INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Polymorphonuclear leukocyte (PMN) influx is a prominent cytologic and histologic feature of pulmonary inflammation in sepsis and the acute respiratory distress syndrome (ARDS) (1, 2). High numbers of neutrophils in bronchoalveolar lavage (BAL) fluid correlate with the degree of abnormality in gas exchange and lung protein permeability (3) and, in patients at risk, the number of PMNs in BAL fluid correlates with subsequent development of ARDS (4). It is likely that extensive migration of activated PMNs into the alveolar spaces is part of the effector arm of an uncontrolled inflammatory response, causing tissue damage via the release of reactive oxygen species and proteases.

One strategy for reducing acute lung injury (ALI) in sepsis is to limit the influx of PMNs into the alveolar space. The paradigm of PMN migration into the lungs involves a two-step process that includes (1) firm tethering to the endothelium and (2) transmigration. After infusion or instillation of lipopolysaccharide (LPS), PMNs adhere rapidly to pulmonary capillaries. This initial margination and subsequent capillary transmigration are regulated by factors that include leukocyte and endothelial adhesion molecule expression, elaboration of chemoattractants, and cytoskeletal changes (5). Three major families of cell surface receptors mediate leukocyte adhesion: the immunoglobulin superfamily, the selectins, and the integrins. One receptor-ligand pair of particular importance in neutrophil-endothelial interactions in the lung consists of the PMN integrin CD11b/CD18 and endothelial CD54, an immunoglobulin-like receptor also known as intercellular adhesion molecule 1 (ICAM-1). PMN migration in the lung is often described as CD11b/CD18 dependent or independent. Migration stimulated acutely by Escherichia coli LPS is CD18 dependent (5). In small animal models of acute lung injury, administration of monoclonal antibodies to CD54 decreases PMN migration into the lung (6).

The endothelial ligand CD54 contains five extracellular immunoglobulin-like domains, a transmembrane region, and a cytoplasmic tail (9). It is constitutively expressed at low levels on several cell types including vascular endothelium, immune cells such as monocytes, T lymphocytes, and PMNs. CD54 expression on all of these cells increases after stimulation with proinflammatory cytokines (10). In addition to facilitating cell adhesion and migration, experimental evidence suggests that CD54 and other cell surface adhesion molecules also have important functions in cell signaling to produce specific immune and inflammatory responses, for example, monocyte activation and PMN cytotoxicity (11, 12).

We hypothesized that administration of a monoclonal antibody to CD54 would decrease PMN sequestration and transmigration in the lung and attenuate lung injury in gram-negative sepsis. We used a baboon model of sepsis, in which the animals are infused with a "priming dose" of heat-killed E. coli before the infusion of live bacteria (13, 14). This effectively creates a host "at risk" for development of ALI, and subsequent infusion of live bacteria results in injury to the lung and other tissues that approximates clinical sepsis, ARDS, and multiple organ failure. In the following study, CD54 was blocked with a murine anti-human antibody that binds to extracellular domain 2 and inhibits neutrophil adhesion to vascular endothelium (15, 16).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Twelve adult male baboons (Papio cyanocephalus) weighing 14 to 20 kg were obtained from the Southwest Foundation for Biomedical Research (San Antonio, TX), quarantined for a minimum of 4 wk, and determined to be tuberculosis free before use. Animals were handled in accordance with AAALAC (American Association for Accreditation of Laboratory Animal Care) guidelines. Animals were divided into treatment and sepsis controls groups (n = 6 each) and prepared as noted below. Treated animals received 1 mg/kg intravenously of monoclonal antibody to CD54 (R6.5; Boehringer-Ingelheim, Danbury, CT) at t = 12 h, immediately before the infusion of live bacteria. The murine anti-human monoclonal antibody used in these studies is a clinical-grade, low-endotoxin whole antibody preparation (isotype IgG_2a) that cross-reacts with baboon tissues.

After an overnight fast each animal was sedated with intramuscular ketamine (20-25 mg/kg) and orotracheally intubated. Heavy sedation was maintained with ketamine (3-10 mg/kg/per hour) and diazepam (0.4-0.8 mg/kg every 2 h). Animals were paralyzed intermittently with pancuronium (4 mg intravenously before respiratory measurements) and mechanically ventilated with a volume-cycled ventilator (model 7200a; Puritan-Bennett, Kansas City, KS). The fraction of inspired oxygen (FI_O2) was 0.21, the tidal volume was 10 ml/kg, the positive end-expiratory pressure was 2.5 cm H₂O, and the rate was sufficient to maintain an arterial PCO₂ close to 40 mm Hg. An indwelling arterial line and a pulmonary artery catheter (5F) were placed via femoral cut-down for hemodynamic monitoring. Ampicillin (1 g) every 6 h intravenously and gentamicin (40 mg) and polymyxin (20,000 units) every 4 h intratracheally were given to prevent nosocomial pneumonia and secondary bacteremia. All instumentation was completed before the initiation of the sepsis protocol. All animals received approximately 10⁹ colony-forming units (CFU) of heat-killed E. coli per kilogram as a 60-min infusion at t = 0 h, 12 h before receiving live E. coli. Sepsis was induced at t = 12 h by infusing E. coli at 1-2 × 10¹⁰ CFU/ kg in a volume of 50 ml over 60 min (infusion pump model 351; Sage Instruments, Cambridge, MA). Gentamicin (3 mg/kg intravenous) was administered 60 min after completion of the live E. coli infusion. Fluids were given at a rate of 100-150 ml/kg over the first 24 h to support blood pressure as needed for sepsis-induced hypotension. Typical volume requirement in the first 6-8 h after live E. coli was 1,000 ml. Dopamine was used to support animals with a mean arterial pressure () below 65 mm Hg and unresponsive to fluids. Animals were killed by KCl injection at t = 48 h, which was 36 h after the live bacterial infusion, or when specific termination criteria were met, for example, for refractory hypotension ( less than 60 mm Hg), hypoxemia (need for FI_O2 greater than 40%), or refractory metabolic acidosis.

Hemodynamic Monitoring

Heart rate, temperature, arterial blood pressure, pulmonary artery pressure, central venous pressure (Pcv), pulmonary capillary wedge pressure (Ppc,we), ventilator parameters, and fluid balance were recorded every hour during the 48-h protocol. Every 6 h, measurements were obtained of cardiac output (thermodilution, Horizon 2000 monitor; Mennen Medical, Clarence, NY), arterial and mixed venous blood PO₂, PCO₂, and pH (pH/blood gas analyzer, model 1304; Instrumentation Laboratory; Lexington, MA), and oxygen saturation, oxygen content, and hemoglobin (co-oximeter, model 482; Instrumentation Laboratory).

Preparation of Escherichia coli

Escherichia coli reconstituted from lyophilized specimens (serotype 086a:K61; American Type Culture Collection, Rockville, MD) and stored at 70° C were thawed, streaked onto sheep blood agar plates, and incubated at 37° C for 18-24 h. Plates were checked for purity of culture and isolated colonies were streaked onto trypticase soy agar slants and incubated for 18 h at 37° C. Before the experiment, organisms were gently washed off the slants with sterile saline, centrifuged, and resuspended in saline. The concentration was measured by transmittance at 550 nm and adjusted with saline to give, in 50 ml, a final dose of 1-2 × 10¹⁰ CFU/kg for each baboon. This dose of E. coli (1 × 10¹⁰ CFU/kg) is a 100% lethal dose (LD₁₀₀). For heat-killed E. coli, the bacteria were heated in a water bath at 65° C for more than 30 min. The number of organisms and efficacy of heat killing were confirmed by colony counting, using pour plates. Efficacy of heat killing was > 99.99%.

Ventilation-Perfusion (A/) Measurements.

A/ distributions were quantified by the multiple inert gas elimination technique (MIGET) of Wagner and coworkers (17). A dilute solution of six inert gases (sulfur hexafluoride, ethane, cyclopropane, enflurane, dimethyl ether, and acetone) in normal saline was infused continuously via a peripheral vein at a rate of 2.5 to 3 ml/min. Duplicate samples of systemic arterial blood, mixed venous blood, and mixed expired gas were collected at least 40 min after the beginning of infusion. Each blood sample was equilibrated with an equal volume of nitrogen in a heated water bath (37° C) for at least 40 min. The equilibrated gas and expired samples were analyzed for sulfur hexafluoride by an electron capture detector and for the other five gases by a flame ionization detector (model 5890 series II chromatograph; Hewlett-Packard, Wilmington, DE). The data were stored digitally on a computer for analysis. Blood-gas partition coefficients of the six inert gases were also determined each day. Retention-partition coefficient and excretion-partition coefficient curves were generated by enforced smoothing and least-squares analysis, and the data were transformed into A/ distribution, using a 50-compartment lung model. Shunt was defined as A/ less than 0.005 and dead space as A/ greater than 100. The logA/ standard deviations for perfusion and ventilation (SDq and SDv) were used to measure dispersion of blood flow and ventilation, respectively. A/ measurements were considered satisfactory if the residual sum of squares was 10 or less. A/ measurements were done at t = 0, 12, 18, 36, and 42 h.

Hematologic Measurements and Antibody Levels

Serial blood samples were drawn at t = 0 (before heat-killed E. coli), 12 (before live E. coli), 13, 18, 24, 36, and 48 h. Complete blood counts were performed (Sysmex-1000 hemocytometer; Sysmex, Long Grove, IL), and serum was stored at 80° C. R6.5 levels were measured quantitatively by an enzyme-linked immunosorbent assay (ELISA).

Tissue Preparation

At the end of the experiments the left mainstem bronchus was ligated and the left lung was removed. Bronchoalveolar lavage (BAL) was performed on the left upper lobe with 240 ml of 0.9% saline. Tissue samples were taken from the remainder of the left lung for wet/dry lung weight determination and biochemical studies. Samples were stored at 80° C before analyses. The right lung was inflation fixed via the endotracheal tube for 15 min at 30 cm fixative pressure with 2% glutaraldehyde in 0.85 M sodium cacodylate buffer (pH 7.4), and then immersed in fixative for 7 to 10 d before processing for light and electron microscopy.

Biochemical Measurements

Myeloperoxidase assay. Tissues were weighed and homogenized, mixed with an equal volume of 1% hexadecyltrimethyl ammonium bromide in phosphate buffer (pH 6.0), and rehomogenized. The suspension was freeze-thawed three times in liquid nitrogen, rehomogenized, and centrifuged at 15,000 × g for 5 min at 4° C. Samples were assayed by comparison at 460 nm with a standard curve of the decomposition of hydrogen peroxide in the presence of o-dianisidine (U-2000 spectrophotometer; Hitachi Instruments, Norcross, GA) (modified from reference 18). MPO activity was expressed as a change in absorbance per minute per gram wet weight tissue.

Thiobarbituric acid-reactive substance assay. Aliquots of the supernatants prepared above were mixed with an equal volume of 3.5% sodium dodecyl sulfate (SDS; Bio-Rad, Hercules, CA) and 0.13 ml of 100% acetic acid and adjusted to pH 3.5. Five milliliters of 0.6% thiobarbituric acid was added to each sample and then heated at 95° C for 30 min. After cooling, 1.5 ml of n-butanol:pyridine (15:1) was added, and samples were vortexed for 30 s and centrifuged at 4,000 rpm for 10 min. Absorbance of supernatants was measured at 525 nm and expressed as absorbance per gram wet weight of tissue.

Protein and LDH. Protein concentration of BAL fluid was measured with the bicinchoninic acid assay, using bovine serum albumin as a standard (19). Lactate dehydrogenase (LDH) was measured by conversion of L-lactate and nicotinamide adenine dinucleotide (NAD⁺) to pyruvate and NADH. Absorbance was measured by time-based spectrophotometry at 339 nm and values expressed in units of activity per liter (U/L).

Histology and Lung Morphometry

Lung tissue from all animals was examined by both light and electron microscopy (EM). The lung was transversely sectioned into 1-cm slices and stratified random samples were obtained by random selection of eight 1-cm³ cubes (three from the upper lobe, one from the middle lobe, and four from the lower lobe). Large bronchovascular structures were not included. For light microscopy, tissue samples were embedded in paraffin, sectioned, and stained with hematoxylin and eosin. For EM, 5 tissue cubes were sectioned into smaller cubes 1 to 2 mm on a side, and 10 to 20 of these small cubes from each site were processed together. The tissue was washed in cacodylate buffer and postfixed with 2% osmium tetroxide. After dehydration in graded ethanol solutions they were immersed in propylene oxide and embedded in Epon resin. Thin sections were cut with a diamond knife, placed on 200 mesh coated copper/rhodium grids, and stained with uranyl acetate and lead citrate. Sections were viewed and photographed on a Zeiss (Thornwood, NY) 10C transmission electron microscope, and then enlarged to ×8,500 on 11- by 14-in. photographic paper for analysis. Micrographs were analyzed morphometrically by EM, using the methods of Weibel and Bolender (20).

Thirty photos were counted from each of the 5 sites, using a 112-line overlay, yielding 224 points per micrograph. Point and intercept data were recorded as follows. Measurements were made for epithelial volume density including alveolar Type 1 and 2 cells, endothelial volume density, inflammatory cell volume densities in intravascular, interstitial, and alveolar compartments, and interstitial volume density. The interstitium was analyzed using both matrix and cell counts. Fibroblast volume density data included other noninflammatory interstitial cells such as myofibroblasts and pericytes, which cannot be differentiated from fibroblasts in fragmentary cytoplasmic profiles present on electron micrographs. Intercept counts were used to calculate surface density of alveolar epithelium and capillary endothelium. Segments of alveolar basement membrane denuded of epithelial covering because of severe injury were recorded as bare basement membrane and added to the calculation of total alveolar basement membrane surface area. Similarly surface intercepts of capillary basement membrane, where endothelium was fragmented or widely displaced, were counted as disrupted capillary basement membrane. These data are expressed as percent coverage of the respective basement membrane. Morphometric data were normalized relative to the surface density of epithelial basement membrane in a standard volume of alveolar tissue, assuming that the alveolar basement membrane remains intact even in badly damaged areas and does not stretch significantly (13, 14).

Statistics

Data are expressed as means ± standard error. Repeated measures data from physiologic measurements were analyzed by linear regression analysis, allowing for both fixed and random effects. The standard repeated measures model was used to account for missing values that resulted from some animals dying before the scheduled termination point (SAS Statistical Software; SAS Institute, Cary, NC). Drug effect and changes over time were evaluated at the 48-h time point. Repeated measures analysis of variance was used for physiologic data that did not fit first- or second-order models. Morphometry data and biochemical data from BAL fluid and tissue obtained at the end of the experiments were analyzed with an unpaired t test. Survival data were expressed as median values after the initiation of live bacterial sepsis at t = 12 h, and were analyzed by the LIFETEST procedure on SAS. The p values are provided; p < 0.05 was considered significant and trends are noted for p < 0.10.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Treatment of septic baboons with anti-CD54 monoclonal antibody did not significantly decrease the extent of lung injury but, unexpectedly, decreased survival. Median survival after the live E. coli infusion was given (at t = 12 h) decreased from 36 h in the sepsis control group to 24 h in septic animals treated with anti-CD54 (p < 0.05, Figure 1). The causes of early death in the treated animals were refractory hypotension (5), acidosis (2), and hypoxemia (2); two animals met multiple criteria. Hypotension in animals that received anti-CD54 antibody was refractory to fluid administration, and five of the six animals in the treatment group required dopamine to support blood pressure before meeting termination criteria. In contrast, sepsis control animals developed hypotension that typically responded to fluid resuscitation alone and only one control animal met early termination criteria, with refractory hypotension. Results of physiologic, biochemical, and histologic measurements are provided in the following sections.

View larger version (11K): [in this window] [in a new window]   Figure 1.   Survival of sepsis control and anti-CD54 treatment groups. Mortality in sepsis was increased after anti-CD54 treatment compared with controls (p  0.05). Heat-killed bacteria were infused at t = 0 h; live E. coli were infused at t = 12 h (indicated by arrow).

Physiologic Measurements

The hemodynamic measurements for the two groups are shown in Table 1, with additional physiologic measurements in Table 2. Heart rate and per kilogram increased similarly in both groups after infusion of heat-killed bacteria. After live E. coli infusion, animals in both treated and control groups developed hyperdynamic cardiovascular responses with progressive increases in per kilogram and heart rate (HR). Systemic vascular resistance (Rva,s) · kg decreased after heat-killed bacteria and declined further after live E. coli administration in both groups. These responses did not differ significantly. Hypotension worsened with anti-CD54 antibody therapy (p < 0.02; Figure 2) and was significant despite the use of dopamine for blood pressure support in these animals. Ppc.we was maintained with intravenous fluids and did not differ between the two groups. Oxygen delivery (O₂)/kg increased after priming in both groups and then decreased similarly after live bacteria were infused. Oxygen consumption (O₂)/kg increased similarly in both treated and untreated septic animals.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Mean arterial pressure (Pa, or MAP) in sepsis control and anti-CD54 treatment groups. Pa decreased over time in both groups, but to a significantly greater extent in septic animals that received anti-CD54 antibody (p = 0.019). This difference was significant despite the use of dopamine in addition to fluids for blood pressure support in the antibody-treated group.

In addition to the worsened hypotension, renal function was worse in septic animals that were treated with anti-CD54. Urine output declined during sepsis in both treated and untreated animals, but dropped more precipitously in the treated group (p < 0.05; Figure 3). All the animals in the group that received anti-CD54 antibody were anuric before the end of the experiment, compared with half the untreated septic controls. In addition, antibody-treated animals developed a more severe metabolic acidosis with greater decrease in serum pH (p < 0.01; Figure 4). The PCO₂ was controlled with mechanical ventilation and did not differ between the two groups, but calculated bicarbonate concentrations decreased significantly after live E. coli administration in the antibody-treated animals, remaining stable in the septic controls (p < 0.01; Figure 4).

View larger version (23K): [in this window] [in a new window]   Figure 3.   Urine output in sepsis control and anti-CD54 treatment groups. Urine output decreased over time in both groups after the infusion of live bacteria but was more severe in the sepsis groups that received anti-CD54 therapy (p  0.05). Complete anuria was more frequent in the antibody-treated group. Heat-killed bacteria were infused at t = 0 h; live E. coli were infused at t = 12 h.

View larger version (14K): [in this window] [in a new window]   Figure 4.   Metabolic acidosis in control and antibody-treated groups. Decreases in arterial pH and serum [HCO3] in sepsis were more severe in the group treated with anti-CD54 (p < 0.01 for each).

Pulmonary vascular resistance and mean pulmonary artery pressure increased during sepsis but were not different between the two groups. Peak airway pressures (Paw) increased almost 50% in both groups (p < 0.01), but the increase was greater in anti-CD54-treated animals surviving to the end of the experiment (p < 0.01). There was no time-drug interaction for this parameter, suggesting that the overall change in Paw was similar in the two groups. Likewise, pulmonary system compliance decreased similarly after heat-killed and live E. coli administration in the two groups. Gas exchange worsened in both treated and untreated animals after live bacterial infusion (p < 0.01) but there were no significant differences in D(A-a)O₂ between the two groups.

Ventilation-perfusion measurements are shown in Figure 5. Shunt (%) increased equally after priming with heat-killed bacteria and after administration of live E. coli sepsis in the two groups. Dead space did not change significantly in either group. Dispersion of ventilation and perfusion was increased after administration of heat-killed and live bacteria (p = 0.03 and p < 0.01, time effects). SDq increased more in antibody-treated septic animals (p = 0.047), although the number of animals remaining alive at the later time points was small.

View larger version (49K): [in this window] [in a new window]   Figure 5.   Ventilation-perfusion (A/) data during sepsis. Shunt, shown as percent cardiac output (% or %CO), and dispersion of ventilation (SDv) and perfusion (SDq) all increased during sepsis in both groups. SDq increased more in the group treated with anti-CD54 (p = 0.047), although the number of animals represented at the final time point was only two. Shunt and SDv increases did not differ significantly between the two groups. Dead space (DS), shown as percent tidal volume (%VT), did not change significantly in either group.

Hematologic and Drug Measurements

Hemoglobin levels dropped gradually during sepsis. Total circulating leukocyte count dropped sharply to near zero in all groups after infusion of live E. coli with subsequent increase back to preinfusion levels. Platelet counts also decreased markedly in both groups after infusion of live bacteria (Figure 6). There were no significant differences between the two groups in any hematologic parameters. Drug levels 1 h after infusion of R6.5 at 1 mg/kg averaged 19.7 ± 3.2 µg/ml and declined progressively throughout the remainder of the experiment.

View larger version (14K): [in this window] [in a new window]   Figure 6.   White blood cell (WBC) and platelet counts from sepsis control and anti-CD54-treated groups. WBC counts dropped sharply in both groups after live bacteria were infused, with progressive recovery over the next 36 h. Platelet count decline was persistent during the experiments. There were no differences between the two groups.

Biochemical Measurements

Lung wet/dry weights were increased in both untreated and treated septic animals (6.19 ± 0.47 versus 6.63 ± 0.88, p = NS). A normal value for the lungs of these baboons in our laboratory is 4.97 ± 0.20. Small bowel wet/dry weights were also increased but were not different between the two groups (6.69 ± 0.57 versus 6.48 ± 0.59, p = NS). BAL biochemical measurements were also unchanged by antibody treatment. LDH was 54.6 ± 27.0 in untreated septic controls and 79.9 ± 19.0 U/L in septic animals treated with anti-CD54. BAL protein was 1.09 ± 0.49 and 1.34 ± 0.73 mg/ml, respectively. Myeloperoxidase (MPO) values for lung, liver, heart, kidney, and small bowel are shown in Figure 7. MPO in all tissues in the treated animals, except in the liver, where MPO was 3.4-fold higher in septic animals treated with anti-CD54 (p < 0.01). Thiobarbituric acid-reactive substance (TBARS) assay results were similar in both groups for all tissues.

View larger version (14K): [in this window] [in a new window]   Figure 7.   Measurements of myeloperoxidase (MPO) in tissues from sepsis control and anti-CD54-treated groups. There were no significant differences for MPO values in lung, small bowel, kidney, and heart, but MPO was significantly higher in liver from septic baboons treated with anti-CD54 antibody (*p < 0.01).

Pathology

At autopsy, lungs, kidneys, and small intestine were grossly abnormal in all animals. Areas of hemorrhage were apparent in lungs and kidneys, and small bowel was edematous. Metastatic abscesses were not seen in either group at autopsy or under light microscopy, although tissues were not cultured postmortem. Under light microscopy the lungs of sepsis control animals had intra-alveolar edema and a mixed inflammatory cell infiltrate. Alveolar septae were thickened with increased cellularity. Findings were similar in septic animals that were treated with monoclonal antibody to CD54 although intra- alveolar cellularity was less extensive.

Quantitative morphometry was done on electron micrographs of the lung from the two groups. A summary of the results is shown in Table 3. Sepsis in primed animals damaged the alveolar epithelium and endothelium, increased interstitial thickness and cellularity, and caused inflammatory cell infiltration into the alveolar spaces. Treatment with anti-CD54 did not prevent the increases in volume density of endothelium or Type 1 epithelium in sepsis. There was also no difference in volume density of Type 2 epithelium between treated and untreated septic animals, although there was a trend toward increased percentage of basement membrane covered by Type 2 alveolar epithelium and decreased percentage covered by Type 1 epithelium in the group treated with anti-CD54 (p = 0.08 for each). The increased volume density of interstitium and fibroblasts in sepsis was unaffected by CD54 antibody treatment. The total volume density of neutrophils in the lung did not change and the distribution of PMNs was also not significantly different between the two groups. In contrast, alveolar macrophage volume density was significantly different in septic animals treated with anti-CD54 (p < 0.01). Alveolar macrophage volume density was increased after live E. coli sepsis but in the treated animals was close to the normal reference value for our laboratory (Figure 8).

View larger version (17K): [in this window] [in a new window]   Figure 8.   Alveolar macrophage and polymorphonuclear leukocyte volumes, determined on the basis of morphometric analysis by electron microscopy. Increase in alveolar macrophage volume in sepsis was attenuated by treatment with anti-CD54 monoclonal antibody (*p < 0.01). The decrease in alveolar PMN volume was not significant.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The anti-CD54 monoclonal antibody used in this study (R6.5) did not prevent lung injury in gram-negative sepsis but did have a profound and unexpected impact on the severity of shock. Septic animals treated with antibody to CD54 developed refractory hypotension, metabolic acidosis, and renal insufficiency. Mortality increased in the treated group, primarily as a result of hypotension. In contrast, measures of acute lung injury did not differ between the two experimental groups. There were no significant group differences in alveolar-arterial oxygen gradient, pulmonary system compliance, or lung water, although two animals from the treatment group did develop hypoxemia that met our termination criteria. In addition, we could not demonstrate any decrease in neutrophil sequestration or migration into the lungs despite therapeutic levels of anti-CD54. These results were contrary to our hypothesis that CD54 blockade would decrease neutrophil migration into the lung and attenuate lung injury in sepsis.

One potential reason for the adverse physiologic outcome after CD54 antibody treatment is a change in proinflammatory cytokine profiles. We have reported that CD54 antibody increased peak levels of interleukin 1 (IL-1) and increased both peak and duration of the elevations in IL-6, IL-8, and tumor necrosis factor receptor 1 in sepsis (21). We have also reported similar physiologic effects and increased mortality after blockade of E and L selectin (CD62E and CD62L) in this animal model, although the effect on cytokine profiles was different (14, 21). In contrast to our findings in sepsis, R6.5 administration to patients with rheumatoid arthritis resulted in a prolonged decrease in peripheral blood monocyte IL-6 mRNA (22), supporting a potential regulatory role for CD54 in cytokine production. Although R6.5 is not known to activate ICAM-1 in vitro, it is possible that unsuspected agonist effects may have occurred in our experiments. If receptor density were increased by prior immune system activation, cross-linking might occur and activate intracellular signaling. In cell lines, specific cross-linking of CD54 induces IL-1 transcription (23), and increases IL-8 production (24). Although CD54 blockade might be expected to decrease proinflammatory cytokine production, preactivation of inflammatory responses as in our experiments may change receptor density, distribution, or affinity and lead to different effects of adhesion molecule blockade on outcome.

Other studies of adhesion molecule blockade have suggested that CD54 is important to survival in localized infection. In mice with staphylococcal bacteremia and arthritis, antibody blockade of CD54 improved local disease but increased mortality at 4 d. The investigators reported similar results in mice genetically deficient in CD54 and speculated that CD54 might help optimize phagocytic defenses (25). Survival was also decreased after blockade of either CD54 or its PMN ligand, CD11b/CD18, in rats with pneumonia. The cause of increased mortality in these studies was proposed to be related to interference with local host defense (26). Also, genetic CD54 deficiency has had variable effects on survival after systemic or local infection (25, 27). We did not detect metastatic abscess formation at necropsy that might suggest a change in bacterial clearance. None of the studies using either blockade or deficiency of CD54, however, showed any correlation between survival and bacterial clearance (25, 28, 29).

CD54 blockade in our studies may have altered host response to E. coli through effects of the antibody on other immune or nonimmune cell function. The CD54 receptor and its ligands are found on many nonendothelial cell types, including monocytes, lymphocytes, epithelial cells, and fibroblasts, and are induced in response to inflammation (10). On these cells CD54 coregulates important immune functions of T lymphoctes, for example, antigen presentation (31) and determination of helper T phenotype and responsiveness (32). Blockade of CD54 may also affect the function of other surface receptors for which it may serve as an accessory molecule. In vitro, CD54 influences target cell sensitivity to membrane-bound Fas ligand and may therefore have an important role in regulating T cell-mediated cytotoxicity (33). Blockade of CD54 might redirect immune responses toward more intense inflammation with excessive "bystander" injury (33). CD54 also stimulates neutrophil respiratory burst and protease release in some in vitro models of cytotoxicity (12).

Monocyte function and migration is also influenced by CD54 (15). In vitro stimulation of the receptor enhances production of macrophage inflammatory protein 1 (MIP-1) (34) and cosignals the oxidative burst (11). Although our histologic analysis did not detect a significant decrease in PMN migration into the alveoli, the alveolar macrophage volume after anti-CD54 antibody was lower than in untreated controls. Interestingly, the presence of increased lung macrophages in patients with acute lung injury correlates with improved survival, especially with underlying sepsis (1). Because of the early mortality in our experiments, we do not know whether the difference in alveolar macrophages might have persisted and eventually influenced the course of acute lung injury.

Although anti-CD54 decreased monocyte accumulation in the lung, it did not affect neutrophil content. Thus it seems likely that alternate pathways of neutrophil migration are induced in the lung by the complex sequence of inflammatory stimuli in sepsis. CD54 antibody in our study also did not affect neutrophil content significantly in other tissues, with the exception of the liver, where MPO actually increased 3.4-fold over sepsis controls. The higher cytokine levels in septic animals treated with CD54 antibody (35) may have altered adhesion interactions among cell types. For example, in vitro studies have shown that CD54-independent adhesion may predominate in hepatocyte-PMN interactions in the setting of high IL-8 concentrations (36).

The changes in cytokine profiles in after CD54 antibody and after E and L selectin blockade suggest that the unexpected physiologic deterioration in this study is a result of interference with specific host defenses in sepsis, although a toxic contribution from the antibody preparation cannot be excluded. Potential toxic effects include complement activation, although in clinical use of monoclonals this is thought to cause only mild and transient effects, occurring primarily at higher antibody doses. R6.5 has been shown in vitro to activate PMNs through an isotype-specific complement-mediated mechanism (37); however, it has been used safely both in animal toxicity studies and in human trials for rheumatoid arthritis and renal transplantation (22, 35, 38). In those studies antibody infusions have been tolerated without physiologic responses that would have predicted the adverse outcomes of our experiments. The serum levels we achieved are comparable to those reported to be safe and efficacious in human trials (35, 38). Immune complex formation is a second potential source of toxicity, although tissue deposition of immune complexes has not been reported in any animal or human studies using monoclonal antibody therapy. Soluble CD54 can be elevated in humans with sepsis and ARDS (39) and is likely to be elevated in our animals as well. The possibility of renal immune complex deposition was not assessed in these animals; however, patients with rheumatoid arthritis have increased soluble CD54 levels (40) and do not exhibit renal toxicity (35). We also did not test F(ab')₂ fragments and cannot exclude the possibility that it would decrease injury in sepsis, but this result seems unlikely because it would require such a huge reversal of effect.

In summary, our results indicate that antibody blockade of CD54 in baboons during E. coli sepsis increased shock and mortality without effect on lung neutrophil migration or acute lung injury. The data suggest that differences in host immune and/or injury responses during sepsis are responsible for the worsened outcome in antibody-treated animals in this study compared with other inflammatory conditions. In clinical trials where the effects of CD54 blockade have been promising, an immunosuppressive paradigm predominates, and the use of adhesion molecule blockade in gram-negative sepsis or in patients at increased risk for sepsis should be approached cautiously.