Bronchopulmonary dysplasia (BPD) is a common adverse outcome of extremely premature birth.1 At an National Institute of Health (NIH) workshop, a consensus was reached that infants should be diagnosed as having BPD if they were oxygen dependent beyond 28 days and then classified at a later date as to whether they had mild, moderate or severe BPD according to their supplementary oxygen and respiratory support requirements at that time.2 Using that classification, BPD status and severity reliably predicted adverse long-term respiratory outcomes.3 Unfortunately, despite advances in neonatal intensive care, a review of 5115 prematurely born infants revealed that the risk of BPD development had remained constant between 1994 and 2004.4 Key to reducing BPD is a better understanding of the pathophysiology of the condition and the means to more accurately identify those at high risk of developing BPD, particularly the more severe forms.

We have demonstrated that exhaled nitric oxide (eNO) levels were higher at 36 weeks postmenstrual age (PMA) in infants who remained dependent on oxygen beyond 36 weeks PMA compared with those without BPD.5 In addition, eNO levels in infants with severe BPD were reduced by dexamethasone administration.6 No significant differences, however, were seen in eNO levels in the neonatal period between infants who subsequently remained on supplementary oxygen beyond 36 weeks PMA and those who did not,7 but very few infants were studied on day 28. A possible explanation for those results5^–7 is that eNO levels may only become raised in infants with an ongoing inflammatory process, that is, those developing the more severe forms of BPD. Infants developing the more severe forms of BPD may have higher levels of inflammation earlier in the neonatal period than infants without BPD or those developing mild BPD and thus have different eNO profiles in the neonatal period. The aim of this study, therefore, was to test the hypotheses that eNO levels on day 28 and changes in eNO levels in the neonatal period would differ according to whether infants developed BPD and its severity.

PATIENTS AND METHODS

Infants born at less than 32 weeks of gestational age whose parents gave informed, written consent were recruited. The study was approved by King’s College Hospital Research Ethics Committee. Measurements of eNO were attempted on days 1, 3, 5, 7, 14 and 28 days after birth. Measurements of eNO levels were not made, for technical reasons, while the infants were supported by high-frequency oscillatory ventilation (HFOV) or nasal continuous positive airways pressure (CPAP) or when they were receiving inhaled nitric oxide.

eNO was measured using a Sievers Chemiluminescence analyser (Sievers Boulder Co USA). The analyser measured accurately to one part per billion (ppb) and had a 90% response time of 400 ms. Prior to each measurement, the NO analyser was calibrated with NO-free air (filtered through a NO scavenger) and NO gas certified at 25 parts per million (ppm) (Bedfont scientific gases, Upchurch, Kent, UK). The NO scavenger used contained potassium permanganate crystals. The colour of the crystals was checked regularly, as any change in colour indicated a deterioration in the efficiency of the scavenger. Once a colour change was noted the crystals were replaced. The analyser behaved linearly between 0 ppm and 25 ppm. The sampling rate used in the study was 100 ml/min. When making eNO measurements in ventilated infants, an NO scavenger was placed within the inspiratory limb of the ventilator circuit. The efficacy of the NO scavenger was regularly checked to ensure that the gas distal to the scavenger was NO free using the calibrated NO meter. A 5 Fr gauge sampling catheter was passed through the suction port of the ventilator manifold. The catheter was positioned so that its tip lay at the lower end of the endotracheal tube. The infants were all ventilated via shouldered endotracheal tubes, the standard practice of the unit. We have previously demonstrated, there is minimal or no leak around shouldered endotracheal tubes.8 In non-ventilated infants, a facemask was held snugly over the infant’s nose and mouth. A 4 l/min bias flow of air from the wall outlet was passed through the facemask via a t-piece, this did not pressurise the circuit and hence did not result in leaks. To ensure the air was free of NO, an NO scavenger was placed into the tubing leading to the mask. A valve was put in the distal part of the facemask. A 5 Fr gauge catheter was fed through the leak free valve and the tip positioned so that it lay at the infant’s lips. The catheter was attached to the NO analyser. After a period of 1 min to allow for stabilisation of eNO levels, the maximum eNO levels were calculated from 10 consecutive breaths and the average taken. The mean intrasubject coefficient of variability of eNO levels had been calculated from the results from each of 10 infants and was 11.7%.7

Demographic data were collected from the patients’ notes, including information on their gestational age, birth weight, whether they were small for gestational age at birth (ie, their birthweight was less than the tenth percentile for their gestational age), were male or had received surfactant or their mother had received antenatal steroids. In addition, their durations of ventilation and oxygen dependency were recorded, as were the number of septic episodes they had had and whether they received postnatal steroids. The routine policy of the NICU was to commence infants on antibiotics if they had respiratory distress at birth and at any time sepsis was suspected. If, however, there were no ongoing abnormal signs and blood cultures were negative, antibiotics were discontinued at 48 h. As a consequence, infants in this study were described as having a septic episode if they had received antibiotics for at least five days. The routine policy of the unit was to prescribe systemically administered corticosteroids to infants who remained ventilator dependent in high inspired oxygen concentrations and were at least two weeks of age.

Analysis

Infants were diagnosed as having BPD if they remained oxygen dependent at 28 days, they were then re-examined at 36 weeks PMA to determine the severity of BPD2; mild BPD was diagnosed if they were no longer oxygen dependent at 36 weeks PMA, moderate BPD if they remained oxygen dependent at 36 weeks PMA and severe BPD if they were dependent on respiratory support at 36 weeks PMA. The results of some of the infants (n = 24) have been previously reported;7 for the purposes of this analysis those infants were also classified regarding BPD status and severity as above. Generalised estimating equations (GEEs)9 with exchangeable correlation structure were used to estimate the change in eNO levels over time in the four groups (no BPD, mild, moderate and severe BPD). The analysis was repeated in two groups—that is, using BPD as a binary variable (no/yes). The distribution of the residuals for the regression analyses was assessed and found to be slightly skewed. We performed the analysis using untransformed NO, log transformed and square root transformed and found the p values very similar; as a consequence, for ease of interpretation, we presented the results for NO on the natural scale. In addition, eNO results at 28 days were assessed using multiple regression and the trend across the four groups determined. The software used was STATA V.9.0.

RESULTS

Eighty infants with a median gestational age of 28 (range 24–32) weeks were examined; 46 developed BPD. The infants who developed BPD were more immature (p<0.001) and of lower birthweight (p<0.001) than those without BPD. The no BPD group was more mature than the other three groups (p<0.001), but the maturity of the three BPD groups was similar (table 1). The birthweight of the no BPD group was greater than the mild BPD group (p<0.05) and both of the other two BPD groups (p<0.001). Similar proportions of each of the four groups had received antenatal steroids and surfactant and were small for gestational age. A greater proportion of the moderate (p<0.001) and severe (p<0.001) BPD groups were boys compared with the no BPD group. The durations of ventilation were significantly longer in both the severe and moderate BPD groups than the no BPD group (p<0.001) and in the severe BPD group compared with both the mild and moderate groups (p<0.05). The duration of oxygen dependency was longer in the three BPD groups compared with the no BPD group (p<0.001), in the moderate compared with the mild BPD group (p<0.05) and the severe compared with the mild (p<0.001) and moderate (p<0.05) BPD groups. There were no significant differences between the four groups with regard to the number of septic episodes per baby experienced, but more of the severe BPD group went on to require postnatal steroids than the no BPD group (p<0.001), the mild group (p<0.01) and the moderate group (p<0.01). None of the infants received postnatal steroids prior to 28 days of age.

The median number of eNO observations per subject was 4 (range 1–6), this did not vary by severity of BPD. Plotting the change in eNO with time in individual subjects according to BPD status demonstrated a flat relationship in most subjects without BPD, but positive relationships in the moderate and severe groups. GEE analysis, however, showed that although the average slope varied slightly by severity of BPD, this did not differ statistically significantly between either the four groups (p = 0.30) or the two groups (p = 0.08) (table 2). Regression of eNO levels at 28 days demonstrated a significant difference in eNO levels between the four groups (p = 0.002) and between infants who did and did not develop BPD (p<0.001) (table 3). The test of trend across the four groups was significant (p = 0.006), showing there was good evidence for a linear trend in eNO levels as BPD severity increased. At 28 days, all those measured in the severe BPD group, but very few of any of the other groups were assessed while ventilated (table 4). Comparison, therefore, was made of the eNO levels at 28 days of the moderate and no BPD groups revealing the eNO level of the moderate group was significantly higher (p<0.0001). Tests of pairs of groups using Gabriel’s method10 showed that the no BPD group differed significantly not only from the moderate but also the severe group (this method does not give actual p values but simply gives “significant” or “non-significant”, while preserving the overall significance level at 5%).

DISCUSSION

In very prematurely born infants, we have demonstrated significant differences in eNO levels on day 28 after birth according to BPD status, particularly in infants developing moderate or severe BPD. There were no significant differences before day 28, suggesting that eNO levels may only be raised in infants with established BPD and ongoing inflammation. This is in keeping with our finding of significantly raised eNO levels at 36 weeks PMA in infants who had moderate BPD (p = 0.028).5

eNO levels were measured using two sampling techniques, either from the tip of the endotracheal tube or, in non-ventilated infants, from within a facemask. During both techniques, NO free gas was administered to the infants. We have previously demonstrated that there is no significant correlation between nasal and lower airway to NO levels,11 hence we deliberately positioned the tips of the catheter so that it lay at the infant’s lips. It is possible, however, when sampling from the facemask that NO produced in the nose, even on the first day after birth11 may have influenced the results. Yet, despite more of the no BPD group than the severe group being studied when non-ventilated, the latter group’s results were significantly higher on day 28. When measuring the non-ventilated infants, there was a bias flow throughout the respiratory cycle, whereas the ventilated infants only experienced the bias flow in inflation which may have influenced the eNO levels. As a consequence, we compared the eNO results at 28 days of the no BPD to the moderate BPD group, as very few of either group were ventilated at that time and the majority of the infants were sampled using the same technique. This demonstrated the eNO levels of the moderate BPD were significantly higher than the no BPD group.

Ideally, eNO measurements should be made while the subject is breathing out at a constant expiratory flow rate. This can be achieved by incorporation of a dynamic flow restrictor in the collection system,12 but it is not appropriate to use such a restrictor for ventilated, very prematurely born infants. In more mature infants, flow can be controlled by using the modified single breath technique, rather than a tidal breathing technique, but comparison of results from the two techniques demonstrated poor agreement between them.13 The results of the single breath technique were more sensitive than those of the tidal breathing techniques, only the single breath technique yielded eNO levels which differed significantly between healthy infants and infants with wheeze. To use the single breath method, however, the raised volume rapid thoraco-abdominal technique was used13 and this is not suitable for the prematurely born infants we studied. We did not relate the eNO levels to minute ventilation as has been done in some studies,14 yet we demonstrated significant differences in eNO results on day 28. Exhaled NO levels are influenced by breathing patterns and expiratory flow rates, as well as nasal contamination.12 15 The ventilated infants were all receiving morphine at the time of measurement and thus more likely than the spontaneous breathing babies to have a regular breathing pattern and less variability in eNO levels. Interestingly, however, we saw significant differences between the groups only on day 28 when except for the severe BPD group, very few of the babies were measured while ventilated. If higher flow rates had influenced our results, as would be anticipated in ventilated infants, it would be expected they would have lower eNO levels, yet on day 28 the severe BPD group (who were all ventilated) had higher eNO levels compared to the no BPD group (none of whom were ventilated). It is possible that BPD infants might adopt a pattern of breathing that resulted in such low flow rates it “raised” eNO levels, even if this were the case, higher eNO levels remain a marker of BPD, particularly moderate or severe, development. Subanalysis demonstrated both the severe (all ventilated) and the moderate BPD (15% ventilated) had significantly higher eNO levels than the no BPD group (4% ventilated). The “severe” BPD infants might be predicted to have high flow rates due to being ventilated (as above), whereas the non-ventilated BPD infants might have adopted a pattern of breathing with low flow rates. It thus seems unlikely that differences in flow rates explained our results as both the moderate and severe BPD groups had significantly higher eNO levels than the no BPD groups.

In older infants16 and children17 18 who had had BPD, similar or lower eNO levels compared with controls have been reported. The children, however, were of school age18 or studied between 4 and 25 months17 and the infants examined at 44 weeks PMA,16 when the authors state they felt the main inflammatory process was over. In contrast, our infants were studied up to and at 28 days after birth, when at the latter age the BPD groups remained oxygen dependent, reflecting an ongoing process. This may explain the difference between our findings and previous studies.16^–18

A possible explanation for the elevated eNO levels on day 28 in the infants with BPD was that the babies had enhanced inducible NO synthase (iNOS) activity. In the baboon model of BPD, there is deficiency of neuronal nitric oxide synthase (nNOS) and endothelial NO synthase (eNOS), but enhanced iNOS.19 iNOS can be induced by proinflammatory cytokines, such as tumour necrosis factor α (TNFα), interferon δ (IFNδ) and interleukin (IL1α).20 In infants who develop BPD, TNFα activity increases late with peak levels between 14 and 28 days,21 which is in keeping with our finding of significantly raised eNO levels on day 28. Exogenous factors also cause transcription activation of iNOS; these include bacterial toxins, virus infection and hypoxia.22 In our study, the infants who developed moderate or severe BPD tended to have more septic episodes and they required significantly longer ventilatory support. Hence we feel they may have had enhanced iNOS activity. Glucocorticoids are thought to have an anti-inflammatory role in part by inhibiting iNOS expression. We have previously reported glucocorticoid administration in babies on long-term ventilation with reduced eNO levels6 which further suggests infants with severe BPD have enhanced iNOS activity.

Randomised trials have demonstrated that administration of systemically administered corticosteroids within 96 h of birth23 and between 7 and 14 days24 reduces BPD. Administration of corticosteroids after 3 weeks of age25 is associated with a significant reduction in need for home oxygen and late rescue therapy. Unfortunately, systemically administered corticosteroids have many side effects and there has been particular concern regarding the possibility of an increased risk of cerebral palsy. A meta-analysis of 20 randomised trials of systemically administered corticosteroids that included 2064 infants, demonstrated the relative risk of cerebral palsy was 1.45.26 Further analysis demonstrated that only early (relative risk 1.70, 95% CI 1.20 to 2.42) and not late (relative risk 1.20, 95% CI 0.83 to 1.74) treatment was associated with a significant excess of cerebral palsy. Thus our findings are clinically important as they highlight that at 28 days it is possible to distinguish by measurement of eNO levels, infants likely to develop the more severe forms of BPD. At that postnatal age, systemic administration of corticosteroids improves outcome25 without significantly increasing the risk of cerebral palsy.26

In conclusion, we have demonstrated raised eNO levels on day 28 in infants with BPD, particularly in those who went on to develop moderate or severe BPD. It is important to confirm those findings in another cohort and hence whether raised eNO levels on day 28 could be used as a reliable criteria to institute therapies aimed at reducing supplementary oxygen requirements and associated chronic respiratory morbidity.