INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Respiratory syncytial virus (RSV)-induced bronchiolitis accounts for a significant proportion of hospital admissions during the winter months. Approximately 8% of RSV-positive infants admitted are subsequently transferred for intensive care, and three-quarters of these require mechanical ventilation for an average of 5 d (1). For those with severe disease the mean duration of ventilation has been quoted as 21 d (2), and infants with a history of prematurity or underlying disease are especially at risk. Therapies aimed at reducing disease severity and hence length of ventilation would have considerable benefit, both in terms of cost savings and reduction in patient morbidity. To date, however, neither ribavirin (3) nor corticosteroids (4) have shown effect, and RSV-specific immune globulin has shown benefit only when used prophylactically (5).

Administration of exogenous surfactant is one potential additional therapy for infants with severe RSV infection. The ability of surfactant to improve lung compliance in neonates with respiratory distress syndrome is well documented (6). Surfactant deficiency has recently been demonstrated in infants with bronchiolitis (7, 8), characterized by increased minimal surface tension activity, a reduced concentration of the major surface-active phospholipid, dipalmitoylphosphatidylcholine (PC16:0/16:0)* (8), and reportedly undetectable phosphatidylglycerol (PG) during the acute phase of the disease (9). Similarly, RSV infection of mice resulted in a dose-dependent inhibition of lung surfactant function and increased concentration of inhibitory proteins in bronchoalveolar lavage fluid (BALF) (10). As theoretical (11) and animal (12) models suggest that surfactant acts to prevent small airway closure, such surfactant dysfunction could exacerbate the increased mucus secretion, desquamation of damaged ciliary cells, and submucosal edema (13) characteristic of the underlying bronchiolar obstruction in RSV bronchiolitis.

The primary purpose of this randomized controlled pilot study was to assess the short-term effects of exogenous surfactant administration to ventilated infants infected with RSV. Our hypothesis was that surfactant would improve lung compliance and resistance, which would be reflected by improvements in indices of gas exchange. In addition, we also used electrospray ionization mass spectrometry to examine the phospholipid molecular species composition in BALF from these infants, to assess the effects of RSV infection and surfactant administration on surfactant composition. Previous work has shown that the ratio of disaturated PC16:0/16:0 to the monounsaturated species palmitoyloleoylphosphatidylcholine (PC16:0/18:1) is predictive of neonatal respiratory distress syndrome and may be an index of surfactant maturity in newborn infants (14). With this in mind, we wished to examine if this ratio (PC16:0/16:0:PC16:0/18:1) was altered in infants with RSV infection, and whether it correlated with parameters of lung function. Demonstration of an inverse relationship between surfactant status and airway resistance would support the concept that normally functioning surfactant is necessary to prevent small airway closure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

Nineteen ventilated infants with RSV-induced respiratory failure were enrolled between December 1996 and March 1998. RSV was confirmed by fluorescent antibody staining on nasopharyngeal aspirate. Exclusion criteria included those with neuromuscular disease, uncorrected congenital heart disease, preexisting pneumothorax, ventilation for more than 24 h, and those with an oxygenation index (OI) below 5, and a ventilation index (VI) below 20 at enrollment (see later in METHODS for definitions). These last criteria were to exclude those infants with mild disease who were unlikely to remain ventilated for more than 48 h. Presence of chronic lung disease or history of prematurity were not exclusion criteria. The primary reasons for intubation were evenly distributed between the groups, and included apnea (surfactant 3, placebo 3), increasing oxygen requirements (usually with the clinical impression that the infant was tiring) (surfactant 4, placebo 5), worsening acidosis (surfactant 2, placebo 2). Some patients had more than one indication.

Infants were randomized to receive either a bovine surfactant (Survanta; Abbot Laboratories, Maidenhead, UK) (n = 9), or air placebo (n = 10). Two doses of surfactant (100 mg/kg = 4 mL/kg) were given, the first at enrollment, and the second 24 h later in the manner suggested by the manufacturers. This involved instillation of 1 ml/kg in each of four positions: head-up left lateral, head-down left lateral, head-up right lateral, and head-down right lateral. Surfactant was instilled using a syringe via a 6-French feeding tube (cut to the length of the endotracheal tube) that was fed through a small hole in the "micromount" endotracheal tube connector (Intersurgical Ltd, Wokingham, UK). This allowed manual ventilation to continue during instillation, after which the connector was replaced so that refluxed surfactant was not visible in the circuit. Patient management was conducted wholly by the clinical and nursing staff who were blinded to the treatment given by the investigators.

All infants were ventilated in time-cycled, pressure-controlled mode, employing a strategy of permissive hypercapnia, tolerating pH down to 7.25, and oxygen saturations of  88%. Positive end expiratory pressure was initially set at 5 cm H₂O. Peak pressures were limited to 30 cm H₂O. If air trapping was the primary problem, a low rate, prolonged expiratory time was employed; if oxygenation was the primary problem fractional inspired oxygen was increased to 0.6, and positive end-expiratory pressure (PEEP) to 7-8 cm before increasing peak pressures.

All infants were sedated with morphine (20-40 µg/kg/h) and chloral hydrate (30-50 mg/kg every four hours) and paralyzed with vecuronium (50-100 µg/kg/h) from admission until 6 h after the second surfactant dose; neuromuscular blockade was discontinued as soon as possible thereafter.

No patients received steroids or bronchodilators during the study period. Patients were nasogastrically fed as soon as practicable (usually within 6 h from admission) and fluid restricted to 65 to 75% of normal maintenance requirements. The study was approved by the Guy's Hospital Research Ethics Committee, and written informed consent was obtained from patients' parents.

Pulmonary Function Tests and Indices of Gas Exchange

Static respiratory system compliance, resistance, and time constants were measured using the single breath occlusion technique (15) with a commercially available device (PEDS 4.1; Medical Associated Services, Hatfield, PA). Pulmonary function and indices of gas exchange (see below) were measured immediately before treatment, and at 1, 2, and 6 h posttreatment. Endotracheal suctioning was not performed for 3 h posttreatment, and was always performed 15 min before the 6-h measurement. Three successive measurements of pulmonary function were made and averaged on each occasion. Compliance was calculated by dividing the expired volume by the change in airway pressure from the end-inspiratory plateau to end expiration. The expiratory time constant was calculated from the slope of the expiratory flow- volume curve between 25% and 75% of expired volume after least-squares regression analysis. Resistance was calculated as the time constant divided by compliance. Measurements were rejected if an endotracheal air leak of > 5% was present, or if the coefficient of determination (r²) of the line of best fit for the time constant was less than 0.9. Indices of gas exchange were calculated using standard formulas:

oxygenation index (OI) = mean airway pressure (cm H₂O) × FI_O2 × 100/arterial PO₂ (mm Hg)

ventilation index (VI) = respiratory rate × peak-inspiratory pressure (cm H₂O) × arterial PCO₂ (mm Hg)/1,000

alveolar arterial gradient (Aa) = (716 × FI_O2)  [arterial PCO₂ (mm Hg)/0.8]  arterial PO₂ (mm Hg)

Bronchoalveolar Lavage and Measurement of Phospholipid

BALF was obtained on Day 1 prior to treatment and 48 h later (24 h after the second treatment dose), by inserting a 6-French feeding catheter through a small hole cut in the endotracheal tube connector and advancing this distally until wedged in a small bronchus, allowing manual ventilation to continue. Two milliliters of normal saline was then introduced via a 10-ml syringe and aspirated in a pumping motion. The catheter was withdrawn and flushed with a further 0.5 ml of saline. Samples were immediately centrifuged at 400 × g for 2° C, and BALF supernatants stored at 80° C.

Total phospholipid was measured as inorganic phosphate after acid digestions (16) on a 50-ml aliquot from each BALF supernatant sample following total lipid extraction with chloroform and methanol (17). Aliquots of BALF containing 25 nmol phosphate (10-340 ml) were taken from the remaining sample, total lipid was again extracted with chloroform and methanol, and phosphatidylcholine and phosphatidylglycerol species present were detected by electrospray ionization mass spectrometry (18). Dimyristylphosphatidylcholine (5 nmol) and dimyristylphosphatidylglycerol (1 nmol), phospholipids not found in surfactant naturally, were added to the sample prior to extraction as internal recovery standards. Lipid extracts were resuspended in 20 ml chloroform:methanol (1:2 vol/vol) containing 5 mM NaOH and 5 ml was then introduced by rheodyne valve injection into a flow of methanol:chloroform:water (70:20:10) pumped at 20 ml/min into the electrospray interface of a Quattro II mass spectrometer (Micromass UK, Wythenshaw, UK). Phosphatidylcholine species were detected preferentially as sodiated adducts under conditions of positive ionization, and phosphatidylglycerol and phosphatidylinositol species as molecular ions under negative ionization. Spectra were processed using Masslynx software (Micromass UK) and converted to area centroid format. Concentrations of each phosphatidylcholine and phosphatidylglycerol species were calculated, after correction for ¹³C isotope effects, by reference to the appropriate internal standard. A sample of Survanta was analyzed using the same method. Results are presented either in absolute concentration terms or as fractional molar compositions within a given phospholipid class.

Statistical Analysis

Demographic data were compared using Fisher's exact and Mann- Whitney tests. Phospholipid ratios were analyzed by the paired t test, and their relationship to pulmonary mechanics by Pearson's correlation coefficient. Repeated measures, two-way ANOVA was used for pulmonary function and gas exchange comparisons, using marginal (type III) sums of squares because of nonorthogonality. Posthoc tests consisted only of comparison of initial (pretreatment) and final measurements within groups using the paired t test, thus the Bonferroni correction dictated an  of < 0.025; elsewhere the level of significance was set at  < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient groups were well matched in terms of demographic variables and disease severity at enrollment (Table 1). All infants had a clinical picture primarily of air trapping, as evidenced by high initial ventilation indices, increased airway resistance (241 cm H₂O/L/s) and hyperinflation on admission chest X-rays. In addition to hyperinflation, infants with chronic lung disease also showed chest X-ray changes consistent with this condition, and one other infant had partial right upper lobe collapse. No complications were noted after surfactant administration.

Pulmonary Function Tests and Indices of Gas Exchange

The mean (SEM) coefficients of variation for compliance and resistance measurements were 3.8% (0.6) and 7.1% (0.5%), which are comparable to previously reported results in this patient population using the same technique (4.4% and 7.3%) (19).

Administration of surfactant caused a transient worsening of OI, VI, and alveolar-arterial (Aa) gradients after both doses, but these all resolved within 2 h (data not shown). The mean OI, VI, and Aa gradient changed from 6.3 to 9.3, 27 to 33, and 198 to 246, respectively, following the first dose, and 6.8 to 7.6, 26 to 32, and 191 to 225 after the second dose. All changes were not significant.

Two-way repeated measures ANOVA showed deterioration in compliance (ANOVA time effect p < 0.01, interaction p = 0.3) and resistance (ANOVA time effect p = 0.01, interaction p = 0.02) over the 30 h following supplementation in the study group as a whole. These changes were seen only among the placebo group, whose compliance fell (0.36 to 0.27 ml/cm H₂O/kg, p < 0.01) and resistance rose (229 to 391 cm H₂O/L/s, p < 0.001) (Figure 1).

View larger version (21K): [in this window] [in a new window]   Figure 1.   Changes in lung mechanics over time. Deterioration in both compliance and resistance was seen only in the placebo group. Data points represent mean values, error bars SEM. Arrows refer to timing of surfactant/placebo. Post hoc t tests all involve comparison of time = 30 h with the corresponding group at time = 0, with a Bonferroni adjusted  level at p < 0.025: * p = 0.4 (NS), ** p < 0.01, # p < 0.001, ## p = 0.9 (NS). See text for ANOVA values.

Interestingly, over the same period the changes in lung mechanics were not reflected in parameters of gas exchange apart from an increase in VI among the placebo group from 25 to 34 (ANOVA interaction p = 0.04). As lung mechanics were not measured beyond the initial 30-h period, post hoc analysis of gas exchange indices over a longer period was undertaken to assess whether early (and perhaps ongoing) changes in lung mechanics were reflected by subsequent alterations in gas exchange. Measurements were made in time-averaged 12-h blocks for 60 h following treatment, until the first patient extubation. Figure 2 shows these changes over time. OI showed a significant improvement over both groups (ANOVA time effect p = 0.01), with the surfactant group trending toward a more rapid drop (ANOVA interaction effect p = 0.07). VI decreased significantly among the surfactant group (ANOVA interaction effect p = 0.05) but not among both groups as a whole (ANOVA time effect p = 0.13). Aa gradient dropped over time with no differences between groups (ANOVA time effect p = 0.02, interaction effect p = 0.33). The improvement in all parameters with time among the surfactant group compared to the placebo group was highlighted by Bonferroni adjusted t tests comparing time points 0 and 60 h, which demonstrated a significant improvement for OI, VI, and Aa gradient for the surfactant infants only (Figure 2).

View larger version (14K): [in this window] [in a new window]   Figure 2.   Changes in parameters of gas exchange over time. Significant improvement in all parameters over time occurred in the surfactant group only. Data points represent mean values, error bars SEM. Time 0 refers to the average values in the period prior to first treatment. Subsequent values refer to averaged values for the 12-h blocks following first treatment. Post hoc t tests all involve comparison at time = 60 h with the corresponding group at time = 0 with a Bonferroni adjusted  level at p < 0.025: * p = 0.04(NS), ** p < 0.01, # p = 0.9 (NS), ## p < 0.001, ~ p = 0.04 (NS), p = 0.01. See text for ANOVA values.

BALF Analysis

There were no differences between surfactant and placebo groups in the molecular species compositions of PC, PG, and PI (PI data not shown) of BALF before treatment, but these compositions were very different from that of Survanta (Table 2). Survanta is supplemented with PC16:0/16:0, which it contained at 56.1 mol% of total PC compared with 34.9% for the placebo group and 32.5% for the surfactant group on entry to the study (Table 2). The composition of BALF phospholipid was significantly altered for infants in the placebo group at 48 h, with decreased amounts of PC16:0/16:0, and increased proportions of all unsaturated PC species and PI18:0/20:4. In contrast, PC, PG, and PI species compositions of BALF remained essentially unchanged 48 h after surfactant treatment. Intriguingly, the observation that the phospholipid composition of BALF 48 h after administration of exogenous surfactant was not dominated by that of Survanta suggests that the major part of the administered surfactant had been metabolized by this time point.

Relative concentrations of PC, PG, and PI did not vary between groups or after either treatment (surfactant pretreatment 81:13:06, posttreatment 81:12:07, versus placebo pretreatment 77:12:11, posttreatment 85:8:7). This finding refutes a previous study suggesting that PG is absent or undetectable during the acute phase of RSV (9).

Calculation of the concentration ratio between PC16:0/16:0 and PC16:0/18:1 provided a sensitive index of surfactant phospholipid status. This ratio decreased with time for placebo patients (mean 2.3 to 1.6, p = 0.04) compared with an increase seen for surfactant-treated infants (mean 2.1 to 2.8, p = 0.09). Additionally, the ratio correlated positively with respiratory compliance (p = 0.02) and negatively with resistance (p = 0.03) for the patient group as a whole at study entry (Figure 3).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Correlation plots for the ratio of PC16:0/16:0 to PC16:0/ 18:1 and (A) compliance and (B) resistance for all patients prior to treatment.

Correlation posttreatment was not examined, as the final lung function measurements and BALF samples were taken at different time points, t = 30 h and t = 48 h, respectively. Nevertheless, these results suggest that the composition of phospholipid in BALF was directly related to lung function in these infants.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The major clinical finding from this pilot study was that administration of exogenous surfactant appeared to prevent the progressive impairment of lung function in infants with RSV bronchiolitis. The deterioration in lung compliance and resistance of placebo infants was not observed for those in the surfactant group, and this translated into a more rapid improvement in both oxygenation and ventilation indices over the first 60 h (2.5 d) following enrollment (Figure 2). In contrast to previous studies, which reported improvements in gas exchange within 1 h of treatment (20, 21), surfactant-treated infants in the present study showed no improvement in lung mechanics or indices of gas exchange over the first 30 h following enrollment. It is possible that administration of surfactant early in the course of the disease process, before significant deterioration of endogenous surfactant has occurred, may be responsible for this discrepancy.

It is worth commenting on the type and severity of disease in our patient group. We studied only the sickest RSV infants seen over a 2-yr period. Infants with RSV-associated pulmonary disease have been subdivided into those with either an obstructive or a restrictive pattern (consistent with acute respiratory distress syndome [ARDS]) on lung function testing (2). Although our patients had air-trapping as the dominant feature, as evidenced by increased airway resistance, high ventilation indices, and hyperinflation on chest X-ray without alveolar consolidation, it must be noted that there was also a restrictive component, as evidenced by the low mean (± SEM) compliance (0.38 ± 0.04 ml/cm H₂O/kg) and reduced PO₂/FI_O2 ratio (182 ± 24) in the group as a whole at study entry. Several possible reasons may exist for this finding. First, it may be that clinical manifestations of RSV disease represent a spectrum, rather than two distinct entities. Second, our patients were very young (median uncorrected age 1.5 mo). This is closer to Hammer's restrictive patients (2.1 mo) who were significantly younger than the obstructive group (5.2 mo) (2). It is worth noting that Hammer's obstructive group also had a reduced compliance that was 50% of normal infants (0.57 versus 1.00 ml/cm H₂O/kg). Thus the reduced compliance may partly be a function of age. And lastly, overdistension as a consequence of air trapping may reduce lung compliance.

We acknowledge the limitations of measuring airway resistance when significant scalloping and hence multiple time constants are present on the expiratory portion of the flow-volume loop. To attempt to standardize measurements, we performed regression analysis for the time constant at a defined portion of the curve (25%-75% of expiratory volume) and excluded data where the r² was < 0.9, or if a significant air leak was present (> 5%). This produced repeatable results, but may reflect trends in airway resistance rather than the true figure.

Three findings from analysis of surfactant phospholipid in BALF from these infants deserve elaboration. First, all infants had a low initial mol% of saturated PC (16:0/16:0), which continued to decline in the placebo group. The fractional concentration of PC16:0/16:0 is maximal at 65 mol% at term delivery (14, 22), decreases to less than 40 mol% in children (18), and then increases to about 50 mol% in adult life, although the precise timing of these changes is uncertain. The initial values for PC16:0/16:0 of 34.9 mol% and 32.5 mol% in our patient groups are somewhat low (Table 2), but may reflect a combination of postnatal age, prematurity, and the initiation of the disease process. The value of 25.1 mol% for BALF PC16:0/ 16:0 in the placebo group after 48 h is extremely low, and is equivalent to that measured in fetal lung at 15 wk gestation before the initiation of surfactant synthesis (23).

The deterioration in surfactant PC composition of BALF in the placebo group at 48 h may have been due to dilution of surfactant PC with membrane material derived from inflammatory cells, which are prominent in the airways of infants with bronchiolitis (24). At this time point, fractional concentrations of PC species characteristic of surfactant (PC16:0/14:0, PC16:0/16:1, and PC16:0/16:0) had all declined, whereas those characteristic of neutrophil membranes (PC16:0a/18:1, PC16:0/ 18:2, PC 16:0/18:1, PC16:0/20:4, PC18:1/18:2, PC18:0/18:2, and PC18:0/18:1) had all increased. Additionally, the fractional concentration of PI18:0/20:4, the predominant PI species of inflammatory cell membranes, increased from 16.2 mol% to 28.1 mol% by 48 h in BALF from placebo infants only.

Other reasons may include altered specificity of surfactant PC synthesis and secretion, preferential uptake of "spent" surfactant components, and hydrolysis by phospholipases. It is not clear, however, how phospholipase activity could result in a selective decrease in PC16:0/16:0, as no phospholipase enzyme has been reported with a molecular selectivity for this PC species.

Second, supplementation with Survanta produced only a slight enrichment in the fractional concentration of PC16:0/ 16:0 in BALF, implying that the administered Survanta had been effectively cleared from the lungs of these infants by 48 h (24 h after the second dose). Animal (25) and human (28) studies suggest there is a postnatal transition between an initial pattern of low surfactant secretion and turnover in newborn infants to an adult pattern, where more active synthesis and rapid turnover enables the lungs to respond more readily to extrinsic factors. The precise timing during postnatal development of these changes in surfactant metabolism is uncertain. However, the rapid clearance of Survanta in the present study suggests that turnover of surfactant phospholipid in the lungs of these infants with RSV infection was already high, having increased substantially from the low values experienced at birth. These findings are consistent with those by Cogo and coworkers (29), who documented mean fractional synthesis rates of phosphatidylcholine of 34-50% per day in a group of critically ill infants of an age similar to those in our study.

Third, the potential physiological significance of surfactant phospholipid composition in RSV bronchiolitis is shown by the ratio of PC16:0/16:0 to PC16:0/18:1, which has been shown to be predictive of neonatal RDS (14). Here, the positive correlation of this ratio with lung compliance (Figure 3A) demonstrates the importance of surfactant phospholipid composition as well as its concentration in the dynamics of lung expansion, whereas the negative correlation with airway resistance (Figure 3B) adds to previous theoretical and animal work suggesting a role for normally functioning surfactant in the prevention of small airway closure (11, 12). Potential contributors to the latter effect include the observation from a capillary tube model that inactivated surfactant, through loss of its ability to maintain a high surface pressure, will not facilitate the removal of fluid from narrowed small conducting airways (30), and that normally functioning surfactant improves mucociliary clearance (31) and decreases transcellular protein leakage (32). It is unfortunate that lung mechanics were not measured at the time of the 48-h BAL, as this would have provided more data points either "firming up" or refuting the correlation analysis.

Despite the small sample size, these findings raise the clinical question as to whether surfactant supplementation has the potential to reduce days of ventilation and length of intensive care unit (ICU) stay. Although determination of such outcome measurements was not the aim of this small pilot study, there were trends in the surfactant-treated group toward reduction in mean duration of ventilation (126 versus 170 h, p = 0.4), ICU stay (161 versus 213 h, p = 0.3), and hospital stay (13 versus 17 d, p = 0.3). However, it must be noted that the power from this study to detect a 48-h reduction in any of the above parameters was 54%, 47%, and 32%, respectively.

Timing and dosage of surfactant are important considerations in the design of future outcome trials. We gave two large doses relatively early in the disease process, which may be important as degree of inhibition is related to dose of RSV inoculation (10) and can be overcome with increasing concentrations of surfactant (33). However, the rapid disappearance of exogenous surfactant from the BALF suggests that administration of doses more frequently may have a role in maintaining therapeutic concentrations of surfactant in these infants' lungs.

Although no major complications were noted during instillation of surfactant, it was not unusual for patients to experience transient rises in PCO₂ and require an increase in FI_O2 for 1-2 h after dosing. Despite this no patient experienced a drop in O₂ saturation below 86%. The need for transient increase in ventilator pressures and/or FI_O2 is reflected by the rise in oxygenation index (see RESULTS). Only one infant required a transient increase in peak pressure from 28 to 36 cm H₂O after the first dose. Despite this, his plateau pressures were 27, and peak pressure was weaned back to 30 by 2 h postdose, so significant barotrauma was unlikely. No other patient required peak pressures above 30 cm H₂O postdose. Again, using a surfactant preparation with a higher phospholipid concentration allowing for lower administration volumes may be a reasonable alternative.

Despite the limitations inherent in a study of such small size, we believe enough evidence is apparent to warrant a larger outcome study, examining the potential for a reduction in length of ventilation and ICU stay following surfactant supplementation in this patient group. Given the current costs involved with modern intensive care ($2,500 per day) compared with an average course of surfactant ($1,900 for Survanta) a 2-d reduction in ICU stay would prove cost effective, especially as cheaper, synthetic surfactants containing essential apoproteins become available.