INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Imaging by thoracic computed tomography (CT) scanner has led to a better understanding of the pulmonary morphology of human acute lung injury (ALI). A precise quantification and regional analysis of positive end-expiratory pressure (PEEP)- induced alveolar recruitment (1) can be performed together with an assessment of PEEP-induced overdistension (8). However, a thoracic CT scanner is not always available and the pressure-volume (P-V) curve is often used to adjust ventilatory settings (9).

In patients with ALI, the P-V curve is markedly altered compared with normal subjects. The total lung capacity is dramatically reduced and the pressure at which overdistention occursthe upper inflection pointis lower. The initial part of the P-V curve is often characterized by a lower inflection point, which is classically considered as the pressure required to reopen distal bronchioles that are collapsed by the excessive lung weight in dependent parts of the lung (9). In addition, the slope of the P-V curve in its linear partbetween the lower and upper inflection pointsis reduced, attesting to the stiffness of the respiratory system. On the basis of hemodynamic and respiratory measurements, it was first recommended to apply a PEEP slightly above the lower inflection point of the P-V curve (9). More recently, it has been suggested that to avoid lung barotrauma and overdistension, plateau airway pressure should remain lower than the upper inflection point of the P-V curve (8, 15). However, clinical experience suggests that many patients fulfilling the criteria for ALI have P-V curves without a lower inflection point. The exact significance of the presence or the absence of a lower inflection point on the P-V curve in terms of pulmonary morphology and its consequences on the respiratory effects of PEEP are unknown.

The goal of this study was to test the hypothesis that the presence or the absence of a lower inflection point on the P-V curve corresponds to differences in lung morphology that influence the response to PEEP. Therefore, the gas-tissue ratio and the distribution of density histograms, obtained with a spiral CT scanner performed on the overall lung, were compared in two groups of patients with ALI, differing by the presence or the absence of a lower inflection point on their P-V curve.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

During a 6-mo period, 14 patients (10 men and 4 women, 61 ± 13 yr of age) with ALI as defined by the American-European Consensus Conference (16) were studied prospectively. The protocol was considered an integral part of clinical care, with direct and immediate beneficial effects for the patients; nevertheless, informed consent was obtained from the patients' next of kin. Patients treated for ALI in the Surgical Intensive Care Unit of La Pitié-Salpêtrière Hospital are routinely transported to the Department of Radiology, which is close to the unit; there a spiral thoracic CT scan is performed to assess pulmonary morphology during the early phase of the disease. It also provides direct insight on the extension of lung hyperdensities with and without PEEP, and on the presence of pleural effusion and/or pneumothorax. Inclusion criteria consisted of the presence of unilateral or bilateral infiltrates as revealed by a bedside chest radiograph, associated with a Pa_O2  300 mm Hg using an FI_O2 of 1.0 and zero end-expiratory pressure (ZEEP). Exclusion criteria included a history of chronic obstructive pulmonary disease, heart failure, acute heart ischemia, or acute neurological disease, or the presence of a chest tube with a persistent air leak.

All patients were orotracheally intubated and mechanically ventilated using volume-controlled mode (César ventilator; Taema, Antony, France). The following ventilatory settings were applied and kept constant throughout the study: tidal volume, 10 ml · kg¹; respiratory rate, 18 breaths · min¹; TI/Ttot, 33%; and FI_O2, 100%. All patients were sedated with a continuous intravenous infusion of fentanyl (250 µg · h¹) and flunitrazepam (1 mg · h¹) and were paralyzed with vecuronium (4 mg · h¹). A fiberoptic thermodilution pulmonary artery catheter (CCO/Sv_O2/VIPTD catheter; Baxter Healthcare Corporation, Irvine, CA) and a radial or femoral arterial catheter were inserted in each patient as an integral part of cardiorespiratory monitoring.

Protocol

Respiratory and hemodynamic parameters, P-V curves, and spiral thoracic CT scans were performed at ZEEP and at two PEEP levels. The first PEEP level (PEEP₁) was equal to the lower inflection point + 2 cm H₂O in the presence of a lower inflection point on the P-V curve or to 10 cm H₂O in the absence of a lower inflection point. The corresponding values for the second PEEP level (PEEP₂) were equal to the lower inflection point + 7 cm H₂O and 15 cm H₂O. In the Surgical Intensive Care Unit, respiratory and hemodynamic parameters and P-V curves were measured after a 1-h steady state under each condition. The patient was then transported to the Department of Radiology within 6 h and a spiral thoracic CT scan was performed under each condition after a 15-min steady state had been obtained.

Hemodynamic and Respiratory Measurements

Systemic and pulmonary arterial pressures were simultaneously measured using the arterial cannula and the fiberoptic pulmonary artery catheter connected to two calibrated pressure transducers (PX-1X2; Baxter SA, Maurepas, France) positioned at the midaxillary line. Measurements were recorded on a Macintosh personal computer (Apple Computer, Cupertino, CA) using a commercially available system MP 100 WS (Biopac System, Goleta, CA) connected to the analog port of the hemodynamic monitor Merlin (Hewlett-Packard, Palo Alto, CA) and on a strip-chart recorder (Gould ES 1000; Gould Instruments, Cleveland, OH). Variables were sampled online with an analog/digital converter at a rate of 100 Hz. Pressures were measured at end expiration. Cardiac output was continuously measured with a Baxter Swan-Ganz catheter. End-tidal CO₂ (PET_CO2) was measured using a 47210A infrared capnometer (Hewlett-Packard) inserted between the Y-piece and the endotracheal tube. Arterial pH, Pa_O2, Pv_O2, and Pa_CO2 were measured using an IL BGE blood gas analyzer (Instrumentation Laboratory, Paris, France). Hemoglobin and arterial and mixed venous oxygen saturations (Sa_O2 and Sv_O2) were measured using a calibrated OSM3 hemoximeter (Radiometer Copenhagen, Neuilly-Plaisance, France). Standard formulas were used to calculate cardiac index (CI), stroke volume index (SI), pulmonary vascular resistance index (PVRI), systemic vascular resistance index (SVRI), true pulmonary shunt (S/T), arteriovenous oxygen difference [D(a-v)_O2], oxygen extraction ratio (Ea_O2), oxygen delivery (DO₂), oxygen consumption (O₂), and alveolar dead space (VDA/VT). Inspiratory and expiratory flow were carefully calibrated and measured with a heated pneumotachograph (Hans Rudolph, Kansas City, KS), which was inserted between the Y-piece and the endotracheal tube. Tidal volume was obtained by integration of the flow signal (Gould Instruments). Airway pressure (Paw) was measured at the proximal tip of the endotracheal tube with a pressure transducer (PX-1X2; Baxter SA). Esophageal pressure was measured using a water-filled catheter (Argil; Sherwood Medical, Belgium) connected to a similar pressure transducer positioned at the level of the lower third of the esophagus and zeroed 2 cm above the posterior axillary line of the ninth intercostal space as previously described (17).

Determination of Lower Inflection Point on the P-V Curve

According to Levy and coworkers, P-V curves were obtained by performing "study breath" occlusions at different tidal volumes (18). Briefly, measurements were made using the end-inspiratory and expiratory pause hold functions of the César ventilator. During control mechanical ventilation and before administering the tidal volume corresponding to a given study breath, intrinsic PEEP (PEEPi) was determined by activating the end-expiratory pause knob. The expiratory pause knob was then released and control mechanical ventilation was resumed for five cycles. At the end of the expiration of the fifth cycle, the tidal volume corresponding to the study breath was set on the ventilator and the inspiratory pause knob was pressed to obtain a 3-s postinspiratory pause. The same sequence was repeated for each study breath corresponding to different tidal volumes. The smallest tidal volume was 100 ml and the highest tidal volume was the tidal volume corresponding to a plateau airway pressure of 30 cm H₂O under ZEEP conditions. Between these two extremes, tidal volumes in 100-ml increments were administered at random. When P-V curves were measured under PEEP conditions, the same tidal volumes were applied at random irrespective of plateau airway pressure. Respiratory frequency was kept constant, independent of the studied tidal volume.

For each patient, the following P-V curves were constructed: (1) the total respiratory system P-V curve, by plotting each tidal volume against the corresponding difference between airway pressure and PEEPi (17); (2) the lung P-V curve, by plotting each tidal volume against the corresponding pressure difference between airway pressure, PEEPi, and esophageal pressure; (3) the chest wall P-V curve, by plotting each tidal volume against the corresponding esophageal pressure (changes in esophageal pressure were referred to the end- expiratory value of the Pes signal). Each P-V curve was transformed by means of graphic software (Microcal Software, Northampton, MA) to determine lung volumes corresponding to standardized increments of airway pressure. For total respiratory system and lung P-V curves, lung volumes corresponding to 2.5-cm H₂O increments of airway pressure were computed, whereas for chest wall P-V curves, lung volumes corresponding to 0.5-cm H₂O increments of airway pressure were computed. Changes in end-expiratory lung volume corresponding to each PEEP level were measured using the scannographic analysis described below (see Equation 3). For tracing P-V curves at the different PEEP levels, the increases in end-expiratory volume measured between PEEP₁ and ZEEP and between PEEP₂ and ZEEP were added to each tidal volume administered during each study breath.

The slopes of each P-V curve corresponding to total respiratory system and to lung and chest wall compliances were determined by linear regression analysis, taking into account only values between 500 and 1,000 ml (19, 20). The determination of the value of the lower inflection point was performed as described by Gattinoni and coworkers (6): the starting compliance was computed as the ratio between the first inflation of 100 ml and its corresponding airway pressure and the lower inflection point was computed as the pressure corresponding to the intersection between the starting compliance and the slope of the P-V curve.

Scanographic Assessment of Pulmonary Morphology

Each patient was transported to the Department of Radiology (Thoracic Division) by two employees. Mechanical ventilation was provided using an Osiris ventilator (Taema) specifically designed for transportation of critically ill patients and delivering 100% oxygen. Electrocardiogram (EKG), pulse oximetry, and systemic arterial pressure were monitored continuously using a Propaq 104 EL monitor (Protocol Systems, North Chicago, IL). Disconnection from the ventilator for 15 s in order to obtain CT sections under ZEEP conditions resulted in a transient desaturation in most patients. The lowest oxygen saturation measured was 85%. Under PEEP conditions, arterial oxygen desaturation was moderate, the lowest hemoglobin oxygen saturation being 90%.

Lung scanning was performed from the apex to the diaphragm using a Tomoscan SR 7000 (Philips, Eindhoven, The Netherlands) as previously described (19). All images were observed and photographed at a window width of 1,600 Hounsfield units (HU) and level of 700 HU. The exposures were taken at 120 kV and 250 mA. An intravenous injection of 80 ml of contrast material was used in each patient to differentiate pleural fluid collections from consolidated lung parenchyma. Airway pressure was monitored to ensure that the PEEP was equal to the preset value and maintained during the 20 s necessary for the CT acquisition.

Lung volumes were quantified by a method previously described and validated (22). Briefly, the radiologist manually traced the right and left lung outlines with a roller ball on each spiral CT section from the apex to the diaphragm. Lung areas and mean lung density values were determined by using the region of interest function. Frequency histograms of the densities in Hounsfield units were subsequently generated for each region of interest by using the analysis function. The frequency distribution of CT numbers was computed for 50 compartments, from 1,000 HU to +100 HU, examining a 22-HU segment for each compartment. The frequency distribution of CT numbers of the entire lung was then calculated by adding the absolute values of each compartment. The lung volume of each compartment was calculated by multiplying the following: number of lung pixels times square pixel size times section thickness. Total lung volume was obtained by adding the lung volume of each compartment. Lung zones with a density less than 900 HU were considered as overdistended, those between 900 and 500 HU as normally aerated, those between 500 and 100 HU as poorly aerated, and those between 100 and +100 HU as nonaerated (6). The threshold of overdistension of 900 HU had been determined previously in healthy volunteers (23).

Under ZEEP conditions, a qualitative assessment of lung morphology was performed. The CT scan aspect was classified by an independent radiologist (P.C.) according to the location of the radiological hyperdensities: if hyperdensities were localized and delineated by an anatomical structure such as the major fissura, then the CT scan was classified as localized hyperdensities; on the other hand, if hyperdensities were diffuse and not limited by an anatomical structure, then the CT scan was classified as diffuse hyperdensities.

In addition to the quantitative assessment of lung volume, the percentage of gas-tissue ratio under ZEEP and PEEP conditions was also assessed using a technique previously described (4, 23). Briefly, it is based on the assumption that the mean CT numberor the mean densityof a given lung volume correlates with the respective proportion of gas and tissue within this lung. The CT number characterizing each individual pixel is expressed in Hounsfield units and defined as the attenuation coefficient of the X-ray by the material being studied minus the attenuation coefficient of water divided by the attenuation coefficient of water. By reference, the CT number of water is 0 (HU) and the CT number is scaled by a factor of 1,000, the CT number of air being 1,000 HU. A lung area characterized by a mean CT number of 500 HU is considered as being composed of 50% gas and 50% tissue. Using this analysis, it is possible to compute the volume of gas and tissue present in the lungs. In a first step, the volume of gas and tissue for each compartment of 22 HU was computed using the following equations:

(1)

(2)

(3)

(4)

(5)

where CT is the mean CT number of the compartment analyzed. In a second step, the volume of gas and the volume of tissue for the whole lung were calculated by adding the values of the volume of gas and the volumes of tissue obtained for each compartment of 22 HU.

Statistical Analysis

The results are expressed as means ± SD in the text and tables and as mean ± SEM in the figures. The qualitative variables were compared by a ² test corrected for small samples. The hemodynamic, respiratory, mechanical and CT scan parameters under ZEEP, PEEP₁, and PEEP₂ conditions were compared between groups by two-way analysis of variance for one within and one between factor. Parameters measured under ZEEP conditions were compared by a Student unpaired t test. The significance level was fixed at 5%.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Ten males and 4 females (61 ± 13 yr of age) were included in the study. Nine patients were admitted for complications after surgery, two patients for multiple trauma, and three for a medical disease. The general characteristics of the patients are shown in Table 1A and B. Bronchopneumonia was the principal cause of ALI. The simplified acute physiologic score (SAPS II) was 45 ± 16 and the lung injury severity score (LISS) was 2.4 ± 0.8. The mean delay between the onset of ALI and the study was 7 ± 4 d. Overall mortality was 43%.

Eight patients had a lower inflection point on their total respiratory system P-V curve. The results were similar if the pulmonary instead of the total respiratory system P-V curve was considered. In six patients the lower inflection point was detected only on the lung P-V curve, whereas in two the lower inflection point was detected both on lung and chest wall P-V curves. Patients with a lower inflection point were younger than patients without a lower inflection point (55 ± 14 versus 70 ± 7 yr, p < 0.05). There was a trend toward an increased mortality rate (63% versus 17%, p = 0.1) and LISS score (2.8 ± 0.8 versus 2.0 ± 0.6, p = 0.07) in patients with a lower inflection point, whereas the SAPS II score was similar between the two groups (44 ± 19 versus 44 ± 13). In contrast, pulmonary morphology was different between the two groups: patients with a lower inflection point had diffuse hyperdensities disseminated more frequently in upper and lower lobes than did patients without a lower inflection point, who had lung hyperdensities localized predominantly in their lower lobes. This difference in regional distribution of hyperdensities was significantly different between the two groups (p < 0.05).

Differences in Lung Morphology under ZEEP Conditions

As shown in Figure 1 and Table 3, the slopes of total respiratory system and lung P-V curves obtained under ZEEP conditions were significantly different between the two groups (p < 0.05). For a given pressure, lung volume was lower in patients with a lower inflection point. When present, the mean value of the lower inflection point was 9 ± 1 cm H₂O. As shown in Table 2 and Figure 2, overall lung volume (gas plus tissue), and respective volumes of air and tissue measured from the mean CT number, were not different between groups. In both groups the volume of tissue was markedly increased when compared with healthy volunteers (23), suggesting an increase in pulmonary inflammation of an equivalent amount in patients with and without a lower inflection point. However, the distribution of aerated, poorly aerated, and nonaerated lung parenchyma was different. The proportion of normally aerated lung was lower (24 ± 22% versus 55 ± 12%, p < 0.05) whereas the proportion of poorly aerated lung was greater (40 ± 12% versus 23 ± 8%, p < 0.05) in patients with a lower inflection point. As shown in Figure 3, lung density histograms of patients with a lower inflection point showed a unimodal distribution with a progressive increase in the volume of the lung along the Hounsfield unit scale to a maximum peak of 7 ± 20 HU. In contrast, lung density histograms of patients without a lower inflection point had a bimodal distribution with a first peak at 727 ± 58 HU and a second peak at 27 ± 27 HU. In addition, the proportion of parenchyma with a CT density less than 800 HU was lower in patients with a lower inflection point as compared with patients without a lower inflection point (4 ± 4% versus 16 ± 12%, p < 0.03).

View larger version (26K): [in this window] [in a new window]   Figure 1.   Total respiratory system (left) and lung (right) P-V curves in patients with (n = 8, open squares) and without (n = 6, solid squares) a lower inflection point. ZEEP = zero end-expiratory pressure; PEEP = positive end-expiratory pressure; PEEP1 = lower inflection point + 2 cm H2O or 10 cm H2O in the absence of a lower inflection point; PEEP2 = lower inflection point + 7 cm H2O or 15 cm H2O in the absence of a lower inflection point. The increase in end-expiratory lung volume resulting from PEEP1 and PEEP2 was measured using the scanographic analysis described in METHODS (see Equation 3).

View larger version (17K): [in this window] [in a new window]   Figure 2.   Volumes of lung with density of air (top) and with density of tissue (bottom) obtained by gas-tissue ratio analysis under ZEEP, PEEP1, and PEEP2 conditions in patients with (n = 8, open bars) and without (n = 6, solid bars) a lower inflection point; *p < 0.05. ZEEP = zero end-expiratory pressure; PEEP = positive end-expiratory pressure; PEEP1 = lower inflection point + 2 cm H2O or 10 cm H2O in the absence of a lower inflection point; PEEP2 = lower inflection point + 7 cm H2O or 15 cm H2O in the absence of a lower inflection point.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Lung density histograms of patients with (left) and without (right) a lower inflection point at ZEEP (open squares), PEEP1 (solid circles), and PEEP2 (open circles). To improve readability, standard errors of the mean were removed. ZEEP = zero end-expiratory pressure; PEEP = positive end-expiratory pressure; PEEP1 = lower inflection point + 2 cm H2O or 10 cm H2O in the absence of a lower inflection point; PEEP2 = lower inflection point + 7 cm H2O or 15 cm H2O in the absence of a lower inflection point.

As shown in Table 3, peak and plateau pressures were higher and total respiratory system and lung compliances were lower in patients with a lower inflection point. In contrast, all of the hemodynamic and respiratory parameters measured were similar between the two groups (Table 4). There was a trend for higher Pa_CO2 levels in patients without a lower inflection point, despite a similar alveolar dead space and minute ventilation.

Effects of PEEP

As shown in Figure 4, PEEP-induced alveolar recruitment, defined as a decrease in nonaerated lung volume, was similar in both groups. However, the pattern of recruitment between PEEP₁ and PEEP₂ was different. In patients without a lower inflection point, nearly maximal alveolar recruitment occurred at PEEP₁, whereas alveolar recruitment continued to increase between PEEP₁ and PEEP₂ in patients with a lower inflection point. Individual density histograms and CT scans of representative patients with and without a lower inflection point as well as their corresponding P-V curves are shown in Figures 5 and 6. Interestingly, as shown in Figure 2, PEEP caused a significant increase in the amount of gas present within the lungs whereas it did not change the volume of tissue, suggesting that PEEP reopened collapsed lung areas without decreasing lung inflammation and edema.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Changes in the volume of overdistended areas (top) and in the volume of nonaerated areas (bottom) induced by the two PEEP levels in patients with (n = 8, open circles) and without (n = 6, solid circles) a lower inflection point. ZEEP = zero end-expiratory pressure; PEEP = positive end-expiratory pressure; PEEP1 = lower inflection point + 2 cm H2O or 10 cm H2O in the absence of a lower inflection point; PEEP2 = lower inflection point + 7 cm H2O or 15 cm H2O in the absence of a lower inflection point.

View larger version (69K): [in this window] [in a new window]   Figure 5.   Total respiratory system P-V curve under ZEEP conditions (top left), and lung density histogram analysis (top right) and tomographic lung scan cuts (bottom) under ZEEP (open squares), PEEP1 (solid circles), and PEEP2 (open circles) conditions of a typical case from the group of patients with a lower inflection point. A lower inflection point was noted at 10 cm H2O and the patient was ventilated with a PEEP1 of 12 cm H2O and a PEEP2 of 17 cm H2O. Further alveolar recruitment was observed in the linear part of the P-V curve, above the lower inflection point, without concomitant alveolar overdistension, as attested by the absence of lung parenchyma with a CT number less than 900 HU.

View larger version (68K): [in this window] [in a new window]   Figure 6.   Total respiratory system P-V curve under ZEEP conditions (top left), and lung density histogram analysis (top right) and tomographic lung scan cuts (bottom) under ZEEP (open squares), PEEP1 (solid circles), and PEEP2 (open circles) conditions of a typical case from the group of patients without a lower inflection point. No lower inflection point was noted, and so the patient was ventilated with a PEEP1 of 10 cm H2O and a PEEP2 of 15 cm H2O. Alveolar recruitment occurred at the two PEEP levels with a simultaneous overdistension, as attested by the increase in the volume of lung parenchyma with a CT number less than 900 HU.

As shown in Figure 1, the implementation of PEEP₁ and PEEP₂ induced in both groups an upward and right shift of the P-V curves. At a given airway pressure, lung volume was always lower in patients with a lower inflection point than in patients without a lower inflection point. As expected, a lower inflection point could not be identified under PEEP conditions. In contrast, an upper inflection point was present in the majority of patients under PEEP conditions. Among the patients without a lower inflection point, it appeared in five patients at PEEP₁ (31 ± 1 cm H₂O) and in six patients at PEEP₂ (32 ± 2 cm H₂O). In patients with a lower inflection point, it appeared in six patients at PEEP₁ (32 ± 4 cm H₂O) and in seven patients at PEEP₂ (33 ± 4 cm H₂O). When PEEP₁ or PEEP₂ was applied, the slopes of total respiratory system, lung, or chest wall P-V curves did not vary in either group (Table 3). PEEP induced a significant increase in Pa_O2, peak and plateau airway pressures, mean pulmonary artery and capillary wedge pressures, and a significant decrease in cardiac output, true pulmonary shunt, alveolar dead space, and oxygen delivery. These changes were similar in both groups (Tables 3 and 4).

As indicated in Table 2, PEEP induced a significant increase in the overall lung volume that was more pronounced in patients without a lower inflection point. As shown in Figure 4, PEEP-induced lung overdistension was observed only in patients without a lower inflection point. When considering all patients together, the volume of overdistension induced by PEEP correlated significantly with the amount of lung parenchyma in ZEEP, with a CT number less than 800 HU (volume of overdistension = 0.52 × volume less than 800 HU + 4 ml; R = 0.87 and p < 0.0001). In both groups, the volume of poorly aerated lung areas did not change significantly after PEEP implementation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In a series of patients with ALI and comparable alveolar damage as assessed by the excess of lung tissue, two groups of patients were identified as differing in terms of lung morphology and P-V curve. The first group was characterized by diffuse lung hyperdensities present in upper and lower lobes, with P-V curves demonstrating a low respiratory compliance and the presence of a lower inflection point. The second group, which demonstrated a distribution of lung hyperdensities predominantly in lower lobes (upper lobes remained normally aerated), had a higher respiratory compliance without any evidence of a lower inflection point. In both groups, PEEP induced significant alveolar recruitment that was accompanied by lung overdistension only in patients of the second group.

Assessment of Lung Morphology by the Scannographic Method

In the present study, a spiral thoracic CT scanner was used to assess pulmonary morphology before and after PEEP implementation. The scanographic method combines volumetric and densitometric analysis. The volumetric analysis, which quantifies the total number of voxels present in the lungs, provides an accurate measurement of total lung volume (gas plus tissue). A normal lung is composed of gas, tissuealveolar septa, pulmonary vessels, and cellsand blood. An injured lung is characterized by an excessive amount of extravascular lung water, which accumulates in interstitial and alveolar spaces, and by an infiltration of lung structures by inflammatory cells. In patients with ALI, alveolar damage and lung inflammation result in an excessive amount of lung with a tissular density, which can be accurately measured by computing the mean CT number provided the overall lung volume has been measured by volumetric analysis. Because healthy as well as diseased lungs are basically composed of air and tissue and because the Hounsfield density scale is linear between 1,000 HU (density of air) and 0 (density of water), it can be assumed that a lung with a volume of 2 L and a mean CT number of 500 HU is composed of 1 L of air and 1 L of water and/or tissue (4, 5). Combining volumetric and densitometric analysis not only allows an accurate determination of the overall volume of air present in the lung but also quantifies the extent of alveolar damage characterizing ALI by determining the excess lung tissue. Of course, this assumption is valid only if volumetric and densitometric analyses are applied to the whole lungs and not solely to individual CT sections.

A different approach, which can be combined with the approach just described, is to divide the lung into four compartments characterized by different degrees of aeration and to assess the lung volume of each compartment by analyzing the distribution of density histograms (6, 23). The volumes of overdistended and normally, poorly (corresponding to radiologic aspects of "ground glass"), and nonaerated lung areas can be accurately measured using this qualitative analysis, which allows an understanding of the distribution of air and tissue within the diseased lungs. Two lungs characterized by identical total lung volume and mean CT number do have the same amount of air and tissue present in lung parenchyma but may have a totally different lung morphology: in one the excess of tissue and the decreased aeration might be homogeneously distributed within the lung; in the other, the excess of tissue might be concentrated in some parts of the lung, which appear nonaerated, whereas the remaining lung parenchyma remains essentially normally aerated. Such differences were observed in the present study and had important therapeutic consequences.

Lower Inflection Point, Lung Morphology, and Respiratory Mechanics

In both groups of patients, the excess of lung tissue and the volume of gas within the lungs were equivalent, suggesting that the amount of alveolar damage was similar. However, the distribution of air and tissue and, therefore, the pulmonary morphology was different between the two groups. In patients with a lower inflection point, lung density histograms showed a unimodal distribution with a predominance of poorly and nonaerated lung areas relative to normally aerated lung areas, which represented less than 25% of the overall lung volume. Radiologically, this corresponded to bilateral diffuse lung opacities and "ground glass" areas involving upper and lower lobes. In contrast, patients without a lower inflection point showed a bimodal distribution of lung density histograms with a predominance of normally aerated lung areas relative to poorly and nonaerated lung areas, which represented less than 45% of the overall lung volume. Radiologically, this corresponded to nonaerated "white" lower lobes coexisting with normally aerated "black" upper lobes. This radiological aspect could be easily differentiated from lobar atelectasis on the basis of several arguments. First, there was no significant loss of regional lung volume. Second, the amount of tissue characterizing lower lobes was excessive when compared with healthy volunteers (data not shown). Atelectasis would have been characterized by a loss of air with a normal amount of lung tissue, resulting in a loss of regional volume. Third, in many patients with "white" lower lobes, divisions of the main lower bronchus filled with air could be identified, ruling out a bronchial obstruction. In fact, patients of both groups differed on the basis of some clinical and respiratory characteristics. Patients with a lower inflection point were younger and had a lower respiratory compliance than did patients without a lower inflection point. There was a trend to a higher Murray score and mortality in patients with a lower inflection point. No differences could be identified in terms of the etiology of ALI.

The reasons for these differences in lung morphology and respiratory mechanics between the two groups have not been elucidated. In fact, under ZEEP conditions, the two groups did not differ in terms of the total amount of excess tissue present in the lung or in terms of reduction of lung aeration, but only in terms of the distribution of gas and tissue within the lung parenchyma. It can be speculated that patients without a lower inflection point had lung lesions involving predominantly lower lobes, whereas patients with a lower inflection point had lung lesions homogeneously distributed and involving the overall lung parenchyma. In the present study, no correlation was found between the etiology of ALI (primary versus secondary lung injury, infectious versus noninfectious lung lesions) and lung morphology. In each group, composed of a limited number of patients, the same proportion of patients had surgery, bronchopneumonia, pulmonary contusion, aspiration pneumonia, and ALI secondary to severe sepsis or extracorporeal circulation. In eight patients of this series, the lower inflection point detected on the total respiratory system P-V curve was present on the lung P-V curve, whereas in two patients it was also detected on the chest wall P-V curve. This result differs markedly from a study in which, among 13 patients with ALI and demonstrating a lower inflection point on their total respiratory P-V curve, 7 patients had a lower inflection point detectable only on their chest wall P-V curve (14). In these patients, Pa_O2 did not improve with PEEP, suggesting that the decrease in the initial respiratory system compliance was related to extrapulmonary factors such as a postsurgical upward displacement of the diaphragm secondary to an increase in abdominal pressure (24). It has been pointed out that the differences in lung morphology found in the present study between patients with and without a lower inflection point detected on their total respiratory system P-V curve are likely to be valid only in patients in whom the lower inflection point can be also detected on the lung P-V curve. Further studies including a greater number of patients are required to understand the underlying mechanisms of these differences in lung morphology and to assess the role of factors causing ALI.

Effects of PEEP in Patients with and without a Lower Inflection Point

In both groups of patients, the first PEEP levelPEEP₁induced an alveolar recruitment of a similar magnitude. Simultaneously, arterial oxygenation improved and pulmonary shunt decreased in the same proportion in both groups of patients. However, in patients without a lower inflection point, PEEP₁ increased the volume of normally aerated lung areas significantly more than in patients with a lower inflection point. When a higher PEEP level was appliedPEEP₂the behavior of each group became even more different. Alveolar recruitment continued to increase in patients with a lower inflection point, and a significant proportion of the lung was re-aerated to a normal level. This result is in accordance with a previous study that demonstrated in patients with adult respiratory distress syndrome (ARDS) a continuous and progressive alveolar recruitment with increasing PEEP levels ranging between 5 and 20 cm H₂O (2). In patients without a lower inflection point, PEEP₂ did not induce any additional recruitment but was associated with the occurrence of overdistension in some parts of the lung. It should be emphasized that to avoid excessive X-ray exposure, a second series of CT sections was not performed at end inspiration. It can be reasonably assumed that overdistension was even more pronounced after the tidal volume had been entirely delivered.

In acute lung injury, lung areas appearing as nonaerated on the CT scan may result from (1) small airway collapse related to an increase in lung weight along the anteroposterior axis (2) and (2) true alveolar atelectasis related to pulmonary edema, lung infection, and/or lung inflammation. It is considered that the pressure necessary to reopen small airways is lower than that required to reopen true alveolar atelectasis (25). This difference might be due to the presence of residual air distal to the point of airway closure, which, by maintaining a certain lung volume, reduces the opening pressure (26). As a consequence, PEEP-induced alveolar recruitment might be biphasic: at low levels, small collapsed airways reopen whereas higher levels are required to reopen true alveolar atelectasis. If the presence of a lower inflection point on the P-V curve is related to the existence of collapsed distal airways (2), it is therefore not surprising to have found that alveolar recruitment continued with increasing PEEP levels. This hypothesis is substantiated by the results of Gattinoni and co-workers, who observed that the plateau airway pressure necessary to reopen the most dependent regions of the lung can be as high as 50 cm H₂O, a pressure much above the pressure required to counterbalance the superimposed pressure, which cannot exceed the anteroposterior size of the lung (2).

The different response to PEEP between the two groups can be easily explained from the observed differences in lung morphology. Because patients without a lower inflection point had fairly well-aerated upper lobes and nonaerated lower lobes in ZEEP, it is likely that the lung compliance of the upper lobes was much higher than the lung compliance of the lower lobes. When PEEP₂ was applied to the entire respiratory system, it first increased the volume of previously aerated lung areas before recruiting nonaerated lung areas, inducing predominantly lung overdistension and not alveolar recruitment. When comparing the density histogram distribution of the two groups of patients, it appears that the volume of lung with a CT number less than 800 HU represented 16% of the normally aerated lung parenchyma in patients without a lower inflection point and only 4% in patients with a lower inflection point. When considering all patients under ZEEP conditions, a significant relationship was found between the amount of lung characterized by lung densities  800 HU and PEEP- induced alveolar distension, suggesting that only the lung parenchyma with a CT number less than 800 HU is at risk of overdistension after PEEP implementation.

Clinical Implications

The possible clinical implications of the present study are as follows: in the absence of a lower inflection point on the lung P-V curve of a patient with ALI, it can be hypothesized that a large amount of normally aerated lung parenchyma is coexisting with other parts of the lung that are nonaerated and that the patient is at risk of lung overdistension at high PEEP levels. A PEEP of 10 cm H₂O should be tested because it represents a good compromise between alveolar recruitment and lung overdistension. When a lower inflection point is present on the lung P-V curve, a PEEP value well above the lower inflection point might be tested because the PEEP-induced alveolar recruitment is predominant over the risk of overdistension. In doing so, great care should be accorded to the tidal volume, which should always be set in order to obtain a plateau pressure below the upper inflection pressure. Any alternative ventilatory strategy giving the opportunity of increasing end-expiratory pressure while limiting plateau airway pressure below the upper inflection pressure is likely to be efficient in this group of patients.