We evaluated associations between prenatal exposure to select PAH metabolites and childhood airway outcomes in a diverse, multi-site cohort with comprehensive covariate characterization. We did not observe strong evidence for associations between concentrations of urinary OH-PAH metabolites—individually or in mixtures—and child asthma at age 8-9 or wheezing trajectories in childhood. Counter to our hypothesis that PAH exposure has adverse impacts on airway health, urinary concentrations of 1-NAP were inversely associated with asthma in fully-adjusted models and most sensitivity analyses. This seemingly protective association was attenuated among girls, though evidence for effect modification was lacking. Concentrations of 1/9-PHEN were positively associated with asthma and persistent wheezing in several sensitivity analyses and among girls, with borderline evidence for effect modification.
Several prior studies have evaluated the association between maternal PAHs and child airway outcomes, though most have focused on early childhood outcomes based on reported respiratory signs and symptoms. Using a similar study design, Loftus et al (2022) explored links between maternal OH-PAHs and asthma and wheeze in early childhood (age 4-6 years) in the CANDLE and TIDES cohorts. The authors identified null associations for most exposure-outcome associations, though with borderline evidence of an adverse effect of 1-NAP and 1/9-PHEN on current asthma at age 4-6 years. Notably, evidence of effect modification by child sex was strong, with elevated associations between OH-PAHs (except for 1-NAP) and current asthma among girls only; interaction p-values were statistically significant for metabolites including 1/9-PHEN, 3-PHEN, 2-PHEN, and 2-NAP.
Other studies of prenatal exposure to PAHs and child airway outcomes have largely focused on HMW parent PAHs, yielding mixed results. Differences in specific analytes measured, study design, timing of exposure assessment, age of outcome assessment, and/or case definition among prior studies complicate comparison with the present analysis. Furthermore, these studies have generally relied on small sample sizes in relatively homogenous populations, derived from three pediatric cohorts. In a birth cohort of 339 mother-infant pairs in Krakow, Poland, researchers measured personal exposure to HMW PAHs including benzo(a)anthracene, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(g,h,i)perylene, benzo(a)pyrene, chrysene/iso-chrysene, dibenzo(a,h)anthracene, indeno(1,2,3-c,d)pyrene, and pyrene over 48 hours during the second trimester of pregnancy (Jedrychowski et al., 2005, 2014, 2015). Increased maternal PAH exposure was associated with increased number and duration of respiratory symptoms (including ear infection, cough, barking cough, wheezing without cold, and sore throat) during the child’s first year of life (Jedrychowski et al., 2005) and with increased incidence risk ratio of wheeze during the first four years of life (Jedrychowski et al., 2014). Prenatal exposure to PAHs was also assessed by DNA adducts of benzo(a)pyrene in umbilical cord blood in the same cohort; elevated benzo(a)pyrene DNA adducts were associated with increased risk of wheezing during the first two years of life, but not with wheezing events during the third and fourth year of life. (Jedrychowski et al., 2010).
In the Columbia Center for Children’s Environmental Health cohort, personal exposure to pyrene and 8 HMW PAHs (benz(a)anthracence, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(g,h,i)perylene, benzo(a)pyrene, chrysene/isochrysene, dibenz(a,h)anthracene, and indeno(1,2,3)pyrene) was measured over 48-hours during the third trimester of pregnancy among a sample of 303 Dominican and African American women in New York City and airway outcomes were measured throughout infancy and early childhood (Jung et al., 2014; Miller et al., 2004; Rosa et al., 2011). The researchers found prenatal PAH exposure in combination with environmental tobacco smoke exposure was associated with increased risk of cough and wheeze at 12 months and probable asthma at age 24 months, with significant interaction between PAH and ETS exposure (Miller et al., 2004). This interaction persisted at age 5-6, with associations between prenatal exposure and asthma observed only among children with exposure to ETS (Rosa et al., 2011). Higher levels of prenatal and postnatal pyrene exposure were also associated with increased asthma and wheeze among non-atopic children (but not children with allergic sensitization) at age 5–6 years (Jung et al., 2014).
Most prior studies have evaluated associations between prenatal PAH exposure and airway outcomes in early childhood. Our analysis additionally considers asthma and wheezing trajectories in middle childhood that may represent distinct etiologies or clinical presentations. We evaluated three wheezing phenotypes identified by The Tucson Children’s Respiratory Study (TCRS) based on clinical observations throughout childhood: early, late, and persistent wheeze (Martinez et al., 1995). Similar phenotypes have also been identified through data-driven analyses of wheezing and asthma symptoms in other cohorts (Henderson et al., 2008; Savenije et al., 2011). Children with these phenotypes have been found to have differences in allergic sensitization, lung function, genetic factors, and risk factors for asthma development (Trivedi & Denton, 2019). For example, early wheeze that resolved by middle childhood is generally associated with only mild lung function impairments that do not usually require medication use (Depner et al., 2014). Late-onset wheezing has been linked to high likelihood of allergic sensitization (Depner et al., 2014; Henderson et al., 2008). Children with both late-onset and persistent wheezing have been found to have decrements in lung function and airway responsiveness, while persistent wheezing is generally associated with the highest rate of asthma diagnosis (Depner et al., 2014; Henderson et al., 2008; Savenije et al., 2011).
Our primary analysis did not identify associations between maternal OH-PAHs and any wheezing phenotype, though 1/9-PHEN was significantly associated with persistent wheezing (but not other wheezing trajectories) in models mutually adjusted for other OH-PAHs and among girls only in effect modification analyses. Thus, we observed similar trends in effect estimates and modification by child sex for the outcomes of persistent wheezing and diagnosed child asthma at age 8-9 years, potentially suggesting a similar underlying mechanism and/or overlap in affected populations due to high rates of asthma diagnosis among children with persistent wheeze. While no prior studies have evaluated associations between prenatal PAH exposure and wheezing trajectories, a recent meta-analysis identified associations between postnatal traffic-related air pollution and transient and persistent asthma/wheezing, but not late-onset asthma/wheezing (Lau et al., 2018). Studies have also linked prenatal exposure to CO, NOx, and PM2.5 to persistent asthma/wheeze specifically (Pennington et al., 2018).
Our analysis of maternal OH-PAHs and secondary asthma phenotypes in children (asthma with recent exacerbation and asthma with atopy) did not yield strong associations. Findings from the Columbia Center for Children's Environmental Health birth cohort suggested that maternal personal pyrene measurements during pregnancy were associated with child asthma and wheeze at 5-6 years only among non-atopic children (Jung et al., 2011, 2014; Jung, Yan, et al., 2012), and that prenatal PAH exposure was not associated with the development of seroatopy at age 5-6 years (Rosa et al., 2011). While few studies have evaluated prenatal PAH exposure in association with asthma exacerbation, Jung et al. (2011) identified associations between high prenatal pyrene levels and emergency room visits for asthma before age 5-6. Further research is needed to characterize the role of prenatal PAH exposure in the development of seroatopy, atopic vs nonatopic asthma, and asthma severity in childhood.
The predominantly null results in our study may indicate that maternal OH-PAH levels in our study population are not associated with child asthma and wheezing trajectories at age 8-9. The OH-PAH concentrations measured in our study were similar to those identified by Cathey et al. (2018) among pregnant individuals in Boston (N = 200) and Puerto Rico (N = 50) with the exception of 2-NAP, for which geometric mean concentrations were higher in participants from Puerto Rico relative to those in Boston and our study population. Associations between OH-PAH levels and airway outcomes may be more modest than our study was powered to detect, particularly for asthma phenotypes with low prevalence in the study population, or may be limited to OH-PAH concentrations higher than those measured in the study population. We did observe an inverse (protective) association between 1-NAP and asthma, which is not easily explained. There may be plausible mechanisms by which prenatal PAH exposure could prevent or reduce airway symptoms (Rosa et al., 2011), for example by inhibiting B cell growth (Mann et al., 2001) or inducing pre-B cell apoptosis (Allan et al., 2006), which could be hypothesized to downregulate IgE production and thereby lead to a decrease in respiratory symptoms related to atopic asthma. However, most prior research has focused on the mechanisms by which prenatal PAH exposure may adversely affect childhood airway outcomes. While these mechanisms are not fully understood, adverse associations have been hypothesized to derive from oxidative stress and inflammation induced by PAH exposure. For example, PAH exposure from diesel exhaust was shown to generate reactive oxygen species in macrophages and epithelial cells that activated inflammatory signaling pathways, leading to upregulation of genes involved in regulating immune response (Li et al., 2003). Both LMW and HMW PAHs are also lipid-soluble and transferred across the placenta, and inflammatory responses and oxidative stress induced by PAH exposure is also hypothesized to adversely affect fetal development (Autrup et al., 1995; Perera et al., 1999). Gestational exposure to PAHs may also trigger transcriptomic and epigenetic changes (Herbstman et al., 2012). For example, a recent study found that maternal urinary OH-PAH metabolites were associated with changes in the placental transcriptome (Paquette et al., 2023). In particular, phenanthrene metabolites showed strong associations with expression of TRIP13 and genes related to vitamin absorption/digestion, consistent with our finding of stronger adverse associations for phenanthrene metabolite 1/9-PHEN relative to other OH-PAHss.
We found some evidence for effect modification by child sex, with statistically significant associations between 1/9-PHEN and the outcomes of age 8-9 asthma and persistent wheeze among girls only. However, the interaction p-values for these associations reached only borderline statistical significance; in contrast, Loftus et al. (2022), identified strong evidence for effect modification by sex in the association between maternal OH-PAH levels and asthma and wheeze at age 4-6 in a similar study population. Few other studies have evaluated effect modification by child sex in the relationship between PAH exposure and airway outcomes. Liu et al. (2016) identified some elevated associations between urinary 2-PHEN and asthma at age 13-19 among boys and between 4-PHEN and asthma at age 13-19 among girls, though most exposure-outcome associations were not significantly different between sexes and no formal tests of interaction were conducted. Majewska et al. (2018) found that lung function growth trajectories were lower among girls exposed prenatally to HMW PAHs, but did not differ by exposure among boys, though the interaction between child sex and exposure was not statistically significant. A recent study of maternal OH-PAHs and placental gene expression may offer a potential mechanism to explain a stronger adverse effect on airway development and function in females; relative to males, placentas from female births had over 60% more genes for which expression was associated with OH-PAH concentrations, including genes related to vitamin absorption (Paquette et al., 2023). More research is needed to clarify potential effect modification by child sex and the role of specific parent compounds and metabolites.
Our findings were somewhat sensitive to model specification and certain stratified analyses. The associations between 1/9-PHEN and multiple outcomes (asthma, persistent wheeze, asthma with recent exacerbation, and asthma with atopy) were elevated in models mutually adjusted for all OH-PAHs, relative to individual metabolite models. Mutually-adjusted models have been used to control for confounding by co-occurring OH-PAHs, which showed moderate to high correlation between metabolite concentrations (rho = 0.34-0.75), but can also amplify bias an lead to inflation of effect estimates and variance (Chiu et al., 2018; Weisskopf et al., 2018). In cohort-specific analyses, the associations between all OH-PAHs except 1-NAP and current asthma were elevated in TIDES relative to CANDLE. However, cohort-specific estimates of associations were also imprecise with overlapping confidence intervals. Thus, the differences in estimates by cohort may be driven by random noise, particularly given the small sample size of TIDES participants relative to CANDLE.
Our study has several notable strengths. As described above, the use of airway data from multiple time points also enabled us to evaluate wheezing trajectories that may provide insight into wheeze phenotypes with distinct environmental risk factors and etiologies. In addition to longitudinal outcome assessment, we also leverage data on reported asthma diagnosis in middle childhood, when asthma persistence can be more easily identified and diagnosed relative to early childhood. Our analysis used prospective data from two cohorts across five cities, representing participants with diverse demographic and socioeconomic backgrounds. These cohorts also collected comprehensive data on covariates and precision variables, enabling a high degree of statistical adjustment for potential confounders.
Our study also has various limitations. First, our estimates of PAH exposure are derived from only one spot urine sample collected during mid-pregnancy. Urinary OH-PAHs represent short-term exposure, with half-lives in the range of 2.5 – 12 hours (Brzeźnicki et al., 1997; Li et al., 2012). Prior studies of urinary OH-PAH concentrations across pregnancy have identified intraclass correlation coefficients (ICCs) that vary widely between OH-PAH analytes and among sub-populations, ranging from 0.04 to 0.73 (Cathey et al., 2018; Dobraca et al., 2018; Gaylord et al., 2022; Zhu et al., 2021). Measurement from a single time point may not represent pregnancy average exposures, and noise introduced by short-term variations in exposure may contribute to null findings and obscure associations with adverse airway outcomes. Our exposures of interest also included only metabolites of LMW PAHs, namely naphthalene and phenanthrene, thus we did not capture the full range of PAHs potentially influencing asthma outcomes. HMW PAHs are predominantly excreted in feces, rather than urine (Agency for Toxic Substances and Disease Registry, 1995). Urinary metabolites of three HMW PAHs, including benz[a]anthracene, benzo[c]phenanthrene, and chrysene, were measured in our study but only detected in a small percentage of samples and were not included in the analysis. In addition, 1-NAP is a metabolite of both naphthalene and the carbamate pesticide carbaryl (Meeker et al., 2007) and so we cannot distinguish associations with either parent compound for this metabolite. We further cannot rule out the potential for residual confounding due to imprecisely measured or unmeasured covariates, including postnatal PAH exposure.
Both asthma and wheezing outcomes relied on maternal report of recent symptoms of airway obstruction and diagnoses and may be affected by imprecise recall. Misclassification of both outcomes would most likely be nondifferential with respect to exposure, potentially contributing to null findings. Longitudinal follow-up of children at more time points (e.g., annually) would result in more consistent and accurate measures of wheezing phenotypes. Nonetheless, the use of data from two time points in middle- and late-childhood to classify wheezing phenotypes has precedent in the literature (Abellan et al., 2022; Jung, Hsu, et al., 2012; Martinez et al., 1995) in situations where more frequent outcome assessment were not possible. Furthermore, access to reported asthma diagnosis in middle childhood is a strength of the study; prior research on PAH exposure and pediatric airway outcomes has largely focused on outcomes in early childhood, when asthma diagnosis is difficult and objective measurements of airway obstruction and hyperreactivity are often not feasible. Finally, clinical histories derived from ISAAC questionnaire data are widely used to classify asthma outcomes in epidemiological studies and clinical settings (NHLBI, 2007).
In conclusion, in this large, multi-city study of maternal urinary OH-PAHs and airway outcomes in middle childhood, we did not identify strong evidence for associations between OH-PAHs and child airway outcomes. We identified borderline evidence for effect modification by child sex, with adverse associations between 1/9-PHEN and asthma and persistent wheezing among girls. Future research into the potential sex-specific impacts of prenatal PAH exposure on airway health throughout childhood are merited. Replication studies, particularly those with repeated exposure assessment throughout pregnancy, would improve scientific understanding of these linkages and the role of specific PAH parent compounds and metabolites.