Post-natal exposure to air pollution is associated with diminished lung growth during school age. The current authors aimed to determine whether pre-natal exposure to air pollution is associated with lung function changes in the newborn.
In a prospective birth cohort of 241 healthy term-born neonates, tidal breathing, lung volume, ventilation inhomogeneity and exhaled nitric oxide (eNO) were measured during unsedated sleep at age 5 weeks. Maternal exposure to particles with a 50% cut-off aerodynamic diameter of 10 μm (PM10), nitrogen dioxide (NO2) and ozone (O3), and distance to major roads were estimated during pregnancy. The association between these exposures and lung function was assessed using linear regression.
Minute ventilation was higher in infants with higher pre-natal PM10 exposure (24.9 mL·min−1 per μg·m−3 PM10). The eNO was increased in infants with higher pre-natal NO2 exposure (0.98 ppb per μg·m−3 NO2). Post-natal exposure to air pollution did not modify these findings. No association was found for pre-natal exposure to O3 and lung function parameters.
The present results suggest that pre-natal exposure to air pollution might be associated with higher respiratory need and airway inflammation in newborns. Such alterations during early lung development may be important regarding long-term respiratory morbidity.
There is growing evidence that air pollution has adverse effects on lung function and development 1. Both cross-sectional as well as longitudinal studies have clearly shown diminished lung function in children exposed to higher levels of air pollution 1–4. A causative association has been suggested in the observation of decreased age-related lung function decline in adults after reduced exposure to air pollution 5. Moreover, short- and long-term exposure to pollutants has been associated with airway inflammation 6, 7.
Growth and development of the respiratory system take place mainly during the pre-natal and early post-natal periods 8, 9 and adverse effects of pre-natal exposures, such as tobacco smoking of the mother, on lung development are well documented 10. Although air pollution might possibly lead to comparable developmental changes, no epidemiological studies have examined potential associations between pre-natal air pollution and lung functional development and inflammation or oxidative stress in the newborn 8, 11. This early developmental phase is thought to be very important in determining long-term lung growth 9. So-called “tracking” of lung function was found in retrospective 12 and prospective cohort studies 13. Therefore, early changes in lung function may have a considerable impact upon long-term respiratory morbidity and even mortality.
The aim of the present study was, thus, to assess in a prospective birth cohort whether increased maternal exposure to air pollution during pregnancy is associated with changes in tidal breathing, lung volume or airway inflammation, measured during natural sleep in 5-week-old infants.
The current prospective birth cohort study comprised a group of unselected, healthy neonates recruited antenatally since 1999 in the region of Bern, Switzerland 14. Exclusion criteria for the study were pre-term delivery (<37 weeks) and significant perinatal disease, including respiratory distress and later diagnosis of chronic respiratory disease. Potential risk factors (sociodemographic status, smoke exposure and parental atopic disease) were assessed by interviews using standardised questionnaires. The Ethics Committee of the Region of Bern approved the study and written consent was obtained at enrolment.
Lung function measurements were performed in unsedated neonates during quiet natural sleep in the supine position with the head mid-aligned, via an infant mask (Size 1; Homedica AG, Hünenberg, Switzerland), according to the European Respiratory Society (ERS)/American Thoracic Society (ATS) standards of infant lung function testing 15. Flow was measured using an ultrasonic flowmeter (Spiroson®; EcoMedics AG, Duernten, Switzerland).
For analysis, the first 100 regular breaths of tidal breathing during non-rapid eye movement sleep from the total 10-min recording were used, and sighs and 10 breaths before and after a sigh were excluded. From these, mean tidal breathing parameters of flow, volume and flow–volume loop were calculated according to the ERS/ATS standards for infant lung function testing 15. The main outcome parameters were minute ventilation (tidal volume multiplied with respiratory rate) and mean tidal inspiratory and expiratory flow 15.
Lung volume and ventilation inhomogeneity were determined using the multiple-breath washout (MBW) technique by ultrasonic flowmeter as previously described 16. The main outcomes were functional residual capacity (FRC) at airway opening, FRC per body weight, and lung clearance index.
Exhaled nitric oxide (eNO) was measured online with a rapid-response chemiluminescence analyser (CLD 77 AM; EcoMedics AG) in the range of 0–100 ppb. Contamination of eNO by ambient NO was avoided using NO-free air for inspiration. The eNO was measured breath by breath, and mean eNO was calculated over 100 breaths 14. As eNO is flow dependent, eNO was adjusted for minute ventilation in the multivariable analysis and results are presented for both eNO concentration and NO output (eNO concentration multiplied with corresponding expiratory flow) 14.
Air pollution data included daily mean levels of particles with a 50% cut-off aerodynamic diameter of 10 μm (PM10) and nitrogen dioxide (NO2), as well as the daily maximum of mean hourly levels of ozone (O3) for the period from January 1999 to July 2007. Air pollution was measured at the monitoring station of Payerne (part of the Swiss National Air Pollution Monitoring Network), which lies within the study area and reflects temporal variability of air pollutants during the study period.
These regional data were used to calculate the mean exposure to each pollutant for each subject during pregnancy, between date of conception and birth date as well as separately for each trimester. In addition, post-natal exposure to air pollution between birth and date of lung function test was calculated. As a proxy for traffic-related air pollution exposure, the distances from the mother’s home coordinates to the closest major road of at least 6-m width (first class road) and to that of at least 4-m width (second class road) were computed. This was performed analogously to another Swiss cohort study 17. Calculations were performed with a geographic information system (GIS; ArcGIS, version 9; Environmental Systems Research Institute, Redlands, CA, USA). Addresses were geo-coded using the building registry of the Swiss Federal Statistical Office (Neuchâtel), and street information was obtained from the VECTOR25 map of the Swiss Federal Office of Topography (Wabern).
The association between mean exposure levels of each pollutant during pregnancy and during the post-natal period, as well as distance to 4- and 6-m roads and pulmonary function data, was assessed by linear regression analysis. First, univariable regression analysis was performed for each exposure variable and, secondly, a multivariable model was used, in which sex, post-natal age, season of birth, outdoor temperature on the measurement day and maternal smoking during pregnancy were adjusted for. Season of birth was parameterised using a cosine function that assumed a value of 0 at July 1 and a value of 1 at January 1. Sensitivity analyses that included length and weight at study date, year and month of birth, infectious season, exposure to air pollution during the last 2 days before lung function testing, paternal educational status and maternal asthmatic disease were performed.
Between 1999 and 2007, the study enrolled 241 infants, with data from 221 (87%) used for tidal breathing analysis, 205 (81%) for eNO analysis and 181 (72%) for MBW analysis. Reasons for exclusion were insufficient duration of quiet sleep during lung function testing (n = 16), lower respiratory tract infection before the measurement (n = 3), technical problems (n = 1 for tidal breathing, n = 17 for eNO) and strict quality control criteria of MBW (n = 41). Anthropometric and lung function data, as well as air pollution exposure data, are given in table 1⇓, and the distribution of possible risk factors is given in table 2⇓. Daily mean values of PM10 and smoothed temporal trends are shown in figure 1⇓.
The association between pre-natal exposure to air pollution and lung function at 5 weeks of age is given in table 3⇓ for both the univariable and the adjusted model. Mean PM10 exposure during pregnancy was associated with changes in tidal breathing parameters. Each increase of 1 μg·m−3 PM10 was associated with an increase in minute ventilation of 24.9 (95% confidence interval (CI) 9.3–40.5) mL·min−1 (p = 0.002). Similar associations were found between PM10 and respiratory rate (fig. 2⇓) and tidal breathing flows, especially the inspiratory flow (table 3⇓). No association was found between air pollution and lung volume or ventilation inhomogeneity assessed by MBW (table 3⇓). The eNO was associated with mean NO2 exposure (fig. 3⇓), increasing by 0.98 (95% CI 0.45–1.51) ppb per μg·m−3 higher mean pre-natal NO2 exposure (p<0.001; table 3⇓).
The association between post-natal exposure to air pollution and lung function is given in table 4⇓. There was no consistent association in either the univariable or adjusted model for any of the examined pollutants.
Both the associations between PM10 and minute ventilation, as well as between NO2 and eNO, were strongest for exposure during the third trimester of the pregnancy (fig. 4⇓). Values (95% CI) for the association between increase in eNO per μg·m−3 increase in NO2 concentration were 0.09 (-0.08–0.25) ppb for the first trimester, 0.18 (-0.09–0.45) ppb for the second trimester, 0.20 (-0.0–0.41) ppb for the third trimester and 0.06 (-0.23–0.34) ppb for the post-natal time period.
No association was found for distance to 4- or 6-m roads and lung function parameters (table 5⇓). However, a trend towards a stronger association of pre-natal PM10 exposure with minute ventilation was found for newborns of mothers living close to major roads. For example, in infants of mothers who lived within 150 m of a 6-m road, minute ventilation was 39.2 (95% CI 17.2–61.1) mL·min−1 higher per μg·m−3 higher pre-natal PM10 exposure, compared with 12.6 (-10.0–35.3) mL·min−1 higher minute ventilation in infants of mothers who lived further than 150 m away (p for interaction 0.06). A comparably weak trend towards a stronger association was found in infants of smoking mothers.
The observed associations remained stable in the sensitivity analyses, as shown in table 6⇓, including when assessing the relationships without adjustment for outdoor temperature or season and with adjustment for paternal education, month or year of birth and exposure to air pollution on the 2 days before the lung function measurement.
To the current authors’ knowledge, the present study is the first to examine air pollution during pregnancy and subsequent lung growth and development in early life. In the present study, exposure to PM10 during pregnancy was associated with higher respiratory need in newborns as reflected by higher minute ventilation and tidal flows. Higher levels of NO2 during pregnancy were found to be related to elevated eNO, indicative of the induction of inflammatory processes.
In light of the cost- and time-consuming nature of infant lung function testing, which has to date hampered such research in healthy infants, the present study has several methodological strengths. Lung function was measured in a standardised way, based on ERS/ATS standards, and used the latest recommendations for analysis 15, 16. The same equipment, the same masks and the same measurement order were used throughout the whole study period, to ensure comparability. The current authors analysed 100 tidal breaths, giving more robust estimates than the 30 breaths recommended by the standards 15. Since all infants were healthy, breast-fed, of a narrow age range and measured during natural sleep, the contributions of these other possible influencing factors were comparable within the cohort. Causes of lung function changes in older children, such as physical activity, obesity or hypoxia, were negligible in the present study 18. Selection bias was not an issue, as participants were recruited pre-natally without knowledge of exposure to air pollution or lung function after birth. Furthermore, both exposure and outcome variables were objective measures and were independently assessed and analysed. It was possible to adjust for known biological and time-variant confounders, such as smoking during pregnancy, season or outdoor temperature. Further possible confounding factors, such as socio-economic status, number of siblings, post-natal exposure to air pollution and seasonal fluctuations (e.g. month and year of birth, or infectious season) were considered in sensitivity analyses and did not affect the present results.
A limitation of the present study was that individual exposure to air pollution was not sampled, which, while possible after recruitment, is expensive and difficult to do. This is a limit of most studies assessing long-term exposure to air pollution. Therefore, in addition to using mean air pollutant levels during pregnancy as a proxy for temporal variability, the current authors also used distance of homes to major roads as a proxy for spatial variability in exposure to traffic-related air pollutants 17. One limitation of the present study in this regard is the fact that only road proximity was used, without information about traffic density.
In contrast to studies in cooperative subjects, measurements of forced expiratory volumes and flows need sedation in infants and are, thus, not well accepted by parents of healthy subjects.
Although clear associations, robust towards a range of sensitivity analyses, may indicate a true association, causality cannot be proved. Furthermore, multiple comparisons were performed due to the nature of the study design, with different exposures and several outcome parameters. It cannot be totally excluded that positive associations may have occurred due to chance and, thus, the current authors recommend replication of the present results, preferably in a different cohort with varying pollution levels.
Comparison with other studies
Studies in cooperative school children showed a reduction of lung growth upon exposure to particulate matter 1, 3. The present authors found no association between PM10 and lung volume in unsedated sleep, possibly because FRC in infants is dynamically regulated to maintain end-expiratory lung volume above airway closure 19. Thus, the increased minute ventilation in infants with higher pre-natal PM10 exposure may be a compensatory mechanism for lower lung volumes. In line with these observations, no study has so far demonstrated an effect of pre-natal smoke exposure on lung volume in infants 10.
The present study found a clear association between NO2 exposure during pregnancy and post-natal eNO levels. Such a causal association seems plausible, as NO2 is known to induce inflammatory processes 20. Nevertheless it cannot be excluded that NO2 acts as a proxy for a complex mixture of combustion pollutants that originates primarily from vehicular traffic 21. The role of upregulated inducible nitric oxide synthase in this process is unknown, but may provide a link between environmental air pollution and the evolution of asthma.
No association between O3 and lung function was found. Results from other studies are contradictory, with an association between O3 and lung volume and inflammation shown by some groups 1, 6, but not by others, e.g. one study from Switzerland with summer daytime O3 exposure levels comparable to those of the present study 22.
Several studies showed that living closer to major roads has adverse effects on lung function and respiratory symptoms 2, 17. However, there is still an ongoing debate about a possible threshold regarding traffic-related pollution, as well as the boundary between traffic-related exposure and background pollution levels 23. In the present study, in contrast to temporal variation (mean pollutant levels), spatial variation during pregnancy alone (distance to roads) was not associated with lung function changes. In this regard, important factors special to the present study have to be mentioned. First, exposure was calculated during a time period of 9 months (pregnancy) for each subject individually, whereas most other studies dealt with mean annual exposure levels. Secondly, females maintain a mobile lifestyle during pregnancy. As such, exposure at home (usually during times with lower pollution levels) does not necessarily contribute to overall exposure as much as, for example, in the less mobile elderly population. Thirdly, the particular environmental situation in urban Switzerland with high population and road network density leads to homogenous distribution of PM2.5 and PM10 24. Taken together, these points may explain why, for exposure assessment over a limited time period (such as pregnancy) in mobile subjects living in areas with homogenously distributed PM10 levels, temporal variation of regional pollution levels may be more important than small-scale spatial variations of air pollutants. However, the relative contribution of background exposure and road proximity to health effects remains controversial 23.
The changes in lung function seen in the present study infants with increased pre-natal exposure to particulate matter are similar to those observed in premature infants with bronchopulmonary dysplasia 25, in infants of smoking mothers 26 and in animal models of pre-natal nicotine exposure 27. Thus, the current findings indicate considerable impact of air pollution on lung development already present at an early developmental stage, and are suggestive of either increased respiratory need due to increased resistance (smaller airways), decreased compliance (smaller or stiffer airways) and/or factors influencing control of breathing (e.g. hypoxia).
The exact mechanisms for these changes are unknown. Some hypotheses have been proposed in the literature 28. Oxidative stress and inflammation of the airways in the mother after exposure to air pollution may affect the blood–air barrier 29, potentially leading to reduced foetal breathing movements and decreased alveolarisation 30. The effect could also be mediated via systemic inflammation in the mother upon air pollution leading to decreased placental blood flow with reduced transfer of nutrients to the foetus 30. Although entirely unknown, the increasing role thought to be played by nanoparticles could also be involved in either process 31. Furthermore, evidence for decreased birthweight upon higher exposure to air pollution suggests that growth factors may be involved 32, which may also be applicable to lung growth. In the current study cohort, all associations were unaltered by birthweight, suggesting that the effect is even independent of birthweight.
Due to the need for an outcome parameter assessable early after birth, the current authors measured well-defined physiological surrogates for lung growth and development in this age group. Early changes in lung function track into later life 13 and are believed to have a huge impact upon long-term respiratory morbidity, e.g. asthma occurrence 33.
The higher respiratory need, as reflected by increased minute ventilation upon higher pre-natal exposure to particulate matter, might be clinically important, especially for pre-morbid infants with an already reduced breathing capacity or infants that are acutely sick.
From a public health point of view, the present results are particularly crucial in areas with higher outdoor pollution, or considerable indoor pollution from biomass fuels where it is practically impossible for an individual to avoid exposure. Especially in these areas, altered lung function may be one mechanism responsible for the association between PM10 and infant mortality, particularly since a stronger association for respiratory rather than nonrespiratory mortality has been observed after PM10 exposure 34. Therefore, it seems worthwhile to investigate a possible effect of activities during pregnancy associated with exposure to pollution (e.g. cooking at open fireplace) on infant morbidity, because exposure to indoor pollution could be substantially reduced by using cleaner fuels and improved stove constructions for cooking 35.
The present study is the first prospective birth cohort study suggesting a relationship between pre-natal levels of particles with a 50% cut-off aerodynamic diameter of 10 μm and nitrogen dioxide exposure with lung function and inflammation after birth. The current findings involving airway mechanics provide additional evidence to epidemiological studies, since they suggest potential mechanisms behind adverse outcomes of pollution. Influences during the vulnerable phase of pregnancy are known to affect lung development and growth 9, 13 and the evolution of asthma and allergy. If the hypothesis of Barker et al. 12 is correct, these early influences on the respiratory system result in a higher burden of respiratory disease in older people and shortened life expectancy. The present results thus provide further rationale for more stringent measures to reduce air pollution.
This study was supported by Swiss National Foundation grant 3200-B0-112099 to U. Frey and P. Latzin, Swiss Federal Office of Public Health Tobacco Prevention Fund 07.005776 to U. Frey and Swiss National Foundation PROSPER grant 3200-069349 to C.E. Kuehni.
Statement of interest
A statement of interest for P. Latzin can be found at www.erj.ersjournals.com/misc/statements.shtml
The authors thank the study nurses C. Becher, M. Graf and B. Hofer (all Division of Respiratory Medicine, Dept of Paediatrics, Inselspital and University of Bern, Bern, Switzerland) for help with the lung function measurements. They also thank L. Sauteur for her help in the analysis of the measurements and C. Thamrin (both Division of Respiratory Medicine, Dept of Paediatrics, Inselspital and University of Bern) and M. Egger (Institute of Social and Preventive Medicine, University of Bern) for critical appraisal of the manuscript. Furthermore they thank the Amt für Geoinformation (Bern) for providing the geo-data VECTOR25 and R. Weber from the Federal Office for the Environment (Bern) for providing air pollution data of the monitoring station in Payerne (Switzerland).
- Received June 4, 2008.
- Accepted October 21, 2008.
- © ERS Journals Ltd