INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: ARDS; VEGF; endothelial; permeability

The syndrome of acute respiratory distress in adults (ARDS) is a devastating disease with a mortality that remains unacceptably high (1). Pulmonary injury in ARDS results in the disruption of the alveolar-capillary membrane. It is now commonly believed that inflammatory mediators create an acute inflammatory response in the microvessels of the lung (and other organs) and that locally released inflammatory cell products damage the endothelial cells resulting in increased permeability. A wide range of vasoactive agents, including eicosanoids, endothelins, and nitric oxide, is released, which modulate vascular tone at a local level. The overall result is a loss of functional and structural vascular integrity (2). The purpose of this study was to determine if vascular endothelial growth factor (VEGF) plays a role in this process.

VEGF has been identified as a key molecule in the control of vascular permeability via interactions with the endothelial cell (3). In vitro studies have estimated that VEGF is 20,000 times more potent than histamine at inducing vascular permeability (4). All normal tissues express VEGF mRNA and protein. The human VEGF gene has eight exons, and alternative splicing of these has resulted in four different mRNA transcripts leading to at least four different proteins, VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉, and VEGF₂₀₆, which may have different functions; VEGF₁₆₅ and VEGF₁₂₁ are soluble products, whereas VEGF₁₈₉ and VEGF₂₀₆ remain cell associated (5, 6).

The biological activity of VEGF is dependent upon its interaction with specific receptors. Two well-defined receptors have been identified: Flt-1 and KDR (7). A soluble form of Flt-1 (sFLT) has been described that may act as a naturally occurring inhibitor of VEGF bioactivity (8). A wide variety of cells express VEGF receptors, including activated macrophages, neutrophils, vascular endothelial cells, and alveolar type II epithelial cells (9, 10, 11). Thus, VEGF is potentially capable of having an effect at both the alveolar epithelial and endothelial barrier.

Clearly, if VEGF has a role in lung injury, the triggering factors for its production would be important. VEGF production by peripheral blood monocytes (PBM) is increased in response to lipopolysaccharide (LPS) and tumor necrosis factor (TNF) treatment (12). Both of these compounds can experimentally induce lung injury and increase vascular permeability, and both TNF and LPS have been strongly implicated in the pathogenesis of ARDS (13). In addition, hypoxia increases VEGF production via the HIF-1a transcription factor (14). Increased VEGF production by the lung and peripheral tissues may, therefore, be an early physiological response to the sort of systemic insults that precipitate ARDS.

The aims of this study were to determine whether plasma levels of VEGF were elevated in ARDS, to establish whether PBMs were an important source of VEGF constitutively or in response to hypoxia in patients with ARDS, and to determine the functional importance of VEGF in the plasma of patients with ARDS.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Blood samples were obtained from patients within 48 h of admission to the Intensive Therapy Unit in Southmead Hospital, Bristol, UK. The severity of lung injury was assessed according to the American-European consensus (15). Patients with bilateral infiltrates on chest x-ray and a Pa_O2:FI_O2 ratio < 200 mm Hg were defined as having ARDS (n = 40). Patients with predisposing risk factors for ARDS, but without lung injury, were classified as at risk (n = 28). These patients did not subsequently develop ARDS. Although some at-risk patients had elevated lung injury scores with low PO₂:FI_O2 ratios, none had radiological evidence of bilateral alveolar shadowing and did not, therefore, fall into the "acute lung injury" category as defined by the consensus statement. Blood samples were also obtained from 14 nonsmoking volunteers as normal control subjects and 9 patients who had been ventilated during abdominal aneurysm repair as ventilated control (VC) subjects. The etiology and severity scores for the patients with ARDS and at-risk patients are shown in Table 1.

VEGF Measurements

Plasma VEGF was measured using a sandwich ELISA kit according to manufacturer's instructions (R&D Systems, Abingdon, UK). Briefly, a monoclonal antibody specific for VEGF was precoated onto a microplate. Standards and samples in duplicate were then pipetted into the wells. After washing, a polyclonal-labeled detection antibody specific for VEGF was added. Following a wash, a substrate solution was added to the wells and color developed in proportion to the amount of VEGF present. The intensity of the color reaction was read spectrophotometrically in a plate reader. In our hands, this assay has an intraassay precision of 5.4% and an interassay precision of 9.7%. Recovery of VEGF spiked into heparinized plasma was 93%. The lower limit of detection was 3.75 pg/ml. This assay measures biologically active VEGF₁₂₁ and VEGF₁₆₅.

Peripheral Blood Mononuclear Cell (PBM) Preparation

Fifty milliliters of blood was collected into heparinized tubes. Plasma was removed by centrifugation at 1000 × g, 4° C for 6 min, and stored at 80° C until required (less than 3 mo in all subjects). The remaining blood was diluted to twice the original volume with RPMI 1640 medium (Sigma, Poole, UK), and carefully layered onto a Ficoll-Hypaque density gradient (specific gravity 1.077, Amersham Pharmacia, Amersham, UK). After 30 min centrifugation at 400 × g the resultant interface of mononuclear cells was removed, washed, and resuspended in RPMI medium supplemented in 10% fetal calf serum (Gibco, Life Technologies, Paisley, UK). The cells were allowed to adhere for 1 h in petri dishes (Nunc, Life Technologies) at 37° C 5% CO₂. The nonadherent cells were then removed by rinsing with phosphate-buffered saline (PBS). Adherent PBMs were then removed by gentle scraping, washed, and resuspended in complete media. Cells were used if > 95% viable as assessed by trypan blue exclusion, and were > 95% PBM as determined by morphology under Diff-Quik staining.

PBMs were obtained for cell culture experiments from 21 patients with ARDS and 12 at-risk patients consecutively recruited. PBMs were cultured in normoxic conditions alone or with 10 ng/ml TNF- (R&D Systems), or in hypoxic conditions in 5% O₂.

Endothelial Cell Culture

Primary human pulmonary artery endothelial cells (HPAEC) (Clonetics; BioWhittaker Inc., Walkersville, MD) were grown to confluence (usually 3-5 d) in large vessel endothelial cell growth medium (LVEGM, Clonetics). At confluence, cells were detached with trypsin/EDTA and transferred to new dishes with a split ratio of 1:4 for further propagation. Endothelial cells were used at passages 3-8.

Establishment of Monolayers for Permeability Assay

Polycarbonate micropore filters (Costar Transwell System, 6.5 mm diameter, 3 µm pore size, Corning Costar, High Wycombe, UK) were coated with 50 µl of 50 µg/ml of collagen type IV (Sigma) ensuring that an even film covered the surface. The filter inserts were left to dry overnight in a laminar airflow cabinet. They were then sterilized by rinsing with 70% ethanol and allowed to air dry. Before being seeded with HPAEC, the filter inserts were rinsed three times with PBS. HPAEC were detached with trypsin/EDTA and washed once with fresh medium. Cells were then seeded onto the transwells at a seeding density of 0.44 × 10⁶ cells/cm². The HPAEC were incubated at 37° C for 4 d with a medium change every 24 h.

Endothelial Permeability

Endothelial cell layer permeability was assessed using a method adapted from Mackarel and coworkers (16). At the start of the experiment, the upper chamber of the transwell was incubated with either medium alone or 10 ng/ml VEGF (R&D Systems). The dose of VEGF had been previously optimized. Plasma from patients with ARDS and normal control subjects was added alone or with 20 ng/ml sFLT-Fc chimera (R&D Systems) to neutralize any VEGF present. The fluid levels in the upper and lower chamber were kept equal to avoid the effect of hydrostatic pressure across the monolayer. Zymosan-activated plasma (25%) in LVEGM was added to transwells without cells as positive controls. Negative control wells contained LVEGM alone. Experiments were performed at 37° C in a humidified atmosphere containing 5% CO₂.

After incubation for 2 h, the upper chamber fluid was aspirated and 300 µl of 1% fluorescein isothiocyanate (FITC)-labeled albumin (Sigma) in LVEGM was added. One milliliter of 1% bovine serum albumin (BSA)/LVEGM was added to the lower well.

For permeability assays, 100-µl aliquots of the lower chamber fluid were aspirated at time 0, 1, 2, 3, and 4 h. This was replaced with 100 µl of fresh LVEGM/1% BSA. The test aliquots were diluted 1:5 with PBS and the absorbance was read in a spectrophotometer at 490 nm. Absolute differences in absorption of lower chamber samples at 4 h compared with 0 h corrected for intrinsic absorption of the medium and used each transwell as its own control. Increases in FITC absorbance was fitted to a straight line relative to time 0 using least-squares linear regression for each transwell experiment (median correlation coefficient r = 0.92, p = 0.03).

Statistical Analysis

The Ryan-Joiner normality test was used to test the distribution of the data. Significant differences between raw data and normal distribution were observed (p < 0.01) for plasma data and culture data. Nonparametric Mann-Whitney tests were therefore used to compare these data. The data for monolayer permeability studies were normally distributed. Student's t test was therefore used to compare these data. Statistics were analyzed using Minitab for Windows. A p value < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma VEGF Levels Are Elevated in Patients with ARDS

VEGF levels were significantly elevated in ARDS plasma (median 243 pg/ml) compared with at risk (median 120, p = 0.007) and VC (Figure 1). Patients with ARDS with persistent systolic hypotension (BP less 100 mm Hg) (n = 23) had elevated VEGF levels (median 307 pg/ml) compared with normotensive patients (n = 17) (147.5 pg/ml, p = 0.01) (Figure 2). There were no significant associations between plasma VEGF levels and admission severity scores (AII, AIII, SAPS II, lung injury score, plasma PO₂ or FI_O2).

View larger version (15K): [in this window] [in a new window]   Figure 1.   Plasma VEGF levels. Individual and median values of plasma VEGF detected by ELISA are shown for patients with ARDS, patients at risk, and normal and ventilated control (VC) subjects.

View larger version (13K): [in this window] [in a new window]   Figure 2.   VEGF levels in patients with ARDS with and without persistent systolic hypotension. Individual and median plasma VEGF values in patients with ARDS, with and without persistent systolic hypotension (blood pressure less than 100 mm Hg).

Changes in Plasma VEGF Reflect Disease Progression

A second plasma sample was collected from patients with ARDS, where possible, 72 h after initial sample (Day 4, n = 26) for VEGF determination. In those patients who subsequently died, VEGF levels on Day 4 were higher (median 353 pg/ml) compared with surviving patients with ARDS (median 130 pg/ml, p = 0.005) (Figure 3).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Temporal changes in plasma VEGF. Individual and median plasma VEGF values on Days 1 and 4 after admission to Intensive Therapy Unit in survivors and nonsurvivors of ARDS.

When absolute changes in VEGF are considered, a decrease in VEGF from Day 1 to Day 4 of ARDS was associated with 12% mortality. If VEGF increased over this time, subsequent mortality was 78%. An increase in plasma VEGF of greater than 100% over the Day 1 value was associated with 100% mortality (n = 7).

PBM Culture

PBMs from patients with ARDS spontaneously produced 17.5 pg/ml VEGF compared with 5.2 pg/ml from at-risk PBM (p = 0.05). At-risk PBMs produced significantly more VEGF after 24 h of hypoxic treatment (Figure 4). In contrast, VEGF production by PBMs from subjects with ARDS was not significantly increased by either hypoxia. TNF- treatment had no significant effect on VEGF production in either group.

View larger version (16K): [in this window] [in a new window]   Figure 4.   VEGF production by PBM. Production of VEGF by cultured PBM under control (normoxic) conditions, hypoxic conditions (5% O2), or stimulated with TNF- (10 ng/ml) in normoxic conditions. The horizontal bar represents the median and the boxes represent interquartile ranges. Vertical lines show minimum-maximum range.

Permeability Assays

Human recombinant VEGF significantly increased the flux of albumin across the HPAEC monolayers compared with basal medium (p = 0.015) (Figure 5). Plasma from patients with ARDS induced significantly more albumin flux across the monolayers than plasma from normal control subjects (p = 0.008) (Figure 6). Neutralization of VEGF in ARDS plasma by the sFLT-Fc chimera reduced the FITC-labeled albumin flux across the monolayers by 48% and abolished the significant difference at 4 h (p = 0.78) when compared with normal control plasma.

View larger version (12K): [in this window] [in a new window]   Figure 5.   Effect of VEGF upon permeability of HPAEC monolayers. Absorbance values of fluid aspirated from the lower chamber of a transwell following addition of FITC-labeled albumin to the upper chamber.

View larger version (20K): [in this window] [in a new window]   Figure 6.   Effect of VEGF neutralization on HPAEC permeability. Changes in absorbance (mean ± SD) of samples aspirated from the lower chamber of a transwell following addition of FITC-labeled albumin, comparing normal control and ARDS plasma with or without sFLT-Fc chimera.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study has shown for the first time that VEGF levels in plasma from subjects with ARDS are elevated compared with appropriate control subjects. VEGF is known to act via specific receptors upon endothelial cells to increase permeability via alterations in both paracellular (tight junctions) and transcellular (caveolae) pathways (17, 18). In animal experiments, VEGF has been demonstrated to induce fenestrations in previously nonfenestrated capillaries (19). These actions are dependent upon intracellular third messenger production and the release of nitric oxide (20). Elevated VEGF levels in ARDS plasma would, therefore, be expected to contribute to abnormal capillary permeability, which is a characteristic feature of ARDS.

This study has demonstrated that VEGF levels at Day 4 are significantly greater in nonsurvivors than survivors of ARDS. Furthermore, changes in VEGF appear to be associated with either a good outcome if VEGF levels fall (12% subsequent mortality), or a bad outcome if VEGF levels rise (78% mortality). Increases in plasma VEGF of over 100% baseline values were associated with 100% mortality in our cohort. In contrast, there was no relationship between VEGF and mortality for the at-risk subjects. Our patients with ARDS consisted, for the most part, of patients with sepsis or infectious causes. This population has the highest mortality from ARDS, but is not representative of all patients (e.g., trauma, which has a better outcome). We considered whether plasma VEGF might be reflecting sepsis by measuring plasma TNF- in our subjects. There was no significant relationship between TNF- and VEGF (data not shown). A great deal of research has focused on identifying predictors of outcome in ARDS. Indices that have not predicted outcome include measurement of markers of acute inflammation (LTB₄, TNF-, IL-6, and IL-8) and endothelial activation (vWF-antigen, E-selectin) (21). These findings, therefore, need further confirmation before they can be considered useful as a clinical marker of poor outcome in patients with ARDS.

Elevated plasma VEGF levels in ARDS probably reflect multiple cellular sources. Normal control PBM spontaneously produce VEGF and the production is known to be cytokine responsive (12). PBM in active Crohn's disease produce more VEGF than inactive disease (24). TNF- did not increase VEGF production from PBM in our population in contrast to these studies. However, we have shown that hypoxia is a potent stimulus to PBM VEGF production. Our data suggest that elevated production of VEGF by PBM may be the source of increased plasma levels in ARDS.

Hypoxia is also likely to be an important stimulus to peripheral tissue production of VEGF. Hypoxia stimulates VEGF production from skeletal and cardiac muscle cells, vascular endothelial cells, cerebral glial cells, and dermal fibroblasts (25). Serum VEGF levels are elevated in human diseases where vascular damage and tissue hypoxia are features (e.g., eclampsia, myocardial and cerebral infarction) supporting a role for increased tissue production in hypoxic human tissues (29). The elevated VEGF levels in hypotensive patients with ARDS would also support a role for tissue hypoxia in determining plasma levels.

Previous studies in ARDS have demonstrated that many cytokines capable of increasing vascular permeability are elevated in patients with ARDS (2). No single mediator has been found to be consistently elevated. In addition, little information about the true functional importance of these mediators in humans is available. Indeed there is no published information as to the effect of VEGF upon human pulmonary endothelial cells. We have shown for the first time that VEGF increases permeability in HPAEC monolayers.

VEGF in vitro significantly increased the permeability of HPAEC monolayers to VEGF. Plasma from patients with ARDS significantly increased endothelial cell monolayer permeability compared with normal control plasma, confirming that circulating factors do play a role in the abnormal capillary permeability found in ARDS. Neutralization of VEGF bioactivity within this plasma reduced permeability by 48% and abolished the statistical difference from similarly treated normal control plasma. This in vitro assay suggests, therefore, that VEGF is of significant functional importance as a determinant of pulmonary endothelial cell permeability in ARDS.

The measured levels of VEGF within ARDS plasma (0.2-1 ng/ml) were much lower than the dose used to prove that HPAEC monolayers respond to VEGF (10 ng/ml). It is important to consider that measured plasma levels of VEGF may not truly reflect local levels at the endothelium due to VEGF₁₆₅ binding to heparin associated with the endothelium (6). In addition, many cytokines present within ARDS plasma are known to increase VEGF secretion or potentiate its actions (32). It is likely, therefore, that cytokine networks augment the biological effect of VEGF present within ARDS plasma. It is also possible that other short-lived vasoactive factors may contribute to vascular permeability in vivo, leading to an overestimate of the role of VEGF from these monolayer studies.

In conclusion, VEGF is elevated in the plasma of patients with ARDS, particularly after 4 d in subsequent nonsurvivors. VEGF is able to influence the function of isolated pulmonary endothelial cells by increasing permeability to albumin. Circulating mediators in plasma derived from patients with ARDS significantly increase the permeability of pulmonary endothelial cells, and VEGF appears to be an important determinant of this. Anti-VEGF therapies are currently under investigation as cancer therapy, but further investigation is required before such therapy could be considered in ARDS (36).