Alterations of exhaled nitric oxide in pre-term infants with chronic lung disease

Study design and subjects

In an unmatched case−control study, eNO was measured in 39 pre-term infants with CLDI, 23 healthy pre-term infants and 127 healthy unselected term infants at a PCA of ∼44 weeks (table 1⇓). Former pre-term infants were recruited from the neonatal unit of the University Maternity Hospital (Berne, Switzerland) between January 1999 and December 2004. The parents of the 62 pre-term infants, of which 23 were classified as healthy, 12 as having mild and 27 as having moderate CLDI according to the definition of the National Institute of Child Health and Human Development/National Heart, Lung and Blood Institute/Office of Rare Diseases Workshop 16, agreed to participate. The clinical characteristics of all the pre-term infants were obtained from the patient medical record. The healthy term infants were recruited antenatally from two maternity hospitals in the Berne region as part of a prospective birth cohort study during the period 1999–2004 14, 15.

Environmental and maternal exposures did not differ significantly between the three groups, with the exception of maternal smoking in pregnancy. This was more common in the CLDI group than in the healthy term group, and it was absent in the healthy pre-term group. Maternal coffee drinking was more common in mothers of healthy pre-term infants compared with healthy term and CLDI infants. CLDI infants had more cardiac malformations (persistent ductus arteriosus (PDA) and atrium or ventricle septum defect), were more likely to have a history of chorioamnionitis and received more post-natal corticosteroids than healthy pre-term infants.

Information on demographics, family history of atopic disease, clinical symptoms and environmental risk factors, including pre- and post-natal tobacco exposure and maternal caffeine intake, were obtained from the mothers by standardised interview. The presence of chorioamnionitis was determined based on placental histology.

Exclusion criteria were as follows: ethnicity other than white, major birth defects, treatment with caffeine and anti-inflammatory treatment or respiratory tract infection within the preceding 3 weeks. Additional exclusion criteria for the healthy controls were: pre-term delivery (<38 weeks), respiratory distress with need for oxygen for >30 min after birth or other significant perinatal disease.

The ethics committee of the university hospital and the State of Berne approved the study. Parental written informed consent was obtained prior to study commencement. Parents were generally present during the measurements.

Measurements of online eNO

All the infants were studied during unsedated quiet sleep, in a supine position with the head in the midline. Heart rate and arterial oxygen saturation (Biox 3700; Datex-Ohmeda, Helsinki, Finland) were monitored throughout the study. A compliant silicon mask (Infant mask, size 1; Homedica, Cham, Switzerland) was placed over the nose and mouth, and flow–volume loops were inspected for leaks prior to commencing the measurement. An NO filter ensured that all infants inhaled NO-free air. Tidal gas flow (V′), volume (V), and eNO and CO₂ levels were measured using commercially available prototype infant lung function equipment (Exhalyser; EcoMedics, Duernten, Switzerland) as previously validated and described in detail elsewhere 14. Briefly, a representative example of online V′, eNO and NO output (V′_NO) shows a steep increase of eNO at the beginning of expiration, which achieves a plateau towards the end of this phase (fig. 1b⇓). During inspiration, eNO rapidly returns to zero, indicating negligible re-breathing of NO from the equipment dead space. V′_NO is calculated by multiplying V′ by eNO (V′_NO = eNO×V′; fig. 1c⇓). Flow (fig. 1a⇓) rapidly approximates to zero at the end of expiration. These rapid flow changes result in variable V′_NO during the fourth quartile of the breath duration. eNO and V′_NO have, therefore, been measured in the third quartile of expiration, since this shows the lowest breath-to-breath variability 14 and corresponds approximately to the phase III slope 17–20. In order to ensure consistency of analysis between infants, only the first 100 breaths in the tidal breathing data were analysed. Since eNO is strongly flow dependent and the CLDI and healthy groups were significantly different in expiratory flow (table 2⇓), it is mandatory to calculate not only eNO but also V′_NO 14.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Representative example of a) online flow (V′), b) exhaled nitric oxide (eNO) and c) NO output (V′_NO). The levels of eNO show a steep increase at the beginning of expiration and then approximate a plateau towards the end of expiration. By multiplying V′ and eNO, V′_NO can be calculated (V′_NO = eNO×V′). V′ rapidly approximates zero at the end of expiration.

Statistical analysis

All continuous variables were normally distributed with the exception of eNO, which was therefore transformed for further analysis.

The comparison of eNO between healthy and sick children was difficult because eNO concentration is influenced by physiological, maternal and environmental factors 15 (fig. 2⇓), some of which are also related to prematurity and disease. In view of these complexities, the present authors undertook a pragmatic approach and systematically analysed the two outcomes using two different statistical models as follows: 1) a simple unadjusted comparison of means using unpaired t-tests; and 2) a multivariable regression model adjusting for the most influential physiological covariates (expiratory time (t_E) and minute ventilation (V’_E)) and also for other factors associated with eNO and V′_NO in healthy term infants, including sex of the infant, maternal atopic disease, pre- and post-natal maternal smoking and caffeine exposure 15. For the regression models, continuous variables (t_E, V’_E) were centred and categorical variables were entered as indicator variables. From these regressions, parameter estimates and their 95% confidence intervals (CI) were reported together with p-values for the null hypothesis, which presumes that these estimates equal zero. In pre-term children, the gestational age (GA) and CLDI were so closely correlated that it was not possible to separate their association with NO by simultaneous inclusion in a multivariate model. Therefore, two series of models were elaborated. In the first instance, the children were categorised according to presence or absence of CLDI (table 3⇓) and then according to their GA, as 24–27 weeks, 28–31 weeks, 32–35 weeks and ≥38 weeks (table 4⇓).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2—

Causal pathway of concomitant factors of exhaled nitric oxide (eNO) and interactions between these factors. CLDI: chronic lung disease of infancy ^#: only in infants of atopic mothers.

Within pre-term infants, it was then tested systematically using an unadjusted model whether NO was associated with any of the following clinical risk factors for CLDI: GA; clinical risk index for babies (CRIB) score; duration of oxygen supplementation; PDA; pre- and post-natal steroid treatment; administration of surfactant; and chorioamnionitis. A second test of these indices was then performed using a multivariable model adjusted for t_E and V’_E.

Clinical, physiological and environmental characteristics of the participating infants

The clinical, physiological and environmental characteristics of the infants are summarised in tables 1⇑ and 2⇑. Each of the three groups were of similar PCA; however, weight and length at study date were highest in healthy term infants. These anthropometric indices were progressively smaller at the time of study for healthy pre-term and CLDI infants. Tidal breathing parameters also differed considerably between healthy term infants, healthy pre-term infants and pre-term infants with CLDI, with the noteworthy exception of V’_E, where the group mean was similar in all three groups (table 2⇑).

NO measurements in pre-term infants with and without CLDI: comparisons with healthy term infants

Mean tidal eNO was 15.2 ppb in healthy term infants and this value was not different to those obtained from healthy pre-term or CLDI infants when examined with either of the models (table 3⇑). V′_NO was 0.63 nL·s^-1 in healthy term infants and 0.58 nL·s^-1 in infants with CLDI. After adjustment for tidal breathing parameters, sex, maternal and environmental factors, the differences between healthy term and CLDI infants increased (model II: term infants 0.63 nL·s^-1, CLDI infants 0.52 nL·s^-1, p = 0.024). Values for healthy pre-term infants were between the two other groups (0.58 nL·s^-1).

NO in pre-term infants according to GA compared with healthy term infants

Categorisation of pre-term children according to GA ranges instead of lung disease revealed lower V′_NO in infants with a GA <28 weeks (table 4⇑). No differences between other GA groups were identified using the two models.

Clinical factors associated with NO outcome parameters within the pre-term group

Within the group of pre-term children, there was no association between eNO and any of the potential clinical confounding factors examined. When adjusted for t_E and V’_E, mean V′_NO for all pre-term infants was 0.59 nL·s^-1. V′_NO decreased by 0.02 nL·s^-1 per week decrease of GA (p = 0.02) and per step increase of CRIB score (p = 0.02). An increase in duration of oxygen supplementation of 1 week resulted in a decrease in V′_NO of 0.01 nL·s^-1 (p = 0.06). Post-natal treatment with corticosteroids was associated with a decreased V′_NO (-0.15 nL·s^-1, p = 0.04). Within the CLDI group, V′_NO values in infants with moderate CLDI were 0.03 nL·s^-1 lower than in infants with mild CLDI, although this finding did not reach statistical significance (p = 0.34).

Both eNO and V′_NO were independent of the duration of ventilation, the presence of PDA, pre-natal corticosteroid therapy and surfactant treatment. V′_NO was 0.09 nL·s^-1 lower in the group of pre-term infants with chorioamnionitis compared with those without (p = 0.11). However, when additional adjustments were performed for GA, the presence of chorioamnionitis and all other clinical risk factors disappeared, since all of them were closely associated with low GA.

Choosing the correct statistical model and reference group

The current analysis illustrates how difficult it is to define the correct statistical model and reference group in complex multi-factorial disease processes such as CLDI. Some biometric factors, such as weight and length, are strongly associated with disease but not with the outcome measures (fig. 2⇑). Adjusting for these factors may mask differences between healthy and pre-term infants. Lung functional parameters, such as tidal breathing indices and flow, are strongly related to disease but also demonstrate a relationship with the level of eNO. Pre-term infants with very low birth weight are more likely to have CLDI, thus healthy pre-term infants are more likely to be of more advanced GA than pre-term infants with CLDI. Environmental factors, such as tobacco or caffeine exposure, and maternal factors, such as atopic disease, significantly influence eNO even in healthy offspring 15. All infants were measured at the same PCA of 44–45 weeks and undertook age adjustment for any remaining differences in PCA between the groups. GA, however, was so strongly correlated with the presence and severity of CLDI that it was unavoidable that the reference groups differed from the CLDI infants regarding this feature. This fact resulted in the inability to statistically disentangle the effects of CLDI and prematurity on eNO and V′_NO. Recruitment of equivalent numbers of healthy pre-term infants with the same very low GA as that observed in the CLDI infants is likely to represent an extremely challenging task and was not feasible in the context of the current study.

Clearance of eNO is influenced by changes in lung mechanical function and the present authors were able to adjust for the effects of altered mechanics using a stepwise approach. In the first instance, flow was accounted for on an individual basis by calculating V′_NO. Additionally, further adjustments were made for t_E and V’_E in a stepwise manner and for genetic, maternal and environmental factors (sex, maternal atopic disease, tobacco, caffeine) known to influence NO in the healthy offspring in a more complex multivariate regression model (model II). All model approaches produced fundamentally similar results. This consistency supported the robustness of the present findings.

Effects of severity and known risk factors for CLDI on eNO

Associations between risk factors and V′_NO were investigated using model II. Stratifying disease severity according to recently published guidelines 16, a stepwise decrease in V′_NO was found from pre-term infants without CLDI through mild and on to moderate CLDI disease severity. V′_NO was lowest in infants with low GA, high CRIB scores, long duration of oxygen therapy and in infants who had suffered from chorioamnionitis, although this was not statistically significant for the latter two factors. The relationships observed between these factors are clearly not independent of each other, since infants with low GA usually required longer oxygen therapy, suffered more often from chorioamnionitis and had higher CRIB scores. It was therefore not possible to separate the individual effects of these components in a group of 62 infants. There was no association between V′_NO and CLDI risk factors, such as duration of ventilation, post-natal infections or the presence of PDA. Similarly, treatment with corticosteroids or surfactant pre- or post-natally was not related to V′_NO.

Interpretation of the findings

Following adjustment for breathing pattern, flow and known maternal and environmental factors, V′_NO was found to be decreased in premature infants with CLDI in comparison with healthy term infants. The present results require careful interpretation since there are likely to be interactions between various factors.

The present data do not support the assumption that increased eNO is purely a marker of airway inflammation in CLDI, as is the case in asthmatics. Other groups have found eNO concentration to be elevated in a small group of younger infants at the PCA of 36 weeks, with a decrease observed following corticosteroid treatment 13, 22. Observed variations between these studies and the present findings might reflect methodological or age differences. It is possible that, in this more acute phase, the induction of iNOS during the inflammatory response is more dominant, potentially leading to an increase of eNO. The present findings are more in accordance with those of Baraldi et al. 12, who found eNO to be low in survivors of CLDI at school age.

Studies in baboons suggest that there might be differences in the development of NOS in premature offspring who subsequently develop CLDI. Developmental changes occurring in foetal baboon lungs during the third trimester may increase NOS expression and activity and, therefore, NO production 23. Further work from that group showed that NOS expression was attenuated in a foetal baboon model of CLDI 11. The group speculated that since NO is thought to play an important role in airway and parenchymal function in the immediate post-natal period, the alterations in NOS expression seen in CLDI baboons may contribute to the pathogenesis of the disease. These findings are supported by eNO measurements of newborn pre-term and term infants that demonstrate very low to absent eNO in pre-term infants compared with term infants in the first 6 h of life, suggestive of a more difficult perinatal adaptation at low GAs 24. The present findings in human infants with CLDI are consistent with this hypothesis 11.

Alterations to NO metabolism occurring in addition to developmental differences in NOS also require consideration. Post-natal inflammatory response dominated by persistent neutrophil activity 25 and the presence of oxidative stress, may play a major role in this process since both are known to alter eNO 7, 8. Neutrophil activity and oxidative stress both promote the oxygenation of NO to soluble peroxy-nitrites, nitrites and nitrates. Oxidative metabolites of NO are increased in both bronchoalveolar lavage (BAL) fluids 26 and plasma 27 during the first 28 days of life. However, nothing is known about the persistence of oxidative stress or neutrophilic inflammation during the late post-acute phase of 44 weeks PCA. Data from BAL samples suggest that neutrophil activity decreases following the peak inflammatory response seen in the first month of life 25, 28.

A final possible explanation for the decrease in V′_NO observed in CLDI infants also requires consideration. Airway epithelial remodelling may lead to epithelial dysfunction and therefore an alteration of NO flux across the airway surface.

Conclusions

Human infants suffering from mild-to-moderate CLDI demonstrate no difference in eNO in comparison to healthy term and pre-term infants when variations in flows and t_Es between these groups are not taken into account.

Accommodation for these differences, through the calculation of exhaled NO output, reveals a small but significant decrease in NO output in infants suffering from chronic lung disease of infancy. Furthermore, less mature infants with more severe chronic lung disease of infancy were observed to have a correspondingly lower NO output in the post-acute phase of their illness. These findings indicate that NO metabolism might be involved in chronic lung disease of infancy even in a very late phase when the acute inflammatory processes are thought to have resolved. Whether NO metabolism plays a causative role in the pathophysiology of chronic lung disease of infancy remains unclear. A decrease in NO output may be the result of inborn developmental differences of NOS expression as suggested by Shaul et al. 23, or the result of an altered balance between NO production and increased oxygenation due to persistent neutrophilic inflammation, persistent oxidative stress or airway epithelial dysfunction. The lack of association between post-natal interventions (ventilation, anti-inflammatory drugs) suggests that NO output may be determined very early (e.g. pre-natally or during the early post-natal period). The present observational studies in human infants cannot distinguish whether the lower NO output is related to developmental differences or to persistent inflammation or oxidative stress in these infants. The relationship between low gestational age with higher disease severity and low NO output, however, supports evidence that NO output is related to the degree of prematurity in infants with chronic lung disease of infancy, the most relevant risk factor for the disease. Although healthy premature infants tended to have lower NO output than term infants, in fact, low NO output was mainly seen in infants born at <28 weeks. The present data indicate that longitudinal bronchoalveolar lavage or biopsy studies commencing at birth may provide important information regarding the role of NO metabolism in the pathophysiology of chronic lung disease of infancy. The present findings also suggest that the post-acute phase of chronic lung disease of infancy is probably more important than initially anticipated and is still prone to the disturbance of airway and alveolar lung development 29–31.