The distributions of PAH measurements in air and urine, BaP–DNA adducts, and DNA methylation in our study population are described in . Median BaP and total PAH concentrations were 0.24 and 2.47 ng/m3, respectively. BaP–DNA adducts were detected in half of all participants. DNA methylation ranged from 0.06 to 5.42 ng/100 ng total DNA. Demographic characteristics in this study sample were similar to those of the underlying cohort (). Two-thirds of the participants were Dominican, and one-third were African American. Most participants were unmarried and had an annual household income < $30,000. Consistent with the study sample selection criteria, half (49.4%) of the study sample had prenatal air PAH levels above the population median of 2.3 ng/dL, and half (50.6%) had air PAH levels below the median (). The proportion of participants with detectable BaP–DNA adducts in this sample (50%) was significantly lower than in the underlying cohort (67%).
Distribution of PAH exposure in air and urine, BaP–DNA adducts, and DNA methylation in our study population.
Characteristics of subjects included in this analysis and the underlying cohort [n (%)].
Indicators of PAH exposure. We used airborne PAH exposure and PAH urinary metabolites measured in maternal urine collected prenatally from a subset of 87 women to estimate prenatal PAH exposure and internal dose, respectively. Demographic characteristics were not associated with PAH exposure indices, with a few exceptions: African Americans were significantly more likely than Dominicans to have high geometric mean levels of 1-hydroxynaphthalene (3117.1 vs. 1497.5 ng/L urine), 2-hydroxyfluorene (337.1 vs. 231.2 ng/L urine), and 4-hydroxyphenanthrene (80.3 vs. 53.0 ng/L urine). Women with higher income had lower levels of 1-hydroxypyrene: Women with annual household incomes < $10,000 had geometric mean 1-hydroxypyrene concentration of 187.8 ng/L urine, compared with 137.2 and 130.2 ng/L urine among women making $10,000–30,000 and > $30,000, respectively.
Mothers exposed to ETS during their pregnancy were, on average, younger, more likely to be African American than Dominican, less likely to be married, and more likely to have lower annual household income than were mothers who were not ETS exposed (data not shown). Those with ETS exposure had higher levels of 3-hydroxyfluorene (56.2 vs. 37.5 ng/L urine, p = 0.04). African Americans tended to consume more PAH-containing meats than did Dominican participants. However, our index of PAHs from smoked, broiled, fried, or barbequed meat was not correlated with any of our PAH metabolite measures (r = –0.11 to 0.19).
The total PAH level in air was positively correlated with pyrene and BaP (Pearson’s r = 0.69 and 0.96, respectively), and individual PAH metabolites were also correlated with one another (r = 0.30–0.89). The total airborne PAH level was only moderately correlated with 2-hydroxyfluorene, 2-, 3-, and 4-hydroxyphenanthrene, and 1-hydroxypyrene (r = 0.20–0.26) and was not correlated with the other measured PAH metabolites. Pyrene in air was moderately correlated with all measured metabolites (r = 0.20–0.49), except for the naphthalene metabolites.
PAH exposure and DNA methylation. Using univariate linear regression, we found that cord blood global methylation was lower (p < 0.01) among African Americans (geometric mean = 1.04 ng/100 ng total DNA) than among Dominicans (1.35 ng/100 ng total DNA). However, no other demographic variables were significantly associated with global methylation levels.
Natural log–transformed methylation was negatively associated with the natural log of prenatal air PAH exposure [β = –0.11; 95% confidence interval (CI): –0.21, 0.00; p = 0.05]. Associations were similar for pyrene (β = –0.18; 95% CI: –0.32, –0.04; p = 0.01) and for BaP (–0.09; 95% CI: –0.18, 0.00; p = 0.06). Methylation levels among participants with BaP, pyrene, and total PAHs above the median (“high”) were lower than methylation levels among those with BaP, pyrene, and total PAHs below the median (“low”) (p = 0.08, 0.03, and 0.01, respectively) (). Participants exposed to prenatal ETS had levels of methylation similar to levels in those who were not exposed (1.25 vs. 1.24 ng/100 ng total DNA, p = 0.90). When ethnicity was added to the multivariate linear regression models to assess confounding, none of these associations (β-coefficients) changed, leading us to conclude that the associations were not confounded by ethnicity.
Figure 2 Global DNA methylation in cord blood according to prenatal air PAH exposure (based on univariate analyses). “High” and “Low” represent the geometric mean methylation (and 95% CIs) for those with prenatal concentrations (more ...)
Participants with PAH urinary metabolite data were similar to the rest of the sample in terms of total air PAH concentrations and proportion of detectable DNA adducts, but they had significantly higher average global DNA methylation (geometric mean of 1.00 vs. 1.51 ng/100 ng DNA, p < 0.01). In this subset of participants (n = 87), a slightly positive but nonsignificant association was observed between metabolite levels and global DNA methylation when two heavily influential observations were removed (all p-values > 0.10) (data not shown).
PAH exposure and BaP–DNA adduct formation. No association was observed between any of the covariates presented in and the presence of BaP–DNA adducts (p > 0.10), except for parity: Nulliparous women were less likely to have detectable adducts (p = 0.06) than were multiparous women. Detectable adducts were not associated with any of the PAH exposure indicators, including all measures of air exposure, ETS, dietary PAHs from meat, and (in the subset for whom they were measured) PAH metabolites (data not shown).
DNA methylation and BaP–DNA adduct formation. Newborns with detectable cord BaP–DNA adducts had higher levels of genomic methylation than did those with nondetectable adducts (odds ratio = 2.35; 95% CI: 1.35, 4.09; p < 0.01) (). None of the covariates presented in confounded this association. In this analysis, global methylation levels did not mediate or modify the relationship observed between PAH exposure and the presence of detectable adducts.
Figure 3 Probability of detectable BaP–DNA adducts as a function of global methylation in cord blood (based on univariate association). Circles represent the predicted probability of adduct detection based on logistic regression analyses where global methylation (more ...)