Congenital heart defects (CHDs) are a commonly occurring group of malformations with a birth prevalence of 1 per 100 live births in the United States (US), impacting pediatric morbidity and mortality (Botto et al., 2007; Moller et al., 1993; Christianson et al., 2006). Although some CHDs occur in association with certain genetic syndromes (e.g., trisomy 21, 22q11 deletion, Alagille syndrome, Noonan syndrome) and teratogenic exposures (e.g., anticonvulsants, maternal pregestational diabetes), approximately 80% are of unknown etiology (Christianson et al., 2006; Jenkins et al., 2007; Caton et al., 2009). While environmental and occupational factors have been suggested as potential risk factors for CHDs, the evidence is still incomplete (Jenkins et al., 2007).

One group of contaminants found in the environment and workplace that is hypothesized to be associated with adverse birth outcomes is polycyclic aromatic hydrocarbons (PAHs) (Perera et al., 1998). The primary sources of exposure to PAHs for most of the US population include tobacco smoke, wood smoke, ambient air pollution (e.g. vehicle exhaust, coal combustion), as well as consumption of charbroiled foods (Mumtaz and George, 1995). PAHs are suspected teratogens because of their ability to readily penetrate cellular membranes, including the embryonic and fetal blood-brain barrier, resulting in increased oxidative stress and DNA damage (Huang et al., 2013; Hanzalova et al., 2010; Patri et al., 2010; Anwer and Mehrotra, 1988). A variety of birth outcomes and anomalies, including CHDs, have been reported following PAH exposure in experimental animal model systems (Anwer and Mehrotra, 1988; Barbieri et al., 1986; Farwell et al., 2006; Incardona et al., 2004; Sestak et al., 2018; Ng et al., 2009). Additionally, recent studies in humans suggest maternal occupational and environmental exposure to PAHs is associated with other structural birth defects such as neural tube defects and gastroschisis (Lupo et al., 2012a; Naufal et al., 2010; Ren et al., 2011). While there is evidence in a few human studies that PAHs are associated with CHDs, these studies have largely been limited by sample size, inability to evaluate specific subtypes, and non-specific PAH exposure assessment (Li et al., 2018; Lupo et al., 2012b).

The National Birth Defects Prevention Study (NBDPS) is the largest population-based case-control study of birth defects and provides a unique opportunity to examine the association between maternal PAH exposure and birth defects. With the goal to update a previous NBDPS study on PAH exposure and CHDs from 1997 to 2002 (Lupo et al., 2012b), we conducted an investigation of the overall association and potential dose-response relationship between PAH exposure and CHDs using NBDPS data from 1997 through 2011.

Of the 12,584 CHD case mothers and 11,829 control mothers included in the NBDPS for the period 1997 through 2011, 69% (8,644 case mothers; 8,140 control mothers) were employed for at least 1 month during the critical window of exposure (the remaining 31% mothers reported no job during the critical window and were not included in this analysis as to limit our inferences to women who were specifically employed during the critical period of development).

Exclusions for CHD case and control infants were based on maternal pregestational diabetes (n = 298), multi-fetal gestations (n = 834), and first-degree family history of CHDs (n = 410). Furthermore, CHD cases with extracardiac defects (multiple or complex) (n = 1,204); CHD cases with defects designated as “association” (n = 1,281); and “complex” CHDs (n = 1,204) were not included because of smaller sample sizes and to maintain homogenous case groups. After exclusions, there were 4,775 CHD case infants and 7,734 control infants eligible for analyses.

Table 1 displays selected characteristics summarized by case-control status. There were significant differences in the distributions of infant sex (45% females in cases and 49% females in controls), maternal education (38.9% case mothers vs. 41.3% controls had ≥16 years education), BMI (20.3% case mothers vs. 18.1% control mothers were ≥30 kg/m² [obese]), and alcohol use (40.2% case mothers vs. 42.4% control mothers). CHD case mothers were also more likely to have gestational diabetes and report exposure to secondhand smoke at home and work than control mothers. The overall prevalence of PAHs exposure was 10.2% among case mothers and 10.2% among control mothers.

Table 2 displays the number of jobs linked to the 23 SOC major job groups for mothers of cases and controls by exposure status. The most frequent jobs in which case mothers were exposed during the critical window of development were reported in “Food Preparation and Serving Related” (n = 299 cases; n = 517 controls) and “Sales and Related Occupations” (n = 65 cases; n = 83 controls). Some jobs could not be coded to a single SOC major job group and are not represented in Table 2 (n = 3).

Tables 3 and 4 show crude and adjusted associations between quartiles of cumulative PAH exposure and risk of CHDs among case and control mothers respectively. The quartiles were classified as Q1 (0.04–7.2 μg/m³-hr), Q2 (7.54–51.43 μg/m³-hr), Q3 (52.11–218.06 μg/m³-hr), and Q4 (≥219.43 μg/m³-hrs). In unadjusted analysis (Table 3), exposure to cumulative PAHs in fourth quartile (Q4) was positively associated with conotruncal heart defects as group (OR 1.43; 95% CI 1.02–2.00), tetralogy of Fallot (ToF) (OR 1.81; 95% CI 1.21–2.72), and atrial septal defects (secundum) (OR 1.56; 95% CI 1.03–2.37). After adjusting for potential confounders (Table 4), the associations remained significant for conotruncal heart defects as a group (OR 1.41; 95% CI 1.00–2.00) and ToF specifically (OR 1.83; 95% CI 1.21–2.78). While not statistically significant, there was a decreasing pattern of risk of right ventricular outflow tract obstruction defects (RVOTO) with increasing exposure (OR 1.19; 95% CI 0.78–1.81 in Q1 vs. OR 0.64: 95% CI 0.37–1.12 in Q4). We did not observe significant linear trends for any of the CHDs subgroups or subtypes.

Supplemental Table S1 shows comparison of adjusted odds ratio for associations between maternal occupational exposure to PAHs (yes/no) and CHDs by three time-periods (1997–2002, 2003–2011, 1997–2011). While there were some effect estimates that changed direction or magnitude, overall these differences did not represent a consistent trend.

We found little evidence that PAHs were strongly associated with CHDs overall or with selected subtypes in this study, which is one of the largest population studies conducted to date. However, our findings suggest that higher estimated exposure levels (≥219.43 μg/m³-hr) were positively associated with conotruncal heart defects, largely driven by the association with ToF. Overall, these conclusions are consistent with an earlier and smaller subset of data from the NBDPS (Lupo et al., 2012b).

While we observed a positive association with one group of CHDs, results from our current study and the previous NBDPS assessment are partially consistent to other studies published in literature. For instance, a recent study by Li and colleagues from China that explored the association of maternal exposure to PAHs and CHDs among 627 infants found significant positive associations between maternal exposure to PAHs and some CHD subtypes including septal defects (OR 2.37, 95% CI: 1.38–4.09), right-sided obstructive malformations (OR 2.42, 95% CI: 1.19–4.93), and left-sided obstructive malformations (OR 2.66, 95% CI: 1.08–6.52), along with conotruncal heart defects (OR 2.35, 95% CI: 1.25–4.42) (Li et al., 2018). The study, however, used a different cut-off point for assessing PAHs exposure in form of a PAH metabolite, 1-Hydroxypyrene-glucuronide, using Youden’s Index. This might account for differences in observed risk across in the current study, in addition to differences in populations. Moreover, two studies reported an association between maternal exposure to particulate matter (both PM2.5 and PM10, which contain significant amount of PAHs variably) (Murillo et al., 2017) and some CHD subtypes, namely atrial septal defects and pulmonary valve stenosis (Padula et al., 2013; Gilboa et al., 2005). Furthermore, another recent study from China reported that CHDs were associated with an interaction between PAHs and a specific genetic polymorphism in the EPHX1 gene (Tao et al., 2019).

To more fully explore the relationship between PAHs and CHDs, a unique aspect of this study was the assessment of increasing level of PAH exposure on the risk of CHDs. In fact, we did not observe a linear trend between increasing quartile of exposure and any CHD subtype. While this could be indicative of a threshold effect, it is not clear from this assessment. We also explored differences by time period, to see if magnitudes of associations varied over the past two decades. Overall, results were not consistent, with some associations being stronger during the earlier time period (1997–2002) and some being stronger in the more recent time period (2003–2011).

The strongest finding in our assessment was between maternal occupational exposure to PAHs and conotruncal heart defects in offspring. The potential mechanisms underlying this association are unclear. However, as PAHs are lipophilic, it has been demonstrated that they freely penetrate cellular membranes, including the placenta (Mumtaz and George, 1995). As a result, we also conducted a secondary analysis to observe interaction effects of BMI and PAHs exposure, but we did not observe any statistically significant interaction. Also, during PAH metabolism, enzymatic activity can result in the formation of reactive intermediates that covalently bind to DNA, forming adducts that may be teratogenic (Wells et al., 2010). Additionally, there is evidence from animal models that maternal exposure to PAHs can lead to CHDs in offspring. Specifically, studies have demonstrated PAH exposure can lead to cardiac dysfunction in embryos of various fish model systems, including zebrafish and Atlantic killifish (Farwell et al., 2006; Incardona et al., 2004; Clark et al., 2010; Wassenberg and Di Giulio, 2004; Wassenberg et al., 2005). The teratogenic potential of PAHs has also been shown recently in other fish model systems including the widow tetra, Mandeli and Pelagic (Clifton and Pandian, 2016; Dhananjayan and Muralidharan, 2012; Romero et al., 2018). While these assessments were not specific to conotruncal heart defects, it does provide some support for the observed association.

Our study should be considered in the light of certain limitations. First, we had 20 unique comparisons of CHD groups and subtypes that might create an issue of multiple comparisons. Based on 20 hypotheses, we had 64% chance of observing at least one significant result, even if all of the tests were actually not significant. Although the probability for getting the positive association is higher, our study findings depend on consistency and biological plausibility based on previous literature. As a result, we did not adjust for multiple comparisons to avoid diminishing the p-value threshold for statistical significance. Second, since exposure was based on estimates calculated by industrial hygienists, there is the potential for exposure (random with respect to case status) misclassification. However, this approach has been used extensively in estimates of occupational exposure to PAHs and is likely to be superior to self-report alone (Olsson et al., 2010). A third limitation is the potential for recall bias (as in any case-control study), and that this might vary in strength between women interviewed closer to delivery compared to those interviewed two years after delivery. However, we suspect that recall bias may have been relatively low in this study overall because the NBDPS has shown little evidence of this bias (Reefhuis et al., 2015) and because most women are unlikely to be aware which occupations have the highest PAH exposure (Rocheleau et al., 2011). Additionally, while we did evaluate a range of CHDs, we were unable to evaluate the role of maternal occupational exposures to PAHs on complex cases (i.e., cases that included three or more distinct CHDs) due to the sample size. It is possible that these exposures could be associated with these more severe phenotypes. Another limitation is our inability to fully assess environmental sources of PAHs in the analysis. Because of this, we could not fully account for the totality of PAH exposure, but it should be noted that exposures to external environmental sources of PAHs are likely to be lower compared to occupational sources (Brandt and Watson, 2003). Additionally, consistent with previous NBDPS assessments on PAHs and birth defects, we were able to assess confounding due to secondhand smoke at home or work and meat consumption, the two major sources of environmental PAHs (Lupo et al., 2012a; Naufal et al., 2010; Ren et al., 2011; Boers et al., 2005; Hansen et al., 2008). The final limitation is unknown validity of the exposure ratings, since direct measurements or biomarkers were unavailable. Although the use of multiple raters, literature reviews, and other tactics were utilized to improve the quality of the exposure assessment—and prior research shows that expert raters can achieve relatively high validity in occupational exposure assessments using expert judgement—we ultimately do not know how much exposure misclassification might be present (Blair et al., 2007; Friesen et al., 2011; Stewart, 1999; Stewart et al., 2003; Wheeler et al., 2013).

Our study also has notable strengths. First, the NBDPS is one of the largest and most comprehensive case-control studies of birth defects in the US with extensive information on occupational history and other important covariates, including maternal nutrition, pre-pregnancy BMI, and smoking. The information on several of these covariates was not only around the time of delivery, but throughout the entire time window prior to conception to end of pregnancy. Additionally, through the NBDPS, there was extensive clinical review of all cases, which limits the inclusion of syndromic cases, resulting in a more homogenous CHD case groups. This also provides an added benefit of inclusion of stillbirths and terminations, unlike majority of the previous studies that only focus on livebirths.

In summary, we found little evidence that maternal occupational exposure increased the risk of most sub-phenotypes of CHDs in offspring. However, we report a moderate association with conotruncal heart defects, which is consistent with some other reports, as well as evidence from animal models. While our study adds to the growing body of evidence on the role of PAHs on the risk of structural birth defects, future studies must consider better estimates of exposure, as well as factors that could modify risk (e.g., genotypes).