Figure 1). In five animals (Animals 1, 4, 7, 8, and 10), UIP (median: 28 cm H₂O; IQR: 4 cm H₂O) appeared on the PV curve recorded immediately after lavage.

Effects of Increasing PEEP

With increasing PEEP, C_2P peaked at a PEEP of 12 cm H₂O (31 ± 10 ml/cm H₂O), and thereafter decreased. C_2P reached a maximum at a PEEP of 9 cm H₂O in two animals, at a PEEP of 12 cm H₂O in nine animals, at a PEEP of 15 cm H₂O in one animal, and at a PEEP of 18 cm H₂O in two animals (Table 1 and Figures 4 and 5).

View larger version (14K): [in this window] [in a new window]   Figure 5.   Volume recruited by increasing VT at PEEP 15 cm H2O. x-axis: set VT (ml), y-axis: gray lower line represents difference in end-expiratory lung volume (EELV) between a PEEP of 15 cm H2O and ZEEP (EELV15  EELVZEEP) plus set VT (ml/kg). y-axis: black upper line represents (EELV15  EELVZEEP) plus the volume recruited by increasing VT (ml/kg). The values were recorded during the immediate postlavage construction of the quasistatic PV curve, and represent mean ± SD (n = 14).

Configuration of C_chord Curves

At PEEP  LIP, C_chord was at a minimum in most lungs. C_chord was low both with high and with low PEEP, but the configuration of the compliance curve differed. At high PEEP (18 to 24 cm H₂O), most lungs displayed a descending C_chord, whereas at low PEEP (0 to 6 cm H₂O) an ascending shape was observed (e.g., Figure 4, 14/3, 3/3, 10/3).

The C_chord curve showed a horizontal or near horizontal shape in 2/6, 2/9, 1/12, 11/9, 9/12, 6/15, 4/15, 4/18, and 4/21. With lung injury of increasing severity, more PEEP was required to obtain a horizontal C_chord curve (1/12, 5/12, 6/15, 10/18, 4/21). Occasionally the C_chord curve was nearly horizontal at low PEEP levels and at a low compliance level (2/3, 2/6, 11/6, 12/3, 13/3).

At intermediate PEEP levels, C_chord was nonlinear over VT, with ascending and descending segments following each other, a finding compatible with both recruitment and overdistension within one and the same VT.

Differential effects of PEEP and VT were discernible in some curves (e.g., 7/9), in which tidal recruitment predominated until a horizontal curve was obtained during midinspiration. The next PEEP steps (7/12 and 7/15) again indicated initial recruitment, but now with overdistension during late inspiration. In 3/12, a relatively long initial horizontal segment indicated a maximum volume change per pressure change, but VT created overdistension at end-inspiration.

Recruiting Effects of VT

The median difference in end-expiratory lung volume between a PEEP of 15 cm H₂O and ZEEP was 15 ml/kg (IQR: 6 ml/kg; minimum: 8 ml/kg; maximum: 17 ml/kg, respectively). At a PEEP of 15 cm H₂O, a VT > 300 ml recruited additional lung volume (Figure 5), but the increase was not linear or parallel with the increase in the preset VT. There was a weak association between recruited lung volume and Pa_O2/FI_O2 ( = 0.58, p < 0.001), but no relationship was observed between recruited lung volume and decrease in S/ T.

Gas Exchange and Oxygen Transport

Oxygenation (Pa_O2/FI_O2) increased with increasing PEEP, but the increase was less pronounced at a PEEP above 12 cm H₂O (Table 1 and Figure 6). Oxygen transport remained essentially unchanged between zero and 12 cm H₂O PEEP, but above that level it tended to decrease (ZEEP: 16.2 ± 4 cm H₂O, versus 24 cm H₂O of PEEP: 13.4 ± 2.9 ml/min/kg, p  0.01). Over the entire PEEP range, C_2P showed no correlation either with Pa_O2/FI_O2 ( = 0.4, p < 0.001) or with CO ( = 0.1, p = 0.9). Calculated venous admixture was weakly associated with CO ( = 0.6, p  0.01).  S/ T decreased with increasing PEEP from a median of 46% during ZEEP to 5% with a PEEP of 18 cm H₂O. No correlation was observed between CO and  S/ T. A/ matching improved with increasing PEEP, as indicated by a decreasing log SDQ (Table 2 and Figure 7). Good agreement was found between measured and calculated Pa_O2 (difference < 5%).

View larger version (34K): [in this window] [in a new window]   Figure 6.   Oxygen delivery (DO2, shaded area), C2P (bold line) and PaO2/ FIO2 (thin line) at PEEP steps from 0 to 24 cm H2O (indicated on the bottom line). Values are mean + SD. For significant differences, refer to Table 1.

View larger version (13K): [in this window] [in a new window]   Figure 7.   MIGET data in one representative animal at different PEEP steps. Note that true shunt ( S/ T) decreases with increasing PEEP, and A/ matching is improved (as indicated by reduction of log SDQ).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main observation in the current study was that instantaneous compliance, measured sequentially over small volume increments, varied over the range of VT and could both increase and decrease within the same breath. Finding a PEEP level that resulted in a constant compliance over the full VT was difficult, and such a level could not be derived from conventional respiratory mechanical analyses, nor did it coincide with maximum gas exchange.

Gas Exchange and Hemodynamics

Pa_O2 increased with increasing PEEP, although (most likely by reason of hemodynamic compromise) the increase was less pronounced at PEEP values above 12 cm H₂O (Figure 6 and Table 1). Arterial Pa_CO2decreased at PEEP values from zero to 12 cm H₂O, possibly as an effect of reduced  S/ T. With a further increase in PEEP, deadspace also increased slightly (Table 2), but Pa_CO2 did not change, presumably because improved matching counterbalanced the effect of an increase in deadspace on Pa_CO2. True  S/ T was reduced and matching improved with PEEP below 12 cm H₂O (Figure 7 and Table 2), but little additional effect was seen with increases in PEEP above this level (Table 2 and Figure 7). The minimum  S/ T did not coincide with maximum C_2P or with a horizontal course of C_chord. The previously observed close relationship between  S/ T and CO (16) was not seen in this model. Although the experiment was not designed specifically to test the association between  S/ T and CO, we consider that a major contribution of extrapulmonary factors to the reduction in  S/ T was unlikely, and that the main effect of PEEP on  S/ T was probably exerted through recruitment of alveolar surface. Although the small number of animals investigated with MIGET does not allow firm conclusions to be drawn, we consider that these results suggest a poor relationship between respiratory system mechanics and pulmonary gas exchange or oxygen delivery variables. The different dimensions of the healthy and the diseased lung may explain this poor relationship (17). A lung at one third of its healthy volume (11) may not be able to accomplish the full task of a normal-sized lung without being mechanically overstressed. It would be surprising if one and the same ventilator setting would both achieve stress reduction and provide optimal gas exchange and hemodynamics.

Respiratory Mechanics

Chord compliance and LIP. C_2P is generally used for finding optimal PEEP (18), and in the present study C_2P reached a maximum at a PEEP of 12 cm H₂O. C_2P therefore suggested that, collectively, a PEEP of 12 cm H₂O would provide a good compromise between the requirements of mechanics and gas transport (Figure 6). In the individual animals, however, maximum C_2P did not coincide with constant C_chord. Except at the extreme limits of PEEP, C_2P did not indicate whether overdistension or underrecruitment was responsible for a low C_2P, whereas C_chord distinguished between tidal recruitment (Figure 4, ascending compliance curve, 14/3, 3/3, 10/3) and overdistension (Figure 4, descending compliance curve; 14/24, 3/24, 10/24). The possibility of distinguishing between different causes of a low compliance was particularly helpful when PEEP was set near or at the LIP. With PEEP  LIP (Figure 4), C_2P was at a minimum in most lungs. In addition to being equally low, C_chord showed a sharp decrease. Only C_chord therefore suggested that most lungs were already overdistended at this PEEP level (although a UIP, another potential indicator of overdistension, appeared in only five animals).

Our finding that overdistension occurs with PEEP set near the LIP is supported by two recent experimental studies (19, 20). Rimensberger and colleagues demonstrated that after a recruitment maneuver and with PEEP set below the LIP of the inflation limb but above the closing pressure of the deflation limb, surfactant-depleted lungs could be ventilated at optimum compliance and that open lung conditions could be maintained. This challenges the widely accepted view that in order to keep the lung open it is necessary to set PEEP at or above the LIP, at least under experimental conditions. The LIP presumably corresponds to the pressure required to open the lung. On the other hand, the pressure necessary to keep the lung open after atelectatic regions have been recruited and the high surface tension in the surfactant-deficient lungs has been overcome is much lower. One might even wonder whether a strategy that combines lower PEEP levels with appropriate intermittent recruitment maneuvers (21) would be safer. In humans, there may be good reasons for using fairly high PEEP levels. We doubt, however, whether the justification for this would lie in the specific appearance of the PV curve or in its interpretation. Moreover, the assumption that the steep linear segment of the PV relationship between the LIP and UIP indicates the zone without further tidal reopening of lung units has been called into question on the basis of recent results (3). The apparent nonlinearity of compliance at almost all PEEP levels in our study is in conformity with those results. In addition, Carney and colleagues (22) found a classic PV curve with wide hysteresis of the whole lung, but no hysteresis of the alveolar PV curve nor an increase in alveolar volume with increased airway pressure. In the light of these and related findings (23), the question arises as to whether the change in lung volume as manifested in the PV loop reflects what occurs at the level of 300 million alveoli.

Variations of C_chord over VT

The shapes of the individual relationships between C_chord and VT (Figure 4) varied considerably, indicating that both the lung injury and the mechanical response to ventilation were markedly inhomogeneous and highly individual. Despite this complexity, patterns of individual mechanical response can be identified, with tidal recruitment or overdistension predominating at low or high PEEP levels, and high compliance and near-horizontal shapes, as well as shapes that indicate tidal recruitment followed by overdistension at a single VT, predominating at intermediate PEEP levels. The interpretation of other shapes is less clear: A horizontal C_chord at low PEEP, for example, probably indicates that the pressure built up during delivery of VT is insufficient to recruit a sufficient lung volume to induce an increase in compliance. Differential effects of PEEP and VT are discernible on curves such as 7/9, where PEEP is probably too low and, therefore, tidal recruitment occurs up to midinspiration. Higher PEEP levels (7/12 and 7/15) did not completely abolish the initial recruitment, but caused overdistension during late inspiration. A pattern such as that in 2/12 or 2/15 might indicate that PEEP is initially too high, since overdistension has already begun at early inspiration. The situation in 3/12 is probably more favorable: here, a relatively long initial horizontal segment appears to indicate a maximum volume change per pressure change, but VT again creates overdistension at end-inspiration. The assumption that recruitment/derecruitment are rapid events taking place within a single breath is supported indirectly by the observation by Neumann and colleagues (24) of some collapse and recruitment even within as short a time span as 0.6 s.

That the lung volume increases in a nonlinear manner during insufflation is further suggested by the disproportionate increase in volume induced by VT in steps of > 300 ml at 15 cm H₂O PEEP (as indicated by the difference between the set VT and the exhaled VT when switching between a VT of 12 ml/kg and the VT steps up to 600 ml; Figure 5). The volume gain per additional 50 ml of VT above 300 ml ranged between 240 ml and 90 ml, with a decreasing tendency toward maximal VT, probably as a result of overdistension.

With respect to recruitment and overdistension, the lung does not behave as a single compartment (25). Rather, the insufflated volume at a given PEEP sequentially recruits compartments with differing elastic properties. Within one and the same inspiration, this includes overdistension of open lung units before the airway pressure reaches a level sufficient to recruit closed units. Depending on the number of lung units that successively "pop" open, this process evolves either in a fairly smooth and linear manner or in a more stepwise fashion. Because of their respective contributions to the overall mechanical behavior of the lung (26), the characteristics of overdistension of the easily recruited units, together with or followed by the characteristics of recruitment of the units that are difficult to recruit, will appear on the same C_chord curve. Smoothly ascending, descending, or near-horizontal shapes indicate that there is a population of lung units with similar elastic properties that is large enough to dominate the overall mechanical response. An uneven shape, with signs of both recruitment and overdistension, indicates that the contribution of lung units with differing elastic properties to the overall mechanical response is more balanced, and that these populations are sequentially addressed by the increasing airway pressure.

Critique of Methods and Significance of the Lavage Model

The considerable variety in individual C_chord over VT plots (Figure 4) found in our study was quite unexpected, as the population studied was homogeneous with respect to body mass and age, as well as onset, type, extent, and mechanism of induced lung injury. We may safely rule out the possibility, however, that the shapes of the C_chord curves are mere artifacts, since the CV for three consecutive determinations of C_chord was below 2% over the entire tidal breath (see Figure 2, bottom).

Generation of the PV curve requires a maneuver whose PV history differs from that of ongoing ventilation (27). Modifications of the original method have been proposed to maintain the PV history of ongoing ventilation (3). The determination of C_chord in the present study was based on a multiple-occlusion method (10, 28), but with the following modifications: (1) instead of being studied at ZEEP, PV relationships were studied at several PEEP levels, including the PEEP setting during ongoing ventilation; (2) instead of analyzing a more extended volume range, we limited the insufflated volume to full VT; (3) instead of applying the volume increments only once, we applied them for seven breaths. The result was a sequence of "frozen" images of the elastic properties of the lung at succeeding fractions of the VT, representing a magnification of conditions that are of a transient nature during ongoing inspiration to full VT.

The impact of PV behavior is also reflected by the difference between C_2P (which corresponds to compliance at 1.0 VT) and C_chord at 1.0 VT (Figure 4): uninterrupted ventilation to full VT over 20 min yielded a compliance (C_2P) that was higher in many animals, especially in the PEEP range of 6 to 15 cm H₂O, than the compliance obtained after repeated application of fractions of VT for seven breaths. In fact, the elastic properties of the respiratory system are greatly influenced by the previous ventilatory pattern employed. The change in PV behavior induced by a very short measurement maneuver with settings that differ from the ongoing ventilation will alter these elastic properties, as recently demonstrated by Liu and colleagues (29). Only methods that evaluate respiratory system mechanics noninvasively at uninterrupted ongoing ventilation (30, 31) have the potential to eliminate this problem. Furthermore, the random order in which the PEEP levels were set in our study apparently also influenced the individual elastic properties in different animals as can be inferred from the rightward shift of the PV curve at ZEEP in six animals (Animals 12, 5, 9, 6, 10, and 4; Figure 1). We also point out that the potential contribution of increased surface tension was not taken into account in our interpretation of C_chord curves.

How relevant, it may be asked, is a model of moderate lung injury induced by surfactant depletion for the understanding of respiratory mechanics in human adult respiratory distress syndrome (ARDS)? The main problem in human ARDS consists of dysfunction and compositional alteration of surfactant (32), rather than of mere lack of surfactant. The lavage model portrays important pathophysiologic changes, such as are seen in early ARDS (e.g., reduced compliance and a specific appearance of the quasistatic PV loop, impaired gas exchange, increased extravascular lung water, pulmonary hypertension, and widespread atelectasis) (33). It also reproduces some of the key histologic changes of ARDS (7, 34). More important, however, than the similarities of the lavage model to human ARDS or to other lung injury models is the opportunity it offers to review our interpretations of pressure, flow, and volume signals and the mathematical model underlying these interpretations.

Conclusions

Compliance was nonlinear during the course of VT over the entire PEEP range studied, with signs of both recruitment and overdistension during the same VT. A PEEP level at which the compliance curve was horizontal or nearly horizontal could be identified in each individual animal. At PEEP  LIP, most lungs were overdistended when this PEEP level was set after a recruitment procedure. The PEEP level at which C_chord was constant over VT did not coincide with the PEEP level at which C_2P, gas exchange, or oxygen transport was at its maximum.