INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary thromboendarterectomy (PTE) for chronic thromboembolic pulmonary hypertension may be complicated by reperfusion lung injury. In this setting, the lung injury usually develops within 48 h of operation and is characterized by hyperoxemia and radiographic infiltrates corresponding to areas that have been endarterectomized and reperfused. This phenomenon has been previously demonstrated to be neutrophil-mediated (1, 2).

Neutrophils mediate reperfusion injury in a variety of tissues. After endothelial activation, recruitment of neutrophils to sites of inflammation occurs in several steps, including rolling of cells along the activated endothelium, activation of neutrophils, firm adhesion of the neutrophils to the endothelium, and migration of inflammatory cells into the surrounding tissue. Rolling of neutrophils is thought to be mediated by the adhesion molecules E- and P-selectin (expressed on the endothelial cells) and L-selectin (expressed on the neutrophil). Prior studies in which E-, P-, or L-selectins have been "blocked" demonstrated that selectin-mediated leukocyte adhesion is essential for the process of reperfusion injury (3, 4).

CY-1503 (Cylexin; Cytel Corporation, San Diego, CA) is an analog of the carbohydrate structure sialyl-Lewis X (SLX), which is expressed on the surface glycoproteins of neutrophils, and serves as the ligand for E- and P-selectins. In previous animal models of ischemia-reperfusion, the administration of SLX analogs has prevented the development of reperfusion injury (5, 6). We hypothesized that blocking selectin-mediated adhesion of leukocytes to the endothelium with Cylexin would prevent reperfusion lung injury in patients undergoing PTE.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Population

Fifty-three patients with operable, chronic thromboembolic pulmonary hypertension (CTEPH) who had consented to undergo PTE were enrolled in this study. There were 28 men and 25 women ranging in age from 19 to 75. Patients were excluded from the study if they were pregnant, had renal function impairment defined as a serum creatinine greater than 2.5 mg/dl, had a history of an acute illness that had resolved within 28 d before the study, had a history of an active or chronic infection, received a blood transfusion (except for autologous blood) within 28 d of the study, or had participated in another investigational drug protocol within 28 d of the study. Each patient gave informed consent to a protocol approved by the Human Subjects Committee, University of California, San Diego.

Study Design

This was a double-blind, randomized, placebo-controlled, parallel study of patients undergoing PTE. Twenty-seven patients were randomized to receive Cylexin and 26 patients were randomized to receive placebo. The investigators, site personnel (with the exception of the investigational pharmacist), and Cytel personnel involved in conducting or monitoring the study were blinded to the assigned treatment group.

Cylexin was supplied by Cytel Corporation as an aqueous solution in 5-ml vials containing 100 mg/ml. The investigational pharamist prepared the Cylexin and placebo (0.9% NaCl) in identically appearing, evacuated containers. Cylexin was given as a bolus of 20 mg/kg diluted in 100 ml 0.9% NaCl and administered over 5 min into the cardiopulmonary bypass circuit 30 min before the start of the pulmonary endarterectomy. The bolus was followed immediately by a continuous infusion through a perpheral intravenous catheter. The continuous infusion dose was 1 mg/kg/h diluted in 0.9% NaCl to a total dose volume of 500 ml given over 10 h (50 ml/h). This dose was selected because it was expected to maintain blood levels of CY-1053 at 10 to 20 µg/ml, concentrations found to be effective in animal models of ischemia-reperfusion injury. The duration of therapy was based on availability of manufactured drug during the study period. Patients randomized to placebo received a 100-ml bolus followed by 500 ml 0.9% NaCl given as a continuous infusion over 10 h. The infusions were administered through the 10 h of therapy except during intraoperative periods of circulatory arrest.

Patients underwent a screening history and physical examination before surgery (Day 0), and a physical examination on the day of surgery (Day 1) and each of the first four postoperative days (Days 2-5). Vital signs were recorded during each physical examination. A 12-lead electrocardiogram was obtained on Days 0-5. Hematology (hemoglobin, hematocrit, total erythrocyte count total leukocyte count with differential, platelet count), serum chemistries (blood urea nitrogen [BUN], creatinine, glucose, calcium, uric acid, total protein, total bilirubin, alkaline phosphatase, lactate dehydrogenase [LDH[, aspartate transaminase [AST], alanine aminotransferase [ALT], and urinalysis were measured preoperatively (Day 0), the first postoperative day (Day 2), and at the conclusion of the study (Day 5). Blood for the measurement of plasma soluble E-, L-, and P-selectins was drawn at baseline (Day 0), immediately before surgery (Day 1), immediately after coming off cardiopulmonary bypass (CPBP), the evening of surgery (approximately 4 h after coming off CPBP), and 24 and 48 h after surgery (Days 2 and 3).

Bronchoscopy with bronchoalveolar lavage (BAL) was performed immediately before surgery on Day 1, approximately 4 h after surgery on Day 1, and on Day 3. During each bronchoscopy, BAL was performed in one lobe thought to be an area at risk for the development of reperfusion lung injury. No segment was lavaged more than once. Each BAL was performed using three 30-ml aliquots of normal saline. After instillation of each aliquot, fluid was aspirated by hand suction and immediately placed on ice. Returned volumes were pooled and total protein, cell count with differential, and soluble E-, L-, and P-selectins were measured.

Chest radiographs were obtained preoperatively and then daily on Days 1-5. A single radiologist evaluated each radiograph, blinded to treatment group as well as the clinical condition of each subject. Each radiograph was divided into four quadrants (right upper, right lower, left upper, left lower) and each quadrant was scored based on the intensity of infiltrates as follows: no infiltrate = 0, under 1/3 of quadrant opacified = 1; 1/3 to 2/3 of quadrant opacified = 2; > 2/3 of quadrant opacified = 3. The sum of the quadrant scores was transformed into a Chest Radiographic Assessment Score as listed in Table 1. Arterial blood gas values, ratio of arterial oxygen pressure to fraction of inspired oxygen (Pa_O2/FI_O2), and extrinsic positive end-expiratory pressure (PEEP) were measured every 6 h for the duration of mechanical ventilation. A lung injury score was calculated using the Pa_O2/FI_O2, PEEP, and chest radiograph score (modified from Murray and coworkers [7]) on Days 1-3.

A binary assessment of lung injury based on clinical impression was also recorded for each patient. For each day of the study, patients were evaluated independently by two blinded investigators for the presence of hypoxemia (Pa_O2/FI_O2 < 300) and/or opacities on chest radiograph in areas that were reperfused. If other causes for the hypoxemia or radiographic abnormalities could not be identified, the diagnosis of reperfusion lung injury was rendered. The clinical impressions were formed independently by the two investigators.

The lung injury that occurs in patients undergoing PTE is localized to anatomic areas that are endarterectomized and reperfused. Areas at risk for lung injury can be estimated by the location of vascular obstruction on preoperative perfusion scans and pulmonary angiograms. Preoperative angiograms were scored for degree of pulmonary obstruction. Postoperative lung perfusion scans were scored for the number of areas at risk for lung injury. Pulmonary segments/lobes were defined as being at risk for lung injury if removal of thrombus from the lobe or segment was achieved at surgery and if there was evidence of reperfusion of the same area on the postoperative perfusion scan.

Processing of Bronchoalveolar Lavage Fluid (BALF)

BALF was filtered through saline-soaked gauze to remove mucus and debris, and cell count and differential was obtained. The fluid was then clarified by centrifugation (1,000 × g, 10 min) and an aliquot was stored at 70° C for the total protein analysis. The remainder of the fluid was concentrated approximately tenfold in a Centriprep-30 concentrator (Amicon Corp., Beverly, MA) having a nominal molecular weight cutoff of 30,000. Concentrated BALF was stored at 70° C until analyzed.

Protein Determinations

Total protein in cell-free BALF was determined by the bicinchoninic acid (BCA) method (8) using bovine serum albumin (BSA) as standard.

Quantitation of Selectins in Plasma and BALF

Concentrations of soluble E-, P-, and L-selectins (sE-, sP-, and sL-selectins) were measured in plasma and concentrated BALF by highly specific ELISA. Purified recombinant human sE- and sP-selectin standards, as well as monoclonal (mouse) and polyclonal (rabbit) antibodies against these proteins, were provided by Cytel Corporation. For the sE-selectin assay, microplates (Immulon-2; Dynatech, Chantilly, VA) were coated overnight at 4° C with monoclonal anti-sE- selectin antibody (CY1787; 2 µg/ml in 0.1 M Na HCO₃, pH 8.5). After blocking residual binding sites with bovine serum albumin (BSA), samples and standards (0-100 pg/ml), prepared in 0.12 M NaCl, 0.02 M imidazole, 0.005 M citric acid, pH 7.3, containing 0.1% BSA (dilution buffer), were incubated in duplicate wells for 2 h at room temperature. Plates were then rinsed three times with dilution buffer and incubated for 2 h with polyclonal anti-SE-selectin antibody (1:5,000 in dilution buffer). After rinsing, immune complexes were detected by a final incubation (1 h) with anti-rabbit IgG-horseradish peroxidase (HRP) conjugate (Jackson Immunoresearch, Westgrove, PA) followed by a final rinse with dilution buffer and development (5 min) with o-phenylene diamine (OPD) as substrate. Absorbances at 490 nm were read in a Vmax microplate reader (Molecular Devices) and sample values were extrapolated from the computer-generated (four parameter fit) standard curve. The sP-selectin assay was performed in the same fashion except for the following: (1) microplates (immulon-3; Dynatech) were coated with polyclonal anti-sP-selectin antibody (1:2,000); (2) monoclonal anti-sP-selectin (CY 1747; 2 µg/ml) was used as secondary antibody; and (3) immune complexes were detected with anti-mouse IgG-HRP conjugate (Jackson Immunoresearch). The sL-selectin assay was performed using a commercially available kit (Biosource International, Camarillo, CA) according to the manufacturer's instructions. Plasma values are reported as the mean of two sample dilutions, and BALF values are corrected for the Centriprep concentration factor.

Data Analysis

All patients who received any study drug were included in the safety analysis. Patients who received at least 75% of the preassigned total drug dose were included in the efficacy analysis. One patient (placebo) was withdrawn from the study for personal reasons and was not included in the analysis. One patient (Cylexin) was withdrawn because of severe pulmonary hemorrhage that at autopsy was felt to be due to severe reperfusion lung injury. Because this patient received < 75% total drug dose, he was not included in the final analysis.

The development of reperfusion lung injury as assessed by clinical impression represented the primary binary endpoint. Differences in incidence of reperfusion lung injury between treatment groups were compared using a chi-square test. A secondary analysis, the degree of lung injury between treatment groups using the lung injury score, was compared using the Wilcoxon rank-sum test. Preoperative and postoperative clinical characteristics, ventilator, intensive care unit (ICU), and hospital days were compared between treatment and injury groups using the Kruskal-Wallis test. Fisher exact test was used to compare incidence rates of adverse events between the two treatment groups. In order to compare data from all three timepoints of plasma and BALF laboratory measurements in a combined analysis, a mixed effects analysis of variance model was performed when comparing treatment versus control and injury versus no injury groups. Comparisons within treatment or injury groups were performed using analysis of variance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The baseline characteristics, age, sex, New York Heart Association (NYHA) functional class, the percent of the pulmonary vasculature obstructed on pulmonary angiogram, and the preoperative pulmonary vascular resistance were similar between the control and treated groups. The duration of CPBP, number of circulatory arrests (2 in all groups), and duration of circulatory arrest were similar between the treated and control groups. These characteristics were also similar between the lung injury and the no lung injury groups (Table 2).

The postoperative characteristics of the treated and control groups were similar. The percent reduction in pulmonary vascular resistance (62% in each group) and percent increase in cardiac output (51% treated, 61% control, p > 0.8) were not significantly different. The mean number of lung regions reperfused (areas at risk for reperfusion lung injury) was 3 per patient in each group.

Lung Injury

There was a statistically significant difference between the proportion of treated and control group patients who developed reperfusion lung injury as assessed by both objective criteria and clinical impression. Eight (31%) of the 26 in the treated group and 15 (60%) of the 25 in the control group (p = 0.036) developed reperfusion lung injury during the trial period. The relative risk of developing reperfusion lung injury was 0.52 (95% confidence interval [CI]: 0.265 to 0.992) in the treatment group when compared with the control group. There was 100% agreement between the two investigators in the assignment of subjects to the clinical impression groups of lung injury versus no lung injury.

Using objective criteria alone, the proportion of patients in each treatment group with hypoxemia (Pa_O2/FI_O2 < 300) and an infiltrate on chest radiograph was determined. On Day 1 (the day of surgery and Cylexin infusion), 13 (52%) of the treated group versus 21 (84%) of the placebo group had hypoxemia and radiographic infiltrates. This difference was statistically significant (p = 0.015). Using the same criteria on Day 2 and Day 3 of the study, there was no statistical difference between the treated and control groups (p = 0.389 Day 2, p = 0.423 Day 3). Considering only hypoxemia (Pa_O2/FI_O2 < 300), significantly fewer subjects treated with Cylexin were hypoxemic on Day 1 compared with control subjects (16 [62%] versus 22 [88%], p = 0.030). There were no statistical differences in the incidence of hypoxemia between the groups on Days 2 and 3.

The total lung injury score and the individual components of the lung injury score are listed in Table 3. Although the differences in scores did not reach statistical significance, there was an overall trend toward a lower lung injury score in the Cylexin treated versus the control group.

The average number of days on mechanical ventilation was lower in the Cylexin treated than the control group (5.3 versus 8.0 d), but the difference did not reach statistical significance (p = 0.076). The average number of days in the ICU (7.7 versus 6.6, p = 0.317) and the duration of hospitalization (15.3 versus 17.7 d, p = 0.177) were not statistically different between the treated and control groups.

Safety Analysis

Adverse events were similar in both treatment groups. In the treatment group, adverse events consisted of cardiac arrest (1), pulmonary hemorrhage secondary to severe reperfusion lung injury (1), sternal dehiscence (2), hemorrhagic pancreatitis (1), acute renal failure (2), disseminated intravascular coagulation (1), superficial wound infection (1), possible Horner's syndrome (1), and bilateral hand edema (1). The adverse events in the control group included cardiac arrest (1), severe reperfusion edema (2), sternal dehiscence (1), upper gastrointestinal bleeding (2), laryngeal edema (1), right heart failure secondary to unrelieved pulmonary hypertension (1), and acute abdominal pain of unknown origin that spontaneously resolved (1).

There were five deaths during the study (3 treated, 2 control) among the 53 enrolled patients. Three died of severe reperfusion edema complicated by cardiac arrest (2 treated, 1 control), one died of hemorrhagic pancreatitis and acute renal failure (treated) which developed 1 mo after receiving study drug, and one died of right heart failure secondary to unrelieved pulmonary hypertension and reperfusion lung injury (control). None of the adverse events or deaths were thought to be study related.

Plasma Selectins

Plasma selectin levels are displayed in Figures 1, 2, and 3. Soluble P-, E-, and L-selectin levels dropped from baseline values in the immediate preoperative period. Soluble P-selectin increased in the postoperative period, peaking with the sample drawn approximately 4 h after the patient was removed from CPBP. Soluble E - and L-selectin levels continued to decline in the postoperative period. The changes seen in all three selectins are similar across both treatment groups and in those who did and did not develop reperfusion lung injury (p > 0.05). The total peripheral blood leukocyte counts and percent neutrophils were similar between the treated and control groups on Days 0, 2, and 5 (p > 0.05).

View larger version (11K): [in this window] [in a new window]   Figure 1.   Plasma sP-selectin in treated and control groups and injured (treated and control) and noninjured (treated and control) groups.

View larger version (10K): [in this window] [in a new window]   Figure 2.    Plasma sE-selectin in treated and control groups and injured (treated and control) and noninjured (treated and control) groups.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Plasma sL-selectin in treated and control groups and injured (treated and control) and noninjured (treated and control) groups.

BALF

There was no significant difference in any of the BALF findings between the treatment and control groups. A comparison of the BALF results in the injured and noninjured groups is summarized in Table 4. Patients who developed reperfusion lung injury (as defined by the clinical impression criteria) developed a statistically significant increase in percent neutrophils recovered in BALF compared with those who did not sustain lung injury. At 48 h, BALF protein in those with lung injury remained significantly elevated above baseline (p = 0.006), whereas protein returned toward baseline in those who did not sustain lung injury (p = 0.105). In addition, patients with lung injury developed a dramatic rise in BALF concentrations of sE- and sP-selectin compared with baseline values and with patients who did not develop lung injury. However, when sE-selectin is normalized to BALF protein, there is no significant difference over time or between injury groups. The BALF sL-selectin also rises significantly in postoperative period, but the change is similar for the injured and noninjured groups.

The BALF results from the treated-injured and control- injured subgroups are summarized in Table 5. A similar statistically significant increase in percent neutrophils, total protein, sP-selectin, and sP-selectin/protein was seen in both injury groups (treatment and control groups).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study demonstrates that the administration of Cylexin to patients undergoing pulmonary thromboendarterectomy is safe and seems effective in reducing the incidence of postoperative reperfusion lung injury. Although reperfusion lung injury was not prevented in all patients treated, the relative risk of developing reperfusion lung injury was reduced by almost one-half in the treatment group. However, despite the reduction in the incidence of reperfusion lung injury, there was no significant impact on the overall clinical outcome. The number of days of mechanical ventilation, days in the ICU and hospital, and adverse events were not significantly different between the two treatment groups.

A significantly higher proportion of the placebo group was hypoxemic on Day 1 compared with the treated group. In addition, a significantly higher proportion of those in the placebo group had both hypoxemia and an infiltrate on chest radiograph on Day 1 compared with the treated group. These differences in oxygenation and chest radiographic abnormalities were not significantly different on Day 2 or Day 3 of the study. These differences correspond to the treatment period. Cylexin was administered for 10 h on Day 1, but then drug administration was discontinued per study protocol. On Days 2 and 3, when no drug was being administered, there were no significant differences between the treated and control groups. These observations suggest that a longer treatment period might prove beneficial.

The lung injury scores were not significantly different between the treated and control groups; however, the overall trend was in favor of the treated group. The greatest difference in lung injury scores was seen on Day 1, coincident with the treatment period.

The use of a "clinical impression" in the establishment of lung injury is not without flaw. Though objective criteria (hypoxemia, opacities on chest radiograph) were constituent to the formation of this decision, the subjective opinions of the investigators were also determinants. Our previous experience in conducting clinical trials in this patient population has led to the assertion that some clinical judgment is necessary to ensure proper assignment of patients to a "lung injury" versus "no lung injury" category. While the objective modified lung injury score is helpful in quantifying lung injury and comparing degree of lung injury between individuals or groups, it is not specific. Postoperative PTE patients frequently have clinical entities other than reperfusion lung injury that cause hypoxemia and/or infiltrates on chest radiograph such as retained secretions, blood, and atelectasis. These too can result in an elevated "lung injury" score. Therefore, the use of objective criteria alone could result in the inappropriate placement of patients into the reperfusion lung injury group. To minimize bias, the clinical impression is registered by two experienced investigators who are blinded to the treatment assignment. In addition, these same two investigators form their clinical impression independently. The 100% agreement between the two investigators supports the reliability of the clinical impression. Further, the statistically higher total protein and percent neutrophils in the BALF of the injured group compared with the noninjured group supports the belief that the clinical impression is able to appropriately separate those with reperfusion lung injury from those without injury. We believe the clinical impression is the most valid indicator of reperfusion lung injury in this study population.

Our findings lend support to other studies examining selectin blockade in preventing reperfusion injury. Animal models of acute lung injury, using either blocking antibodies to E-, P-, and L-selectins or oligosaccharide analogues of SLX, have demonstrated that selectin blockade can protect against lung injury as assessed by changes in lung vascular permeability, hemorrhage, and neutrophil accumulation in the lungs. Mulligan and coworkers demonstrated that the administration of a pentasaccharide derivative of SLX reduced the amount of lung injury in rat models of cobra venom factor or IgG immune complex-induced lung injury (5). Sekamp and coworkers showed that local as well as remote lung injury was prevented by the administration of specific selectin blocking antibodies or the infusion of SLX pentasaccharide in a rat hind limb ischemia-reperfusion model (9). The administration of an E-, and P-selectin blocking antibody (CY-1747) prevented the accumulation of neutrophils in the lung and liver of rats subjected to burn injury (10). In a canine coronary artery thrombosis model, the administraiton of CY-1503 as an adjunct to recombinant tissue plasminogen activator (rTPA)- mediated thrombolysis significantly reduced the infarct size and myocardial neutrophil infiltration when compared with thrombolysis alone (11). Presumably, selectin blockade in these models as well as in our own study results in inhibition of leukocyte activation and migration, and consequently leukocyte-induced tissue injury.

One of the more intriguing findings resulting from this study on the BALF came from the comparison of those patients who developed reperfusion lung injury with those who did not suffer lung injury. Those with lung injury developed a significant increase in protein concentration and percent neutrophils compared with those without lung injury. This is consistent with other studies and implicates a high-permeability, neutrophil-mediated lung injury (1, 2). Soluble E-selectin also increased significantly in those with lung injury, but remains without significant change in the noninjured group. However, when sE-selectin was corrected for protein, there was no significant difference between the injured and noninjured groups. This may suggest passive "leak" of E-selectin across the capillary-aveolar basement membrane due to increased permeability rather than an active increase in expression of sE-selectin in the alveolar space. Soluble P-selectin found in BALF increased significantly from baseline values in both those who did and did not sustain lung injury. However, sP-selectin rose much more dramatically in those in the injury group. This significant rise in sP-selectin persisted, even when corrected for protein level. The basis for the increase in sP-selectin in this setting is unclear. P-selectin is constitutively found in alpha granules of platelets and Weibel-Palade bodies of endothelial cells. After stimulation by inflammatory and thrombogenic mediators (histamine, thrombin), P-selectin is rapidly mobilized to the cell surface. P-selectin mediates the rolling of myeloid cells on activated endolthelium and adhesion of platelets to monocytes and neutrophils by binding SLX-like structures (cell surface carbohydrates) found on myeloid cells (12, 13). A soluble form of P-selectin has been demonstrated in the normal plasma, but it is not known whether it is derived from endothelium, platelets, or both sources (14). That sP-selectin may serve an anti-inflammatory function is also conceivable. Exposure of neutrophils activated by tumor necrosis factor-alpha (TNF-) to sP-selectin inhibits CD-18-dependent adhesion to endothelium (15) and superoxide production (16).

The origin of the elevated P-selectin found in the BALF cannot be determined from these data. Increased production by platelets or vascular endothelium is quite likely in these patients who have been on CPBP and have undergone PTE. The fact that plasma sP-selectin levels are not significantly different between the lung injury and noninjured groups suggests that the elevated sP-selectin levels are not significantly different between the lung injury and noninjured groups suggests that the elevated BALF sP-selecting in the injured group may be generated at the alveolar capillary level. Sakami and coworkers measured sP-selectin concentrations in the plasma of patients with acute respiratory distress syndrome (ARDS), pneumonia, and normals. Whereas those with pneumonia had levels elevated above control subjects, those with ARDS had sP-selectin concentrations significantly greater than those with pneumonia. They suggested that diffuse pulmonary inflammation is needed rather than focal injury to elevate plasma sP-selectin levels (17). Because lung injury in our population is a confined to areas of lung that are reperfused, this may be why a significant difference in plasma sP-selectin levels in those who developed lung injury and those who did not was not seen.

In conclusion, reperfusion lung injury after PTE is a high-permeability edema accompanied by an influx of protein and neutrophils into the alveolar space. A rise in sP-selectin is also found in BALF of patients sustaining reperfusion lung injury. In this small study, the administration of Cylexin was safe, and reduced the risk of developing reperfusion lung injury in patients undergoing PTE for chronic thromboembolic disease. However, the administration of Cylexin had no impact on the overall clinical course as assessed by days of mechanical ventilation, ICU days, or length of hospital stay. Larger scale studies using a longer duration of therapy are needed to further evaluate the efficacy of Cylexin in preventing reperfusion lung injury and acute lung injury from other causes.