Environmental Exposures and Congenital Heart Disease

A literature search was performed in PubMed by using the search terms “exposome,” “environmental exposure,” “congenital heart disease,” and “congenital heart defect.” Terms were entered in different combinations, with reference list search performed with selected articles. Inclusion criteria were case-control or population-based studies, systematic reviews, and meta-analyses examining associations between maternal or paternal exposures and the incidence of CHD overall or a specific subtype of CHD. Articles included were published between January 1970 and December 2019, which was the time of index search. Articles were initially screened by title and abstract; preliminarily eligible articles were read in completion by reviewers for confirmation of eligibility. Given the diversity of studies included in this review, the emphasis of data extraction from each article were study type and methodology, population studied, major outcomes, strengths, and limitations.

Hypoplastic left heart syndrome (HLHS) and coarctation of the aorta (CoA) have been shown to exhibit seasonality in incidence.^9,10  Peak birth months of HLHS have been noted in the summer months, with peak birth months of April to May for CoA.⁹  Others have shown bimodal seasonal peaks for CoA from September to November and January to March.¹⁰  Geographical clustering and variation led to the conclusion that multiple different environmental factors were implicated; variations in nutrition, UV light exposure, and seasonal viral infections were posited as possible etiologies.^9,10 

Multiple different classes of medication have been investigated as potential causative agents of CHD. These categories include antihypertensives, antiinfectives, insulin, fertility drugs, NSAIDs, anticonvulsants, antidepressants, and other classes of psychotropic medication. Although moderate associations have been found between each of these drug classes and CHD (Table 1), the results have been largely inconsistent and nonreproducible.^10–20  The potential for confounding by underlying maternal disease and the risk of untreated disease renders maternal medication usage an exposure ideally suited to measurement via a comprehensive exposome model.²¹ 

Ingestion of alcohol or illicit drugs and tobacco use have been widely studied as potential causal factors in CHD. The literature has been largely inconsistent, and most of the positive associations have been moderate at best (Table 2).^22–29  Dose-response relationships noted in tobacco and marijuana use, alcohol consumption, concomitant alcohol and tobacco use, as well as associations noted only when both parents smoke, highlight the importance of measuring the totality of these types of prenatal exposure through an exposome framework rather than as individual exposures.^23,26–29 

Much attention has been paid to the correlation between chronic and acute maternal disease and CHD. Obesity, diabetes mellitus (DM) type I or II, hypertension, acute febrile illness, thyroid disorders, metabolic derangement (such as hyperlipidemia or hyperhomocysteinemia), connective tissue disease and mental health disorders have all been implicated in the incidence of CHD to varying degrees (Table 3).^4,22,30–43  Similar to alcohol and tobacco consumption, maternal obesity has been noted repeatedly to have a dose-response relationship with CHD; however, studies examining the synergistic effect of obesity with diabetes, hypertension, and other metabolic or endocrine derangements are lacking.^30–32  The majority of studies of maternal disease have not differentiated between treated and untreated disease, further obscuring the relationship between medication use, pathophysiologic changes due to disease, and the incidence of CHD.

Carbon monoxide, sulfur dioxide, and nitric oxide are gaseous pollutants that have been implicated in the development of CHD. Particulate matter are air pollution components particularly associated with proximity to high-traffic geographic areas; most adverse outcomes reported to date are associated with the smallest size of particulate matter: particulate matter <10 μm in diameter (PM10) and/or PM2.5.^44–50  Methods used to assess air pollution exposure vary; in purely epidemiological approaches, researchers use maternal residence location but cannot assess transplacental exposure or account for duration of exposure, indoor versus outdoor time, or intralocation heterogeneity.^44–48  Alternatively, biological assay-based models of exposure rely on direct measurement of levels of exposure through placental and umbilical cord sampling (Table 4).^49,50  Polycyclic aromatic hydrocarbons (PAHs) are environmental pollutants from coal and tobacco smoke: maternal PAH inhalation and/or consumption leads to fetal exposure, as evidenced by PAH presence in placental tissue and umbilical cord blood.⁴⁹  Limitations that must be addressed to maximize the benefit of human placental measurement for fetal toxin exposure include the need for standardization of placental sampling.⁵¹ 

The World Health Organization estimates that 25% of congenital disorders can be attributed to environmental pollution exposure.⁵²  Maternal heavy metal exposure, including lead, nickel, arsenic, cadmium, and manganese, has been associated with the incidence of CHD, in addition to pesticide and organic chemical solvent exposure (Table 5).^53–60  Hair and fetal placental sampling avoids the self-reporting bias of other occupational exposure studies; differences in the level of nickel in samples among septal defects, conotruncal defects, and outflow tract obstruction have been observed, suggesting varying levels of exposure may confer a risk to specific types of CHD.^54,55  A strong synergistic interaction between arsenic and cadmium exposure has also been noted, conferring a ninefold, positive association with CHD, emphasizing the importance of studying the degree and interplay of environmental exposures.⁵⁴ 

Paternal occupational exposures, such as metal and jewelry manufacturing, welding, lead soldering, paint stripping and exposure to ionizing radiation, medication use, tobacco, marijuana, and wine use have also been shown to have associations with the incidence of CHD in their offspring (Table 6).^61–66 

Given the broad, heterogenous nature of environmental exposures that have been studied, clinicians’ understanding of what risk factors are the most common and potentially modifiable becomes paramount. Although the exact prevalence of these risk factors remains largely unknown, certain exposures, such as maternal obesity and diabetes, are known to be among the most commonly encountered. Exposure to pollution is also ubiquitous, with an estimated 92% of the world’s population affected; however, exact knowledge of levels of pollution at exact days and windows of pregnancy to determine true exposure remains challenging. Of the risk factors studied, those with the highest correlations to incidence of CHD overall include binge drinking in tandem with cigarette use (odds ratio [OR] 12.65, 95% confidence interval [CI] 3.54–45.25),²⁸  maternal diabetes (OR 6.75 [95% CI 1.85–24.58]),³⁸  and maternal medication exposure, including nitrofurantoins (adjusted odds ratio [aOR] 4.2, 95% CI 1.9–9.1),¹³  benzodiazepines (OR 5.4, 95% CI 1.5–19.11),¹⁵  and dual selective serotonin reuptake inhibitor (SSRI) exposure (OR 4.7, 95% CI 1.74–12.7).²⁰  The majority of these risk factors are modifiable with perinatal interventions, including weight loss and avoidance of known toxins, such as tobacco and alcohol. However, the clinical picture is less clear when differentiating between the risk of maternal disease versus the risk of maternal medications for disease management. Approaching this question from the framework of the exposome, in which maternal disease, medication usage, and other external environmental factors can be accounted for simultaneously, will further clinicians’ ability to predict and modify risk in vulnerable populations.

Cardiac development begins shortly after implantation of the fertilized embryo in the uterine wall; differentiation of cardiomyocytes begins by postconception day 16, and circulation, including a beating heart, presents by day 21 and completion of human cardiac morphologic development by 8 weeks.³  Therefore, many CHDs occur before knowledge of pregnancy, which obfuscates the ability to screen and intervene before the development of cardiac anomalies. Despite this barrier, there has been great effort to study potential causal pathways beyond genetic mutations in the incidence of CHD, including the recent development of validated machine learning models for predicting periconceptional risk of CHD, and identification of potential microRNA biomarkers for the early detection of ventricular septal defect.^67,68 

There is a growing awareness of the interplay between placental and cardiac development, a phenomenon termed the “heart-placental axis,” which signifies the parallel development of both the placenta and the heart with use of common genes and molecular pathways.^3,69–72  Maternal disease, ingestion of medications, drugs, exposure to environmental chemicals and toxins, and nutritional imbalances may play a role in disrupting these pathways; however, the exact mechanisms remain unelucidated.³  Two pathways that have been implicated as having key importance in both placental and cardiac development are the canonical Wnt/β-catenin pathway and the folate metabolism pathway.^3,69–72  Even a single exposure to lithium, ethanol, or homocysteine during days 16 to 19 postconception can lead to abnormal blood flow in the umbilical artery, leading to misexpression of Wnt-mediated genes and placental, heart, and valvular defects.^3,69  Multiple studies have shown the effect of intersection of genes and environment, including maternal obesity; tobacco, occupational, and solvent exposure; and folate use.^22,73–77  This intersection may lie in as simple a concept as how their synergistic effects change placental blood flow.⁷⁸ 

The yolk sac becomes vascularized at ∼24 days postconception, with vascularization of the chorioallantoic placenta happening shortly thereafter; variations in either of these extraembryonic circulations can impact cardiac development.^71,79  During the early stages of pregnancy, hypoxia prevents the invasion of trophoblasts into the uterus; in later stages, the trophoblast cells migrate and replace the maternal vascular wall within the uterine wall, creating a low-resistance state that encourages oxygen and nutrient exchange.⁷⁹  The maintenance of a hypoxic environment and prohibition of premature trophoblast invasion has been shown to be mediated by numerous cytokines, such as tumor necrosis factor α, interleukin 1α, and interleukin 1β.⁷⁹  This suggests that an acute infectious illness leading to upregulation of these cytokines at the right gestational time period could lead to alteration in proper transition to the invasive trophoblast phenotype, ultimately leading to impaired placental angiogenesis. Deficiency in endothelial nitric oxide synthase has been shown to lead to decreased uterine blood flow, placental oxygenation, and spiral artery elongation in murine models. Certain classes of antidepressants modulate the hippocampal nitric oxide level in vivo; however, no studies have been done to date to explore the effect of antidepressants on placental levels of endothelial nitric oxide synthase.^80,81  Notably, both chronic and acute prenatal exposure to bupropion may be linked to reduced uterine blood flow, focusing on the role of serotonin as a known uterine vasoconstrictor.⁸²  Inactivation of placental p38 mitogen-activated protein kinase has also been shown to reduce maternal-fetal vascularization of the placenta causing severe cardiac defects in a murine model; pathways by which exposure to environmental chemicals and toxins that may exert an effect on p38 mitogen-activated protein kinase should be explored further.⁸³  Other possible mechanisms of placental vascular insufficiency include deficiency of peroxisome proliferator-activated receptor γ, a nuclear hormone receptor, which has been shown to interfere with trophoblast differentiation and vascularization of the placenta, leading to severe myocardial thinning and death in knock-out mice models.⁸⁴  Notably, peroxisome proliferator-activated receptor γ is a target of many medications used in the treatment of DM type II and is heavily involved in metabolic derangements, such as obesity and DM type II.⁸⁴ 

Additionally, the overexpression of growth arrest and DNA damage-inducible 45 α, a small acidic protein involved in DNA repair and mitogen-activate protein kinase modulation, in syncytiotrophoblasts, extravillous trophoblast cells, and trophoblast columns of placental villi has been associated with impaired migration and invasion of trophoblasts, leading to decreased angiogenesis.^85,86  Given that the growth arrest and DNA damage-inducible 45 family of proteins are susceptible to acute environmental and physiologic stresses, such as hypoxia, transient ischemia, and angiotensin II, there are multiple ways that maternal environmental exposures may lead to high expression of placental growth arrest and DNA damage-inducible 45 α.^85,86  Knock-out of small ubiquitin-related modifier–specific protease 2 (SENP2) is associated with cardiac dysgenesis, specifically the absence of SENP2 in the developing placenta leading to thinning of the myocardium, absence of atrioventricular cushions, and CHD with hypoplastic chambers.⁸⁷  Interestingly, in a recent report, the Epstein-Barr virus (EBV) latent membrane protein-1 has been described as independently decreasing SENP2 activity and turnover, in addition to altering the localization of SENP2.⁸⁸  Given the high prevalence of EBV infection, this presents a unique opportunity to study the correlation between maternal previous EBV infection, SENP2 expression within the placenta, and the incidence of CHD.

These proposed mechanisms of impaired placental development and vascularization may be the key to further understanding the heart-placental axis and the interplay between general and specific external factors and the incidence of CHD. Further study of these associations and mechanisms is required through clinical case-control studies of exposed cohorts, basic science studies focusing on the association between individual and collective environmental exposures and the hypothesized mechanism, and basic science studies of the association between the hypothesized mechanism and the incidence of CHD in in vitro and small animal in vivo models. Further elucidation of these mechanisms could identify targets for drug development to optimize periconceptional placental development and vascularization; this would complement preventive public health interventions to decrease CHD occurrence, including decreasing prenatal exposure and improving prenatal education on risk-factor avoidance.

One of the greatest barriers to evaluating the effect of abnormal blood flow on the development of CHD is the lack of technological ability to study the internal vasculature and morphology of the placenta in early stages of pregnancy. In a study of infants born with HLHS, placental analysis revealed reduced surface area for oxygen and nutrient exchange and reduced vasculature and abnormal parenchyma, suggesting that the immaturity of the placenta may be due to vascular abnormalities.⁸⁹  However, others have noted normal umbilical artery blood flow by Doppler ultrasound in pregnancies complicated by HLHS, implying the need for alternative ways to assess actual placental vascularization and functionality.^89–91  This argument is particularly compelling when noting that in a study of the uterine artery pulsatility index at 11 to 13 weeks’ gestation in 68 cases of isolated fetal CHD, there was no significant difference between cases and controls.⁹²  However, that same study revealed a significantly lower level of a member of the vascular endothelial growth factor family, maternal serum placental growth factor, indicating that even in the absence of identifiable impairment of placental perfusion, placental angiogenesis can be affected.⁹²  Similarly, reduced expression of vascular endothelial growth factor receptor 1 and vascular endothelial growth factor receptor 3 in preeclamptic and preeclamptic with hemolysis–elevated enzyme–low platelet count placentas has been documented.⁹³  Microcomputed tomography has been used to visualize the fetoplacental arterial tree in ex-vivo murine placentas, with identification of decreased arterial vascularization in those with exposure to cigarette smoke; however, more advances will be needed to safely use this technique in vivo.⁹⁴ 

To date, our understanding of the effect of general and specific external factors on CHD has been limited by adopting a siloed approach of studying specific exposures and failing to account for the totality and synergy of both maternal and paternal exposures. There are multiple areas in which these factors may synergistically influence proper placental vascularization and subsequent fetal growth and development; however, as of yet, a comprehensive model to assess the complete exposome has not been used. The inherent complexity and resources required to generate such a model have been barriers in furthering our understanding of the mechanisms contributing to the incidence of CHD. Although completion of such a complex project may be beyond the realm of current possibility, continued advances in the emerging field of the exposome and establishment of a bank of exposomic data from epidemiological studies will be required to make significant inroads in development of this model. The process itself of adopting such a framework is imperative in advancing our understanding of the causes of CHD because the model will constantly require scrutinization and additions.

aOR

adjusted odds ratio

CHD

congenital heart disease

CI

confidence interval

CoA

coarctation of the aorta

DM

diabetes mellitus

EBV

Epstein-Barr virus

HLHS

hypoplastic left heart syndrome

NSAID

nonsteroidal antiinflammatory drug

OR

odds ratio

PAH

polycyclic aromatic hydrocarbon

PM2.5

particulate matter <2.5 μm in diameter

PM10

particulate matter <10 μm in diameter

SENP2

small ubiquitin-related modifier–specific protease 2