We investigated whether inhibiting an endothelial adhesion molecule [intracellular adhesion molecule 1 (ICAM-1)] would alter outcome and lung injury in a similar fashion to inhibition of a leukocyte adhesion molecule (integrin CD11b) in a rat model of gram-negative pneumonia. Inhibition of ICAM-1 with monoclonal antibody (MAb) 1A29 (1 mg/kg sc or 0.2 or 2 mg/kg iv, q 12 h × 3) or of CD11b with MAb 1B6 (1 mg/kg sc, q 12 h × 3) were compared against similarly administered placebo proteins in rats challenged with intrabronchial Escherichia coli. After challenge, all animals were treated with antibiotics. ICAM-1 MAb (6 mg/kg, iv, total dose) increased mortality vs. control (P = 0.03). CD11b MAb (3 mg/kg, sc, total dose) did not significantly (P = 0.16) increase mortality rates, but this was not in a range of probability to exclude a harmful effect. All other doses of MAb had no significant effect on survival rates. ICAM-1 and CD11b MAbs had significantly different effects on the time course of lung injury, circulating white cells and lymphocytes, and lung lavage white cells and neutrophils (P = 0.04–0.003). CD11b MAb decreased, whereas ICAM-1 MAb increased these measures compared with control from 6 to 12 h after E. coli. However, from 144 to 168 h afterE. coli both MAbs increased these measures compared with control rats but to a greater level with CD11b MAb. Thus both ICAM-1 and CD11b appear to be necessary for survival during E. coli pneumonia. Although these adhesion molecules may participate differently in early lung injury, with CD11b increasing and ICAM-1 decreasing inflammation and injury, both are important for the resolution of later injury. During gram-negative pneumonia the protective roles of ICAM-1 and CD11b may make their therapeutic inhibition difficult.
- adhesion molecule
- acute lung injury
leukocyte integrin molecules (CD11/18) and immunoglobulin intercellular adhesion molecules (ICAMs) present on endothelial cells are being investigated as potential therapeutic targets to reduce lung injury (1, 30). These two adhesion receptors act as coligands in the lung during neutrophil adhesion, migration, and activation. Although important to normal lung function, these receptors also participate in the pathogenesis of inflammatory lung injury. One possible application for adhesion receptor inhibitors would be to reduce lung injury caused by inflammatory mediators released during pulmonary infection. Such inhibitors would be given in conjunction with antibiotics to ensure elimination of the microbial focus.
Many studies have now evaluated the effects of inhibiting CD11/18 in animal models of inflammation. In the absence of infection, CD11/18 inhibition in most studies reduces inflammatory tissue injury and improves outcome (4, 15, 17, 25, 45). However, in models employing infectious challenges, inhibiting CD11/18 has frequently been detrimental (9, 12, 34, 35, 38, 40). Similarly, our laboratory found that inhibiting CD11b during inflammatory injury in the absence of infection [e.g., tumor necrosis factor (TNF) challenge in canines], improved lung injury and survival (8). We also found in canines and rats challenged with viable bacteria, however, that both CD18 (R15.7)- and CD11b (1B6)-directed monoclonal antibodies (MAbs) worsened survival (9, 13). In combination, these studies by us and others suggest that although integrins may contribute to proinflammatory injury, they clearly have an important role in host defense. Thus inhibiting integrin function to limit inflammatory tissue injury during infection may be difficult. The potential risk is underscored by the increased rate of bacterial infection found in humans with a congenital CD11/18 deficiency (2).
Less preclinical data are available evaluating the effects of inhibiting ICAM receptors during inflammation and infection. One important member of the ICAM family of adhesion receptors is ICAM-1 (3,7). Expression of this receptor is upregulated by a number of the mediators thought to be involved in the pathogenesis of infectious lung injury (29). Of note, ICAM-1-knockout mice are resistant to the lethal effects of endotoxin (48).
MAbs directed against ICAM-1 have been shown to reduce pulmonary injury related to complement (26) and immune complex deposition (28), as well as nonpulmonary injury related to organ transplant rejection (5, 11), arthritis (16), glomerulonephritis (20), and reperfusion injury (10). Antibodies directed against ICAM-1 primarily target endothelial cells in lung and other tissues but not the neutrophil as do antibodies directed against parts of the CD11/18 complex (24). Thus ICAM-1 inhibition might be beneficial during pneumonia when CD11/18 inhibition was not because it could potentially reduce inflammatory lung injury with infection and not adversely affect leukocyte-host defense function (36). Previous studies have shown that MAbs directed against CD18 but not ICAM-1 worsen abscess formation during soft-tissue bacterial infection (24). ICAM-1 inhibitors have been proposed for use clinically during aspiration pneumonia, which is frequently associated with bacterial infection (22).
In the present study we investigated whether inhibiting ICAM-1 on endothelial cells would prevent inflammatory lung injury and improve survival rates better than inhibiting CD11b on leukocytes during acute infectious lung injury. Previously, we had shown CD11b MAb to be harmful in a total dose of 3 mg/kg subcutaneously (sc) administered over 36 h (13). Therefore, in this study we initially compared the effects of ICAM-1 MAb at 3 mg/kg sc with this same dose of CD11b Mab. We also studied lower and higher doses (0.6 and 6 mg/kg total) of ICAM-1 MAb administered via a different route [intravenously (iv)] to see whether altering these two parameters would better enable us to find a beneficial effect of inhibiting endothelial adhesion molecules.
MATERIALS AND METHODS
Male Sprague-Dawley rats (200–250 g) with chronic indwelling central venous catheters (Zivic-Miller) were randomized in blinded fashion to receive 1 mg/kg sc of either CD11b MAb (total dose 3 mg/kg,n = 54); ICAM-1 MAb (total dose 3 mg/kg, n = 106); CD11b MAb (0.5 mg/kg) together with ICAM-1 MAb (0.5 mg/kg; total dose 3 mg/kg,n = 41); or a control protein (total dose 3 mg/kg, n = 108), 12 h before, at the time of, and 12 h after intrabronchialEscherichia coli challenge. CD11b MAb at 3 mg/kg in a previous investigation had been shown to be harmful in this model (13). To assess whether route or dose could alter the effects of ICAM-1 MAb in the present study, animals received not only the same doses of ICAM-1 and CD11b MAbs given sc but also ICAM-1 MAb in lower and higher doses given iv. ICAM-1 MAb was given iv at a dose of 2 mg/kg (total dose 6 mg/kg, n = 65) or 0.2 mg/kg (with 1.8 mg/kg of mouse serum protein, total MAb dose 0.6 mg/kg, n = 68). In these studies all control rats received 2 mg/kg of mouse serum protein (total dose 6 mg/kg, n = 70) in a similar regimen. Beginning 4 h after bacterial inoculation, all animals were treated with ceftriaxone (100 mg/kg body wt, daily × 4 days; Rocephin, Roche Laboratories, Nutley, NJ) intramuscularly (im). Randomly selected animals were killed 6–12 h or 144–168 h afterE. coli inoculation for blood and lung laboratory measurements (see Blood and lung measurements).
Intact ICAM-1 MAb (MAb 1A29, Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT) and CD11b MAb F(ab′)2 fragments (MAb 1B6, Repligen, Cambridge, MA) were prepared from murine sources. These two MAbs have fewer than five endotoxin units per milligram of protein (13, 41). Lyophilized mouse serum protein with immunoglobulin (Sigma 53509, Sigma Chemical, St. Louis, MO) was used as a control in all experiments. MAbs 1A29 (ICAM-1 MAb) and 1B6 (CD11b MAb) have been previously shown to be reactive and specific for rat endothelial cells and leukocytes, respectively, and to be protective in rat-injury models (10, 20, 26-28, 41). Antibody levels were determined both early, at 6 and 12 h, and late, at 144 and 168 h, after bacterial challenge by using labeled anti-murine globulin antibody.
A suspension of E. coli 0111 was prepared for inoculation as previously described (35). Bacteria (E. coli strain 0111:B4) were stored in 1-ml aliquots of bacto-peptone broth (DIFCO, Detroit, MI) and glycerol at −70°C. Bacterial doses were prepared by inoculating 500-ml bottles of brain-heart infusion (DIFCO), which were incubated then for 18 h at 37°C, followed by centrifugation with two washes and suspension in sterile PBS. The doses of bacteria were quantitated turbidometrically by using a standard curve on the basis of actual viability counts. Animals were anesthetized with ketamine (20 mg/kg im; Fort Dodge Laboratories, Fort Dodge, IO) and xylazine (20 mg/kg im; Miles, West Haven, CT). Each animal then underwent endotracheal intubation under direct visualization with a 20-gauge plastic arterial catheter followed by intrabronchial instillation withE. coli (4.0 × 1010 colony-forming units/kg body wt in 0.5 ml of PBS). Animals were then placed in the prone position on absorbent tissue and observed until full recovery from anesthesia had occurred. They were then housed in Plexiglas cages, observed every 3 h for the first 12 h and then daily, and were considered survivors if still alive after 6 (first study) or 7 (second study) days.
Blood and lung measurements.
After ketamine and xylazine anesthesia in selected animals, tail artery blood was obtained for complete blood count and arterial blood-gas tensions (13). By using sterile techniques, a cardiac puncture was then used to obtain quantitative blood cultures, plasma endotoxin (during sc MAb administration studies only), and serum TNF bioactivity levels (9). Briefly, quantitative blood cultures were collected in 1.5-ml isolator tubes (DuPont Medical Products, Wilmington, DE), with serial dilutions of lysed samples plated for bacterial colony counts. Plasma endotoxin concentrations were determined by using a kinetic, chromogenic limulus lysate assay (MA Bioproducts, Walkersville, MD). Serum TNF bioactivity was measured by using the WEHI 164 cell line assay. Next, the animals were killed by cervical dislocation, and their lungs were removed. One-half of the lungs were used for quantitative bacteria counts, lung lavage cell and protein analysis, and the other one-half were used for wet weight-to-dry weight ratios and histological analysis (13).
Lung lavage was performed by cannulating the trachea with a plastic arterial catheter and sequentially lavaging the lung with four 2.5-ml aliquots of normal saline. These aliquots were combined and centrifuged, the supernatant was removed, and the cell button was resuspended. Cell counts were performed with an electronic cell counter (ZB1, Coulter Electronics, Hialeah, FL). Slides for differential cell counts were prepared (Cytospin 2, Shannon Southern Instruments, Sewickley, PA) and stained (Hematek Slide Stainer, Baxter Scientific, Columbia, MD). Cell differentials were performed on each slide. Lavage supernatants were passed through a 45-μm filter, and protein determinations were made by using a Folin-Lowry technique.
To perform histological analysis, an upper lobe bronchus was cannulated and the lobe was infused and fixed with formaldehyde (4%; Mallinkrodt Specialty Chemicals, Paris, KY), dehydrated, embedded in paraffin, and sectioned at 6 mm. Lung tissue sections were then microscopically analyzed without knowledge of treatment group, and the degree of injury was numerically graded (0, absent, to 5, severe) for six different parameters including perivascular edema (local or diffuse collections of fluid within the perivascular connective tissue) and intra-alveolar edema (local or diffuse collections of fluid within alveoli), hemorrhage (blood within alveoli), fibrin (amorphous material within alveoli) congestion (excessive accummulation of blood within distended alveolar capillaries), and hyperplasia/swelling (hyperplasia and swelling evident in alveolar duct or bronchiolar epithelium). The number of intra- and extravascular pulmonary neutrophils was determined by randomly examining 10 fields/slide at a magnification of ×480 (oil immersion). Neutrophils were classified according to either their vascular or alveolar location.
This experiment was approved by the Animal Care and Use Committee of the Clinical Center and conducted according to the guidelines published by the National Institutes of Health (13). During the study, every effort was made to minimize animal suffering.
Survival data were analyzed by a Cox proportional hazards model (6) for ICAM-1 MAb- and CD11b MAb-treatment effects. The Cox hazard model was stratified so that animals were always compared only with concurrent control animals receiving similar doses and routes of control protein. Relative risks were estimated for each treatment condition. The analysis showed that the effects of ICAM-1 MAb and CD11b MAb were additive relative to the effect on survival, so only the relative risk for the main effects of the agents were reported.
A lung injury score for each group was calculated as described previously (13). For each of nine physiological and histological parameters (i.e., alveolar-arterial oxygen gradient; wet weight-to-dry weight lung ratio; lung lavage protein concentration; histologically determined perivascular edema and alveolar edema, hemorrhage, fibrin, and congestion; and alveolar duct and epithelial hyperplasia/swelling), the mean score in the group of interest was subtracted from the mean score in its concurrent control group. To ensure that the scale was similar across the nine lung injury parameters, this difference for each parameter was divided by a measure of variability (i.e., the root mean square error from an ANOVA). The lung injury scores were analyzed by a four-way ANOVA (39) with effects for route of administration, type of treatment, time, and lung injury parameter as main effects. In addition, two- and three-way interactions were included in the model. The ANOVA indicated that the data could be summarized across the nine parameters to produce a single lung injury score, as well as across route of administration. An illustration of the interaction between time and type of treatment, the only interaction which was significant in the ANOVA, is shown below (see Fig. 2). Data for TNF and endotoxin levels, blood and lung cultures, monoclonal antibody levels, as well as circulating, lung lavage, and histological cells were also analyzed like the lung injury scores. For all clinical variables, the dose of an agent and the route of administration were examined to investigate the impact on the interactions reported. In all cases, with the exception of MAb levels comparing routes of administration, no higher order interactions were observed, so the data were pooled for all other parameters to increase the power of the study. Although data are shown for the groups receiving a combination of ICAM-1 and CD11b MAbs, this group was not included in the data analysis.
The number of animals, treatments, and groups studied are shown in Table 1. The effects of individual MAbs on all parameters were similar [P = not significant (ns)] in a comparison of measures early at 6 vs. 12 h after infection. This was also true late at 144 vs. 168 h after infection (P = NS). Measures at the early time points were averaged, and the late time points were averaged to increase our ability to find significant effects.
Clinical manifestations and survival.
After E. coli challenge, animals appeared weak and lethargic. ICAM-1 MAb at a total dose of 6 mg/kg iv significantly increased mortality compared with control animals (relative risk of death 3.46; 95% confidence interval, 1.32–9.03, P = 0.03). At lower doses, ICAM-1 MAb given iv [relative risk of death 1.34; 95% confidence interval (CI), 0.48–3.7] and sc (relative risk of death 0.87; 95% CI, 0.65–1.13) had no significant effect (P = NS) on mortality. CD11b MAb at a total dose of 3 mg/kg sc did not significantly increase mortality (relative risk of death 1.19; 95% CI, 0.82–1.72). However, this effect on mortality was not in a range of probability values (P = 0.16) to exclude a harmful effect. Furthermore, this effect of CD11b MAb on mortality was very similar to one noted in a previous study using more animals, in which the same dose of CD11b MAb significantly increased mortality rates (P = 0.003) (13).
The effects of CD11b and ICAM-1 MAbs on the time course of lung injury were significantly different (P = 0.007, Fig. 1, Tables 2 and3). From 6 to 12 h after bacterial challenge, CD11b MAb was associated with a reduced lung injury score, whereas ICAM-1 MAb during this same time period was associated with an increased score, compared with control rats. At 144–168 h, CD11b MAb was associated with marked increases in lung injury compared with control animals. At 144–168 h, ICAM-1 MAb also produced increases in lung injury, but not to the same degree as CD11b MAb.
Circulating, lung lavage, and histological cell analysis.
The effects of ICAM-1 and CD11b MAbs on the time course of changes in circulating total white blood cells and lymphocytes and lung lavage total cells and neutrophils were significantly different (P = 0.04, 0.01, 0.02, and 0.003, respectively) and in a similar pattern to that observed with lung injury scores (Fig. 2, Table4). In addition, from the 6- to 12-h period to the 144- to 168-h period, CD11b MAb produced greater mean (±SE) percent increases in hematocrit (−4.3 ± 2.5 to 6.8 ± 3.0%) and lung lavage lymphocytes (−1.6 ± 9.9 to −38 ± 12.7 cells × 104/ml) than ICAM-1 MAb (−0.5 ± 1.5 to 1.2 ± 1.4% and 2.3 ± 5.1 to 3.3 ± 5.1 cells × 104/ml, respectively; both P < 0.03).
The overall effects of each MAb on TNF levels in a comparison of early and late time points were not different (P = NS). When the data from these time points were combined and compared for each MAb and the control animals, there were ordered effects on TNF levels such that CD11b MAb ≥ ICAM-1 MAb ≥ control (5.6 ± 0.6, 4.3 ± 0.3, 3.9 ± 0.2 ng/ml, respectively; P < 0.05) (Table 5). In a comparison of control and ICAM-1 and CD11b MAb groups, there were no significant differences (P = NS) in any other laboratory parameter measured throughout, including quantitative blood and lung cultures and serum endotoxin levels (Table 5).
Murine immunoglobulin serum levels were measured 6 h after sc injections for control mouse serum (276 ± 552 ng/ml), CD11b MAb (1,460 ± 4,089 ng/ml), and ICAM-1 MAb (106 ± 87 ng/ml) and at 144 h after sc injections for control mouse serum (146 ± 133 ng/ml), CD11b MAb (160 ± 80 ng/ml), and ICAM-1 MAb (140 ± 86 ng/ml). Murine immunoglobulin serum levels were measured 6 h after iv injection for control mouse serum (1,210 ± 790 ng/ml) and ICAM-1 MAb in low doses (1,480 ± 1,160 ng/ml) and in high doses (3,220 ± 1,330 ng/ml) and at 168 h for control mouse serum (120 ± 46 ng/ml) and ICAM-1 MAb in low and high doses (undetectable for both). Control protein and MAb levels differed significantly (P < 0.05) from 6 to 168 h in a comparison of sc and iv routes of administration.
Effects of MAbs in animals without infection.
In animals without infection, compared with placebo-treated animals, administration of ICAM-1 MAb did not alter lung function [mean ± SE, alveolar-arterial oxygen gradient (Torr); wet weight-to-dry weight lung ratio; or lavage protein concentration (mg/dl)] at early (control vs. ICAM-1 MAb, 21 ± 2 vs. 21 ± 2 Torr; 1.30 ± 0.04 vs 1.2 ± 0.03; 10 ± 0 vs. 10 ± 0 mg/dl) or late (control vs. ICAM-1 MAb, 11 ± 6 vs. 11 ± 5 Torr; 1.20 ± 0.03 vs. 1.2 ± 0.02; 10.0 ± 0 vs. 25 ± 15 mg/dl) time points (P = NS for all comparisons). This was also true in a previous study using CD11b MAb in normal rats (13).
ICAM-1 MAb increased lung injury and mortality rates in an antibiotic-treated rat model of gram-negative pneumonia. In contrast, CD11b MAb reduced early lung injury in this model. However, as with ICAM-1 MAb, CD11b MAb was associated with a trend in increased mortality rates in this study. Both antibodies were associated with increases in TNF levels during the study, which, although not large, were significant overall. In a previous study our laboratory had also found, using this same model, that CD11b MAb (MAb 1B6) reduced early lung injury but increased mortality rates (13). In the present study, we examined for the first time the late effects of this CD11b MAb on lung injury. We found that the early beneficial effects of CD11b MAb on lung function are transient and that, with time, this antibody also increases lung injury. These studies show that, in gram-negative pneumonia, both ICAM-1 and CD11/18 have protective roles on survival, lung injury, and the regulation of TNF release.
Inhibition of CD11b or ICAM-1 was not associated with significant increases in bacteremia, endotoxemia, or tissue bacteria counts. This is consistent with the previous study from our laboratory of inhibition of CD11b in rat E. colipneumonia (13). In other laboratories, infection withStaphylococcus aureus orStaphylococcus pneumonia in ICAM-1-knockout mice also increased mortality rates without significantly increasing bacteremia (42, 46). Thus either the effect of inhibition of adhesion molecules on bacteremia and endotoxemia are relatively small or the harmful effects caused by inhibition of adhesion receptors during infection occur by mechanisms other than significant reductions in microbial clearance.
Although not significant at the two individual time points investigated, during the overall study, MAb-treated animals had small but significant increases in serum TNF levels. ICAM-1 has been shown in other studies to have an important role in immunological cell interactions (3). Furthermore, isolated spleen cells from ICAM-1-knockout mice have been shown to produce increased amounts of TNF in response to S. aureus, toxic shock syndrome toxin, and concanavalin A challenges (46). This suggests that adhesion molecules or the cells they control may play a role in downregulating the production of TNF. In our study, inhibition of adhesion molecules with MAbs could have interfered with downregulation of TNF, causing increased TNF levels and worsened tissue injury and outcome. In previous studies our laboratory showed that CD11b-directed MAb reduced lung injury and mortality related to intravenous TNF challenge (8). Thus adhesion molecules may have a complex role, downregulating the production of TNF on the one hand but also augmenting its direct effects on the other.
Our laboratory showed previously in a canine model ofE. coli peritonitis that CD11/18 inhibition increased blood endotoxin and TNF levels and worsened outcome (9). In other studies, treatment of rats with ICAM-1 MAb (MAb 1A29) in combination with an MAb directed against a subunit of CD11/18 (CD11a/18) reduced intracellular clearance ofListeria monocytogenes (44). Thus it is clear that at least some adhesion molecule families have an important role in microbial toxin clearance. It is possible that neutrophil adhesion molecules play a more important role than endothelial adhesion molecules in microbial clearance. It is also possible that our culture techniques and assays for endotoxemia may be insensitive to detect important but small changes in bacteremia and endotoxemia. It is also conceivable that differences in results in our present and previous studies with neutrophil vs. endothelial adhesion molecule inhibition relate to other variables such as the type of animal or site of infection.
Serum levels of CD11b MAb were higher when administered via the iv vs. sc route. The smaller size of the CD11b MAb F(ab′)2may have resulted in higher initial concentrations compared with the larger size of the whole ICAM-1 MAb (23). These differences may be responsible, in part, for the trend toward worsened survival with sc CD11b MAb but not sc ICAM-1 MAb, as well as the more pronounced effects of intravenous vs. sc ICAM-1 MAb.
CD11b MAb, over the first 6 h after infection, reduced lung neutrophils and injury, whereas ICAM-1 MAb was associated with increases in these measures during this same period. Thus mechanisms mediated by CD11/18 but independent of ICAM-1 may play an important role in early lung recruitment of neutrophils and injury in this model. This independence may be because CD11/18 interacts with receptors other than ICAM-1, such as ICAM-2, in the early stages of lung injury (1, 21, 32). The data also suggest that ICAM-1 may have a protective effect associated with it at this early time point. Studies in mice have suggested that, under some circumstances, distribution of ICAM-1 may be greater in the interstitium than in adjacent alveoli. Such a distribution might be protective, serving to limit migration of injurious neutrophils into or to stimulate their movement out of the alveolar space. If so, inhibition of ICAM-1 in early lung injury might antagonize this anti-inflammatory mechanism, resulting in greater alveolar accumulation of neutrophils (18). Thus CD11b and ICAM-1 may have different, and in fact opposite, roles in the development of acute infectious lung injury.
Most models of pneumonia have not assessed the effects of new therapeutic agents late in the course of the infection, when resolution of injury and lung repair might be prominent in surviving subjects. In our study, circulating levels of both of these MAbs as well as worsened lung injury were present 6–7 days after infectious challenge. These findings are consistent with a contribution of both CD11/18 and ICAM-1 to resolution of late injury and of lung repair. Inhibition of CD11a and CD11b has also been shown to prolong the course of bleomycin-induced lung injury, such that early neutrophil recruitment persists and a subsequent reparative lymphocyte recruitment is delayed (31).
The control in this study was not a sham antibody. In the absence of a nonbioactive binding control protein in the present study, it is not possible to exclude a harmful, nonspecific effect of CD11b and ICAM-1 MAbs. Importantly though, these MAbs were not associated with lung injury or alterations in survival rates in normal animals in previous studies or the present one (13).
MAbs directed against either CD11/18 or ICAM-1 have been proposed as therapeutic agents in a number of conditions associated with inflammatory injury. ICAM-1-directed MAbs are under study in transplant patients and in those with rheumatoid arthritis (14, 19). In addition, use of ICAM-1-directed MAbs after gastric aspiration, a condition frequently complicated by infection, has been considered (22). In this study and in a previous one using iv E. coli challenge in baboons, inhibition of ICAM-1 increased lethality rates (47). Similarly, whereas CD11/18-directed MAbs have been proposed for use in trauma patients to minimize reperfusion injury, inhibition of this complex during infection was associated with significant increases in lethality in a previous study conducted by our laboratory, as well as a similar trend in the present study (13, 43). It may be challenging to find a beneficial role for inhibiting these receptors in clinical situations complicated by pulmonary bacterial infection, because these receptors have critical roles in survival and the resolution of lung injury during pneumonia.
The authors thank Christine Knuth and Julie Friedman for editorial assistance with this manuscript.
Address for reprint requests and other correspondence: P. Q. Eichacker, Critical Care Medicine Dept., National Institutes of Health, Bldg. 10, Rm. 7D43, 9000 Rockville Pike, Bethesda, MD 20892-1662 (E-mail:).
Portions of this work have been presented at the annual meeting of the American Thoracic Society, New Orleans, LA, May 1996 (Am. J. Respir. Crit. Care Med. 153, 1996).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 1999 the American Physiological Society