INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchopulmonary dysplasia (BPD) was originally described almost 30 yr ago as a progression of characteristic radiographic findings correlated with pathologic changes of acute and chronic inflammation, fibrosis, and bronchial smooth muscle hypertrophy in premature respirator-dependent infants (1, 2). Previous clinical observations and investigations using animal models suggest that BPD is the chronic phase of neonatal lung injury caused by oxidant injury and barotrauma in susceptible premature infants (3, 4). Despite improved survival of extremely low birthweight infants ( 1,000 g) since the introduction of exogenous surfactant, BPD remains a major NICU morbidity.

Neonatal lung injury is mediated by complex interactions of factors promoting inflammation (5) and fibrosis (12,13) that are relatively unopposed by inhibitory regulatory mechanisms (14, 15). We previously found significantly higher IL-6 activity (16), IL-1 concentration and IL-1 activity (15), and depressed fibrinolytic activity (14) during the first week of life in lung lavage of infants who developed BPD compared with activities in lung lavage of infants with limited RDS or controls without lung disease. Tumor necrosis factor- (TNF-) was low in lung lavage in BPD infants on the first day, but subsequently increased with peak activity on day 14 (16). In contrast, the levels of IL-1 receptor antagonist (IL-1ra), which blocks the deleterious effects of IL-1, remained relatively unchanged during the first month of life in infants who developed BPD (15). The ratio of IL-1/IL-1ra increased during the first week in the BPD infants (15). Infants who developed BPD were distinguished from infants who had self-limited RDS by a prolonged pulmonary neutrophil influx during the first week of life (17, 18). Since the differences between infants with limited RDS and those who progressed to BPD were most marked during the first few days of life, interventions to prevent or decrease the severity of BPD need to be initiated early in the first week of life before the processes leading to the development of BPD become established.

Lung inflammation in BPD shares many of the characteristics of inflammation that occurs in asthma (19, 20). Indeed, many drugs used in asthma therapy such as methylxanthines, corticosteroids, and inhaled ₂-agonists, have been used in the clinical management of established BPD (21). Each of these drugs has significant undesirable side-effects, precluding their use for prophylaxis therapy in infants at risk for BPD. We propose that disodium cromoglycate (DSCG), which is a drug with a wide margin of safety and is effective therapy for asthma in children and adults (22, 23), is an appropriate drug for study to prevent and/or ameliorate the course of BPD in at risk infants. Cromolyn prevents the release of inflammatory mediators from mast cells (22), inhibits the influx of neutrophils (24) and inhibits the assembly of an active NADPH oxidase in the neutrophil thereby preventing oxygen radical-induced tissue damage (25). Initiating DSCG therapy in the first few days of life in infants at risk for BPD may inhibit the neutrophil influx and subsequent release of inflammatory mediators in the lung in response to acute injury.

Designing appropriate studies of potential interventions for the prevention of BPD has been hampered by the previous inability to identify infants who will develop the chronic lung disorder. The classic radiographic stages first described by Northway (1) are now uncommon and the clinical diagnosis is often not confirmed until an infant reaches  1 mo of age. Recently, Sinkin and colleagues (26) developed a BPD score, which is assessed at 12 h of age and predicts a  75% probability risk for oxygen-dependency at 28 d. Use of this score will allow prophylaxis intervention studies in a select high risk group. Moreover, a smaller number of subjects will be needed to demonstrate efficacy.

We proposed to conduct a prospective, randomized, blinded, pilot study of DSCG prophylaxis in a high risk group as defined by the Sinkin BPD 12 h score to assess the changes in inflammatory cell populations and the cytokines IL-1, TNF-, and IL-8 in lung lavage in response to DSCG therapy. Because the biologic effects of cytokines are dependent on the relative concentrations of the cytokines and their inhibitors, we also measured the inhibitors, soluble TNF receptors, sTNFR1 and sTNFR2, and IL-1ra. We hypothesized that DSCG will inhibit pulmonary neutrophil influx and inflammatory cytokines.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

This study was a prospective double-blind placebo-controlled trial and was approved by the University of Maryland at Baltimore Institutional Review Board. Eligible subjects were premature infants born at University of Maryland Hospital who (1) required mechanical ventilation on day 1 for respiratory distress syndrome, (2) had a high probability ( 75%) of oxygen dependency at 28 d predicted by a BPD score at 12 h of age (8.12 [1.89 × BW]  [0.21 × GA]  [0.25 × 5 min Apgar] + [0.13 × PIP at 12 h] > 1.1) (26) and (3) survived for the first 48 h. Infants with documented infection or congenital cardiopulmonary anomalies were excluded. After informed parental consent was obtained, infants were randomized to nebulized saline placebo (2 ml) or DSCG 20 mg (2 ml) q 6 h beginning on Day 3 and continued until Day 28. Cromolyn sodium solution is colorless and indistinguishable from NS. Randomization of treatment assignment was made in the pharmacy using a computer-generated table of random numbers. Only the pharmacist was aware of treatment assignment.

Cromoyln Administration

Cromolyn or saline was delivered using a Misty Neb^TM medication nebulizer (Baxter Healthcare Corp., Valencia, CA) at a flow rate of 6 lpm. For ventilated subjects, the nebulizer was connected to the proximal airway temperature port of the ventilator circuit using a neonatal nebulizer adaptor kit (Hudson Respiratory Care Inc., Temecula, CA). For extubated subjects, drug or placebo was delivered by the nebulizer held cupped close to the infant's face. All nebulized treatments were administered by respiratory therapists and the infants assessed after each treatment for possible adverse responses.

Lung Lavage

Subjects' lungs were lavaged by instilling two 1 ml aliquots of normal saline using a standardized technique (14, 16) on Day 3 prior to first treatment of drug or placebo and on Day 7 as long as the infant was intubated. While the infant was supine with the head midline, 1.0 ml sterile saline was instilled endotracheally, three to five breaths were delivered with a Laerdal self-inflating resuscitation bag (Laerdal Medical Corp., Armonk, NY) or ventilator, the endotracheal tube was suctioned using a 6.5 or 8.0 French catheter, and the lavage was collected into a sputum trap. Suctioning was repeated with the head turned to one side and then to the other. The procedure was repeated with a second 1.0 ml saline instillation. The suction catheter was rinsed with 0.5 ml saline and the fractions pooled. The recovered volume per suction procedure did not change over time (0.52 ± 0.18 ml, mean volume ± SEM). Lavage samples were kept on ice until processed. The specimens were then centrifuged at 900 g for 10 min at 4° C. The supernatant was removed and frozen at 70° C until cytokine analysis. Cells were counted using a hemacytometer. Cytospin preparations were made, stained with Diffquick (Baxter Healthcare Corp., McGaw Park, IL), and cell differentials determined by counting a total of 100 cells.

ELISA Assays

Antigenic TNF-, IL-1, and IL-1 ra were measured by commercial enzyme-linked immune absorbent (ELISA) assay kits from Research and Diagnostic Systems (Minneapolis, MN). The ELISA sensitivities were 4.4 pg/ml, 0.3 pg/ml, and 6.5 pg/ml for TNF-, IL-1, and IL-1ra, respectively. Antigenic IL-8 was measured in lavage samples diluted 1:10 to 1:50 in PBS containing 4% bovine serum albumin using a two-antibody ELISA with paired antibodies obtained from Medgenix Diagnostics (Fleurus, Belgium). The IL-8 ELISA sensitivity was 7.8 pg/ ml. Soluble TNFR1 and sTNFR2 were measured using a two-antibody ELISA with paired antibodies obtained from Medgenix Diagnostics. The sensitivities of the sTNFR1 and sTNFR2 ELISAs were 31.25 pg/ ml and 62.5 pg/ml, respectively. The intra-assay and interassay coefficient of variations were < 10% for each ELISA. There was no observable cross reactivity with any other known cytokines for these assays.

Clinical Outcomes

BPD was defined as oxygen dependency and moderate-to-severe radiographic score on Day 28 of life. For each subject a frontal chest radiograph performed at 28 d of age was reviewed by a single observer (A.B.C.) who was blinded to the treatment regimens. A subjective score was assigned to each radiograph, using the modified Edward's roentgenographic severity scoring system (27). The images were evaluated for degree of hyperinflation, cardiomegaly, the presence of focal cystic change, and the presence of coarse linear interstitial densities. Lung disease was characterized as mild (score 0-2), moderate (score 3-6), or severe (score 7-10) (27). Other clinical variables that were recorded included (1) duration of ventilator and oxygen support; (2) ventilator settings; (3) use of other medications; and (4) other adverse outcomes such as air leak, intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC) and sepsis. All decisions about ventilator settings, other medications and duration of oxygen therapy were made by physicians other than the investigators.

Statistical Analysis

Analysis was based on intent to treat. Discrete variables were analyzed using the Fisher's exact test. Continuous variables were analyzed using the two-tailed Student t test if normally distributed, or the Mann-Whitney U test or Wilcoxon rank sum test if not normally distributed. Correlations were tested using the Spearman Rank correlation. Because there are no satisfactory reference denominators to normalize lung lavage data (28), and to compare results of this study with previous reports (6, 14), lung lavage variables are expressed as concentration per volume. Nonparametric analysis was used for lung lavage variables because these variables are not known to be normally distributed. A p value < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Thirteen infants were enrolled in each group. One infant in the DSCG group was withdrawn by parental request at 7 d of age. Because the study design was an intent to treat, the clinical data from this subject are included in the outcome analyses. Comparisons of lung lavage cytokine concentrations were only made between Day 3 (pretreatment) and Day 7 (prior to steroids in all cases) because other drugs were administered during the study period that may have affected inflammatory cells and cytokine levels (e.g., steroids) and elevated pro-inflammatory cytokine levels at one week of age have been previously demonstrated to be associated with the development of BPD (15, 16).

Lung Lavage

The total number of inflammatory cells recovered from lung lavage was similar in the two groups on Day 3 prior to treatment and remained relatively unchanged during the study period (Table 1). The percent neutrophils recovered was higher in the DSCG group on Day 3 (71%) compared with the percent neutrophils in the placebo group (45.3%) (p = 0.02). After 4 d of therapy, the percent neutrophils in lavage decreased 41% in the CS group (p < 0.05), but increased 12% in the placebo group compared with Day 3 pretreatment levels (Table 1). Macrophages were the other predominant inflammatory cells recovered in lavage.

Although the groups' prestudy (Day 3) lavage cytokine concentrations were similar, TNF- was 3.6-fold lower on Day 7 in lavage from the DSCG-treated infants (median, 19.95 pg/ ml) compared with levels in lavage from controls (median, 70.9 pg/ml p = 0.04) (Figure 1a). The soluble TNF receptors were unaffected by DSCG therapy (Figure 1b and c), but there was a trend toward lower ratio of TNF/sum (TNFR1 + TNFR2) in DSCG-treated infants (median, 17 pg/ng) compared with controls on Day 7 (median, 37 pg/ng) (p = 0.09). IL-8 concentrations were 4.9-fold lower on Day 7 in lavage from DSCG-treated infants (median, 2245 pg/ml) compared with levels in lavage from controls (median, 11,009 pg/ml) (p = 0.051) (Figure 2). IL-8 concentrations increased 1.7-fold in control lavage from Day 3 to 7 (p = 0.04), but IL-8 concentration in DSCG lavage decreased 2.45-fold during the same interval. There was a trend toward lower lavage IL-1 on Day 7 in DSCG-treated infants compared to control lavage levels (control median, 208.2 pg/ml; DSCG median, 46.75 pg/ml, p = 0.13), but IL-1ra concentrations were unaffected by DSCG treatment (Day 7 control median, 23 ng/ml; Day 7 DSCG median, 21.5 ng/ml). The IL-1/IL-1ra ratio was lower in Day 7 lavage from cromolyn-treated infants compared to the ratio in control infant lavage (control median, 0.005; cromolyn median, 0.002, p = 0.09). There were significant correlations between Day 7 lavage IL-1 concentration and Day 7 concentrations of TNF- (r = 0.945, p < 0.0001) and IL-8 (r = 0.561, p = 0.02) and Day 7 lavage TNF- and IL-8 concentrations (r = 0.575, p = 0.02). There were also a direct relationship between Day 7 TNF- concentration and Day 7 %neutrophils (r = 0.660, p = 0.004) and Day 7 IL-1 concentration and Day 7 %neutrophils (r = 0.619, p = 0.008).

View larger version (19K): [in this window] [in a new window]   View larger version (21K): [in this window] [in a new window]   View larger version (23K): [in this window] [in a new window]   View larger version (21K): [in this window] [in a new window]   Figure 1.   Comparisons of lung lavage antigen concentrations of TNF- (A), sTNFR1 (B), sTNFR2 (C ) and the ratio of TNF/(TNFR1 + TNFR2) (D) between placebo and DSCG groups on Day 3 (pretreatment) and Day 7. Antigenic TNF-, sTNFR1, and sTNFR2 were determined in duplicate for each sample by ELISA. Data expressed as pg/ml are presented as box graphs (open box, placebo; hatched box, DSCG) with medians as horizontal lines and the range from the 25th to 75th quartiles included. The antigen concentrations of sTNFR1 and sTNFR2 for each sample were added and the ratio of TNF/sum of sTNFR1 + sTNFR2 (pg/ng) presented in the same format. *p = 0.04 compared with placebo group

View larger version (22K): [in this window] [in a new window]   Figure 2.   Comparison of lung lavage IL-8 antigen concentration between placebo and DSCG groups on Day 3 (pre-treatment) and Day 7. Antigenic IL-8 was determined in duplicate for each sample by ELISA. Data are expressed as pg/ml in the same format described in Figure 1. *p = 0.04 compared with Day 3 placebo group.

Clinical Variables

As shown in Table 2, the two groups were similar for prenatal and neonatal variables. All infants had birthweights  1,100 g (placebo, 702 ± 35 g; DSCG, 687 ± 46 g; mean ± SEM) and were premature (placebo, 25.1 ± 0.4 wk; DSCG, 24.6 ± 0.4 wk; mean ± SEM). All infants in the placebo group and 10 (77%) of DSCG-treated infants received rescue exogenous surfactant (Survanta, Ross Laboratories, Columbus, OH). The 12-h BPD predictive score was similar in the two groups. There were no differences between groups for FI_O2, alveolar-arterial oxygen gradients, and ventilatory settings on the day of study entry (data not shown).

All DSCG-treated infants survived the first 28 days of life, but three infants in the placebo group (23%) died during the study period (p = 0.07) (Table 3). Two of the deaths in the placebo group were due to respiratory failure (Day 7 and Day 8 of age) and one was due to necrotizing enterocolitis (Day 17 of age). One DSCG-treated infant died after the study period due to a pneumothorax while on dexamethasone therapy. There were no significant differences between survivors of the two groups for oxygen-dependency at 28 d (100% control subjects; 85% DSCG). Sixty percent of placebo survivors and 62% DSCG-treated infants met criteria for BPD at 28 d of age. Compared to infants who did not meet criteria for BPD in both groups at 28 d (n = 9), the infants who developed BPD (n = 14) had a higher incidence of preterm labor (93% versus 44%, p = 0.02), lower incidence of prenatal treatment with betamethasone (36% versus 89%, p = 0.03) and higher incidence of PDA (86% versus 33%, p = 0.02). There were no differences between treatment groups for number of ventilator days, duration of oxygen supplementation and length of hospital stay (Table 3).

Cromolyn appeared to be a well-tolerated drug without apparent adverse effects (Table 4). There were no reported adverse responses to nebulization. The frequency of other neonatal complications including PDA, air leak, sepsis, NEC, and IVH were similar in the two groups. During the study period (Days 3-28), theophylline, nebulized albuterol and diuretics were used in similar frequencies in the two groups (Table 4). Six infants in the placebo group (46%) and five infants in the DSCG group (38%) were treated with systemic dexamethasone during the study period. An additonal two placebo infants and four DSCG-treated infants received dexamethasone after the study period. Attending physicians made decisions about other medication use and there was no consensus among attendings about criteria for steroid administration.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cromolyn prevents neutrophil influx (24) and activation (25) and stabilizes mast cells in the lung (22). Recently, Bissonnette and coworkers (33) demonstrated that DSCG inhibited rat peritoneal mast cell TNF--dependent cytotoxicity and TNF- release in vitro. Although DSCG therapy has been shown to be beneficial in the management of established BPD (34), its mechanism of action in the newborn lung has not been previously investigated. In the present study we demonstrated for the first time that DSCG prophylaxis therapy during the first week of life in infants at high risk for the development of BPD was associated with a decrease in lung lavage percent neutrophils and concentrations of the inflammatory cytokine TNF-. There was a trend towards lower IL-1 and IL-8 concentrations in DSCG-treated infants. There were significant correlations among these interrelated cytokines and between TNF- and IL-1 and percent neutrophils suggesting that the effect of DSCG on neutrophil influx in the preterm lung may be mediated by an inhibition of these cytokines.

We cannot determine from this study whether DSCG directly inhibits one or more of these cytokines, the mechanism involved (e.g., effect on synthesis, secretion, and/or turnover) or which cells are affected by the drug. We previously found that IL-1 concentration and IL-1 activity peaked on Days 5 and 7 in infants who developed BPD, while TNF- antigen concentration and TNF activity increased in lung lavage with peak levels on Day 14 (16). Groneck and colleagues (11) described an increase in IL-8 in tracheal aspirates of BPD infants with peak levels on Day 10. The time sequence of the appearance of these cytokines in lavage suggest that IL-1 may be central to up-regulation of TNF- and IL-8. IL-1 plays a pivotal role in the host response to infection and inflammation and it exerts many of the same actions as TNF- (37). IL-1 induces TNF- and the two cytokines have supraadditive effects (37, 39). IL-1 and TNF- induce the neutrophil chemotactic factor IL-8 (40). Therefore, lung injury may be augmented when both IL-1 and TNF- are present.

It has been proposed that the deleterious effects of cytokines are due to an imbalance of the cytokines and their natural inhibitors (15, 39, 41, 42). There are two distinct soluble TNF receptors, sTNFR1 (mw 55,000) and sTNFR2 (mw 75,000) released by epithelial cells, neutrophils, monocytes, and alveolar macrophages (43). At high concentrations, the soluble receptors block biologic effects by competing with TNF- for binding to cell surface receptors (42, 44). At low concentrations, binding of sTNFRs to TNF- may enhance TNF activity by stabilizing its structure (45). In children with meningococcemia, the ratios of TNF/TNFR1 and TNFR2 were higher in nonsurvivors than survivors (42). The soluble TNF receptors have not been measured in lung lavage fluid before. In this study, lavage sTNFRs were detectable in high concentrations but did not change over time in either group. There was a trend towards lower ratios of TNF/sum of the sTNFRs in DSCG-treated infants compared to controls at Day 7. Similarly, the ratio of IL-1/IL-1ra was lower in the DSCG-group. The inability to increase inhibitors in response to increases in inflammatory cytokines may contribute to the susceptibility of the preterm lung to injury.

In this present study, we demonstrated that prophylactic DSCG therapy is well tolerated and is not associated with an increased mortality or morbidity in extremely low birthweight infants ( 1,000 g) at high risk for the development of BPD. Cromolyn sodium is nontoxic and is excreted unmetabolized in bile and urine (46, 47). The lack of side-effects of DSCG contrasts with the known side-effects of steroid therapy for established BPD including systemic hypertension and biventricular hypertrophy, glucose intolerance, adrenal suppression and increased risk of infection (21). In the small sample of the present study, there was a trend towards an improved survival in DSCG-treated infants. Our results differ from those from the DSCG prophylaxis trial reported by Watterburg and associates (48) in which all study infants < 1,000 g birthweight died. However, in the previous study, infants did not receive prenatal steroids or exogenous surfactant and were randomized to treatment started on the first day of life. Since the majority of deaths in the  1,000 g birthweight category occur in the first 48 h (49), their sample may have included nonviable infants. We did not obtain consent and randomize subjects until 48 h to ensure a sample likely to survive the study period and began therapy on Day 3 of life. Improved survival in our study is also attributable to the frequent use of prenatal betamethasone and exogenous surfactant.

In this study, TNF- and IL-8 concentrations were reduced in lung lavage of DSCG-treated infants at one week of age but, the incidences of oxygen-dependency and BPD defined by combined oxygen requirement and radiographic criteria were the same in survivors in both groups. These cytokines have been clearly implicated in lung injury (50); this paradox may be explained by a failure to reduce these cytokines below a threshold for inducing lung injury or the persistent expression of other inflammatory factors. The inability to demonstrated clinical efficacy in this pilot study may have been due to other factors, including: (1) small sample size; (2) concomitant use of other medications; (3) variable nebulized drug delivery (46, 47); and (4) the multifactorial nature of BPD. Subsequent studies will need to develop defined criteria for use of systemic steroids, since the frequent use of dexamethasone in both groups in this study may have obscured any effect of DSCG on outcomes at 28 d and 36 wk post conceptional age. Since perinatal factors such as preterm labor and exposure to antenatal betamethasone (3) and neonatal factors such as PDA and infection affect risk for BPD (51), DSCG therapy started on the third day of life may not prevent lung injury potentially initiated before birth or secondary injury from additional insults (e.g., capillary leak, infection).

We describe a possible mechanism by which cromolyn may ameliorate neonatal lung injury. Combining inhaled DSCG with other anti-inflammatory agents may improve its efficacy in modifying the development of BPD. Future studies are recommended to: (1) establish efficacy; (2) determine the optimal dose, delivery system, time of initiation, and duration of therapy; (3) compare DSCG prophylaxis to other potential therapies; and (4) further define the interactions of DSCG and inflammatory cytokines.