Prenatal Exposure to Nitrogen Oxides and its Association with Birth Weight in a Cohort of Mexican Newborns from Morelos, Mexico

Background The Child-Mother binomial is potentially susceptible to the toxic effects of pollutants because some chemicals interfere with placental transfer of nutrients, thus affecting fetal development, and create an increased the risk of low birth weight, prematurity and intrauterine growth restriction. Objective To evaluate the impact of prenatal exposure to nitrogen oxides (NOx) on birth weight in a cohort of Mexican newborns. Methodology We included 745 mother-child pair participants of the POSGRAD cohort study. Information on socio-demographic characteristics, obstetric history, health history and environmental exposure during pregnancy were readily available and the newborns’ anthropometric measurements were obtained at delivery. Prenatal NOx exposure assessment was evaluated using a Land-Use Regression predictive models considering local monitoring from 60 sites on the State of Morelos. The association between prenatal exposure to NOx and birth weight was estimated using a multivariate linear regression models. Results The average birth weight was 3217 ± 439 g and the mean of NOx concentration was 21 ppb (Interquartile range, IQR = 6.95 ppb). After adjusting for maternal age and other confounders, a significant birthweight reduction was observed for each IQR of NOx increase (ß = –39.61 g, 95% CI: –77.00; –2.21; p = 0.04). Conclusions Our results provides evidence that prenatal NOx exposure has a negative effect on birth weight, which may influence the growth and future development of the newborn.


Background:
The Child-Mother binomial is potentially susceptible to the toxic effects of pollutants because some chemicals interfere with placental transfer of nutrients, thus affecting fetal development, and create an increased the risk of low birth weight, prematurity and intrauterine growth restriction. Objective: To evaluate the impact of prenatal exposure to nitrogen oxides (NO x ) on birth weight in a cohort of Mexican newborns. Methodology: We included 745 mother-child pair participants of the POSGRAD cohort study. Information on socio-demographic characteristics, obstetric history, health history and environmental exposure during pregnancy were readily available and the newborns' anthropometric measurements were obtained at delivery. Prenatal NO x exposure assessment was evaluated using a Land-Use Regression predictive models considering local monitoring from 60 sites on the State of Morelos. The association between prenatal exposure to NO x and birth weight was estimated using a multivariate linear regression models. Results: The average birth weight was 3217 ± 439 g and the mean of NO x concentration was 21 ppb (Interquartile range, IQR = 6.95 ppb). After adjusting for maternal age and other confounders, a significant birthweight reduction was observed for each IQR of NO x increase (ß = −39.61 g, 95% CI: −77.00; −2.21; p = 0.04).

Conclusions:
Our results provides evidence that prenatal NO x exposure has a negative effect on birth weight, which may influence the growth and future development of the newborn. dioxide (NO 2 ) and nitric oxides (NO x ) on birth weight in a cohort of newborns from Morelos, Mexico.

Study Design and Population
This analysis is based on the POSGRAD (Prenatal Omega-3 Supplementation on child GRowth And Development) cohort study; a large double blind randomized controlled trial (RCT) of prenatal DHA supplementation. Pregnant women (n = 1094) were randomized to receive a daily supplement of 400 mg of DHA or placebo from 18-22 weeks of pregnancy until delivery. A detailed description of the design and methods has been published elsewhere [23]. Briefly, eligible women were 18-35 years of age, at 18-22 weeks of gestation, planned to deliver at the Mexican Institute of Social Security General Hospital in Morelos, Mexico and planned to live in the area for 2 years after delivery. A total of 1094 women were randomly assigned to the clinical trial and for the present report we included 745 binomials, which had the complete information.
The study protocol was approved by the National Institute of Public Health Biosafety, Investigation, and Ethics Committees and for the Emory University Institutional Review Board. Written informed consent was obtained from participating mothers after they received a detailed explanation of the study.

Collection of Information
Prenatal Period: At the first prenatal check-up, trained staff members administered a structured questionnaire to collect information about sociodemographic characteristics, obstetric history, and maternal health, including: consumption of drugs, active smoking, pre-gestational weight and size, consumption of maternal vitamins, and information about the pregnancy evolution. During the prenatal period, a visit to the participants' homes was performed and through a questionnaire and direct home observation, we obtained information regarding the characteristics of the household, indoor (passive smoking, fuel used for cooking and/or heating the home, etc.), and outdoor exposure to pollutants (whether they lived in industrial or agricultural zones, distances from roads with heavy vehicular traffic, etc.).
At birth: Using an ad hoc form, personnel staff obtained information about the newborn characteristics, including: condition at birth (live or stillbirth), gender, type of birth, complications at birth, presence of congenital abnormalities, and gestational age. Birth weight was obtained according to the technique proposed by Lohman [24], using a TANITA "mommy and baby" scale with a precision of ±20 grams.

Exposure Assessment
For each participant, we estimated the individual NO x exposure during pregnancy at home, using a standardized area specific land-use regression (LUR) models. The LUR models were based on measurements of NO 2 , NO and NO x for a continuous period of 15 days in 60 different sites throughout the State of Morelos. Ambient levels of air pollutants were measured with Ogawa and 3M passive samplers. The samplers were positioned outdoors, near the participants' homes (e.g. roofs, light-poles) making sure they were not being blocked by any object that could obstruct the flow of air (e.g. trees, buildings). The samplers were cleaned before use, and transported in sealed ambercolor containers before and after the measurement. After 2 weeks of continuous monitoring, the samples were collected and again placed in resealable bags in ambercolored containers and transported at 5°C at laboratory of the Mexico National Institute of Public Health where, inside a glove box, the pads were extracted and stored in refrigeration until their analysis. As part of quality control, 10% were blanks and duplicates. NO x concentrations were determined at the Harvard School of Public Health, using spectrophotometry [25].
LUR models were developed for each pollutant in each study area to predict air pollution levels at the residences of the cohort participants using information about: traffic variables (type of roads and highways avenues), weather data (precipitation, wind speed, and temperature), geography (elevation and coordinates), population density and land-use variables (household, industrial, commercial, and services) [26] obtained from Geographic Information Systems (ArcGIS 9.1) [27]. Air pollution measurements were performed in 2009, but the relevant exposure window (second or third trimester of pregnancy) for development of birth outcomes extends further back in time. We therefore extrapolated air-pollution concentrations predicted by the LUR models around 2005-2007. After finding the best predictive LUR models, these were validated using goodness and fit tests (evaluation of residuals, influence points, etc.). Additionally, we evaluate the correlation between the values predicted by the LUR models and those obtained from the monitoring fixed stations.

Statistical Analysis
To evaluate the quality and consistency of the data, an exploratory and univariate analysis was performed, and measures of central tendency and frequency were estimated for each of the study variables. A bivariate and multiple linear regression models were run to evaluate the impact of prenatal exposure of air pollutants (NO, NO 2 and NO x ) on the weight of newborns. As potential confounders, we evaluated: parity, maternal passive or active smoking during pregnancy, pre-gestational maternal anthropometric characteristics, intervention group among others. All of them remained in the final model except those who did not change the crude association between NO x with birthweight. All of the statistical analyses were performed using Stata 13 statistical analysis software for Windows [28]. Table 1 shows selected maternal and infant characteristics. Maternal mean age was 26.3 ± 4.7 years, 39.8% of women had 7 to 12 years of education, which is consistent with the national average. 63.5% were multiparous and most of them used vitamin supplements (96.78%) and were non-smokers (98.26%). Since our exclusion criteria, the mean weight at birth was 3,217 ± 440 g and gestational age mean was 39.1 ± 1.7 weeks. In terms of age, parity, occupation, and education, the women included were not significantly different than those were excluded due to the lack of information needed to estimate exposure. Table 2 shows the pollutant concentrations estimated and weather variables for the study period. The median of NO, NO 2 and NO x were 2.01 ppb, 16.5 ppb and 21.04 ppb, respectively, and the correlation between the measurements and the duplicates was .9985 (p = 0.0015) for NO x and 0.9211 (p = 0.0789) for NO 2 . The values of the blanks used during the monitoring campaign ranged from 0 to 0.2. The average wind speed was 2.51 m/sec and the temperature was 20.3°C during the study period. The correlation between predicted and observed values for each of the pollutants was statistically significant (p < 0.01), and were 0.91 for NO, 0.79 for NO 2 , and 0.94 for NO x (data not shown). Table 3 presents the results from the evaluation of the association between prenatal exposure to NO x and birth weight. After adjusting for mother's age, height and passive smoking, gestational age, and gender of child, a significant birth weight decrease was observed by each increment in the interquartile range of NO x (ß = −39.61, 95% CI −77.0; −2.21 g; p = 0.04) and NO (ß = −42.5, 95% CI −82.73; −2.18 g; p = 0.04). No association between prenatal NO 2 and birth weight was observed.

Discussion
The results from the present study suggest that prenatal exposure to NO 2 and NO x , as estimated by land-use regression model, significantly decreases the birth weight of newborns residing in Morelos, Mexico. To our knowledge, this is the first prospective study performed in Mexico that analyzes the effects of prenatal exposure to nitrogen oxides (NO, NO 2 and NO x ) on birth weight using this methodology to evaluate the exposure to air pollutants. This approach provides stronger results given that most of previous studies evaluated NO x exposure according to data from fixed monitoring stations. Previous studies have reported an association between prenatal NO x exposure per trimester or throughout the pregnancy and adverse effects on birth, including birth weight and/or low fetal weight and some of those studies estimated exposure using land-use regression and particulate matter variants, while others have used other types of dispersion models and local monitoring systems [29][30][31][32][33][34][35][36].
As part of INMA, a multicentric cohort study, Aguilera et al. found a decrease in fetal weight of 74.7 g per increase in interquartile range of NO 2 (IQR = 12 μg/m 3 ) adjusting by NO x exposure of each trimester. That association was observed when the models had been adjusted for the three trimesters of pregnancy and for women who spent over 2 hours/day in non-residential outdoor areas. Nonetheless, when analyzing this data individually, no association was found between exposure during the entire pregnancy nor per trimester [35]. In addition, within the same cohort (INMA) but in Catalonia, Spain, researchers evaluated the effect of NO 2 exposure on fetal weight during different weeks using ultrasound measurements from a sample of 562 women. When limiting the analysis to women who spent over 2 hours/day in non-residential outdoor areas (n = 255), a statistically significant weight decrease of −5.5 g was found in week 32 of gestation, and a similarly significant decrease (4.78 g) was found between weeks 20 and 32, for each increase in the IQR (13.23 μg/m 3 ) [34]; however, this was not the case for the entire sample. Meanwhile, a study in California, United States, to evaluate the effect of NO, NO 2 , and NO x on the risk of low birth weight at term births, using a LUR exposure model and with and without seasonal adjustment [36]. Researchers found a statistically significant increase of 5% and 7% in the risk of low birth weight for NO and NO x , respectively. Nevertheless, the data was obtained from an electronic database of birth certificates.
Although the mechanism through which prenatal exposure to NO x affects birth weight is unclear, the association between exposure to these pollutants and decreased birth weight can be explained by some of the proposed mechanisms that involve placental circulation. NO x may affect birth weight because NO x can promote blood coagulation and viscosity [30], pulmonary and placental inflammation, and endothelial and vascular changes in the placenta, which can decrease uteroplacental blood flow and inhibit the oxygen and nutrients transfer [37]. As oxidants, these compounds increase the lipid peroxidation in both the maternal and fetal tissue and also stimulate the formation of methemoglobin, which would likely lead to hypoxia and hypoxemia [32]. In addition, the resultant systemic inflammation can eventually trigger sub-optimal placentation and increase the mother's susceptibility to infections [38].
This study, however, has some limitations that should be considered when interpreting the results. First, the number of sites monitored and the duration of the measurements NO x limits the possible variations that could occur in the concentrations of the pollutants over the study period, assuming that the air concentrations had the same behavior throughout during the entire time of pregnancy. However, the use of a LUR model to evaluate exposure strengthened the findings and made it possible to detect small-scale variations, thereby limiting classification errors in the assignment of exposure (Figure 1). It is also important to note that the correlation between the concentrations estimated by the models for the entire study period and the values obtained from the monitoring sites during the same time ranged from 0.79 to 0.95, which is statistically significant, as this information implies a correlative relationship. Furthermore, the generation of the different LUR models included weather variables for the entire monitoring period, information about variables that do not change over time, and individual characteristics of the participants.
Second, we did not obtain detailed information about the indoor and outdoor activities of each participants. Instead, we presumed that exposure was primarily associated with the amount of time spent in outdoors. However, for both cases, the measurement error was non-differential and the observed association was underestimated.  Finally, the possibility that the results were a consequence of poor control of confounders is unlikely because the design of the study excluded women with high-risk pregnancies and/or pre-existing illness and the models were adjusted for variables that were considered to have both a possible effect on birth weight and a relationship to NO x exposure.
In summary, our results provide valuable information about the adverse health effect of prenatal exposure to NO x , especially on birth weight. These results could have significant public health implications given the role of birth weight as a determinant of childhood mortality, and more recently as a determinant of chronic illnesses during adulthood. However, it is important to continue with the investigations that allow for the identification of critical windows of exposure and that strengthen the evidence related to these associations.

Funding Information
Supported by the National Council of Sciences and Technology (CONACYT) grant 87721 and the NIH Eunice Kennedy Shriver National Institute of Child Health and Human Development grants R01HD058818.