Nitrogen dioxide exposure during pregnancy and risk of spontaneous abortion: a case-control study in China

Abstract Background Evidence on the relationship between prenatal exposure to NO2 and CO and spontaneous abortion (SAB) is insufficient. We investigated whether there is an association between maternal exposure to nitrogen dioxide (NO2) and carbon monoxide (CO) before and during pregnancy and SAB. Methods We conducted a case-control study using medical records of 2445 pregnant women who admitted for abortion prior to 20 weeks of gestational age from January 2014 to December 2019 at a tertiary-care hospital in Shanghai, China. Of the 2445 participants, 1075 were SAB cases and 1370 were healthy controls (underwent elective abortions). Maternal exposure to NO2 and CO before and during pregnancy was estimated using daily air pollution concentration data. Multivariable logistic regression models were constructed to quantify the relationships between maternal exposure to NO2 or CO and the risk of SAB while controlling for potential confounders. Results NO2 exposure levels during pregnancy were significantly higher in SAB cases than in healthy controls (42.26 vs. 40.67, p < .01). NO2 exposure during pregnancy was positively associated with the risk of SAB. An interquartile range (16 μg/m3) increase in NO2 exposure was associated with 68% increase in the odds of SAB (OR = 1.68, 95% CI, 1.28, 2.21). Analyses of associations by quartile of NO2 exposure showed that elevated NO2 exposure during pregnancy was associated with increased odds of SAB in linear dose–response manners. Compared with the lowest quartile of NO2 exposure, the odds of SAB in the fourth quartile of NO2 exposure increased 61% (OR = 1.61, 95% CI, 1.03–2.53). No associations of CO exposure with SAB risk were observed. Conclusions Our study suggested that exposure to NO2 during early pregnancy was associated with increased risk of SAB. Further studies are needed to confirm our results and explore the potential biological mechanism underlying these associations.


Introduction
Ambient air pollution is the most serious environmental threat and represents a major mortality risk factor worldwide [1]. According to a recent global diseases burden estimates, exposure to ambient air pollution from particulate matter caused about 4.2 million deaths and 103.1 million disability-adjusted life-years (DALYs) in 2015 [2].
The cardiovascular and respiratory effects of air pollution have been well documented [3,4]. Recently, its reproductive and developmental toxicity has raised concern. Numerous epidemiological studies have reported signi cant associations between prenatal exposure to air pollution and increased risk of adverse pregnancy outcomes [5]. Those adverse pregnancy outcomes included low birth weight (LBW), fetal growth restriction (FGR), preterm birth (PTB) and birth defects [6][7][8].
Spontaneous abortion (SAB) or miscarriage, a common and serve complication of pregnancy, de ned as the spontaneous loss of a pregnancy before 20 weeks of pregnancy [9]. At least 25% of all women experience one or more sporadic miscarriages [10]. Detrimental effects of air pollution on SAB have been previously suggested with exposure to particulate matter (PM) and some gaseous pollutants [11. Most of the previous studies reported strong associations of high PM [12][13][14][15], sulfur dioxide (SO 2 ) [16,17] and ozone (O 3 ) [12,15,18] exposure with increased risk of SAB. However, evidence regarding nitrogen dioxide (NO 2 ) and carbon monoxide (CO) are insu cient and less inclusive, suggesting more studies are warranted to better understand their associations with SAB risk [11].
In this study, using clinal records of abortions in a tertiary hospital in Shanghai, China, we conducted a case-control study to examine whether maternal exposure to NO 2 and CO before and during pregnancy was associated with increased risk of SAB.

Study Design and Subjects
A retrospective review of the medical records of abortions was conducted at Shanghai First Maternity and Infant Hospital, which is a tertiary-care hospital serving approximately 45,000 inpatients per year in Shanghai, China. The electronic medical record system of our hospital was used for data collection. From January 2014 to December 2019, a total of 2445 singleton pregnant women, aged between 14 to 48 years, admitted for abortion prior to 20 weeks of gestational age (GA). Of the 2445 pregnant women, 1075 were diagnosed as having missed abortion and were enrolled as SAB cases. While 1370 women with normal pregnancies who requested induced abortion due to unplanned or unwanted pregnancy were enrolled as healthy controls.
Maternal demographic characteristics including maternal age (MA), menstrual cycles, weight and height, the use of assisted reproduction technology (ART) and reproductive history were collected. GA was calculated based on the last menstrual period (LMP) of pregnant women and con rmed by ultrasound. The study protocol was approved and monitored by the medical Ethics Committee of the Shanghai First Maternity and Infant Hospital (NO. KS2008). Since all data were from medical record and used anonymously, there was no informed consent.

Air Pollution Exposure Assessment
Daily (24-h) NO 2 and CO concentration data for each of the 16 administrative districts in Shanghai city, from May 1, 2013 to December 30, 2019, were obtained from the database of the Shanghai Environmental Monitoring Center (SEMC). The daily concentrations of NO 2 and CO in each district were averaged from the available data of all xed-site monitoring stations. Ambient NO 2 and CO measurements in each monitoring stations relied on a chemiluminescence detection method (API 200e, Thermo 42i) and a gas lter correlation method (API 300e, Thermo 48i), respectively. All ambient measurements of NO 2 and CO were operated under the China National Quality Control Automated methods [(GB3095-2012) and (HJ/T 193-2005)] for ambient air quality monitoring. In order to adjust for the potential confounding effects of weather on SAB, we also obtained daily 24-h mean temperature and relative humidity from the database of the Shanghai Meteorological Bureau.
To explore the critical exposure time windows of NO 2 and CO on SAB, we examined four exposure time windows for each subject (Phases 1-4). The four exposure windows were de ned as follows: Phase 1, from the rst day of LMP to the date of abortion; Phase 2, 30 days before the rst day of the LMP; Phase 3, 60 days before the rst day of the LMP; Phase 4, 90 days before the rst day of the LMP. The phaseaveraged values of daily concentrations of NO 2 and CO during various phases in the district where the subject's residence is located were calculated.

Statistical Analyses
In descriptive analysis, the demographic characteristics of all participants were shown as mean ± standard deviation (SD) or percent (%). Parametric t-tests and χ 2 tests were used to compared the difference in demographic characteristics between SAB cases and healthy controls. As NO 2 and CO exposure levels were not normally distributed, medians (25th -75th percentile) were presented to characterize their distribution in descriptive analysis. Mann-Whitney U-test was used to examine the potential differences in NO 2 and CO exposure between SAB cases and healthy controls.
Multivariate logistic regression models were used to examine the association between SAB risk and exposure to NO 2 and CO with adjustment for potential confounders. According to the data accessibility and literature studies [15,19], we denoted the following factors to be covariates: GA, MA, body mass index (BMI), maternal parity, the use of ART, temperature and relative humidity. We rst conducted single pollutant model with only NO 2 or CO in the multivariate logistic regression models. Then we conducted two-pollutants model to examine the joint effects of NO 2 and CO exposure on SAB. We reported the odd ratio (OR) and 95% con dence interval (CI) in association with an interquartile range (IQR) increase in NO 2 or CO concentrations.

Page 5/15
In order to explore the potential dose-response relationships between NO 2 exposure and risk of SAB, the distribution of NO 2 exposure levels was divided into quartiles and an OR and 95% CI was calculated for each higher quartile compared with the lowest quartile using multivariate logistic regression analysis. All statistical analysis was performed by using SPSS16.0 software (SPSS Inc., Chicago, IL, USA) and a twosided p < 0.05 was considered statistically signi cant.

Result
Population Characteristics   the correlation coe cients ranging from 0.633 to 0. 680 (Fig. 1). Analyses of potential differences in NO 2 and CO exposure levels during each exposure window between SAB cases and health controls were conducted using the Mann-Whitney U-test. As shown in Table 2  a Adjusted for GA, MA, BMI, maternal parity, ART, temperature and relative humidity.
Phase 1: form the last menstrual period to the date of missed abortion, Phase 2: 30 days before pregnancy, Phase 3: 60 days before pregnancy, Phase 4: 90 days before pregnancy. **P < 0.01.
Given NO 2 exposure during phase 1 was found to be signi cantly associated with increased risk of SAB, we examined the dose-response associations between NO 2 exposure during phase 1 and the risk of SAB.
As shown in Fig. 2, higher NO 2 exposure was associated with increased odds of SAB in linear doseresponse manners.
In both single-pollutant model, compared with the lowest quartile of NO 2 exposure, being in the fourth quartile of NO 2 exposure was signi cantly associated with increased odds of having SAB (OR = 1.54, 95% CI, 1.01-2.34). Similar results were observed in two-pollutants model. Compared with the lowest quartile of NO 2 exposure, the odds of SAB in the fourth quartile of NO 2 exposure increased 60% (OR = 1.60, 95% CI, 1.03-2.49). The trend tests for the above dose-response relationships were statistically signi cant (single pollutant model, p-trend = 0.053; Two-pollutants model, p-trend = 0.038).

Discussion
SAB is one of the most frequent traumatic life events a woman may encounter. It is associated with various negative psychological outcomes including depression, grief, anxiety and marital con ict [20]. The causes of SAB are likely to be multifactorial [9,21,22]. Along with genetic and socioeconomic factors, environmental pollution, including air pollution, may possibly play a role in the development of SAB.
In this study, on the basis of a case-control study design, we investigated the adverse effects of NO 2 and CO exposure on SAB. We found that exposure to NO 2 during pregnancy was associated with increased risk of SAB. Moreover, we observed signi cantly linear dose-response relationships between NO 2 exposure and SAB risk. However, we did not nd any association between CO exposure and SAB. Our ndings indicate that pregnant women who exposed to higher levels of NO 2 during pregnancy might be at higher risk for SAB.
Some studies have investigated the associations between maternal exposure to NO 2 and SAB. Most of those reported studies used xed air monitoring station data to estimate individual NO 2 exposure and reported signi cant positive associations between NO 2 exposure and SAB risk [18,19,23,24]. Our ndings were consistent with the results of those previous studies. Moreover, one prospective cohort study found that pregnant women living within 50 m of a road with higher annual average daily tra c was statistically signi cantly associated with increased risk of SAB [25]. The results of this prospective cohort study also supported our ndings since vehicle emissions is a major source of ambient NO 2 .
For air pollutant of CO, only three studies examined the association of maternal exposure to CO with the risk of SAB. In agreement with ndings from this presented study, both a time-series study and a prospective cohort study did not nd any association between SAB risk and CO exposure [15,26]. However, a recent study failed to support our ndings. Zhang and colleagues examined the records of 255,668 pregnant women from 2009 to 2017 in Beijing, China, and quanti ed the link between CO exposure and SAB risk. They found long-term exposure to CO before pregnancy was associated with signi cant increased risk of SAB [17]. Inconsistent results may mainly be ascribed to different study designs and exposure ranges.
The mechanisms of action of NO 2 exposure on SAB are not fully understood. Some studies demonstrated that exposure to NO 2 during pregnancy could induce structural or chromosomal anomalies, which are relevant for SAB [27]. Moreover, some previous studies have shown that NO 2 exposure could induced oxidative stress and in ammatory responses [28][29][30], which have been hypothesized to play a role in the development of SAB [31]. Some limitations should be acknowledged. Firstly, we estimated maternal exposure to NO 2 and CO based on xed-site monitoring station data and ignored the spatial distribution of those two pollutants, which may cause exposure misclassi cation. Secondly, due to the data inaccessibility, we were unable to collect information about some potential risk factors for SAB and thus could not rule out the role of those unmeasured confounders. Thirdly, this is only an exploratory study, we were unable to examine the biological pathways underlying the association between maternal exposure to NO 2 and increased risk of SAB.
In summary, our ndings suggest that maternal exposure to NO 2 during early pregnancy is associated with increased risk of SAB. Further studies are needed to con rm our nding and to explore the biologic mechanisms underlying this association. Plots of the correlations between NO2 exposure levels and CO exposure levels. Scatterplots depict the correlation of NO2 exposure levels (Y axis) and CO exposure levels (x axis), with the spearman correlation coe cient (r) provided. All spearman correlation coe cients were signi cant at p<0.01. Phase 1: form the last menstrual period to the date of missed abortion, Phase 2: 30 days before pregnancy, Phase 3: 60 days before pregnancy, Phase 4: 90 days before pregnancy.