Prenatal exposure to PM2.5 and Congenital Heart

Science of the Total Environment 655 (2019) 880–886

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Prenatal exposure to PM2.5 and Congenital Heart Diseases in Taiwan☆ Ching-chun Huang a, Bing-yu Chen b,c, Shih-chun Pan d, Yi-lwun Ho e, Yue Leon Guo a,c,d,⁎ a

Department of Environmental and Occupational Medicine, National Taiwan University College of Medicine and Hospital, Taipei, Taiwan Department of Medical Research and Development, Chang Gung Memorial Hospital, Keelung, Taiwan c National Institute of Environmental Health Sciences, National Health Research Institutes, Miaoli, Taiwan d Institute of Occupational Medicine and Industrial Hygiene, National Taiwan University College of Public Health, Taipei, Taiwan e Division of Cardiology, Department of Internal Medicine, National Taiwan University College of Medicine and Hospital, Taipei, Taiwan b

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Prenatal exposure to air pollution increases the risk of congenital heart diseases (CHDs). • Studies on the effects of PM2.5 were limited at an exposure range higher than the WHO air quality guidelines. • PM2.5 exposure during weeks 3–8 of pregnancy increased the risks of overall and some subtypes of CHDs occurrence in Taiwan. • Results provide crucial information about the effects of PM2.5 on CHDs applicable to the majority of the global population.

a r t i c l e

i n f o

Article history: Received 9 October 2018 Received in revised form 12 November 2018 Accepted 19 November 2018 Available online 20 November 2018 Editor: Lidia Morawska Keywords: Pregnancy Air pollution PM2.5 Congenital Heart Disease Kriging Prenatal exposure

a b s t r a c t Gestational exposure to ambient air pollution has been associated with Congenital Heart Diseases (CHDs). However, only a few studies, with inconsistent results, have investigated the effects of PM2.5 exposure during early pregnancy. This study aims to evaluate the association between prenatal exposure to PM2.5 and CHDs occurrence. We selected 782 births reported to have CHDs between 2007 and 2014 from the Taiwanese Birth Registry and randomly selected 4692 controls without any birth defects using a population-based case-control design. Data of exposure to ambient air pollutants, mainly PM2.5, PM10, CO, SO2, NO2, and O3 during weeks 3–8 of pregnancy were retrieved from air quality monitoring stations and interpolated to every township using ordinary kriging. We applied unconditional logistic regression models adjusted for potential confounders to evaluate the associations. The results revealed a positive correlation between increased PM2.5 exposure (adjusted odds ratio [aOR] = 1.21, 95% confidence interval [CI] = 1.03–1.42, per interquartile range change = 13.4 μg/m3) during early pregnancy and overall CHDs occurrence. Furthermore, we found that atrial septal defect (aOR = 1.43, 95% CI = 1.01–2.02), endocardial cushion defect (aOR = 2.37, 95% CI = 1.01–5.58), and pulmonary artery and valve stenosis (aOR = 1.71, 95% CI = 1.06–2.78) were significantly associated with PM2.5 exposures. No similar effects were observed for the other air pollutants. This study has demonstrated some positive associations between increased PM2.5 exposure during the critical period of cardiac embryogenesis and certain CHDs occurrence. © 2018 Elsevier B.V. All rights reserved.

☆ Conflicts of interest: None declared. ⁎ Corresponding author at: Rm. 341, 3F., No.17, Xuzhou Rd., Zhongzheng Dist., Taipei City 10005, Taiwan. E-mail address: [email protected] (Y.L. Guo).

https://doi.org/10.1016/j.scitotenv.2018.11.284 0048-9697/© 2018 Elsevier B.V. All rights reserved.

1. Introduction Congenital Heart Diseases (CHDs) are among the most common types of birth defect and account for nearly 30% of all infant deaths at

C. Huang et al. / Science of the Total Environment 655 (2019) 880–886

birth (Dolk and Vrijheid, 2003; Rosamond et al., 2007). Human embryonic development of cardiovascular system occurs early during pregnancy (Dhanantwari et al., 2009), but the precise etiology of CHDs remains incompletely elucidated. Only one-fifth of all CHDs can be attributed to chromosomal anomalies, Mendelian syndromes, or nonsyndromal single-gene disorders, while the remainders are proposed to follow a multifactorial gene-environment inheritance model (Ferencz et al., 1993; Gorini et al., 2014; Vrijheid et al., 2011). Epidemiological evidence has suggested that prenatal exposure to ambient air pollution can adversely affect fetal cardiac development. Since the first publication successfully established the association between exposure to high concentrations of carbon monoxide (CO) during the second month of pregnancy and ventricular septal defects (VSD) in 2002 (Ritz et al., 2002), an increasing number of air pollution studies have used CHD occurrence as the primary health outcome. Gestational exposure to sulfur dioxide (SO2) was reported to increase the risk of VSD (Gilboa et al., 2005), tetralogy of Fallot (ToF; Dolk et al., 2010), and aortic artery and valve defects (Hansen et al., 2009). Nitrogen dioxide (NO2), as well as nitric oxide, was demonstrated to have

881

deleterious effect on the occurrence of ToF during early pregnancy (Dadvand et al., 2011a; Dolk et al., 2010). Ozone (O3) exposure during weeks 3–8 of pregnancy was also found to correlate with pulmonary artery and valve defect (Hansen et al., 2009). Although a recent meta-analysis reported a positive correlation between particulate matter (PM) b 10 μm (PM10) and atrial septal defects (ASD) (Vrijheid et al., 2011), only a few studies, with inconsistent results, have focused on PM b 2.5 μm (PM2.5) (Padula et al., 2013; Schembari et al., 2014; Stingone et al., 2014; Zhang et al., 2016). Moreover, most of these investigations were conducted in areas where median exposure levels of PM2.5 were under 25 μg/m3, 24-hour limit suggested by the World Health Organization (WHO) Air Quality Guidelines (24). According to a recent WHO estimate, in 2016, approximately 91% of the global population inhabited areas where the Air Quality Guidelines were not met (World Health Organization, 2016). Additionally, the detrimental impact was larger in low- and middle-income than that in high-income countries. Whether the presence of CHDs and types of anomalies involved is related to gestational exposure to ambient PM2.5 at an exposure range applicable to most of the inhabitants

Fig. 1. Geographic locations of air monitoring stations in Taiwan. Shaded areas represent the townships included in the study.

882


worldwide remains not fully explored. Hence, the study aims to evaluate the association between prenatal exposure to PM2.5 and CHD occurrence in Taiwan. 2. Material and methods 2.1. Study population We conducted a population-based case-control study using the Taiwanese Birth Registry database. Each live birth, stillbirth with ≥500 g of birth weight or ≥20 weeks of gestational age is compulsorily reported to the Health Promotion Administration within 7 days of delivery. Previous validation study of the Taiwanese Birth Registry database had revealed a high validity (sensitivity and specificity were 92.8–96.9% and 99.6–99.7%, respectively) and reliability (Cohen's kappa measure of agreement was 0.92–0.96), and a very low percentage of missing information (1.6%) (Lin et al., 2004). A total of 1,612,976 births were reported between 2007 and 2014. Birth records with incomplete residential information (n = 518), records of multiparous births (n = 43,725), and records of births in families residing in mountainous areas (n = 193,917) or at locations beyond 20 km from the nearest air monitoring station (n = 960) were excluded. In addition, birth records in the east coastal region, which has relatively few monitoring stations, were also excluded (n = 60,441) to ensure thorough cross-validation during the interpolation processes (Fig. 1). Finally, in total, 1,313,415 eligible births were included for further analysis. The study was reviewed and approved by the Institutional Review Board of the National Taiwan University Hospital. 2.2. Definition of cases and controls We classified all the eligible births using the European surveillance of congenital anomalies malformation coding guides and excluded minor anomalies in accordance with the exclusion list (EUROCAT, 2016). Births with chromosomal abnormalities were also excluded. We then selected 782 births with major CHDs (The International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, ICD-10, Q chapter, Q20–Q26) as cases, as well as its six major

subtypes namely VSD (Q210), ASD (Q211), ToF (Q213), endocardial cushion defect (ECD; Q212), transposition of the great arteries (TGA, Q203), and pulmonary artery and valve stenosis (PVAS; Q221). All cardiac defects were confirmed by autopsy, echocardiogram, or cardiac catheterization. Potential controls were defined as births without any congenital anomalies (n = 1,193,458). For each CHD case, we conducted a 1:6 random sampling by matching birth year. 2.3. Exposure to air pollutants Complete air quality-related data from 2006 to 2014 were retrieved from 66 fixed-site stations of Air Quality Monitoring Networks in Taiwan (Taiwan EPA, 2006–2014). Each station routinely monitors hourly criteria air pollutants, including PM2.5, PM10, NO2, SO2, O3, and CO. We calculated the daily average of each air pollutant at every station when monitoring data were available for 18 h or longer. When the measurement was available for b18 h, the data was estimated by multiplying the daily ratio by the average concentration recorded at the other stations on that day, similar to the approach proposed by Hoek et al. (2001). The proportion of daily missing air quality data for PM10, PM2.5, SO2, O3, CO, and NO2 during 2006–2014 (3287 days in total) were 0.75%, 2.13%, 0.29%, 2.74%, 1.53%, and 0.39%, respectively. Thereafter, the daily average of each air pollutant at every station was used for further spatial interpolation. 2.4. Spatial interpolation and cross-validation We used the ArcGIS Desktop (ESRI Inc., Redlands, CA, version 9.3) software with its geostatistical analyst tool, ordinary kriging, to interpolate the daily average concentrations of PM2.5, PM10, NO2, SO2, O3, and CO to the level of the township. To obtain the optional semivariogram parameters for ordinary kriging model of each air pollutant, we utilized three frequently applied covariance models (spherical, exponential, and Gaussian) coupled with data transformation (original vs. logtransform), and chose the best-fitting model based on the results of cross-validation (Huang et al., 2015). We also geocoded each mother's residential address during pregnancy to the level of the township and

Table 1 Findings from previous studies investigating the associations between prenatal exposure to PM2.5 and the occurrence of CHDs. Study

Setting

Design

Exposure assessment

Exposure range of PM

Main findings of PM

Padula et al., 2013 (Padula et al., 2013)

California (USA), 1997–2006

Case-control, 822: 849

CO, NO, NO2, PM10, PM2.5, and 8-hr-max O3 of nearby stations during the first 2 months of pregnancy

p25–p75: PM10, 25.3–44.1 μg/m3 PM2.5, 10.9–26.1 μg/m3

Agay-Shay et al., 2013 (Agay-Shay et al., 2013)

Israeli, 2000–2006

Cohort, 135,527 births

Schembari et al., 2014 (Schembari et al., 2014)

Barcelona (Spain), 1994–2006


Inverse weighting modeling of CO, NO2, O3, SO2, PM10, and PM2.5 during weeks 3–8 of pregnancy Land use regression models of NOx, NO2, PM10, PMcoarse, PM2.5, and PM2.5 absorbance during weeks 3–8 of pregnancy

Stingone et al., 2014 (Stingone et al., 2014)

Nine states (USA), 1997–2006


Median (IQR): PM10, 43 (27.4) μg/m3 PM2.5, 22.1 (13.4) μg/m3 Median (IQR): PM10, 55.7 (12.2) μg/m3 PMcoarse, 21.7 (3.6) μg/m3 PM2.5, 16.6 (2.6) μg/m3 p10–p90: PM10, 14.9–40.6 μg/m3 PM2.5, 7.8–19.7 μg/m3

PM10 increased the risk of pulmonary valve stenosis and perimembranous VSD. PM2.5 was associated with TGA and inversely associated with perimembranous VSD and secundum ASD. PM10 was associated with overall CHDs. PM2.5 was inversely associated with isolated patent ductus arteriosus.

Zhang et al., 2016 (Zhang et al., 2016)

Wuhan (China), 2011–2013

Cohort, 105,988 births

CO, NO2, O3, SO2, PM10, and PM2.5 concentrations at the nearest air monitor during the first 7 weeks of pregnancy PM10 and PM2.5 concentrations at the nearest air monitor during the first 3 months of pregnancy

Median (IQR): PM10, 101.7 (74.8) μg/m3 PM2.5, 65.6 (47.2) μg/m3

Negative association between PM10, PMcoarse, PM2.5, and VSD. No association between overall CHDs and any PM.

PM2.5 was positively associated with hypoplastic left heart syndrome but inversely associated with ASD.

PM2.5 increased risk of overall CHDs, particularly VSD. No similar effect was observed for PM10.

Abbreviations: ASD, atrial septal defect; CHDs, Congenital Heart Diseases; IQR, interquartile range; PM, particulate matter; TGA, transposition of the great arteries; VSD, ventricular septal defect.


assigned to each case an average exposure level to every air pollutant during weeks 3–8 of pregnancy when organogenesis occurs. The concentration of PM2.5–10 (PM sized between 2.5 and 10 μg/m3) was estimated by subtracting the concentration of PM2.5 from that of PM10. The cross-validation of the daily kriging model was performed uisng leave-one-out method, which took one data point out of the fitting and then predicted its concentration using data from the remaining monitoring stations. Difference between the actual and predicted value was claculated using five indices, including prediction error (PE), standardized prediction error (SPE, averaging the standardized prediction errors), root mean squared error (RMS), standardized root mean squared error (RMSS, dividing RMS by the average of the prediction standard errors), and coefficient of determination (R-squared). For a prediction to be unbiased, both PE and SPE should be near zero. Additionaly, RMSS represents the variability of the prediciton and should be close to one, meaning that the average prediction standard error is close to RMS (Kevin et al., 2003; Szpiro et al., 2007).

2.5. Statistical analysis The dependent variables were overall CHDs and its 6 major subtypes, and the principal independent variables were the estimated exposure levels of 7 criteria air pollutants during weeks 3–8 of pregnancy (treated as continuous variables). Potential confounders available from routine birth registration and considered in the study were infant sex, maternal age, the season of conception, maternal diabetes and hypertension, maternal smoking and alcohol consumption during pregnancy, and birth year. The 2012 nationwide statistics of individual income tax were used to determine the socioeconomic status of the residential township. We applied unconditional logistic regression models to evaluate the association between gestational exposure to each air pollutant and the occurrence of CHDs. The inclusion of covariates was based on previous studies or followed a 10% change-in-estimate principle assessed by using the goodness of fit with likelihood ratio tests. All analyses were performed using software R (version 3.3.3, R Core Team, 2017, Vienna, Austria), and the statistical significance was set at P b 0.05 based on a two-tailed calculation.

883

3. Results A total of 782 cases and 4692 controls were enrolled for final analysis. The most common CHD subtype was VSD (n = 218, 27.9%), followed by ASD (n = 94, 18.8%), ToF (n = 123, 15.7%), PAVS (n = 68, 8.7%), TGA (n = 66, 0.4%), and then ECD (n = 22, 2.8%). Table 2 presents the demographic characteristics of the selected cases and controls. The birth weight and gestational age of the cases were significantly lower than those of the controls, whereas the proportion of maternal diabetes, hypertension, smoking, and alcohol consumption during pregnancy did not differ between the two groups. Moreover, the percentage of birth season and annual household income of the residential area were also not significantly different. The results of cross-validation for the selected ordinary kriging models of the five air pollutants are demonstrated in Table 3. In general, the mean prediction errors and standardized prediction errors are very close to zero, meaning the predictions were un-biased. The root-meansquare standardized is also very close to one, indicating that the assessment of uncertainty was valid. Furthermore, the coefficient of determinations, R squared, range from 0.70 to 0.93. Table 4 lists the distribution of ambient air pollutants estimated for all the study subjects during 2006 and 2014 in Taiwan. The mean (standard deviation, SD) exposure concentrations for PM10, PM2.5, and PM2.5–10 were 53.0 (17.0), 30.6 (9.8), and 22.4 (9.3) μg/m3, respectively. The results of logistic regression models are illustrated in Table 5, and per interquartile range (IQR) increase in PM2.5 concentrations (IQR = 13.4 μg/m3) during weeks 3–8 of pregnancy is significantly associated with overall CHDs occurrence (adjusted odds ratio [aOR] = 1.21, 95% CI = 1.03–1.42). In subtype analysis, we found that early gestational exposure to PM2.5 was significantly correlated with ASD (aOR = 1.43, 95% CI = 1.01–2.02), ECD (aOR = 2.37, 95% CI = 1.01–5.58), and PAVS (aOR = 1.71, 95% CI = 1.06–2.78). No similar association was observed for the other air pollutants. 4. Discussion In this study, we have examined the relationship between early prenatal exposure to ambient air pollution and CHDs occurrence, and found associations between higher residential exposure to PM2.5 and occurrence of overall CHDs. In subtype analysis, ASD,

Table 2 Demographic characteristics of the cases and controls included in the study.

Neonatal factors

Maternal factors

Annual household income of living township

Season of conception

Birth year

Sex (male) Birth weight (g) (mean, SD) Gestational age (week) (mean, SD) Stillbirth Natural spontaneous delivery Diabetes Hypertension Smoking during pregnancy Alcohol during pregnancy Low (b25,000 USD) Medium (25,000–30,000 USD) High (≧30,000 USD) Spring Summer Autumn Winter Y2007 Y2008 Y2009 Y2010 Y2011 Y2012 Y2013 Y2014

Cases (N = 782)

Controls (N = 4692)

P

441 (56.4) 2008 (1201) 33.2 (7.5) 280 (35.8) 527 (67.4) 2 (0.3) 4 (0.5) 2 (0.3) 0 (0) 228 (29.2) 309 (39.5) 245 (31.3) 178 (22.8) 192 (24.6) 214 (27.4) 198 (25.3) 125 (16.0) 91 (11.6) 94 (12.0) 80 (10.2) 88 (11.3) 93 (11.9) 86 (11.0) 125 (16.0)

2612 (55.7) 2790 (843) 36.8 (4.8) 426 (9.1) 3112 (66.3) 13 (0.3) 13 (0.3) 11 (0.2) 0 (0) 1439 (30.7) 1825 (38.9) 1428 (30.4) 1247 (26.6) 1115 (23.8) 1180 (25.1) 1150 (24.5) 750 (16.0) 546 (11.6) 564 (12.0) 480 (10.2) 528 (11.3) 558 (11.9) 516 (11.0) 750 (16.0)

0.73 b0.001 b0.001 b0.001 0.59 0.99 0.46 0.99 0.69

0.15

0.99

884


ECD, and PAVS were revealed to be significantly associated with prenatal PM2.5 exposure. Previous studies examining the effects of PM2.5 on CHDs have produced inconsistent results (Table 1). In California, PM2.5 (median = 14.8 μg/m3, IQR = 15.2 μg/m3) was found to inversely correlate with perimembranous VSD and secundum ASD (Padula et al., 2013). Another study involving 9 states in the United States demonstrated that exposure to PM2.5 (median = 12.1 μg/m3, 10th–90th percentile = 11.9 μg/ m3) during weeks 2–8 of pregnancy was positively correlated with hypoplastic left heart syndrome but negatively correlated with ASD (Stingone et al., 2014). In Spain, prenatal exposure to PM2.5 (median = 16.6 μg/m3, IQR = 2.6 μg/m3) during weeks 3–8 of pregnancy was reported to be inversely correlated with VSD but not with overall CHD or other CHD subtypes (Schembari et al., 2014). Nevertheless, a recent study in China reported an increased risk of overall CHDs and VSD occurrence resulting from PM2.5 exposures during weeks 7–10 of pregnancy at a median (IQR) PM2.5 concentrations of 65.6 (47.2) μg/m3. In the present study, we also found that prenatal exposure to PM2.5 was positively correlated with the occurrence of overall CHDs, ASD, ECD, and PAVS; a higher exposure concentration of PM2.5 than most of the previous investigations could have enabled our study to identify an association that may not have been observed when exposure levels were low. Gilboa et al. found that exposure to PM10 during weeks 3–8 of pregnancy increased the risk of isolated ASD (Gilboa et al., 2005); the median concentration of PM10 was 23.8 μg/m3. However, other studies, with median exposure levels between 18 and 33 μg/m3, have not shown significant associations between prenatal PM10 exposure and ASD occurrence (Dadvand et al., 2011b; Dolk et al., 2010; Hansen et al., 2009; Stingone et al., 2014). In a previous study in Taiwan, PM10 (median = 56.6 μg/m3, IQR = 24.7 μg/m3) exposure during the first trimester was not observed to increase the risk of CHDs (Hwang et al., 2015). Furthermore, Zang et al. did not report any significant association between CHDs and exposure at high PM10 concentrations (median = 101.7 μg/m3, IQR = 74.8 μg/m3) during the first 3 months of pregnancy. In a population-based surveillance in Atlanta, apart from septal defects, Georgia et al. found a positive association between PM10 (median = 25.8–43.2 μg/m3, IQR = 14.2 μg/m3) exposure during weeks 3–7 of pregnancy and the occurrence of patent ductus arteriosus, but not the other CHD subtypes (Strickland et al., 2009). In our study, we did not observe any significant association between prenatal PM10 exposure and overall CHD or CHD subtypes, which is consistent with most of the previous investigations. For the other air pollutants, one previous investigation observed positive associations between prenatal exposure to CO (median = 0.5 ppm, IQR = 0.3 ppm) and ToF, and SO2 (median = 1.9 ppb, IQR = 1.4 ppb) and VSD in the USA (Gilboa et al., 2005). Dadvand et al. (2011a, 2011b) reported that CO (median = 0.41 ppm, IQR = 0.22 ppm) and NO (median = 13.9 ppb, IQR = 13.3 ppb) exposure during pregnancy increased the risk of VSD and cardiac septa malformations, while black smoke was associated with congenital malformation of cardiac chambers and connections. In Australia, gestational exposure to O3 (mean = 25.8 ppb, range = 50.1 ppb) was reported to increase the risk of pulmonary artery and valve defects, and SO2 (mean = 1.5 ppb, range = 7.1 ppb) increased the risk of aortic artery and valve defects (Hansen et al., 2009). Furthermore, Schembari et al. (2014) demonstrated a positive correlation between NO2 (median = 28.5 ppb, IQR = 14.0 ppb) exposure during weeks 3–8 of pregnancy and coarctation of the aorta in Spain. In this study, however, no similar association was observed, and the discrepancy could be related to the diverse exposure levels or the complex mixtures of ambient environment in different studies. Embryological evidence indicates that the critical stages of cardiac development before 8 weeks of pregnancy include migration of crest cells, the formation of the endocardial tube, and septation of the ventricles and outflow tracts (Gittenberger-de Groot et al., 2005). Neural crest

Table 3 Results of cross-validation for the selected ordinary kriging models in the study.

PM2.5 PM10 CO O3 NO2 SO2

PE

SPE

RMS

RMSS

R squared

0.75 1.32 0.02 0.09 0.72 0.12

0.06 0.05 0.04 0.01 0.08 0.04

7.07 12.0 0.21 5.25 5.12 1.74

0.97 0.99 1.15 0.96 0.92 0.97

0.91 0.90 0.70 0.90 0.90 0.93

Abbreviations: PE, prediction error; SPE, standardized prediction error; RMS, root-meansquare; RMSS, root-mean-square standardized.

cells were demonstrated to orchestrate cardiac valve formation from endocardial cushions (Jain et al., 2011), whereas an increase in oxidative stress in diabetic mice was found to alter neural crest cell migration, resulting in outflow tract defects (Morgan et al., 2008). Exposure to PM in pregnant women was found to induce oxidative stress (Nagiah et al., 2015); this may have resulted in the observed association in our study. Furthermore, inhaled PM was reported to increase blood viscosity (Seaton et al., 1999), which may adversely affect placental function and contribute to cardiac malformation during the critical window of embryonic development. Other proposed pollutant-induced mechanisms include tissue hypoxia, interference of metabolism or detoxification of other xenobiotics, altered apoptosis, and modified functions of trophoblast cells and early fetal growth (Barroso et al., 1998; Chen et al., 2001; Gabrielli and Layon, 1995; Inoue et al., 2004). The present investigation has the advantage of including population from a nationwide birth registry database. Reporting of all newborn infants has been mandatory since 1995; thus, our study (2007–2014) was also subject to mandatory reporting. Therefore, the study population acceptably represents all newborns in Taiwan. Furthermore, the coverage rate of National Health Insurance program is high (N99%) in Taiwan, and every pregnant woman can receive at least 10 free visits from obstetrician-gynecologists for prenatal care, including one ultrasound examination during the second trimester, and 3–5 days of hospitalization after delivery. The diagnosis of congenital cardiac anomalies is exclusively performed by obstetricians, pathologists, and pediatricians, most often by pediatric cardiologists (Hwang et al., 2011; Lin et al., 2004). The study also has limitations. First, since we focused on birth registration rather than clinical examination, underreporting of cardiac defects was unavoidable due to a limited period for ascertainment (1.47 per 1000 births) (Chen et al., 2009). The previous investigation of the age-specific prevalence of clinically diagnosed CHDs in Taiwan revealed that the prevalence rates for severe and simple CHDs were around 1 and 6 per 1000 children through 1 year of age, respectively. The prevalence of simple CHDs then increased rapidly till 3 years of age, followed by a plateau at 3–10 years. The restriction of birth registration may have introduced both random and systematic error causing either over- or under-diagnosis (Hwang et al., 2011). In this study, we assumed that the misclassification was random and non-differential between high and low exposed pregnant women. Second, false positive findings may result from multiple comparisons that tested overall and 6

Table 4 Distribution of air pollutants estimated using ordinary kriging for all the cases and controls between 2006 and 2014 in Taiwan.

PM10 (μg/m3) PM2.5 (μg/m3) PM2.5–10 (μg/m3) O3 (ppb) SO2 (ppb) CO (ppm) NO2 (ppb)

Mean

SD

53.0 30.6 22.4 28.4 3.74 0.499 17.5

17.0 9.8 9.3 5.4 1.07 0.126 4.3

Median

25tile

IQR

75tile

Range

49.4 28.5 20.8 27.7 3.51 0.504 17.8

39.5 23.1 16.7 24.1 3.06 0.414 14.7

23.2 13.4 9.4 7.9 1.11 0.174 5.8

62.8 36.5 26.0 32.1 4.17 0.588 20.4

109.0 67.2 97.2 30.7 9.07 0.697 24.5

Abbreviations: SD, standard deviation; IQR, interquartile range; tile, percentile.


885

Table 5 Adjusted odds ratio (95% confidence interval) of exposure to ambient air pollutants during weeks 3–8 of pregnancy and occurrence of Congenital Heart Diseases.a,b Controls (n = 4692)

Overall CHDs (n = 782)

VSD (n = 218)

ASD (n = 147)

ToF (n = 123)

ECD (n = 22)

TGA (n = 66)

PAVS (n = 68)

PM10 PM2.5 PM2.5–10 O3 SO2 CO NO2

1.15 (0.97–1.36) 1.21⁎ (1.03–1.42) 1.02 (0.91–1.14) 1.06 (0.94–1.21) 1.00 (0.89–1.13) 1.04 (0.86–1.25) 0.97 (0.82–1.15)

1.08 (0.81–1.44) 1.20 (0.92–1.57) 0.97 (0.81–1.15) 1.22 (0.98–1.51) 0.87 (0.72–1.06) 1.05 (0.76–1.44) 0.95 (0.71–1.28)

1.36 (0.95–1.94) 1.43⁎ (1.01–2.02) 1.09 (0.87–1.36) 0.98 (0.75–1.29) 1.14 (0.85–1.52) 0.73 (0.47–1.11) 0.70 (0.47–1.05)

1.22 (0.84–1.78) 1.15 (0.80–1.65) 1.11 (0.88–1.40) 1.04 (0.79–1.38) 0.89 (0.68–1.18) 1.21 (0.79–1.85) 0.97 (0.65–1.45)

2.38 (0.94–5.99) 2.37⁎ (1.01–5.58) 1.40 (0.78–2.53) 1.70 (0.90–3.22) 0.80 (0.46–1.39) 1.30 (0.46–3.68) 1.07 (0.43–2.66)

1.10 (0.67–1.81) 1.14 (0.71–1.81) 1.02 (0.75–1.38) 1.24 (0.86–1.78) 1.35 (0.99–1.84) 0.98 (0.58–1.68) 0.96 (0.58–1.58)

1.53 (0.92–2.55) 1.71⁎ (1.06–2.78) 1.10 (0.79–1.53) 1.26 (0.88–1.82) 1.07 (0.73–1.56) 1.41 (0.81–2.45) 1.20 (0.71–2.04)

Abbreviations: ASD, atrial septal defect; CHDs, Congenital Heart Diseases; ECD, endocardial cushion defect; TGA, transposition of the great arteries; ToF, Tetralogy of Fallot; PAVS, pulmonary artery and valve stenosis; VSD, ventricular septal defect. a Odds ratio obtained for per IQR increase in: PM10 = 23.2 μg/m3, PM2.5 = 13.4 μg/m3, PM2.5–10 = 9.4 μg/m3, O3 = 7.99 ppb, SO2 = 1.11 ppb, CO = 0.174 ppm, and NO2 = 5.72 ppb. b Logistic regression model adjusted for infant sex, maternal age, season of conception, maternal diabetes and hypertension, maternal smoking during pregnancy, socioeconomic status of the residential township, and birth year. ⁎ P b 0.05.

subtypes of CHDs for 7 air pollutants. At least 2 false positives may result from 49 tests given the preselected alpha value of 0.05; therefore, the observed significant associations should be interpreted with caution. Third, we only had information on the air pollutants routinely measured at the air quality monitoring stations; therefore, the possibilities that other unknown chemicals, emitted along with PM into the ambient air, contributed to the observed positive association cannot be completely excluded. Forth, we could only access the participants' residential addresses to the level of township because of confidentiality considerations. We used average values in the township to represent long-term exposure within the residential area. This might have caused misclassification in exposure characterization, and consequently, our observed effects were less detective for the actual association between exposure to the air pollutant and CHDs occurrence. Finally, the etiology of Congenital Heart Diseases comprises both genetic and environmental factors. In this study, we focused on evaluating the effects from ambient air pollution. Information regarding the daily activity of the pregnant subjects that can help differentiate the contributing effects from both ambient and indoor air pollution was unavailable. Furthermore, Stingone et al. (2017) reported an interaction between NO2 exposure and methyl nutrients intake during early pregnancy on the occurrence of Congenital Heart Diseases. Again, information was lacking in the Taiwanese Birth Registry database. However, in this study, we assumed that these errors did not differ between the cases and controls, and the potential misclassification might have underestimated, but not biased, the observed associations. 5. Conclusions Our findings suggest some associations between prenatal exposure to high ambient PM2.5, around 5 μg/m3 above the current WHO air quality guideline, during the critical window of embryogenesis and the occurrence of CHDs, mainly ASD, ECD, and PAVS. Further studies are needed to confirm the consistency of the effects of PM2.5 exposure at different exposure ranges. CRediT authorship contribution statement Ching-chun Huang: Conceptualization, Methodology, Formal analysis, Writing - original draft. Bing-yu Chen: Methodology, Data curation, Project administration. Shih-chun Pan: Software, Formal analysis. Yi-lwun Ho: Conceptualization, Writing - review & editing. Yue Leon Guo: Conceptualization, Methodology, Writing - review & editing, Supervision. Acknowledgments This study was supported by the Environmental Protection Administration, Taiwan (Grant No. NSC-102-EPA-F-002-002); National Health Research Institutes, Taiwan (Grant No. NHRI-105-EMSP08); and

Ministry of Science and Technology, Taiwan (Grant No. MOST-1072621-M-002-008). References Agay-Shay, K., Friger, M., Linn, S., Peled, A., Amitai, Y., Peretz, C., 2013. Air pollution and congenital heart defects. Environ. Res. 124, 28–34. Barroso, R.P., Osuamkpe, C., Nagamani, M., Yallampalli, C., 1998. Nitric oxide inhibits development of embryos and implantation in mice. Mol. Hum. Reprod. 4, 503–507. Chen, H.W., Jiang, W.S., Tzeng, C.R., 2001. Nitric oxide as a regulator in preimplantation embryo development and apoptosis. Fertil. Steril. 75, 1163–1171. Chen, B.Y., Hwang, B.F., Guo, Y.L., 2009. Epidemiology of congenital anomalies in a population-based birth registry in Taiwan, 2002. J. Formos. Med. Assoc. 108, 460–468. Dadvand, P., Rankin, J., Rushton, S., Pless-Mulloli, T., 2011a. Ambient air pollution and congenital heart disease: a register-based study. Environ. Res. 111, 435–441. Dadvand, P., Rankin, J., Rushton, S., Pless-Mulloli, T., 2011b. Association between maternal exposure to ambient air pollution and congenital heart disease: a register-based spatiotemporal analysis. Am. J. Epidemiol. 173, 171–182. Dhanantwari, P., Lee, E., Krishnan, A., Samtani, R., Yamada, S., Anderson, S., et al., 2009. Human cardiac development in the first trimester: a high-resolution magnetic resonance imaging and episcopic fluorescence image capture atlas. Circulation 120, 343–351. Dolk, H., Vrijheid, M., 2003. The impact of environmental pollution on congenital anomalies. Br. Med. Bull. 68, 25–45. Dolk, H., Armstrong, B., Lachowycz, K., et al., 2010. Ambient air pollution and risk of congenital anomalies in England, 1991–1999. Occup. Environ. Med. 67 (4), 223–227. EUROCAT, 2016. EUROCAT Guide 1.4. Instructions for the Registration of Congenital Anomalies. 2017. Ferencz, C., Rubin, J.D., Loffredo, C.A., M, C.M., 1993. Epidemiology of Congenital Heart Disease: The Baltimore-Washington Infant Study 1981–1989. Futura Publishing Co., Mount Kisco, New York. Gabrielli, A., Layon, A.J., 1995. Carbon monoxide intoxication during pregnancy: a case presentation and pathophysiologic discussion, with emphasis on molecular mechanisms. J. Clin. Anesth. 7, 82–87. Gilboa, S.M., Mendola, P., Olshan, A.F., Langlois, P.H., Savitz, D.A., Loomis, D., et al., 2005. Relation between ambient air quality and selected birth defects, seven county study, Texas, 1997–2000. Am. J. Epidemiol. 162, 238–252. Gittenberger-De Groot, A.C., Bartelings, M.M., Deruiter, M.C., Poelmann, R.E., 2005. Basics of cardiac development for the understanding of congenital heart malformations. Pediatr. Res. 57, 169–176. Gorini, F., Chiappa, E., Gargani, L., Picano, E., 2014. Potential effects of environmental chemical contamination in congenital heart disease. Pediatr. Cardiol. 35, 559–568. Hansen, C.A., Barnett, A.G., Jalaludin, B.B., Morgan, G.G., 2009. Ambient air pollution and birth defects in Brisbane, Australia. PLoS One 4, e5408. Hoek, G., Fischer, P., Van Den Brandt, P., Goldbohm, S., Brunekreef, B., 2001. Estimation of long-term average exposure to outdoor air pollution for a cohort study on mortality. J. Expo. Anal. Environ. Epidemiol. 11, 459–469. Huang, C.C., Wen, H.J., Chen, P.C., Chiang, T.L., Lin, S.J., Guo, Y.L., 2015. Prenatal air pollutant exposure and occurrence of atopic dermatitis. Br. J. Dermatol. 173, 981–988. Hwang, B.F., Lee, Y.L., Jaakkola, J.J., 2011. Air pollution and stillbirth: a population-based case-control study in Taiwan. Environ. Health Perspect. 119, 1345–1349. Hwang, B.F., Lee, Y.L., Jaakkola, J.J., 2015. Air pollution and the risk of cardiac defects: a population-based case-control study. Medicine (Baltimore) 94, e1883. Inoue, T., Kaibara, M., Sakurai-Yamashita, Y., Kawano, M., Ishimaru, T., Taniyama, K., 2004. Increases in serum nitrite and nitrate of a few-fold adversely affect the outcome of pregnancy in rats. J. Pharmacol. Sci. 95, 228–233. Jain, R., Engleka, K.A., Rentschler, S.L., Manderfield, L.J., Li, L., Yuan, L., et al., 2011. Cardiac neural crest orchestrates remodeling and functional maturation of mouse semilunar valves. J. Clin. Invest. 121, 422–430. Kevin, J., Jay, M.H., Konstantin, K., Neil, L., 2003. Using analytical tools when generating surfaces. ArcGIS 9 Using ArcGIS Geostatistical Analyst. ESRI Inc., Redlands, CA.

886


Lin, C.M., Lee, P.C., Teng, S.W., Lu, T.H., Mao, I.F., Li, C.Y., 2004. Validation of the Taiwan Birth Registry using obstetric records. J. Formos. Med. Assoc. 103, 297–301. Morgan, S.C., Relaix, F., Sandell, L.L., Loeken, M.R., 2008. Oxidative stress during diabetic pregnancy disrupts cardiac neural crest migration and causes outflow tract defects. Birth Defects Res. A Clin. Mol. Teratol. 82, 453–463. Nagiah, S., Phulukdaree, A., Naidoo, D., Ramcharan, K., Naidoo, R.N., Moodley, D., et al., 2015. Oxidative stress and air pollution exposure during pregnancy: a molecular assessment. Hum. Exp. Toxicol. 34, 838–847. Padula, A.M., Tager, I.B., Carmichael, S.L., Hammond, S.K., Yang, W., Lurmann, F., et al., 2013. Ambient air pollution and traffic exposures and congenital heart defects in the San Joaquin Valley of California. Paediatr. Perinat. Epidemiol. 27, 329–339. R Core Team, 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL. https://www.R-project.org/. Ritz, B., Yu, F., Fruin, S., Chapa, G., Shaw, G.M., Harris, J.A., 2002. Ambient air pollution and risk of birth defects in Southern California. Am. J. Epidemiol. 155, 17–25. Rosamond, W., Flegal, K., Friday, G., Furie, K., Go, A., Greenlund, K., et al., 2007. Heart disease and stroke statistics—2007 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 115, e69–171. Schembari, A., Nieuwenhuijsen, M.J., Salvador, J., de Nazelle, A., Cirach, M., Dadvand, P., et al., 2014. Traffic-related air pollution and congenital anomalies in Barcelona. Environ. Health Perspect. 122, 317–323. Seaton, A., Soutar, A., Crawford, V., Elton, R., McNerlan, S., Cherrie, J., et al., 1999. Particulate air pollution and the blood. Thorax 54, 1027–1032.

Stingone, J.A., Luben, T.J., Daniels, J.L., Fuentes, M., Richardson, D.B., Aylsworth, A.S., et al., 2014. Maternal exposure to criteria air pollutants and congenital heart defects in offspring: results from the national birth defects prevention study. Environ. Health Perspect. 122, 863–872. Stingone, J.A., Luben, T.J., Carmichael, S.L., Aylsworth, A.S., Botto, L.D., Correa, A., et al., 2017. Maternal exposure to nitrogen dioxide, intake of methyl nutrients, and congenital heart defects in offspring. Am. J. Epidemiol. 186, 719–729. Strickland, M.J., Klein, M., Correa, A., Reller, M.D., Mahle, W.T., Riehle-Colarusso, T.J., et al., 2009. Ambient air pollution and cardiovascular malformations in Atlanta, Georgia, 1986–2003. Am. J. Epidemiol. 169, 1004–1014. Szpiro, A.A., Sheppard, L., Sampson, P.D., Kim, S.Y., 2007. Validating national kriging exposure estimation. Environ. Health Perspect. 115 (7), A338 (author reply A338-9). Taiwan EPA, 2006–2014. Air Quality Monitoring System. 2017. Taiwan Environmental Protection Administration, Taipei, Taiwan. Vrijheid, M., Martinez, D., Manzanares, S., Dadvand, P., Schembari, A., Rankin, J., et al., 2011. Ambient air pollution and risk of congenital anomalies: a systematic review and meta-analysis. Environ. Health Perspect. 119, 598–606. World Health Organization, 2016. Ambient (outdoor) air quality and health. September. http://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-qualityand-health, Accessed date: 5 November 2018. Zhang, B., Liang, S., Zhao, J., Qian, Z., Bassig, B.A., Yang, R., et al., 2016. Maternal exposure to air pollutant PM2.5 and PM10 during pregnancy and risk of congenital heart defects. J. Expo. Sci. Environ. Epidemiol. 26, 422–427.