INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Group B streptococci (GBS) are a common cause of systemic infections in the neonatal period. One to four per thousand newborn infants suffer from early-onset GBS septicemia and although nowadays only few term newborn infants die from the disease, mortality is disturbingly high in infected premature babies (1). Most premature infants with systemic or pulmonary GBS infections have respiratory symptoms (1, 2), and data from animal experiments and limited clinical experience seem to indicate that in these patients respiratory failure is in part caused by surfactant deficiency. We have previously identified a GBS strain pathogenic to rabbits and developed an animal model to study the effect of surfactant in ventilated newborn rabbits after intratracheal infection with GBS (3). In preterm newborn rabbit fetuses with experimentally induced GBS pneumonia, surfactant replacement therapy improved lung function and mitigated intrapulmonary bacterial proliferation (4). There is also evidence from mostly uncontrolled clinical trials that surfactant treatment can improve gas exchange in babies with congenital GBS pneumonia (5).

Infants of mothers who lack type-specific antibodies to the GBS polysaccharide capsule are especially prone to severe infections (8). Thus active and passive immunization against GBS has been proposed (9, 10). The present study was designed to test the hypothesis that combined intratracheal treatment with surfactant and a specific antibacterial immunoglobulin (IgG) could improve lung function and mitigate bacterial proliferation in experimental neonatal GBS pneumonia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bacteria

An abundantly encapsulated low-density (LD) phase variant of GBS (gift from Stellan Håkansson, University of Umeå, Umeå, Sweden) was processed from the reference strain 090 Ia Colindale by repeated gradient centrifugation. It was stored in aliquots at 70° C, precultured, washed, centrifuged, and suspended in normal saline at a concentration of approximately 10⁹ live bacteria per milliliter as determined by spectrophotometric measurement of the optical density at 595 nm. Additionally, the actual number of colony-forming units (cfu) in the suspension was determined for each individual experiment by colony counting (3).

Antibody

A polyclonal antibody to GBS 090 Ia LD was generated in adult rabbits by repeated intravenous injection of heat-killed bacteria as previously described (11). The IgG fraction was purified by affinity chromatography (Protein A, Sephadex 6 MB; Kabi Pharmacia, Stockholm, Sweden) and ultrafiltration (Amicon Centriplus 100; Amicon, Witten, Germany). The final product contained > 99% IgG as judged by cellulose/acetate foil serum electrophoresis (LMB, Bad Godesberg, Germany). The capacity of this antibody to stimulate the release of reactive oxygen metabolites during phagocytosis of GBS by polymorphonuclear neutrophils (PMN) was tested using the nitroblue tetrazolium (NBT) test (12). The antibody was dissolved in normal saline at a concentration of 5 mg/ml and stored in aliquots at 70° C until use.

Surfactant

Curosurf (Chiesi Farmaceutici, Parma, Italy) is a modified natural surfactant isolated from minced pig lungs that has been characterized (13) and tested in various clinical trials before. Curosurf and the specific antibody were mixed in proportion 1:1 and incubated for 30 min at 37° C before use. The final concentrations of surfactant and IgG in the suspension given to the animals were 40 mg/ml and 2.5 mg/ml, respectively.

In Vitro Experiments

Influence of IgG on surface tension. The influence of rabbit IgG (Sigma Chemicals, St. Louis, MO) on the surface activity of Curosurf, suspended in normal saline at a concentration of 2 mg/ml or 5 mg/ml, was tested in a Pulsating Bubble Surfactometer (Electronetics Corporation, Buffalo, NY). IgG was added at concentrations ranging from 1 to 40 mg/ml and the sample was incubated at 37° C for 30 min before analysis. Measurements were made at 37° C during 50% cyclic area compression at a rate of 40 min¹. Surface tension at minimum bubble size (_min) was recorded after 5 min of pulsation.

Influence of surfactant and antibody on release of reactive oxygen metabolites by polymorphonuclear leukocytes. Phagocytes can kill bacteria by release of reactive oxygen metabolites. PMN were isolated and purified from healthy adult donors as described previously (12). Both the encapsulated (GBS LD) and the nonencapsulated phase variant (GBS HD) of the GBS strain were used for these studies. The bacteria (3 × 10⁸) were heat-killed and incubated with 5 × 10⁵ PMN, specific IgG antibody (8.5 mg/ml), and/or surfactant (4 mg/ml) on ELISA plates (12). Commercially available nonspecific rabbit IgG and normal saline were used as controls. Oxygen metabolite release was measured spectrophotometrically by means of the NBT reduction test (12).

Animal Experiments

Animals. Eleven pregnant New Zealand White rabbits were obtained from local suppliers. Rabbit fetuses were delivered by cesarean section (3 to 11 animals per litter) at a gestational age of 29.5 d as previously described (3). Term gestation for rabbits is 30 to 31 d. Previous studies on lung function demonstrated that dynamic compliance at 29.5 d is not significantly different from that of term animals (14). During the operation the doe was anesthetized and supplemental oxygen was delivered via a face mask.

Experimental protocol. The rabbit fetuses were anesthetized, paralyzed, and tracheotomized at birth and allocated in a random order to five different treatment groups. Using a specially designed tracheotomy tube with an indwelling wedged plastic catheter tubing (inner diameter, 0.75 mm; Portex, Hythe, Kent, UK), 5 ml/kg body weight (bw) of the antibody preparation (corresponding to 12.5 mg/kg bw of IgG) with or without surfactant (200 mg/kg bw) were instilled immediately into the liquid-filled lungs before the onset of ventilation. Controls received the same volume of the nonspecific rabbit IgG preparation, surfactant, or normal sterile saline. At 30 min all experimental groups received an intratracheal bolus injection of 5 ml/kg of the GBS suspension. Before reconnecting the animals to the ventilator system, the instilled liquid was moved from the central airways to peripheral airspaces by injecting three times 10 ml/kg bw of air with a microsyringe. The exact number of cfu given to each individual rabbit was calculated from the injected volume and the determination of cfu/ml in the infectious inoculum. The five treatment groups that originated from this procedure will be referred to in the following as: (1) specific IgG, (2) nonspecific IgG, (3) saline, (4) surfactant, and (5) surfactant + specific IgG.

All animals were then transferred to a warmed multiplethysmograph system (14) and ventilated in parallel in sealed Plexiglas chambers with a common ventilator as previously described (3, 4, 14). Peak inspiratory pressure was individually adjusted for each animal to obtain a tidal volume of 6 to 10 ml/kg bw. Stabilization of tidal volumes was obtained within 15 min after birth. Lung-thorax compliance (ml × kg¹ × cm H₂O¹) was calculated from the quotient of tidal volume and peak inspiratory pressure. The animals were ventilated for 5 h. Recordings were obtained at 0, 15, 30, 45, 60, 90, 120, 150, 180, 210, 240, 270, and 300 min. Electrocardiogram (ECG) was recorded at the same intervals and animals were counted as survivors if the heart rate was > 100 beats per min without evidence of arrhythmia or atrioventricular block. At the end of the experiments the rabbits were killed and the chest was opened. Blood from the right cardiac ventricle was aspirated for blood culture (Bactec Plus blood culture system; Becton Dickinson, Sparks, MD). A heparinized sample was taken for blood gas analysis.

The left main bronchus was tied and the left lung was excised, divided in two parts of approximately equal size by a frontal section, and weighed. The peripheral part was placed immediately into the sterilized tube of a tissue homogenizer (Kontes Scientific Glassware Instruments, Vineland, NJ) and stored on ice until further processing.

Bronchoalveolar lavage. The right lung remained in situ and was lavaged via the tracheal cannula. 0.9% saline with heparin (2 U/ml) at room temperature was instilled into the lungs via the tracheal cannula at a volume of 20 ml/kg bw. This volume was washed in and out three times, and the procedure was repeated four times with an average recovery of 83 ± 9%. The lavage fluid from each animal was pooled and centrifuged at 150 × g for 10 min. The resulting pellet was collected and saline was added to obtain a final volume of 1 ml of cell suspension. One hundred microliters of this suspension was stained with Giemsa-Romanowski stain and the cells were counted in a Bürker chamber and a flow cytometer. Cytospin preparations from the rest of the suspension were stained with May-Grünwald-Giemsa stain and used for differential counting of inflammatory cells. The supernatant was stored at 70° C and its protein content was analyzed (protein assay; Bio-Rad Laboratories, Munich, Germany).

Bacterial counting. The weight of the lung specimens was adjusted to 1 g with sterile normal saline. The samples were homogenized with a high-speed (15,000 rpm) nylon microchamber tissue homogenizer (Sorval Omnimix; Dupont Instruments, Newton, CT). A serial dilution was performed and the diluted suspensions were spread on blood agar plates for colony counting after 24 h incubation. As bacterial proliferation follows a logarithmic growth curve, the results were expressed as mean log₁₀ cfu/g lung (wet weight). The calculated number of cfu given to the rabbit (cfu/rabbit) represents a good estimate of the number of cfu in the lung of the rabbit (cfu/g lung) at the beginning of the experiments, as the bacteria were administered intratracheally and the total lung weight of a term rabbit is approximately 1 g (3). In a previous study we observed a very close correlation between the number of GBS instilled into the airways and the number of live bacteria detected in the left lung of animals killed 1 min after the injection (3). A positive difference between cfu/g lung after 5 h of ventilation and cfu/rabbit at the beginning of the experiment thus indicates bacterial growth, a negative difference indicates a decrease in the number of viable bacteria in lung homogenate.

Histologic examination. The central part of the left lung was fixed in 4% formalin and subsequently embedded in paraffin. Longitudinal sections were cut in a standardized fashion. The slides were coded, so that the investigator was unaware of the experimental condition of the animals. The specimens were examined by light microscopy with special reference to intra-alveolar edema, hyaline membranes, and airway epithelial necrosis. The influx of inflammatory cells was estimated using a semiquantitative grading system as previously described (3). Severe pneumonia was defined as an inflammatory reaction involving more than 30% of total lung parenchyma.

The study design and the management of the animals complied with national legislation. The trial protocol was approved by the local ethics committee for animal research.

Statistical Analysis

Data are given as mean ± SD or median and range. Values for lung weight and physiological data were subjected to analysis of variance (ANOVA) using the CRISP software program (Crunch Software, San Francisco, CA). Between-group differences were evaluated by Student-Newman-Keuls' test. Differences in the incidence of complications between the groups were analyzed with the ² test. The limit level of statistical significance was defined as p = 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Vitro Experiments

Surface tension measurements. Addition of IgG at concentrations  1 mg/ml did not influence the low minimum surface tension (_min) of Curosurf diluted to 2 mg/ml (Figure 1). Only at concentrations  2 mg/ml did IgG cause an elevation of _min, indicating surfactant inactivation. When Curosurf was tested at a concentration of 5 mg/ml, an IgG concentration of  40 mg/ml was necessary to inhibit surface activity. These data indicate that the presence of IgG in surfactant at a weight ratio of 1:20 as used for our animal experiments probably does not interfere with the surface properties of Curosurf.

View larger version (10K): [in this window] [in a new window]   Figure 1.   Influence of various concentrations of rabbit IgG on minimum surface tension (min) of porcine surfactant (Curosurf ) suspended in normal saline at a concentration of 2 or 5 mg/ml. Values are mean and SD from five measurements in the Pulsating Bubble Surfactometer. ***p < 0.001 versus 5 mg/ml.

Oxygen metabolite release by PMN. The nonencapsulated (HD) variant of GBS 090 Ia strongly stimulated the release of reactive oxygen species from PMN (Figure 2). The encapsulated strain (GBS LD) had no such effect in itself; only after incubation with the specific IgG antibody was stimulation observed. The increased NBT reduction was not suppressed by adding surfactant to the incubation medium (Figure 2).

View larger version (33K): [in this window] [in a new window]   Figure 2.   Oxygen metabolite release from PMN stimulated by encapsulated (GBS LD) or nonencapsulated group B streptococci (GBS HD) as measured spectrophotometrically (OD540 = optical density at 540 nm) by NBT reduction test. GBS LD need opsonization by specific antibodies for PMN stimulation. Surfactant does not suppress the antibody-mediated response. The bars represent mean and SD. ***p < 0.001 versus unstimulated PMN.

Animal Experiments

Characterization of the experimental groups. Of 79 delivered fetuses, seven demonstrated ECG abnormalities already at the beginning of the experiments and were therefore excluded from further evaluation. Twenty-five other animals with normal initial ECG developed bradycardia < 100 beats/min or ECG abnormalities before the end of the 5-h experimental period and were also excluded from the final data analysis. No significant differences in survival rate (range, 56 to 67%) were found between the groups. Pneumothorax was found in one animal. Mean birth weight, final heart rate, tidal volume, and PCO₂ of the remaining 47 rabbits are shown in Table 1. Final PCO₂ in heart blood was significantly lower in both groups receiving surfactant, reflecting improved ventilation.

Bacterial proliferation. A similar dose of bacteria was given to all groups (Figure 3). Over the study period there was prominent bacterial proliferation (all figures are log₁₀ cfu differences between values representing the beginning and the end of the experiment) in both the saline (+1.29) and the nonspecific IgG controls (+0.92). Slightly less bacterial growth was observed in the animals receiving the specific anti-GBS-IgG (+0.76, not significant). Significantly less bacteria (p < 0.01) were detected in the lungs of infected rabbits treated with surfactant than in those receiving saline at the same timepoint (Figure 3), but bacterial proliferation also occurred in this group (+0.55). In contrast, animals receiving the mixture of surfactant and the specific IgG showed no increase in bacterial number over the 5-h study period (0.05, p < 0.05 versus all other groups). Blood cultures were positive for GBS in all animals except four receiving the combined treatment with surfactant and specific IgG (Table 1).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Bacterial proliferation in left lung homogenate expressed as log10 cfu per gram lung tissue obtained from different groups of animals at the end of the experiments (black bars).There is significant reduction in bacterial growth in the surfactant group (**p < 0.01 versus saline group) and in the surfactant + specific IgG group (***p < 0.001 versus saline group). A similar number of cfu was given to all animals at the beginning of the experiment (white bars). Values are mean and SD.

Lung-thorax compliance. At 15 min no significant differences in dynamic lung-thorax compliance were found between the groups. Compliance increased between 15 and 30 min in most animals, probably reflecting resorption of fetal lung liquid. Animals treated with saline or specific or nonspecific IgG showed a subsequent gradual decrease in dynamic compliance over the observation period. Animals receiving surfactant had significantly higher compliance values and maintained improved lung function throughout the experiments. Rabbits treated with specific IgG together with surfactant had similar compliance values as those receiving only surfactant. At 300 min compliance in each of the two surfactant-treated groups was significantly higher than in the three other groups (Figure 4).

View larger version (15K): [in this window] [in a new window]   Figure 4.   Lung-thorax compliance in different experimental groups during the course of mechanical ventilation. At 300 min compliance values (mean, SD) in the surfactant and the surfactant + specific IgG groups are significantly higher than those in the other three groups (*p < 0.05).

Bronchoalveolar lavage fluid (BALF). No significant differences in the total number of inflammatory cells were noted between the groups (Table 2). PMN were the predominant cell type in the majority of animals. About 10% of the inflammatory cells were represented by alveolar macrophages with only few lymphocytes being present. Total protein content was similar in lavage fluid in all groups (Table 2).

Histological findings. All animals except five showed evidence of pneumonia with accumulation of inflammatory cells (mostly PMN) in alveoli and peripheral conducting airways (Table 2). In 29 rabbits the inflammatory reaction was classified as severe, i.e., involving more than 30% of the lung parenchyma, but there were no differences in the degree of inflammation between the groups. In 17 of the 47 rabbits there was prominent airway epithelial necrosis, again without differences between the groups. Hyaline membranes, absent or discrete in all but one of the 19 surfactant-treated rabbits, were prominent in 17 of 28 rabbits in the three groups that did not receive surfactant (p < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Encapsulated GBS need opsonization by specific antibodies in order to be phagocytized by polymorphonuclear neutrophilic granulocytes (PMN). Systemic disease in neonates has been associated with lack of these specific antibodies that are normally transferred in utero from the mother (8). Animal studies using monoclonal antibodies have demonstrated the benefit of specific, parenterally applied antibodies in the treatment of GBS infections (15). Clinical studies are far more difficult to evaluate as the commercially available IgG preparations often lack sufficient levels of type-specific opsonizing antibodies (16). Prevention strategies such as selective chemoprophylaxis or active immunization of pregnant women are currently being evaluated (9, 10). Our in vitro experiments confirm that nonencapsulated GBS are effectively taken up by PMN even in the absence of specific antibodies, whereas an encapsulated but otherwise identical strain needs opsonization by specific IgG. Surfactant does not interfere with the release of oxygen metabolites from PMN, stimulated by GBS in the presence of specific IgG.

On the other hand, immunoglobulins, like other proteins, may inhibit surfactant function. However, the inhibitory capacity of immunoglobulins is far below that of fibrinogen or albumin (17). The hyperimmune serum prepared for the present study can be used in extremely low concentrations and did not influence dynamic surface tension in our pulsating bubble measurements at IgG:surfactant weight ratios much larger than those used in our animal experiments. This is in keeping with our observation that admixture of IgG to surfactant did not inhibit its therapeutic effect on compliance in GBS-infected newborn rabbits. Also it is reassuring that data from a recent small clinical study, evaluating the use of aerosolized IgG in infants with respiratory syncytial virus (RSV) infections, did not demonstrate serious side effects on lung function (18). The possibility to administer specific antibodies as an aerosol with or without surfactant could have obvious clinical implications, as it might allow treatment of spontaneously breathing patients at an early stage of pulmonary infection, not yet requiring endotracheal intubation.

GBS often invade the neonatal lung via colonized amniotic fluid. Intra-alveolar accumulation of leaking plasma proteins, bacterial degradation products, and substances released from neutrophils (e.g., oxygen radicals, proteases including elastase) can result in a secondary surfactant dysfunction comparable to the acute respiratory distress syndrome (ARDS) in older patients (19). Data from clinical pilot studies suggest that surfactant can be used successfully in newborns with congenital pneumonia (5) but that the response to surfactant treatment is slower than in babies with uncomplicated respiratory distress syndrome. Although initial data from animal experiments published by Sherman and coworkers (20) indicated impaired macrophage function after surfactant treatment, more recent data from the same group demonstrated no stimulatory effects of commercially available modified natural surfactant preparations on the proliferation of GBS neither in vitro nor in newborn rabbits (21). In previous experiments using the same animal model we have shown that surfactant treatment actually mitigates streptococcal growth (3, 4). The mechanisms behind this effect are not clearly understood. Besides a direct bacteriostatic effect of surfactant, prevention of atelectasis and reduced accumulation of proteins in the alveoli might be important factors. Keeping the airways open throughout the respiratory cycle is probably crucial for effective mucociliary clearance. In the present study, histological examination of the lungs revealed no differences in the degree of cellular inflammatory reaction between the groups, although hyaline membranes were less prominent in animals receiving surfactant. For technical reasons (cutting of the left lung) no perfusion fixation under standardized distending pressures was performed. We therefore made no attempts to evaluate the alveolar expansion pattern in various treatment groups. Noninfected animals receiving the same volume of saline were not studied in the present series, but we know from previous experience that such control animals do not show any significant recruitment of granulocytes to the airspaces during a 5-h period of mechanical ventilation (3).

Besides its well-characterized biophysical properties, surfactant is probably of major importance in both the antibacterial and the antiviral defense system of the lung (22) and beneficial effects of exogenous surfactant on lung function have been reported in animal models of bacterial (4, 23), protozoal (24), or viral (25, 26) pneumonia. To some extent, these effects have been linked to the hydrophilic surfactant proteins SP-A and SP-D, which are both important stimulants of macrophage function. An increased susceptibility to GBS was recently described in genetically SP-A-deficient mice exposed to a relatively low dose of bacteria (27). However, these animals did not develop severe pneumonia and continued spontaneous breathing. In contrast, blocking of endogenous SP-A with a monoclonal antibody did not enhance bacterial proliferation in an experimental model of GBS pneumonia, similar to that employed in the present study (28). Further work is therefore required to define the precise role of the hydrophilic surfactant proteins (SP-A and SP-D) in the antibacterial defense system of the neonatal lung.

Intravenous administration of antibiotics is standard treatment for neonates with respiratory distress until bacterial infection can be ruled out by means of negative cultures. However, with conventional treatment drug levels might be low in the alveolar spaces and bacterial resistance against antibiotics is a widespread clinical problem, especially in adults. Bacteria opsonized by specific IgG antibodies are phagocytized and consequently killed intracellularly. This is a potential advantage as bacterial toxins or cell wall components of extracellularly killed microorganisms can cause severe adverse systemic reactions such as endotoxin shock or pulmonary hypertension. In patients affected by severe pneumonia, topical immunoglobulins might therefore serve as an adjunct to standard antimicrobial therapy.

It seems likely that simultaneous instillation of surfactant facilitates spreading of drugs in the peripheral airspaces, as previously shown for radiolabeled sulfur colloids administered via the airways together with surfactant (29). Other investigators documented the effectiveness of surfactant as a vehicle for a variety of pharmacological substances including antibiotics (30), lysozyme (31), antioxidants (32), and corticosteroids (33). A recent study describing the use of surfactant as a carrier for an adenovirus vector (34) introduces the possible use of surfactant in gene therapy.

We conclude that combined treatment with surfactant and a specific antibacterial antibody improves lung function and mitigates bacterial growth in experimental GBS pneumonia in ventilated newborn rabbits. The possible role of surfactant as a "vehicle" for various drugs administered via the airways deserves further evaluation.