Intact Epithelial Barrier Function Is Critical for the Resolution of Alveolar Edema in Humans1- 3
MICHAEL A. MATTHAY and JEANINE P. WIENER-KRONISH
Introduction Both physiologic and morphologic studies have demonstrated that the normal alveolar epithelial barrier is very tight, offering substantial resistance to the passive movement of both small and large molecules (1-4). Recent experimental studies have provided new insights into the mechanisms that regulate removal of excess liquid across the alveolar epithelium from the air spaces of the normal lung. After instillation of autologous serum into the distal air spaces of normal lungs in sheep, dogs, or rabbits, we found that the protein concentration of the residual instilled alveolar fluid progressively increased, indicating that rapid alveolar liquid clearance had occurred across the tight alveolar epithelium (5-8). We and other investigators have provided evidence that the primary mechanism for this rapid alveolar liquid clearance depends on active sodium transport. The data supporting this conclusion have been based on a variety of experimental preparations, including isolated alveolar epithelial type II cells (9-11), the fetal lamb and guinea pig lung (12, 13), isolated perfused rat lungs (14, 15), as well as our own work in intact adult sheep and rabbits (16-18). In spite of the new data from experimental studies on the mechanisms of alveolar liquid clearance, there have been few studies of the resolution of alveolar edema in humans. A number of investigators have reported the usefulness of measuring the protein concentration in pulmonary edema fluid relative to plasma protein concentration as a method for determining whether the pulmonary edema is caused primarily by an elevation of hydrostatic pressure (transudate) or an increase in vascular permeability (exudate) in the pulmonary circulation (19, 20). However, these studies did not follow the change in the protein concentration of pulmonary edema fluid as the alveolar edema resolved. Therefore, this clinical study was designed with the following three objec1250
SUMMARY Within 15 min of endotracheal Intubation, the I8SOlutlon of pulmonary edema wesstudled over the next 12 h In 34 mechanically ventllatad patients by (1) serial measurements of the alveolararterial oxygen difference, (2) the extent of edema on the Initial and follow-up chest radiograph, and (3) by an Initial and final measurement of total protein and albumin concentretlon In sequential samples of pulmonary edama fluid. Based on the oxygenation and chest radiographic data, 24 patients clinically Improved and 10 patients did not Improve. In the 10 patients who did not clinically improve (3, hydrostatic edema; 7, permeability edema), there was no change In the final edema fluid protein concentration (4.1 ± 1.1 gl100 ml) compared with the initial edema fluid protein concentration (4.2 ± 1.0 gl100 ml) (p = ns). However, In the 24 patients who clinically Improved (15, hydrostatic edema; 9, permeability edema), there was an Increase In every patient's final edema protein concentration (5.6 ± 2.3 g/100 ml) compared with their Initial edema protein concentration (3.8 ± 1.2 gl100 ml) (p < 0.01). In 13 of these 24 patients, the final edema fluid concentration (7.3 ± 1.6 g/100 ml) exceeded the final plasma protaln concentration (5.6 ± 0.8 gl100 ml) by a mean velue of 1.7 g/100 ml protein. The data provide the first evidence In humans to support the hypothasis that active Ion transport across the alveolar epithelial barrier is the primary mechanism for clearance of edema fluid from the air spaces of the lung. In addition, the use of sequential pulmonary edema fluid protein concentration as an Index of alveolar epithelial barrier function provided a good prognostic Indicator of survival In the 16 patients with Increased permeability pulmonary edema. AM REV RESPIR DIS 1990; 142:1250-1257
tives. The first objective was to test the hypothesis that alveolar protein concentration would increase as edema liquid was cleared from the air spaces of the human lung. If this hypothesis were correct, then patients who showed clinical improvement would have a rise in the concentration of protein in the edema fluid, even to levels above the plasma protein concentration. The second objective was to compare the time course for the resolution of alveolar edema in patients with either hydrostatic or increased permeability pulmonary edema. Because patients with hydrostatic edema should have an intact alveolar epithelium, the rate of clearance of alveolar edema would be expected to proceed at a faster rate than in patients with increased permeability edema. However, recent experimental work has suggested that some types of increased permeability pulmonary edema are not associated with major injury to the epithelial barrier (21), and therefore the rate of clearance of alveolar edema in patients with increased permeability pulmonary edema might be variable, depending on the severity of acute lung injury. The third objective was to determine the prognostic significance of in-
tact alveolar epithelial function, as measured by the ability of the epithelium to remove liquid from the air spaces of the lung because intact epithelial function could be a major determinant of survival after acute lung injury in patients with increased permeability pulmonary edema. Methods Entry Criteria The study was designed to examine prospectively patients with pulmonary edema that was severe enough to require endotracheal intubation and mechanical ventilation. The study
(Received in originalform February 28, 1990 and in revised form June 14, 1990) 1 From the Departments of Medicine and Anesthesia, Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California. 2 Supported by Pulmonary Vascular SCOR Grant No. HL-19155 and HL-40626 (RO!) from the National Heart, Lung, and Blood Institute. 3 Correspondence and requests for reprints should be addressed to Michael A. Matthay, M.D., Box 0130, Cardiovascular Research Institute, University of California at San Francisco, San Francisco, CA 94143.
1251
INTACT EPITHELIAL BARRIER FUNCTION
group was restricted to this population because the initial and a follow-up edema fluid sample could only be obtained in patients with severe alveolar edema. On the basis of our experimental studies (5, 6, 8), it was anticipated that the clearance of alveolar edema would occur rapidly in some patients. Thus, it was decided at the outset that consecutive patients would qualify for entry into the study only if they were intubated by or in the presence of one of the two authors of this study. This procedure guaranteed that all initial pulmonary edema fluid and plasma samples would be obtained within 15 min of intubation. All patients were initially ventilated with an Flo, of 1.0, a tidal volume of 12 to 15 ml/kg, and positive end-expiratory pressure (PEEP) of 3 to 8 cm H,O of water. An arterial blood gas was obtained within 5 min of obtaining the first pulmonary edema fluid samples so that an alveolar-arterial oxygen difference could be calculated (22). A portable anterior-posterior chest radiograph was also taken within 30 min of endotracheal intubation. This study was approved by the Committee on Human Research at the University of California at San Francisco.
Pulmonary Edema Fluid Sampling: Protocol and Validation Studies The initial pulmonary edema fluid sample was obtained within 15 min of intubation and mechanical ventilation. A 14 French suction catheter (Becton-Dickinson, Lincoln Park, NJ) was passed through the endotracheal tube and wedged into the distal airways; then, gentle suction was applied to obtain edema fluid (at least 1 to 2 ml); the fluid was collected into a specimen trap (Sherwood Medical, St. Louis, MO) and 10 to 100 U of heparin was added to each sample. Then, the heparinized plasma and edema fluid samples were centrifuged at 3,000 X g for 10 min. Hourly arterial blood gases were obtained. None ofthe patients were treated with inhaled or parenteral beta-adrenergic agonists. The protocol for obtaining a subsequent pulmonary edema fluid sample was designed to separate those patients who had clinical improvement as demonstrated by improving oxygenation versus those patients who had no clinical improvement as indicated by stable or worsening oxygenation in the first 12 h after endotracheal intubation. Therefore, a second edema fluid and plasma sample was obtained in any patient who had a decrease in the alveolar-arterial oxygen difference of 50 mm Hg or more based on the arterial blood gases that were sampled every 1 to 2 h. These patients (n = 24) were combined together into the clinical improvement group. There were no changes made in the ventilation parameters (tidal volume, level of PEEP, or Flo,) in these 24 patients before obtaining the second pulmonary edema fluid sample. Therefore, the improvement in oxygenation was not dependent on a change in tidal volume or the level of PEEP. The remaining patients with no clinical improvement (n = 10) had a second edema flu-
id and plasma sample obtained when either the alveolar-arterial oxygen difference increased by 50 mm Hg or they required an increase in the level of positive end-expiratory pressure of 3 cm H,O or more to maintain oxygenation; if there was no change in oxygenation or in the level of positive endexpiratory pressure over the 12 h period of the study, then repeat edema fluid and plasma samples were obtained at 12 h. These 10 patients were combined together into the noclinical-improvement group. In all patients, a second chest radiograph was taken within 30 min of obtaining the second edema fluid and plasma samples. We assumed that the pulmonary edema fluid that we sampled with the wedged 14 French suction catheter was a reasonable reflection of alveolar edema fluid for the following reasons. First, edema fluid has been sampled by simple suctioning through an endotracheal tube during the formation of pulmonary edema in experimental studies, and the protein concentration of this edema fluid correctly characterized the type of pulmonary edema as either hydrostatic or increased permeability (21, 23, 24). Second, a number of clinical studies have been published that validate the use of the protein concentration in edema fluid as a method to categorize the type of pulmonary edema (19, 20, 25, 26). Third, we have recently validated the use of aspirated edema fluid by comparing it to alveolar micropuncture samples. We found that a 1- to 2-ml sample of fluid aspirated from the distal air spaces of a dog's lung, using a catheter similar to the one used in this clinical study, accurately reflected the protein concentration in micropuncture samples of alveolar fluid obtained 4 h after 3 ml/kg of plasma had been instilled into one lung (7). Fourth, although followup samples of pulmonary edema fluid in the resolution phase could theoretically include the presence of newly secreted proteins, a previous polyacrylamide gel electrophoresis study of low, intermediate, and high molecular weight proteins in initial and follow-up pulmonary edema fluid samples showed no change in the protein concentration of the different molecular weight fractions (26).
Measurements Measurement of the total protein and the albumin fraction on the pulmonary edema fluid and plasma samples was done by the Biuret and the bromcresol green dye-binding technique (25). The measurement of protein concentrations was done by a technician who had no knowledge of the patient's clinical course. Because the final pulmonary edema fluid total protein concentration exceeded the plasma protein concentration in 13 patients, we wanted to be certain that the higher protein concentrations in the pulmonary edema fluid in fact reflected a higher protein osmotic pressure compared to the plasma sample. Therefore, we measured the protein osmotic pressure (Wescor Osmometer, Logan, UT) on the final pulmonary edema fluid sample (at
least 0.5 ml was required) and the concurrent final plasma sample in the 6 patients who had an adequate volume of edema fluid for the measurement. In addition, the pulmonary edema fluid and plasma protein data was analyzed as a ratio of the protein concentration in the edema fluid to the simultaneously obtained protein concentration in the plasma. This approach has been adopted in previous studies because the ratio has been used to classify the pulmonary edema as hydrostatic or increased permeability, and because the use of the ratio accounts for any changes in the pulmonary edema fluid protein concentration that may be influenced by changes in the plasma protein concentration from diuresis with concentration of plasma protein or from intravascular expansion with crystalloid therapy that may dilute the plasma protein concentration (19, 20, 25, 26). The evaluation of the portable anteriorposterior chest radiographs was done by a physician who had no knowledge of the clinical course of the patient nor the results of the pulmonary edema fluid and plasma protein concentration measurements. The second chest radiograph was compared with the initial chest radiograph to determine whether there was evidence for either an increase in edema, a decrease in edema, or no change in the degree of edema. The criteria for extent of pulmonary edema were the same that we used in a recent study of patients with severe hydrostatic and increased permeability edema (27).
Patient Classification Patients were classified as either hydrostatic or increased permeability pulmonary edema (or the adult respiratory distress syndrome) on the basis of careful review of all clinical and laboratory data. There were a total of 16 patients classified as increased permeability pulmonary edema. This classification was confirmed by the availability of pulmonary hemodynamic measurements in 14 of these 16 patieJlfs, all of whom had a pulmonary arterial wedge pressure less than 16 mm Hg. Pulmonary hemodynamic measurements were made at end-expiration and recorded on the oscilloscope monitor paper, and recalibration was done every 3 to 4 h to be certain of the reliability of the measurements (28). Furthermore, bedside echocardiograms demonstrated normal left ventricular function in these patients. There were 18 patients classified as hydrostatic pulmonary edema, all of whom had a history of either chronic or acute cardiac failure or volume overload and clinical evidence of left ventricular heart failure. Also, the mean initial pulmonary edema/plasma protein concentration ratio was 0.54 ± 0.11 for these 18 patients, which provides independent confirmation that these patients had a transudative edema from elevated hydrostatic pressures (19, 23, 25, 26).
Estimate of Alveolar Liquid Clearance In order to estimate the rate of alveolar liq-
1252
MATTHAY AND WIENER·KRONISH
uid clearance in the resolution of alveolar edema in patients, we compared the percent concentration of the final pulmonary edema fluid protein concentration to the initial edema fluid protein concentration and calculated the percent concentration as an index of the clearance of edema fluid from the air spaces of the lung. This approach has been validated as a reliable method for estimating alveolar liquid clearance in studies in normal dog and sheep lungs (7, 17) as well as in one experimental study of increased permeability pulmonary edema (21).
Statistics All data are shown as mean ± standard deviation. A Student's paired t test was used to analyze for significant differences in the initial and final albumin and total protein concentrations, the initial and final ratios of edema fluid to plasma total protein concentration, and the initial and final alveolar-arterial oxygen difference in patients classified as clinical improvement (n = 24) or no clinical improvement (n = 10). A Student's unpaired t test was used to test for significant differences between the 24 patients with clinical improvement compared to the 10 patients with no clinical improvement relative to the initial and final total protein concentration, the ratio of the edema fluid to plasma total protein concentration, and the alveolar-arterial oxygen difference. Also, chi-square analysis was used to analyze for significant differences in mortality in the Group A (n = 9) and Group B (n = 7) patients with increased permeability pulmonary edema. We accepted p < 0.05 as significant (29).
TABLE 1 ALVEOLAR-ARTERIAL OXYGEN DIFFERENCE IN PATIENTS WITH PULMONARY EDEMA Alveolar-Arterial Oxygen Difference (mm Hg) Classification Clinical improvement Hydrostatic Increased permeability No clinical improvement Increased permeability Hydrostatic edema
No.
Initial
Final
Time Interval (h)
15 9
545 ± 74 553 ± 40
366 ± 124368 ± 112-
4.5 ± 2.9 6.8 ± 5.1
7 3
518 ± 84 588 ± 23
538 ± 95 616 ± 22
5.4 ± 4.1 5.3 ± 5.8
Data shown as mean ± SO. • P < 0.01 comparing the final with the initial alveolar-arterial oxygen difference.
TABLE 2 EVALUATION OF THE FOLLOW·UP CHEST RADIOGRAPH COMPARED WITH THE INITIAL CHEST RADIOGRAPH FOR THE EXTENT OF PULMONARY EDEMA Chest Radiograph Analysis Classification Clinical improvement Hydrostatic Increased permeability No clinical improvement Increased permeability Hydrostatic
No.
Less Edema
Unchanged
More Edema
15 9
13 8
2 1
0 0
7 3
0 0
5 2
2 1
TABLE 3 INITIAL AND FINAL PULMONARY EDEMA FLUID PROTEIN CONCENTRATIONS IN PATIENTS WITH CLINICAL IMPROVEMENT Pulmonary Edema Fluid
Results
Overall, 34 patients were entered into the study, ::md 24 of these patients clinically improved on the basis of a decrease in the alveolar-arterial oxygen difference. Of these 24 patients, 15 had hydrostatic pulmonary edema and 9 had increased permeability pulmonary edema (table 1). A total of 10 patients were classified as no clinical improvement, all of whom required an increase in positive endexpiratory pressure to maintain oxygenation. Seven of these 10 patients had increased permeability pulmonary edema, and 3 had hydrostatic edema (table 1). The patients who were classified into the clinical-improvement or the no-clinicalimprovement groups had comparable oxygenation defects at the time of entry into the study with the initial mean alveolar-arterial oxygen gradient greater than 500 mm Hg in all groups (table 1). The improved oxygenation in the 24 patients with clinical improvement was matched by a reduction in the extent of pulmonary edema on the second chest radiograph in 21 of the 24 patients (table 2). In the 10 patients without clinical im-
Patient No.
Classification of Edema
Time Interval (h)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Permeability Permeability Permeability Permeability Permeability Permeability Permeability Permeability Permeability
4 9 4 3 2 2 2 3 9 6 6 10 2 1 4 12 12 1 1 3 3 12 12 5
Mean ± SO
5.3 ± 3.9
Total Protein (g/dL)
Albumin (g/dL)
Initial
Final
Initial
Final
3.3 1.4 3.5 2.8 3.t 3.4 5.2 2.7 4.5 4.0 3.4 3.4 2.0 1.8 4.4 4.5 5.0 5.0 6.6 3.4 4.5 4.0 5.0 4.0
4.1 2.6 3.9 3.3 4.2 4.9 7.4 3.1 6.8 5.0 5.1 6.1 2.5 2.5 11.1 6.8 7.9 6.2 7.7 3.7 6.8 5.9 9.6 7.0
2.2 .7 2.5 2.0 2.1 1.8 3.8 1.9 3.2 2.4 2.4 1.4 1.3 1.1 2.9 3.1 2.3 2.6 4.2 2.2 2.6 2.9 3.8
2.7 1.3 2.7 2.1 2.8 2.9 5.2 2.5 4.5 3.0 3.2
3.8 ± 1,.2 5.6 ± 2.3- 2.4 ± 0.9
Data shown as mean ± SO. • P < 0.01 comparing final with initial total protein or albumin concentrations.
1.7 7.2 5.4 3.4 4.9 2.3 3.8 4.2 6.8 3.6 ± 1.6-
1253
INTACT EPITHELIAL BARRIER FUNCTION
TABLE 4
TABLE 5
INITIAL AND FINAL PULMONARY EDEMA PROTEIN CONCENTRATION IN PATIENTS WITH NO CLINICAL IMPROVEMENT
COMPARISON OF THE MEASURED PROTEIN OSMOTIC PRESSURE IN THE FINAL PULMONARY EDEMA FLUID SAMPLE WITH THE CONCURRENT FINAL PLASMA SAMPLE IN SIX PATIENTS
Pulmonary Edema Fluid Patient No.
Classification of Edema
25 26 27 28 29 30 31 32 33 34
Hydrostatic Hydrostatic Hydrostatic Permeability Permeability Permeability Permeability Permeability Permeability Permeability
Mean
Time Interval
Total Protein (g/dL)
Albumin (g/dL)
(h)
Initial
Final
Initial
Final
12 12 1 3 6 11 3 1 3 11
3.3 3.3 3.1 5.7 4.5 6.1 4.1 3.6 4.3 3.7
3.2 3.5 3.1 5.8 4.3 6.0 3.9 3.1 4.6 3.4
1.6 1.5 2.2 3.0 3.2 4.0 2.4 2.0 2.9 2.5
1.5 1.5 2.2 3.0 3.0 3.5 2.2 1.9 3.1 2.3
6.3 ± 4.7
4.2 ± 1.0
4.1 ± 1.1
2.5 ± 0.8
2.4 ± 0.7
Protein Osmotic Pressure (mm Hg)* Patient No.
6 11 13 15 18 22 Mean ± SO (n = 6)
Final Edema Fluid
Plasma
29 23 17 72 31 31
25 18 16 25 20 21
34 ± 19
21 ± 4
- Measurements made at 26° C on a Wescor osmometer. Data shown as mean ± SD.
provement, the second chest radiograph showed no change or an increase in the extent of edema (table 2). Radiographically significant atelectasis was not observed on any of the chest radiographs. Also, the level of positive end-expiratory pressure (PEEP) at the time of the sampling of the initial and final pulmonary edema fluid was the same (5 ± 3 cm H 2 0) in the 24 patients with clinical improvement, whereas the level of PEEP was 5 ± 3cmH 2 0and 10 ± 4 cm H 2 0, respectively, in the 10 patients who did not clinically improve. In all 24 patients who had clinical improvement, the final pulmonary edema fluid sample had a higher edema fluid total protein concentration (mean value, 5.6 ± 2.3 g/l00 ml) compared with their initial pulmonary edema fluid total protein concentration (3.8 ± 1.2 g/l00 ml) (table 3) (p < 0.01). Also, the albumin concentration in the final edema fluid sample was always higher than the initial edema fluid albumin concentration (table 3) (p < 0.01). In contrast, in the group of 10 patients who did not clinically improve, there was no difference in their final edema fluid total protein or in the albumin concentration compared with their initial total protein or with the initial albumin concentration in their edema fluid (table 4) (p = ns). In addition, in all 24 patients with clinical improvement, the final edema fluidto-plasma protein concentration ratio (mean value, 1.1 ± 0.37) was higher than the initial edema fluid protein concentration ratio (mean value, 0.65 ± 0.20) (p < 0.01) (figure 1). In contrast, the edema fluid-to-plasma protein concentration ratio did not change between the final (mean value, 0.79 ± 0.18) and the ini-
tial ratio (mean value, 0.80 ± 0.17) in the 10 patients with no clinical improvement (figure 1). In 13 of the 24 patients who clinically improved, the final edema fluid total protein concentration (7.3 ± 1.6 g/100 ml) exceeded the final plasma protein concentration (5.6 ± 0.8 g/100 ml) by a mean value of 1.7 g/l00 ml. In addition, the results of the protein osmotic pressure measurements on the final pulmonary edema fluid sample and the concurrent final plasma sample in 6 of these 13 patients indicated that the measured protein osmotic pressure in the edema fluid in these patients always exceeded the plasma protein osmotic pressure (table 5). In one patient, the total protein concentration of the final edema fluid was 11.1 g/l00 ml compared with a simultaneous plasma protein concentration of 7.2 g/100 ml, and the edema fluid protein osmotic pressure was 72 mm Hg compared with the plasma protein osmotic pressure of 25 mm Hg. Because in all animal experiments the increased concentration of protein in the air spaces reflects the quantity of edema fluid removed from the alveoli (7, 17), the percent concentration of the final to the initial edema fluid protein concen-
Fig. 1. Individual data points are shown for the initial and final pulmonary edema fluid-to-plasma total protein concentration ratio in the 24 patients with clinical improvement compared with the 10 patients with no clinical improvement. The time interval between the initial and final sample was 5.3 ± 3.9 h for the clinical-improvement group and 5.3 ± 4.7 h for the no-clinical-improvement group.
tration was used to compare the rates of alveolar liquid clearance between the 15 patients with hydrostatic edema and the 9 patients with increased permeability edema, all of whom clinically improved (table 6). Interestingly, the data demonstrated that the mean concentration of protein in the edema fluid, adjusted for the time between the initial and final edema sample, was similar between the two groups. There was considerable variability in the time course for resolution of alveolar edema, but both the range and the mean clearance rates as estimated by the concentration of edema fluid protein were nearly identical in patients with resolving hydrostatic or increased permeability edema. As further support for this conclusion, the improvement in oxygenation and the time interval for thif improvement was similar in patients with resolving hydrostatic or increased permeability edema (table 1). Finally, the relationship of alveolar epithelial function to outcome was analyzed in the 16 patients with increased permeability pulmonary edema. In the first 12 h after endotracheal intubation and the onset of mechanical ventilation, there were 9 patients (Group A) who had a rise in their edema fluid protein concentration
NOCUNICAL
1.6
Edema Fluid Plasma (total protein)
IMPROVEMENT
1.2
===== ..
0.8
0.4
p
MICHAEL A. MATTHAY and JEANINE P. WIENER-KRONISH
Introduction Both physiologic and morphologic studies have demonstrated that the normal alveolar epithelial barrier is very tight, offering substantial resistance to the passive movement of both small and large molecules (1-4). Recent experimental studies have provided new insights into the mechanisms that regulate removal of excess liquid across the alveolar epithelium from the air spaces of the normal lung. After instillation of autologous serum into the distal air spaces of normal lungs in sheep, dogs, or rabbits, we found that the protein concentration of the residual instilled alveolar fluid progressively increased, indicating that rapid alveolar liquid clearance had occurred across the tight alveolar epithelium (5-8). We and other investigators have provided evidence that the primary mechanism for this rapid alveolar liquid clearance depends on active sodium transport. The data supporting this conclusion have been based on a variety of experimental preparations, including isolated alveolar epithelial type II cells (9-11), the fetal lamb and guinea pig lung (12, 13), isolated perfused rat lungs (14, 15), as well as our own work in intact adult sheep and rabbits (16-18). In spite of the new data from experimental studies on the mechanisms of alveolar liquid clearance, there have been few studies of the resolution of alveolar edema in humans. A number of investigators have reported the usefulness of measuring the protein concentration in pulmonary edema fluid relative to plasma protein concentration as a method for determining whether the pulmonary edema is caused primarily by an elevation of hydrostatic pressure (transudate) or an increase in vascular permeability (exudate) in the pulmonary circulation (19, 20). However, these studies did not follow the change in the protein concentration of pulmonary edema fluid as the alveolar edema resolved. Therefore, this clinical study was designed with the following three objec1250
SUMMARY Within 15 min of endotracheal Intubation, the I8SOlutlon of pulmonary edema wesstudled over the next 12 h In 34 mechanically ventllatad patients by (1) serial measurements of the alveolararterial oxygen difference, (2) the extent of edema on the Initial and follow-up chest radiograph, and (3) by an Initial and final measurement of total protein and albumin concentretlon In sequential samples of pulmonary edama fluid. Based on the oxygenation and chest radiographic data, 24 patients clinically Improved and 10 patients did not Improve. In the 10 patients who did not clinically improve (3, hydrostatic edema; 7, permeability edema), there was no change In the final edema fluid protein concentration (4.1 ± 1.1 gl100 ml) compared with the initial edema fluid protein concentration (4.2 ± 1.0 gl100 ml) (p = ns). However, In the 24 patients who clinically Improved (15, hydrostatic edema; 9, permeability edema), there was an Increase In every patient's final edema protein concentration (5.6 ± 2.3 g/100 ml) compared with their Initial edema protein concentration (3.8 ± 1.2 gl100 ml) (p < 0.01). In 13 of these 24 patients, the final edema fluid concentration (7.3 ± 1.6 g/100 ml) exceeded the final plasma protaln concentration (5.6 ± 0.8 gl100 ml) by a mean velue of 1.7 g/100 ml protein. The data provide the first evidence In humans to support the hypothasis that active Ion transport across the alveolar epithelial barrier is the primary mechanism for clearance of edema fluid from the air spaces of the lung. In addition, the use of sequential pulmonary edema fluid protein concentration as an Index of alveolar epithelial barrier function provided a good prognostic Indicator of survival In the 16 patients with Increased permeability pulmonary edema. AM REV RESPIR DIS 1990; 142:1250-1257
tives. The first objective was to test the hypothesis that alveolar protein concentration would increase as edema liquid was cleared from the air spaces of the human lung. If this hypothesis were correct, then patients who showed clinical improvement would have a rise in the concentration of protein in the edema fluid, even to levels above the plasma protein concentration. The second objective was to compare the time course for the resolution of alveolar edema in patients with either hydrostatic or increased permeability pulmonary edema. Because patients with hydrostatic edema should have an intact alveolar epithelium, the rate of clearance of alveolar edema would be expected to proceed at a faster rate than in patients with increased permeability edema. However, recent experimental work has suggested that some types of increased permeability pulmonary edema are not associated with major injury to the epithelial barrier (21), and therefore the rate of clearance of alveolar edema in patients with increased permeability pulmonary edema might be variable, depending on the severity of acute lung injury. The third objective was to determine the prognostic significance of in-
tact alveolar epithelial function, as measured by the ability of the epithelium to remove liquid from the air spaces of the lung because intact epithelial function could be a major determinant of survival after acute lung injury in patients with increased permeability pulmonary edema. Methods Entry Criteria The study was designed to examine prospectively patients with pulmonary edema that was severe enough to require endotracheal intubation and mechanical ventilation. The study
(Received in originalform February 28, 1990 and in revised form June 14, 1990) 1 From the Departments of Medicine and Anesthesia, Cardiovascular Research Institute, University of California at San Francisco, San Francisco, California. 2 Supported by Pulmonary Vascular SCOR Grant No. HL-19155 and HL-40626 (RO!) from the National Heart, Lung, and Blood Institute. 3 Correspondence and requests for reprints should be addressed to Michael A. Matthay, M.D., Box 0130, Cardiovascular Research Institute, University of California at San Francisco, San Francisco, CA 94143.
1251
INTACT EPITHELIAL BARRIER FUNCTION
group was restricted to this population because the initial and a follow-up edema fluid sample could only be obtained in patients with severe alveolar edema. On the basis of our experimental studies (5, 6, 8), it was anticipated that the clearance of alveolar edema would occur rapidly in some patients. Thus, it was decided at the outset that consecutive patients would qualify for entry into the study only if they were intubated by or in the presence of one of the two authors of this study. This procedure guaranteed that all initial pulmonary edema fluid and plasma samples would be obtained within 15 min of intubation. All patients were initially ventilated with an Flo, of 1.0, a tidal volume of 12 to 15 ml/kg, and positive end-expiratory pressure (PEEP) of 3 to 8 cm H,O of water. An arterial blood gas was obtained within 5 min of obtaining the first pulmonary edema fluid samples so that an alveolar-arterial oxygen difference could be calculated (22). A portable anterior-posterior chest radiograph was also taken within 30 min of endotracheal intubation. This study was approved by the Committee on Human Research at the University of California at San Francisco.
Pulmonary Edema Fluid Sampling: Protocol and Validation Studies The initial pulmonary edema fluid sample was obtained within 15 min of intubation and mechanical ventilation. A 14 French suction catheter (Becton-Dickinson, Lincoln Park, NJ) was passed through the endotracheal tube and wedged into the distal airways; then, gentle suction was applied to obtain edema fluid (at least 1 to 2 ml); the fluid was collected into a specimen trap (Sherwood Medical, St. Louis, MO) and 10 to 100 U of heparin was added to each sample. Then, the heparinized plasma and edema fluid samples were centrifuged at 3,000 X g for 10 min. Hourly arterial blood gases were obtained. None ofthe patients were treated with inhaled or parenteral beta-adrenergic agonists. The protocol for obtaining a subsequent pulmonary edema fluid sample was designed to separate those patients who had clinical improvement as demonstrated by improving oxygenation versus those patients who had no clinical improvement as indicated by stable or worsening oxygenation in the first 12 h after endotracheal intubation. Therefore, a second edema fluid and plasma sample was obtained in any patient who had a decrease in the alveolar-arterial oxygen difference of 50 mm Hg or more based on the arterial blood gases that were sampled every 1 to 2 h. These patients (n = 24) were combined together into the clinical improvement group. There were no changes made in the ventilation parameters (tidal volume, level of PEEP, or Flo,) in these 24 patients before obtaining the second pulmonary edema fluid sample. Therefore, the improvement in oxygenation was not dependent on a change in tidal volume or the level of PEEP. The remaining patients with no clinical improvement (n = 10) had a second edema flu-
id and plasma sample obtained when either the alveolar-arterial oxygen difference increased by 50 mm Hg or they required an increase in the level of positive end-expiratory pressure of 3 cm H,O or more to maintain oxygenation; if there was no change in oxygenation or in the level of positive endexpiratory pressure over the 12 h period of the study, then repeat edema fluid and plasma samples were obtained at 12 h. These 10 patients were combined together into the noclinical-improvement group. In all patients, a second chest radiograph was taken within 30 min of obtaining the second edema fluid and plasma samples. We assumed that the pulmonary edema fluid that we sampled with the wedged 14 French suction catheter was a reasonable reflection of alveolar edema fluid for the following reasons. First, edema fluid has been sampled by simple suctioning through an endotracheal tube during the formation of pulmonary edema in experimental studies, and the protein concentration of this edema fluid correctly characterized the type of pulmonary edema as either hydrostatic or increased permeability (21, 23, 24). Second, a number of clinical studies have been published that validate the use of the protein concentration in edema fluid as a method to categorize the type of pulmonary edema (19, 20, 25, 26). Third, we have recently validated the use of aspirated edema fluid by comparing it to alveolar micropuncture samples. We found that a 1- to 2-ml sample of fluid aspirated from the distal air spaces of a dog's lung, using a catheter similar to the one used in this clinical study, accurately reflected the protein concentration in micropuncture samples of alveolar fluid obtained 4 h after 3 ml/kg of plasma had been instilled into one lung (7). Fourth, although followup samples of pulmonary edema fluid in the resolution phase could theoretically include the presence of newly secreted proteins, a previous polyacrylamide gel electrophoresis study of low, intermediate, and high molecular weight proteins in initial and follow-up pulmonary edema fluid samples showed no change in the protein concentration of the different molecular weight fractions (26).
Measurements Measurement of the total protein and the albumin fraction on the pulmonary edema fluid and plasma samples was done by the Biuret and the bromcresol green dye-binding technique (25). The measurement of protein concentrations was done by a technician who had no knowledge of the patient's clinical course. Because the final pulmonary edema fluid total protein concentration exceeded the plasma protein concentration in 13 patients, we wanted to be certain that the higher protein concentrations in the pulmonary edema fluid in fact reflected a higher protein osmotic pressure compared to the plasma sample. Therefore, we measured the protein osmotic pressure (Wescor Osmometer, Logan, UT) on the final pulmonary edema fluid sample (at
least 0.5 ml was required) and the concurrent final plasma sample in the 6 patients who had an adequate volume of edema fluid for the measurement. In addition, the pulmonary edema fluid and plasma protein data was analyzed as a ratio of the protein concentration in the edema fluid to the simultaneously obtained protein concentration in the plasma. This approach has been adopted in previous studies because the ratio has been used to classify the pulmonary edema as hydrostatic or increased permeability, and because the use of the ratio accounts for any changes in the pulmonary edema fluid protein concentration that may be influenced by changes in the plasma protein concentration from diuresis with concentration of plasma protein or from intravascular expansion with crystalloid therapy that may dilute the plasma protein concentration (19, 20, 25, 26). The evaluation of the portable anteriorposterior chest radiographs was done by a physician who had no knowledge of the clinical course of the patient nor the results of the pulmonary edema fluid and plasma protein concentration measurements. The second chest radiograph was compared with the initial chest radiograph to determine whether there was evidence for either an increase in edema, a decrease in edema, or no change in the degree of edema. The criteria for extent of pulmonary edema were the same that we used in a recent study of patients with severe hydrostatic and increased permeability edema (27).
Patient Classification Patients were classified as either hydrostatic or increased permeability pulmonary edema (or the adult respiratory distress syndrome) on the basis of careful review of all clinical and laboratory data. There were a total of 16 patients classified as increased permeability pulmonary edema. This classification was confirmed by the availability of pulmonary hemodynamic measurements in 14 of these 16 patieJlfs, all of whom had a pulmonary arterial wedge pressure less than 16 mm Hg. Pulmonary hemodynamic measurements were made at end-expiration and recorded on the oscilloscope monitor paper, and recalibration was done every 3 to 4 h to be certain of the reliability of the measurements (28). Furthermore, bedside echocardiograms demonstrated normal left ventricular function in these patients. There were 18 patients classified as hydrostatic pulmonary edema, all of whom had a history of either chronic or acute cardiac failure or volume overload and clinical evidence of left ventricular heart failure. Also, the mean initial pulmonary edema/plasma protein concentration ratio was 0.54 ± 0.11 for these 18 patients, which provides independent confirmation that these patients had a transudative edema from elevated hydrostatic pressures (19, 23, 25, 26).
Estimate of Alveolar Liquid Clearance In order to estimate the rate of alveolar liq-
1252
MATTHAY AND WIENER·KRONISH
uid clearance in the resolution of alveolar edema in patients, we compared the percent concentration of the final pulmonary edema fluid protein concentration to the initial edema fluid protein concentration and calculated the percent concentration as an index of the clearance of edema fluid from the air spaces of the lung. This approach has been validated as a reliable method for estimating alveolar liquid clearance in studies in normal dog and sheep lungs (7, 17) as well as in one experimental study of increased permeability pulmonary edema (21).
Statistics All data are shown as mean ± standard deviation. A Student's paired t test was used to analyze for significant differences in the initial and final albumin and total protein concentrations, the initial and final ratios of edema fluid to plasma total protein concentration, and the initial and final alveolar-arterial oxygen difference in patients classified as clinical improvement (n = 24) or no clinical improvement (n = 10). A Student's unpaired t test was used to test for significant differences between the 24 patients with clinical improvement compared to the 10 patients with no clinical improvement relative to the initial and final total protein concentration, the ratio of the edema fluid to plasma total protein concentration, and the alveolar-arterial oxygen difference. Also, chi-square analysis was used to analyze for significant differences in mortality in the Group A (n = 9) and Group B (n = 7) patients with increased permeability pulmonary edema. We accepted p < 0.05 as significant (29).
TABLE 1 ALVEOLAR-ARTERIAL OXYGEN DIFFERENCE IN PATIENTS WITH PULMONARY EDEMA Alveolar-Arterial Oxygen Difference (mm Hg) Classification Clinical improvement Hydrostatic Increased permeability No clinical improvement Increased permeability Hydrostatic edema
No.
Initial
Final
Time Interval (h)
15 9
545 ± 74 553 ± 40
366 ± 124368 ± 112-
4.5 ± 2.9 6.8 ± 5.1
7 3
518 ± 84 588 ± 23
538 ± 95 616 ± 22
5.4 ± 4.1 5.3 ± 5.8
Data shown as mean ± SO. • P < 0.01 comparing the final with the initial alveolar-arterial oxygen difference.
TABLE 2 EVALUATION OF THE FOLLOW·UP CHEST RADIOGRAPH COMPARED WITH THE INITIAL CHEST RADIOGRAPH FOR THE EXTENT OF PULMONARY EDEMA Chest Radiograph Analysis Classification Clinical improvement Hydrostatic Increased permeability No clinical improvement Increased permeability Hydrostatic
No.
Less Edema
Unchanged
More Edema
15 9
13 8
2 1
0 0
7 3
0 0
5 2
2 1
TABLE 3 INITIAL AND FINAL PULMONARY EDEMA FLUID PROTEIN CONCENTRATIONS IN PATIENTS WITH CLINICAL IMPROVEMENT Pulmonary Edema Fluid
Results
Overall, 34 patients were entered into the study, ::md 24 of these patients clinically improved on the basis of a decrease in the alveolar-arterial oxygen difference. Of these 24 patients, 15 had hydrostatic pulmonary edema and 9 had increased permeability pulmonary edema (table 1). A total of 10 patients were classified as no clinical improvement, all of whom required an increase in positive endexpiratory pressure to maintain oxygenation. Seven of these 10 patients had increased permeability pulmonary edema, and 3 had hydrostatic edema (table 1). The patients who were classified into the clinical-improvement or the no-clinicalimprovement groups had comparable oxygenation defects at the time of entry into the study with the initial mean alveolar-arterial oxygen gradient greater than 500 mm Hg in all groups (table 1). The improved oxygenation in the 24 patients with clinical improvement was matched by a reduction in the extent of pulmonary edema on the second chest radiograph in 21 of the 24 patients (table 2). In the 10 patients without clinical im-
Patient No.
Classification of Edema
Time Interval (h)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Hydrostatic Permeability Permeability Permeability Permeability Permeability Permeability Permeability Permeability Permeability
4 9 4 3 2 2 2 3 9 6 6 10 2 1 4 12 12 1 1 3 3 12 12 5
Mean ± SO
5.3 ± 3.9
Total Protein (g/dL)
Albumin (g/dL)
Initial
Final
Initial
Final
3.3 1.4 3.5 2.8 3.t 3.4 5.2 2.7 4.5 4.0 3.4 3.4 2.0 1.8 4.4 4.5 5.0 5.0 6.6 3.4 4.5 4.0 5.0 4.0
4.1 2.6 3.9 3.3 4.2 4.9 7.4 3.1 6.8 5.0 5.1 6.1 2.5 2.5 11.1 6.8 7.9 6.2 7.7 3.7 6.8 5.9 9.6 7.0
2.2 .7 2.5 2.0 2.1 1.8 3.8 1.9 3.2 2.4 2.4 1.4 1.3 1.1 2.9 3.1 2.3 2.6 4.2 2.2 2.6 2.9 3.8
2.7 1.3 2.7 2.1 2.8 2.9 5.2 2.5 4.5 3.0 3.2
3.8 ± 1,.2 5.6 ± 2.3- 2.4 ± 0.9
Data shown as mean ± SO. • P < 0.01 comparing final with initial total protein or albumin concentrations.
1.7 7.2 5.4 3.4 4.9 2.3 3.8 4.2 6.8 3.6 ± 1.6-
1253
INTACT EPITHELIAL BARRIER FUNCTION
TABLE 4
TABLE 5
INITIAL AND FINAL PULMONARY EDEMA PROTEIN CONCENTRATION IN PATIENTS WITH NO CLINICAL IMPROVEMENT
COMPARISON OF THE MEASURED PROTEIN OSMOTIC PRESSURE IN THE FINAL PULMONARY EDEMA FLUID SAMPLE WITH THE CONCURRENT FINAL PLASMA SAMPLE IN SIX PATIENTS
Pulmonary Edema Fluid Patient No.
Classification of Edema
25 26 27 28 29 30 31 32 33 34
Hydrostatic Hydrostatic Hydrostatic Permeability Permeability Permeability Permeability Permeability Permeability Permeability
Mean
Time Interval
Total Protein (g/dL)
Albumin (g/dL)
(h)
Initial
Final
Initial
Final
12 12 1 3 6 11 3 1 3 11
3.3 3.3 3.1 5.7 4.5 6.1 4.1 3.6 4.3 3.7
3.2 3.5 3.1 5.8 4.3 6.0 3.9 3.1 4.6 3.4
1.6 1.5 2.2 3.0 3.2 4.0 2.4 2.0 2.9 2.5
1.5 1.5 2.2 3.0 3.0 3.5 2.2 1.9 3.1 2.3
6.3 ± 4.7
4.2 ± 1.0
4.1 ± 1.1
2.5 ± 0.8
2.4 ± 0.7
Protein Osmotic Pressure (mm Hg)* Patient No.
6 11 13 15 18 22 Mean ± SO (n = 6)
Final Edema Fluid
Plasma
29 23 17 72 31 31
25 18 16 25 20 21
34 ± 19
21 ± 4
- Measurements made at 26° C on a Wescor osmometer. Data shown as mean ± SD.
provement, the second chest radiograph showed no change or an increase in the extent of edema (table 2). Radiographically significant atelectasis was not observed on any of the chest radiographs. Also, the level of positive end-expiratory pressure (PEEP) at the time of the sampling of the initial and final pulmonary edema fluid was the same (5 ± 3 cm H 2 0) in the 24 patients with clinical improvement, whereas the level of PEEP was 5 ± 3cmH 2 0and 10 ± 4 cm H 2 0, respectively, in the 10 patients who did not clinically improve. In all 24 patients who had clinical improvement, the final pulmonary edema fluid sample had a higher edema fluid total protein concentration (mean value, 5.6 ± 2.3 g/l00 ml) compared with their initial pulmonary edema fluid total protein concentration (3.8 ± 1.2 g/l00 ml) (table 3) (p < 0.01). Also, the albumin concentration in the final edema fluid sample was always higher than the initial edema fluid albumin concentration (table 3) (p < 0.01). In contrast, in the group of 10 patients who did not clinically improve, there was no difference in their final edema fluid total protein or in the albumin concentration compared with their initial total protein or with the initial albumin concentration in their edema fluid (table 4) (p = ns). In addition, in all 24 patients with clinical improvement, the final edema fluidto-plasma protein concentration ratio (mean value, 1.1 ± 0.37) was higher than the initial edema fluid protein concentration ratio (mean value, 0.65 ± 0.20) (p < 0.01) (figure 1). In contrast, the edema fluid-to-plasma protein concentration ratio did not change between the final (mean value, 0.79 ± 0.18) and the ini-
tial ratio (mean value, 0.80 ± 0.17) in the 10 patients with no clinical improvement (figure 1). In 13 of the 24 patients who clinically improved, the final edema fluid total protein concentration (7.3 ± 1.6 g/100 ml) exceeded the final plasma protein concentration (5.6 ± 0.8 g/100 ml) by a mean value of 1.7 g/l00 ml. In addition, the results of the protein osmotic pressure measurements on the final pulmonary edema fluid sample and the concurrent final plasma sample in 6 of these 13 patients indicated that the measured protein osmotic pressure in the edema fluid in these patients always exceeded the plasma protein osmotic pressure (table 5). In one patient, the total protein concentration of the final edema fluid was 11.1 g/l00 ml compared with a simultaneous plasma protein concentration of 7.2 g/100 ml, and the edema fluid protein osmotic pressure was 72 mm Hg compared with the plasma protein osmotic pressure of 25 mm Hg. Because in all animal experiments the increased concentration of protein in the air spaces reflects the quantity of edema fluid removed from the alveoli (7, 17), the percent concentration of the final to the initial edema fluid protein concen-
Fig. 1. Individual data points are shown for the initial and final pulmonary edema fluid-to-plasma total protein concentration ratio in the 24 patients with clinical improvement compared with the 10 patients with no clinical improvement. The time interval between the initial and final sample was 5.3 ± 3.9 h for the clinical-improvement group and 5.3 ± 4.7 h for the no-clinical-improvement group.
tration was used to compare the rates of alveolar liquid clearance between the 15 patients with hydrostatic edema and the 9 patients with increased permeability edema, all of whom clinically improved (table 6). Interestingly, the data demonstrated that the mean concentration of protein in the edema fluid, adjusted for the time between the initial and final edema sample, was similar between the two groups. There was considerable variability in the time course for resolution of alveolar edema, but both the range and the mean clearance rates as estimated by the concentration of edema fluid protein were nearly identical in patients with resolving hydrostatic or increased permeability edema. As further support for this conclusion, the improvement in oxygenation and the time interval for thif improvement was similar in patients with resolving hydrostatic or increased permeability edema (table 1). Finally, the relationship of alveolar epithelial function to outcome was analyzed in the 16 patients with increased permeability pulmonary edema. In the first 12 h after endotracheal intubation and the onset of mechanical ventilation, there were 9 patients (Group A) who had a rise in their edema fluid protein concentration
NOCUNICAL
1.6
Edema Fluid Plasma (total protein)
IMPROVEMENT
1.2
===== ..
0.8
0.4
p
