INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: prematurity; mechanical ventilation; pulmonary neuroendocrine cells; lung injury

Bronchopulmonary dysplasia (BPD) (1), also known as chronic lung disease of premature infants, is the most prevalent of the long-term sequelae that affect surviving preterm infants. This disorder is often clinically defined as oxygen dependence with continuing radiologic or clinical evidence of lung disease at 36 weeks postmenstrual age, a definition that has the greatest relevance to long-term prognosis (2). The pathophysiology of BPD is multifactorial, with contributing factors including barotrauma, oxygen toxicity, infection, inflammation, and pulmonary immaturity (3-5). BPD is associated with increased risk of postneonatal mortality, impaired growth, and neurodevelopmental and sensory disabilities (6). Paradoxically, the rate of BPD among surviving very low birth weight infants has increased in association with the availability of exogenous surfactant therapy for treating respiratory distress syndrome and the resulting improved survival of the most vulnerable infants (3, 5, 7, 8). However, no biological marker has been identified that predicts which infants are at greatest risk for BPD (9), including vulnerability factors (gestational age, birth weight, race, sex) and early physiologic risk factors (Apgars, blood gases, early ventilator settings) (9). Consequently, it has not been possible to intervene therapeutically early in the course of the disease. Identification of such a predictive marker would provide the opportunity for both early preventive therapies and possibly a treatment specifically directed at one or more steps in the pathophysiologic cascade leading to BPD.

Bombesin-like peptide (BLP) immunoreactivity represents a family of neuropeptides derived from the pulmonary neuroendocrine cells, which play a critical role in normal lung growth and maturation (10). BLP levels normally peak around midgestation (20-24 weeks) in human fetal lung, thereafter dropping off to near-adult levels by birth. Excessive BLP levels, however, appear to be associated with injury to the developing postnatal lung. Increased numbers of BLP-positive pulmonary neuroendocrine cells occur in infants dying with BPD (11), possibly induced by proinflammatory cytokines and/or oxidant injury (10).

We began to test the hypothesis that BLP can mediate lung injury using two distinct baboon models of BPD, developed by Coalson and coworkers (12, 13). Preterm animals delivered by Caesarean section at 140 days gestation (term = 185 days) and ventilated with 100% O₂ (140 days/100% O₂) for 10 days develop clinical and pathological features typical of moderate to severe BPD ("old BPD"), similar to those described by Northway and colleagues in 1967 (1). In contrast, control 140-day gestation preterm baboons ventilated for 10 days on oxygen pro re nata (PRN) (to keep arterial blood hemoglobin above approximately 90% O₂ saturation) (140 days/PRN) do develop acute respiratory distress syndrome with hyaline membrane disease between 1 and 48 hours after delivery, but recover from the acute injury and do not develop subsequent clinical or pathologic changes reminiscent of BPD. More recently, Coalson and colleagues have developed a model of milder BPD occurring in extremely premature baboons (125 days gestation) ventilated with O₂ PRN and requiring exogenous surfactant to survive, which is much more similar to the version of BPD currently seen in human infants (6). In this model of "new BPD," over 90% of 125 days/PRN animals develop characteristic clinical and pathological features of chronic lung disease found in ventilated extremely low birth weight human infants (13).

Elevated BLP levels in 24-hour urine collections during the early postnatal period were demonstrated in premature baboons in the BPD models, apparently derived from hyperplastic pulmonary neuroendocrine cells in both models (14). Unlike control animals, those destined to develop BPD manifested increased urine levels of BLP between 2-3 days after birth. Further, the severity of the BPD, as assessed by oxygenation index at 10 to 14 days of age, directly correlated with this early increase in urine BLP (14). Finally, the blocking anti-BLP antibody 2A11 given shortly after birth abrogated BPD in both models (14), suggesting that BLP is an early mediator of BPD.

The aim of the current study was to determine whether an elevated urine BLP level measured in the first few postnatal days is associated with increased risk for subsequently developing BPD in preterm infants.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Population

The 132 infants enrolled were born at 28 weeks gestation or less at Brigham and Women's Hospital between July 1997 and December 1998 and had an initial urine bombesin measurement between postnatal day (PND)-1 and PND-4. BPD was defined as a requirement for supplemental oxygen at 36 weeks postmenstrual age. The Brigham and Women's Hospital Institutional Review Board approved this protocol.

Demographic information and perinatal history were obtained from medical records. An experienced neonatal intensive care unit research nurse collected clinical data using standardized forms including perinatal treatments such as corticosteroids, antibiotics, bronchodilators, and ventilation.

For statistics, gestational age was dichotomized at 26 weeks, which is when the risk of BPD changes markedly, in our experience, being nonlinear across the age range of the study infants. Mechanical ventilation was defined as intubation with either intermittent mandatory or high-frequency oscillating ventilation on PND-0 (day of birth), but not positive airway pressure alone. Patent ductus arteriosus (PDA) was defined as a requirement for indomethacin and/or surgical ligation. Chorioamnionitis was defined clinically as the presence of fever or uterine tenderness. Postnatal sepsis was defined clinically.

Urine BLP Measurement

We decided to measure BLP in urine samples for several reasons. First, BLP is degraded quickly in human plasma with t_1/2 of less than 4 minutes, precluding accurate blood testing (15). Second, renal excretion is the major clearance route for neuropeptides, which are concentrated in urine (16, 17), and BLP levels in urine and bronchoalveolar lavage fluid are positively correlated (18). The content of BLP in 24-hour urine collections from preterm baboons is high, approximately 5 nanograms per 24 hours, corresponding to greater than 30% of the total lung BLP content (14). Considering the low BLP levels in extrapulmonary tissues, more than 90% of urine BLP should be from the lung (14).

The study protocol included collection of urine specimens, ideally starting on PND-0, using a cotton ball placed in the diaper for 1-2 hours (greater than 85% BLP recovery in urine at 37° C for 3 hours or less) (18). Urine was squeezed into a 2-ml tube with acetic acid (final concentration 2 N). Urine specimens were frozen for up to 6 hours at 20° C (greater than 90% BLP recovery), then cooled to 80° C (BLP indefinitely stable) until analyzed (within 12 months of being collected).

Our aim was to determine whether a single urine BLP level taken in the first postnatal days might predict which infants would subsequently develop BPD. The first urine sample from each infant was analyzed, if taken between PND-1 and PND-4. Urine BLPs were determined by radioimmunoassay (Peninsula Laboratories, Belmont, CA) in duplicate, normalized for urine creatinine content to correct for urine concentration (Sigma Diagnostics, St. Louis, MO). These methodologies for directly measuring urine BLP were previously published (18, 19). Results were expressed as picograms BLP per milligram creatinine. Bombesin in the radioimmunoassay is a frog peptide, so we converted these numbers to fmol/mg creatinine by multiplying by 0.625 to represent moles of immunoreactive BLPs (20).

Statistical Analysis

Univariable, stratified, and multivariable analyses were conducted using Stata, Version 6 (Stata Corporation, College Station, TX). Univariable associations were examined with Fisher's exact test and the Wilcoxon rank-sum test. Multivariable analyses were conducted using logistic regression.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

One hundred fifty-one infants had a first urine BLP measurement between PND-1 and PND-4. Of these, 19 were excluded from further analyses because their BPD status was unknown. For nine infants, there was no record of BPD status (these infants were transferred out of Brigham and Women's Hospital and lost to follow-up), and 10 infants died before 36 weeks postmenstrual age and could not be assigned a BPD status. There were no significant differences between the study population and the 19 infants excluded from analysis with regard to gestational age, birth weight, sex, race, receipt of antenatal glucocorticoids, mechanical ventilation, or median urine BLP level (Table 1). The median urine BLP level for the nine infants lost to follow-up was 13,704 (range 2,755-28,140) and for the 10 who died was 17,019 (range 4,564-38,128). These levels were not significantly different from that of the study population.

BLP analyses were conducted on 132 infants for whom outcome data were available. Fifty (38%) of these infants were oxygen-dependent at 36 weeks postmenstrual age. After adjusting urine samples for creatinine level to correct for urine concentration, we compared urine BLP levels in infants with BPD to those without (Figure 1). The median urine BLP level of infants with BPD (median: 20,730; lower and upper quartiles: 8,593; 24,282) exceeded the 90th percentile of the distribution for the non-BPD infants (median: 11,718; lower and upper quartiles: 6,439; 15,585) and correctly classified 54% of the infants who went on to develop BPD (p = 0.0003) (Table 2). The specificity of urine BLP above 20,000 pg/mg creatinine (= 12,500 fmol/mg) for BPD was 90%, the positive predictive value was 77%, and the negative predictive value was 76% (21).

View larger version (13K): [in this window] [in a new window]   Figure 1.   Box and whiskers plot of the first urine BLP in infants with and without BPD. The median urine BLP levels of infants with BPD (left) and without BPD (right) are indicated by a horizontal line across the box, which is defined by the lower and upper quartiles (21). The "whiskers" are the vertical lines extending above and below the box, indicating the range of the 5th to 95th percentile values. The median urine BLP level of infants with BPD (median: 20,730; lower and upper quartiles (8,593, 24,282) exceeded the 90th percentile of the distribution for the non-BPD infants (median: 11,718; lower and upper quartiles: 6,439; 15,585).

The complex interactions among biological immaturity, superimposed clinical conditions and treatments, and BPD raised the possibility that the association between BLP and BPD might reflect other coexisting variables (those related to both elevated BLP and BPD) (18), i.e., confounding factors. We evaluated, as potential confounders, sex, gestational age, race, antenatal receipt of corticosteroids, clinical chorioamnionitis, antenatal antibiotic therapy, mechanical ventilation, PDA, and neonatal sepsis (Table 2).

Multivariable logistic regression analysis was performed, employing all identified potential confounders, including demographic and cardiorespiratory variables. Those that were not significant at the alpha 0.05 level were dropped until the most parsimonious model was reached. In the final model, which included race, gestational age, PDA, and mechanical ventilation, a urine BLP level greater than 20,000 pg/mg creatinine (12,500 fmol/mg) was associated with a 10-fold increase in requirement for supplemental oxygen at 36 weeks postmenstrual age (odds ratio 9.9, 95% confidence interval 3.4, 29, p  0.001). In contrast, PDA and mechanical ventilation were each associated with only a 2-3-fold increased risk of BPD in the final model.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study suggests that infants born at or before 28 weeks of gestation, who have a first urine BLP level greater than 20,000 pg/mg creatinine (12,500 fmol/mg) before PND5, are at increased risk of developing BPD. This relationship does not appear to be explained by other factors, such as gestational age, sex, PDA, or mechanical ventilation. In our final multivariable model, the 10-fold increased risk of BPD associated with urine BLP greater than 20,000 pg/mg creatinine (12,500 fmol/mg) far exceeded the 2-3-fold increased risk associated with PDA or mechanical ventilation. However, to determine the best cut-off value of urine BLP to make this a universal tried-and-true diagnostic test, a larger multicenter trial will need to be performed with a much larger group of infants.

BLP is detected in human pulmonary neuroendocrine cells at about 8 weeks gestation, with peak numbers of positive cells occurring around midgestation (10). BLP plays a significant role in lung development, promoting fetal lung branching morphogenesis, epithelial and mesenchymal cell proliferation, and differentiation of type II cells and neuroendocrine cells (10). BLP also promotes surfactant phospholipid secretion via a mechanism involving protein kinase C (PKC) (10). Effects of BLP that are blocked by the antibombesin antibody, 2A11 (10, 14), include: BLP-stimulated proliferation of human lung cancer cell lines in a colony-forming assay or unstimulated tumorigenesis of two BLP-positive cell lines in nude mice in vivo; and fetal lung cell proliferation and type II cell differentiation in vitro and in vivo (10).

In human and rodent lung, there is a normal decline in levels of mRNA encoding the major known pulmonary gastrin-releasing peptide, after midgestation, which occurs in parallel with the observed decrease in relative numbers of pulmonary neuroendocrine cells (10). However, more than a decade ago, Johnson and colleagues made the observation that the lungs of infants dying with BPD have profound hyperplasia of BLP-positive neuroendocrine cells (11). In a small clinical series of preterm infants studied during the presurfactant era, we observed marked elevation of BLP (gastrin-releasing peptide) gene expression (mRNA) in lung tissue from one of four infants dying with hyaline membrane disease at 1-2 days of age (22).

Similarly, urine BLP levels have been reported to be higher among infants with established BPD (19). In that earlier study, urine samples were analyzed during the first 7 weeks of life, but included only six infants of less than 30 weeks gestation with BPD and six premature infants of more than 30 weeks gestation with BPD. There was no information given about the nature of the five premature control infants. There was no significant difference in urine BLP levels between the BPD and non-BPD groups during any time period, although there was a trend toward elevated urine BLP levels during the first 3 weeks. The absence of demographic information about the patients and the small number of patients being analyzed make these data difficult to interpret.

BLPs are known to be potent growth factors for normal adult bronchial epithelial cells, embryonic and pulmonary fibroblasts, and lung cancer cell lines (22). BLPs also induce profound bronchoconstriction (23), macrophage chemotaxis, phagocytosis and activation (24, 25), and fibroblast chemotaxis and proliferation (22, 26). Thus, high levels of BLP during postnatal lung development could induce excessive peribronchiolar and interstitial fibrosis, reactive airways disease, and abnormal alveolar structure, which are all hallmarks of BPD. It appears that excessively high levels of BLP in a subset of preterm infants might contribute to the pathogenesis of BPD (10, 14).

Investigation of the two baboon models of BPD, the hyperoxic model and the interrupted gestation model (12, 13) initiated the hypothesis for the present study. We demonstrated that elevated urine BLP levels correlate with clinical evidence of lung injury in both baboon models and that anti-BLP blocking antibody 2A11 attenuated the injury, by both clinical and histopathologic measures (14). These observations suggest that BLP plays a pivotal role as an early mediator in the pathogenesis of BPD among preterm baboons. These data also support the theory that BPD is an inflammatory process that disrupts normal lung development (4).

The presence of elevated BLP levels in more than half of infants who subsequently develop BPD suggests that genetic factors could regulate BLP overproduction during the early postnatal period. Studies of twins suggest there may be genetic susceptibility to BPD, with the disorder in a first twin conferring a 12-fold increased risk of BPD in the second twin (27). Associations have been reported between BPD and other pulmonary disorders with familial clustering, such as asthma or atopy (28, 29). Pulmonary neuroendocrine cell hyperplasia and degranulation also occur in animal models of reactive airways disease (30). Pulmonary neuroendocrine cells function postnatally as oxygen chemosensors and degranulate in response to airway hypoxia (31). Increased numbers of these cells can occur within 4 to 24 hours after the onset of either hypoxia or hyperoxia (30, 32), probably due to a combination of cell differentiation and proliferation (33, 34). Maternal smoking is a risk factor for pediatric asthma, and abnormally large neuroendocrine cells have been observed in the lungs of fetuses exposed to maternal smoking (30).

We speculate that the two most likely causes of neuroendocrine cell differentiation and/or proliferation in premature infants would be oxidant lung injury (35) and/or cytokine production, such as occurs with tumor necrosis factor- (33). Airway hypoxia is not likely to occur in infants on respiratory support, but hyperoxia could similarly trigger neuroendocrine cell degranulation, a reasonable hypothesis that awaits formal testing. Barotrauma from mechanical ventilation might also contribute, with mechanical forces inducing neuroendocrine cell degranulation, similar to mast cell degranulation. Over half of the neuroendocrine cells are localized as clusters at airway branch points, making them especially vulnerable to changes in airflow, turbulence, and airway pressures. Greater susceptibility of a subset of infants may be due in part to genetic factors predisposing to neuroendocrine cell hyperplasia and/or excessive degranulation and secretion of BLP.

BPD is a multifactorial disease, possibly explaining why only a subset of very low birth weight infants develop this chronic lung disease (2, 6), similar to our present observation of a subset that develops elevated urine BLP levels. Our results suggest that infants born at or before 28 weeks gestational age, and who have elevated levels of urine BLP in the first few postnatal days, are at substantially increased risk of developing BPD. This holds as true even after correcting for the known risk factors for BPD, including low gestational age, sex, PDA, and mechanical ventilation. This urine test provides the opportunity to identify infants at greatly increased risk of developing BPD and potentially to develop a targeted preventive therapy, such as anti-BLP blocking antibody. As our understanding of the precise role that BLP plays in the process of lung maturation, alveolarization, and lung inflammation improves, new insights will be gained into the critical signaling pathways operating during this period of development.

Our cumulative observations provide compelling evidence that pulmonary neuroendocrine cells can function in a proinflammatory cascade by secreting proinflammatory peptides, such as BLP. These studies are leading to a paradigm shift in our understanding of BPD, and thus are opening new avenues for investigation of lung injury. Similarly, chronic inflammatory and/or fibrotic diseases in other organs might be mediated via peptides derived from the neuroendocrine cells, which are distributed throughout most mammalian organs. Thus, the diffuse neuroendocrine system of "amine precursor uptake and decarboxylation" cells described by Pearse in the 1970s (36) could have a generalized proinflammatory function in postnatal life.