Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Dec 1.
Published in final edited form as: Environ Res. 2019 Oct 24;179(Pt B):108854. doi: 10.1016/j.envres.2019.108854

Exposure to 17 trace metals in pregnancy and associations with urinary oxidative stress biomarkers

Stephani S Kim 1, John D Meeker 2, Alexander P Keil 1,3, Max T Aung 2, Paige A Bommarito 4, David E Cantonwine 5, Thomas F McElrath 5, Kelly K Ferguson 1
PMCID: PMC6907890  NIHMSID: NIHMS1058731  PMID: 31678726

Abstract

Background:

Exposure to some toxic metals, such as lead and cadmium, has been associated with increased oxidative stress. However less is known about other metals and metal mixtures, especially in pregnant women who are a vulnerable population.

Methods:

To study the relationship between exposure to trace metals and oxidative stress, we analyzed a panel of 17 metals and two oxidative stress biomarkers (8-isoprostane and 8-hydroxydeoxyguanosine [8-OHdG]) in urine samples collected at ~26 weeks gestation from pregnant women in Boston (n=380). We used linear regression models to calculate percent differences and 95% confidence intervals (CI) in oxidative stress markers for an interquartile range (IQR) increase in each urinary metal with adjustment for other metals. In addition, we applied principal components analysis (PCA) and Bayesian kernel machine regression (BKMR), to examine cumulative effects (within correlated groups of exposures as well as overall) and interactions.

Results:

We estimated 109% (95% CI: 47, 198) higher 8-isoprostane and 71% (95% CI: 45, 102) higher 8-OHdG with an IQR increase in urinary selenium. We also estimated higher 8-isoprostane (47%, 95% CI: 20.5, 79.4) and 8-OHdG (15.3%, 95% CI: 5.09, 26.5) in association with urinary copper. In our PCA, we observed higher 8-isoprostane levels in association with the “essential” PC (highly loaded by Cu, Se, and Zn). In BKMR analyses, we also estimated higher levels of both oxidative stress biomarkers with increasing Se and Cu as well as increasing levels of both oxidative stress biomarkers in association with cumulative concentrations of urinary trace metals.

Conclusion:

We observed higher 8-isoprostane and 8-OHdG levels in association with urinary trace metals and elements, particularly Se and Cu, in linear models and using mixtures approaches. Additionally, increasing cumulative exposure to urinary trace metals was associated with higher levels of both oxidative stress biomarkers. The beneficial effects of these compounds should be carefully questioned.

Keywords: metals, prenatal exposure, oxidative stress, mixtures

1. Introduction

Growing evidence suggests that exposure to toxic metals, such as lead (Pb), cadmium (Cd), and arsenic (As), during pregnancy is associated with adverse birth outcomes, such as preterm birth and lower birth weight (Cantonwine et al. 2010; Ferguson and Chin 2017; Jelliffe-Pawlowski et al. 2006; Myers et al. 2010; Perkins et al. 2014; Torres-Sanchez et al. 1999; Wang et al. 2016; Yang et al. 2016). At the same time, deficiency in essential metals and elements, such as zinc (Zn) and selenium (Se), during pregnancy has also been linked to adverse birth outcomes (Hurst et al. 2013; Rayman et al. 2011; Wilson et al. 2016). However, the mechanisms behind these relationships are not well established or understood. One potential mechanism linking metals exposure and adverse birth outcomes could be oxidative stress. Oxidative stress, caused by a pro- and anti-oxidant imbalance in the body, can damage DNA, lipids, and proteins (Betteridge 2000; Cuffe et al. 2017; Halliwell and Cross 1994). Toxic metals, including Pb, Cd, and As, may directly generate reactive oxygen species (ROS) and often deplete or interfere with antioxidants, such as glutathione (Munzel and Daiber 2018; Valko et al. 2005). Meanwhile, metals that are commonly considered essential (e.g., copper [Cu]), can lead to the creation of excess ROS if levels in the body are too high (Jablonska and Vinceti 2015; Valko et al. 2016). Several human studies have found associations between increased exposure to toxic metals and increased biomarkers of oxidative stress (Munzel and Daiber 2018; Munzel et al. 2018). However, other metals remain understudied in this context and may confound existing estimates of metals and oxidative stress due to correlations between them.

In the present study, we examined associations between a panel of 17 trace metals and two biomarkers of oxidative stress measured in urine from pregnant women collected at the beginning of the third trimester. We sought to estimate associations between each metal and oxidative stress marker after controlling for co-exposures. Additionally, we leveraged the availability of multiple metals measurements to address two unique questions using mixtures methodologies. First, using principal components analysis (PCA), we examined associations between oxidative stress and a metals exposure index created based on correlations within exposures. These indices, or principal components (PCs), potentially represent a shared exposure source that could be intervened upon. Second, using Bayesian Kernel Machine Regression (BKMR), we examined whether there was an overall effect of the mixture (i.e., if perhaps beneficial effects of essential metals counterbalanced detrimental effects of toxic metals) and whether associations between individual metals and oxidative stress varied by levels of exposure to other metals in the mixture (i.e., to investigate interaction).

2. Methods

LIFECODES Birth Cohort

The LIFECODES birth cohort is an ongoing prospective birth cohort that began in 2006. The study recruits pregnant women who are ≥18 years old, <15 weeks of gestation, and plan to give birth at Brigham and Women’s Hospital in Boston, MA. At enrollment, participants provide written informed consent and complete a questionnaire covering demographics, lifestyle, and reproductive health. Urine samples are collected at enrollment and at three subsequent study visits (i.e., visits 1–4); the median gestational age at each visit is roughly 10, 18, 26, and 35 weeks of gestation. This study focuses on a subset of pregnant women who delivered from 2006–2008 who were part of a nested case-control study of preterm birth, since the primary objective was to examine the association between urinary trace metals and prematurity (Ferguson et al. 2014; Kim et al. 2018). This case-control study included 482 participants who were selected from all singleton deliveries that occurred between 2006 and 2008. We selected 130 individuals who delivered preterm as well as 352 unmatched controls from the remaining singleton term births from the same time period in an approximately 3:1 ratio (Ferguson et al. 2014). Of this 482, only 390 participants had an available third visit urine sample available for metals analysis. In addition, 10 of the 390 were missing oxidative stress biomarkers. Therefore, our final sample size for this study was 380. The study was approved by the Institutional Review Board at Brigham and Women’s Hospital.

Urinary trace metals

In collaboration with the Children’s Health Exposure Analysis Resource (CHEAR), NSF International (Ann Arbor, MI, USA) analyzed urine samples from the third visit for 17 trace metals and elements using Thermo Fisher (Waltham, MA, USA) ICAPRQ inductively coupled plasma mass spectrometry and CETAC ASX-520 autosampler. Standards of known purity and identity were used during preparation of the calibration, quality control and internal standards. The quality control samples were analyzed at three levels throughout the study (n=40). In the low-level QC, the % CV ranged from 2.2 to 8.5 for all analytes; in the mid-level QC, the % CV ranged from 0.9 to 3.4 for all analytes; and in the high-level QC, the % CV ranged from 1.5 to 3.6 for all analytes. Additional details of the analytic method and quality control procedures are described elsewhere (Kim et al. 2018). We measured the trace metals arsenic (As), barium (Ba), beryllium (Be), cadmium (Cd), chromium (Cr), copper (Cu), lead (Pb), mercury (Hg), manganese (Mn), molybdenum (Mo), nickel (Ni), tin (Sn), thallium (Tl), tungsten (W), uranium (U), and zinc (Zn), and the trace element selenium (Se). Although Se is not a metal, we refer to the group collectively as trace metals throughout the manuscript for simplicity. For concentrations below the limit of detection (LOD), we kept machine-read values, and replaced those reported as zero or missing with the LOD divided by the square root of two. We treated a metal as detect/non-detect in all analyses if detection was below the LOD for >70% of the samples analyzed.

Oxidative stress biomarkers

Cayman Chemical (Ann Arbor, MI, USA) analyzed oxidative stress biomarkers, including 8-isoprostane and 8-hydroxydeoxyguanosine (8-OHdG), in urine samples from the third visit using enzyme immunoassay. 8-isoprostane is a stable measure of oxidative stress that is sensitive to oxidative damage and specific to arachidonic acid peroxidation by ROS, while 8-OHdG is a commonly used measure of oxidative stress that is released when DNA damage is repaired (Roberts and Morrow 2000; Wu et al. 2004). The collection and analytical methods for this analysis have also been described elsewhere (Ferguson et al. 2015). We replaced oxidative stress marker concentrations below the LOD with the LOD divided by the square root of two.

Statistical analyses

We examined demographic and pregnancy factors including maternal age, education, race/ethnicity, type of health insurance, pre-pregnancy body mass index (BMI), and sex of the neonate. We calculated the geometric mean (GM) and interquartile range (IQR) for the urinary metals and oxidative stress biomarkers to present distributions in our study population.

We used linear regression models to calculate unadjusted and adjusted percent differences and 95% confidence intervals (CI) in oxidative stress markers for an IQR increase in urinary metals. Since metal concentrations were not highly correlated in our study population (Pearson coefficient range: 0.12 – 0.77) (Kim et al. 2018) and to compare with the results from the mixtures methods, we mutually adjusted for co-exposure by including all the metals simultaneously in the models. We natural log-transformed metals and oxidative stress biomarkers for normality and to improve model fit.

To adjust for urine dilution, we measured urinary specific gravity in all samples with a digital handheld refractometer. We included specific gravity as a covariate in all models, as recommended by Barr et al. (Barr et al. 2004). Results were similar in models of specific gravity-corrected exposure and outcome biomarkers (not shown).

We used forward stepwise selection to build adjusted models for each oxidative stress biomarker with variables identified a priori as confounders or precision variables. We retained variables in final models if they had a greater than 10% change on the effect estimates, and selected the same set of adjustment factors for models of both oxidative stress biomarkers. Our final adjusted models included specific gravity, maternal age, pre-pregnancy BMI, maternal race (White, Black, Other), maternal education (high school or less, technical school, junior college/some college, ≥ college graduate), type of health insurance (private, public) and gestational age at the time of sample collection as covariates. We modeled all continuous covariates linearly. We examined parity, use of assisted reproductive technology, smoking during pregnancy, and alcohol use in pregnancy for inclusion but did not retain them for final models as they had a minimal influence on effect estimates. We excluded participants with missing covariates from adjusted models. We used inverse probability weights calculated from the inverse probability of case or control selection from the overall LIFECODES cohort to account for the preterm birth case-control study design. Weights did not account for any other characteristics of the study participants. In sensitivity analyses, we examined associations where: 1) we removed preterm births (n=92) and did not include inverse probability weights; and 2) we categorized essential metals and elements into tertiles.

Mixtures Analyses

We used two chemical mixtures methods to address our secondary research questions. First, we applied PCA to reduce our relatively high number of exposures into groupings (i.e., principal components or PCs) based on correlation structure which tend to reflect a shared exposure source (i.e., an opportunity for intervention). To create the PCs, we used specific gravity-corrected urinary biomarkers and excluded the metals with low levels of detection (Be, Cr, U, and W). We provided detailed explanation of how we identified PCs for modeling in our dataset previously (Kim et al. 2018). Briefly, we identified PCs of interest for our analysis based on Eigenvalues, the scree plot, and proportion of the variance explained (O’Rourke and Hatcher 2013). Subsequently, we fit the selected PCs simultaneously into a linear regression model adjusting for the same covariates as our main model, except we removed specific gravity since corrected biomarkers were used to create the PCs. We included inverse probability weights in models of PCs as well.

We then used BKMR to estimate the overall effect of the metals mixture on each oxidative stress biomarker and to assess interaction between each metal and the rest of the metals in the mixture. We fit BKMR using defaults of the “kmbayes” function in R. We additionally allowed for probabilistic Bayesian variable selection in the BKMR fit, which has been demonstrated to have excellent shrinkage properties in many other settings (Bobb et al. 2015; Valeri et al. 2017). We examined associations and corresponding credible intervals, which are a Bayesian analogue of confidence intervals, in a post-processing step which allows simple graphical assessment. We used this post-processing step to first estimate an overall effect of the mixture, which we characterize using the posterior predicted outcome distribution under scenarios in which all metals concentrations are at a certain percentile (e.g., 75th percentile) compared to all metals concentrations held at the median. Estimates are calculated for all percentiles between the 25th and 75th in 5% increments, which allows us to approximate a flexible regression line that characterizes the effect of the mixture as a whole. Second, to determine whether individual metal effects differed based on levels of other metals within the mixture (i.e., effect modification by the rest of the mixture), we used post-processing of the BKMR fit to calculate the expected outcomes over the range of each metal while fixing all other metals at different levels (25th, 50th, 75th percentile) of the other metals in the mixture. Evidence of modification is present if the effect of one metal differs across levels of the other metals.

As in PCA, we excluded the detect/non-detect metals from the BKMR, but we did not use specific gravity-corrected biomarkers. To facilitate efficient Markov chain Monte Carlo (MCMC) sampling, we scaled the outcome variables, exposure variables, and continuous covariates, and created indicator variables for the categorical covariates. We ran BKMR models for 8-isoprostane and 8-OHdG with 100,000 iterations each and included the same covariates as in the linear regression models from our primary analyses. We did not use the inverse probability weights because BKMR software does not accommodate weighting. We examined trace plots and standard Bayesian diagnostics to assess convergence and mixing of the Markov chain (Gelman et al. 2013), and to select a chain length that resulted in adequate precision.

We performed several sensitivity analyses to assess the robustness of our BKMR results to our Bayesian priors. To do so, we examined three additional models under different prior distributions. Whereas the R package defaults include variable selection with a Beta(1,1) prior (an “ignorance” prior that places equal prior probability on a variable being in or out of the model) and an inverse-uniform prior on the “r” parameters of the kernel, which control the smoothness of the flexible function h(Z), we varied these priors to assess whether results changed under 1) no variable selection (prior certainty that all exposures can influence the outcome), 2) stronger priors on the length-scale “r” parameters of the covariance kernel (a Gamma(shape=1,scale=0.2) hyperprior, which expresses prior information that associations between exposure and covariates are closer to linear, relative to defaults), and 3) skeptical variable selection (using a Beta(100,900) hyperprior, which quantifies strong prior information that only 1/10 of the included exposure variables actually influence the outcomes of interest). These scenarios span plausible extremes of prior information allowed under BKMR, and similarity of results across these scenarios would imply that our results are robust to our choice of priors.

We used R version 3.5.1 and the ‘bkmr’ package (Bobb 2017) for the BKMR analysis and SAS 9.4 (Cary, NC, USA) for all other analyses.

3. Results

Table 1 presents the weighted demographic characteristics for the participants included in the present analysis (n=380). Over half of our population was ≥30 years of age (69%), White (60%), had some college (72%), used private health insurance (83%), and had a normal or below normal BMI before pregnancy (57%). Slightly over half of the infants were male (55%). These characteristics were similar to what we observed in the overall case-control study (Ferguson et al. 2015) and in the overall LIFECODES birth cohort (Cantonwine et al. 2010).

Table 1.

Weighted demographic characteristics (n=380)

Demographic Characteristic N Weighted %
Maternal age
 24 years or younger 39 10.9
 25–29 years 74 19.9
 30–34 years 155 40.2
 35+ years 112 29.0
Race/ethnicity
 White 230 60.2
 African American 55 14.5
 Other 95 25.3
Education
 High school degree or less 53 13.7
 Technical college 56 14.6
 Junior college or some college 113 30.9
 ≥ College graduate 148 40.8
 Missing 10
Health insurance
 Private/HMO/self-pay 308 82.5
 Public 63 17.5
 Missing 9
Pre-pregnancy BMI
 <25 kg/m2 209 56.9
 25–30 kg/m2 93 25.7
 >30 kg/m2 68 17.4
 Missing 10
Sex of neonate
 Male 215 55.1
 Female 165 44.9
Open in a new tab

The GM and IQR for all the metals and oxidative stress biomarkers are presented in Table 2 and are generally similar to levels observed in the US and Canada, as reported previously (Kim et al. 2018). Briefly, the participants in our study had higher urinary concentrations of As and Mn compared to women from the National Health and Nutrition Examination Survey (NHANES) and higher Cu, Ni, and Zn compared to women from a nationally representative population in Canada (CDC 2018; Health Canada 2010; Kim et al. 2018). Be, Cr, U, and W had >70% of the samples below the LOD so they were treated as detect/non-detect. Be, in particular, had over 90% below the LOD. Despite the low levels of detection, we included metals with low detection in our analysis for completeness; however, these associations should be interpreted with caution. Approximately 4.5% (n=17) of the participants had values below the LOD for 8-isoprostane, which were replaced with LOD divided by the square root of two. All of the participants had levels above the LOD for 8-OHdG. 8-isoprostane and 8-OHdG were lowly correlated with one another (Ferguson et al. 2015).

Table 2.

Weighted distributions of urinary trace metals and oxidative stress biomarkers from ~26 weeks gestation (n=380)

Urinary Biomarker % < LOD Geometric Mean (IQR)
Metals (ppb)
As 0 15.0 (8.19 – 31.7)
Ba 1.00 1.43 (0.83 – 3.04)
Be 91.3 0.02 (0.01 – 0.03)
Cd 55.9 0.04 (0.04 – 0.13)
Cr 84.6 0.20 (0.09 – 0.28)
Cu 8.20 8.56 (4.60 – 13.6)
Hg 8.20 0.46 (0.20 – 0.98)
Mn 1.50 0.67 (0.42 – 1.07)
Mo 0 48.7 (25.1 – 78.7)
Ni 13.9 2.46 (1.34 – 4.13)
Pb 23.6 0.30 (0.10 – 0.58)
Se 0.77 35.4 (18.0 – 57.4)
Sn 6.15 0.54 (0.26 – 1.27)
Tl 15.6 0.10 (0.05 – 0.18)
U 87.7 0.007 (0.006 – 0.007)
W 79.2 0.14 (0.08 – 0.17)
Zn 0 203.6 (98.1 – 409.5)
Oxidative Stress
8-isoprostane (pg/mL) 4.47 182.5 (85.3 – 341.7)
8-OHdG (ng/mL) 0 122.6 (62.8 – 216.2)
Open in a new tab

Abbreviations: LOD, limit of detection; GSD, geometric standard deviation; ppb, parts per billion; As, arsenic; Ba, barium; Be, beryllium; Cd, cadmium; Cr, chromium; Cu, copper; Hg, mercury; Mn, manganese; Mo, molybdenum; Ni, nickel; Pb, lead; Se, selenium; Sn, tin; Tl, thallium; U, uranium; W, tungsten; Zn, zinc; 8-OHdG, 8-hydroxydeoxyguanosine; pg/mL, picograms per milliliter; ng/mL, nanograms per milliliter

In adjusted models, we observed increases in 8-isoprostane with an IQR increase in urinary Cu (47.0%, 95% CI: 20.5, 79.4%), Mn (12.9%, 95% CI: 1.73, 25.3%), and Se (109.1%, 95% CI: 46.5, 198.4%) (Table 3). We also observed higher 8-OHdG in association with IQR increases of Cu (15.3%, 95% CI: 5.09, 26.5%), Se (70.9%, 95% CI: 44.8, 101.7%), Mo (9.78%, 95% CI: 0.29, 20.2%), and with detection of Be (22.9, 95% CI: 5.27, 43.4%). In addition, we observed an 18.5% decrease in 8-isoprostane in association with an IQR increase of Tl (95% CI: −28.8, −6.75%) and a 36.0% decrease in association with detection of Be (95% CI: −54.1, −10.9%). Some of the associations in the unadjusted model were attenuated, such as Mn and 8-isoprostane and Mo and 8-OHdG, while others were strengthened (Supplemental Table 1). Otherwise, associations in unadjusted and adjusted models were similar. We added a quadratic term of Cu and Se, but saw no improvement in model fit (data not shown). We observed similar patterns between the urinary metals and oxidative stress biomarkers when we removed all the preterm births (Supplemental Table 2). In the tertile analysis of essential metals and elements, associations appeared to be primarily linear for Cu, Se, and Zn, with individuals in the 1st tertile of exposure having lower concentrations of oxidative stress biomarkers compared to individuals in the 2nd tertile of exposure (Supplemental Figure 1).

Table 3.

Weighted adjusted1 difference (95% CI) in oxidative stress biomarkers in association with an IQR increase in each urinary trace metal (n=361)

Metal 8-isoprostane 8-OHdG
As 0.90 (−11.4, 14.9) 1.19 (−4.73, 7.48)
Ba −2.54 (−14.8, 11.5) −1.19 (−7.14, 5.14)
Cd −1.99 (−13.5, 11.0) −4.62 (−9.96, 1.04)
Cu 47.0 (20.5, 79.4) 15.2 (4.98, 26.4)
Hg 9.67 (−4.42, 25.9) 2.83 (−3.54, 9.63)
Mn 12.9 (1.73, 25.3) −1.26 (−5.92, 3.63)
Mo 17.2 (−3.46, 42.3) 9.66 (0.21, 20.0)
Ni −7.32 (−22.0, 10.1) 5.67 (−2.46, 14.5)
Pb 4.01 (−9.43, 19.5) −4.20 (−10.1, 2.09)
Se 109.1 (46.5, 198.4) 71.2 (45.3, 101.7)
Sn 10.2 (−4.81, 27.5) 4.56 (−2.29, 11.9)
Tl −18.5 (−28.8, −6.75) −4.71 (−10.5, 1.44)
Zn −0.62 (−20.0, 23.4) 0.42 (−9.07, 10.9)
Be −36.0 (−54.1, −10.9) 23.2 (5.84, 43.5)
Cr −6.10 (−29.5, 25.1) −3.96 (−16.0, 9.78)
U 31.9 (−5.71, 84.6) −0.89 (−14.8, 15.3)
W −12.0 (−30.6, 11.6) −9.42 (−18.9, 1.17)
Open in a new tab
1

Adjusted for urinary specific gravity, co-exposures, maternal age, race/ethnicity, education, type of insurance, pre-pregnancy BMI, gestational age at sample collection, infant sex. Shading denotes metals analyzed as detect/non-detect. Bold denotes p<0.05. Abbreviations: CI, confidence interval; IQR, interquartile range; 8-OHdG, 8-hydroxydeoxyguanosine.

Mixtures analyses

We identified three PCs that met our criteria for examination in subsequent models (Kim et al. 2018). PC1 had higher loading factors for Cd, Pb, and Mn, PC 2 had higher loading factors for Cu, Se, and Zn, and PC 3 had higher loadings of As, Hg, and Sn (Kim et al. 2018). These components are consistent with what we would expect based on the correlations between metals, where the highest correlations were observed between essential (Zn and Cu, Zn and Se, Se and Cu) and toxic (Pb and Cd, Cd and Mn) pairs (Kim et al. 2018). The first three PCs accounted for approximately 46% of the variance and each had an Eigenvalue greater than one. Based on our knowledge of the metals that loaded highest onto each PC, we labeled PC 1 “toxic”, PC 2 “essential”, and PC 3 “seafood-related.” When we fit the PCs in the linear regression model, the “essential” PC and “seafood-related” PC were both associated with an increase in 8-isoprostane, 55.7% (95% CI: 34.0, 81.0%) and 32.3% (95% CI: 14.2, 53.4%), respectively (Table 4). The “essential” PC was also associated with an increase in 8-OHdG (23.1%, 95% CI: 15.1, 31.7). The “toxic” PC was associated with a 10.6% decrease (95% CI: −16.4, −4.48) in 8-OHdG but was not associated with 8-isoprostane.

Table 4.

Weighted adjusted1 difference (95% CI) in oxidative stress biomarkers in association with principal components (PCs) (n=361)

Metal 8-isoprostane 8-OHdG
PC1: “Toxic” 1.60 (−12.4, 17.8) −10.6 (−16.4, −4.48)
PC2: “Essential” 55.7 (34.0, 81.0) 23.1 (15.1, 31.7)
PC3: “Seafood-related” 32.3 (14.2, 53.4) 6.82 (−0.02, 14.1)
Open in a new tab
1

Adjusted for urinary specific gravity, maternal age, race/ethnicity, education, type of insurance, pre-pregnancy BMI, gestational age at sample collection, infant sex. Bold denotes p<0.05. Abbreviations: CI, confidence interval; 8-OHdG, 8-hydroxydeoxyguanosine; PC, principal component.

In BKMR analyses, we observed increases in 8-isoprostane and 8-OHdG in association with urinary Se, and the association appeared non-linear (with an apparent plateau at higher levels of exposure) in the exposure-response plots (Figure 1). We also observed increases in 8-isoprostane, but not 8-OHdG, as urinary Cu increased in a non-linear fashion where the slope also became level at a certain threshold. Posterior inclusion probabilities (PIPs) indicated that, among all the metals included in the model, Cu and Se contributed most to the overall fit for BKMR models of 8-isoprostane and 8-OHdG (Supplemental Table 3) which was consistent with our primary results. Cu and Se had the highest PIPs with 0.989 and 0.999, respectively, in the 8-isoprostane model. Se had a PIP of 1.0 and the next highest PIP was Cu with 0.277 in the 8-OHdG model.

Figure 1.

Figure 1.

Open in a new tab

Univariate exposure-response function with the 95% confidence bands for the effect of a log unit increase in each urinary trace metal on 8-isoprostane (A) and 8-OHdG (B) while all other trace metals are held at their median1.

1The exposure and outcome variables were scaled and standardized.

When characterizing the overall effect of the mixture, we observed positive associations with both 8-isoprostane (Figure 2A) and 8-OHdG (Figure 2B). While the relationship with 8-isoprostane appeared to be linear, the plot for 8-OHdG suggested a possible non-linear association, where the association leveled off at higher percentiles of the overall metals in the mixture, which is consistent with univariate regression lines in which linear associations with both Cu and Se appeared to diminish at higher exposures.

Figure 2.

Figure 2.

Open in a new tab

Effect estimates and 95% credible intervals of the overall metals mixture on 8-isoprostane (A) and 8-OHdG (B) for increasing (5% increments) percentile of all metals in the mixture compared to all metals being held at the median (reference).

Finally, we used BKMR to examine individual trace metal-response associations (i.e. the difference in the posterior predicted outcomes between the 75th versus the 25th percentile of the index metal) when all other metals were held at the 25th, 50th, or 75th percentile (i.e., to examine whether the effect of the index metal was stronger or weaker at different levels of all other metals combined). For Cu, increasing levels were associated with higher 8-isoprostane while all other metals were held at the 25th, 50th, and 75th percentile, and effect estimates appeared to be dependent on the levels of other urinary trace metals in the mixture (Figure 3A). When all other metals were held at the 75th percentile, there was a greater effect estimate for Cu compared to when all other metals are held at the 25th, which is consistent with a synergistic effect of the metals (though other interpretations are also possible). The interaction was not statistically significant but was suggestive (Supplemental Figure 2), and, in bivariate plots, it appeared that this could have been driven by an interaction between Cu and Se (data not shown). We observed a similar pattern for the interaction between Cu and other trace metals with 8-OHdG; however, associations at each percentile of the metal mixture were null (Figure 3B). For Se, we observed increased 8-isoprostane (Figure 3A) and 8-OHdG (Figure 3B) when other metals were held at the 25th, 50th, or 75th percentiles, but we did not observe evidence for any interaction. Models from sensitivity analyses were very similar to our primary results (Supplemental Figure 3).

Figure 3.

Figure 3.

Open in a new tab

Individual metal risk estimates (95% credible intervals) for each trace metal while all other metals are fixed at the 25th, 50th or 75th percentiles for 8-isoprostane (A) and 8-OHdG (B).

4. Discussion

In a population of pregnant women from the Boston area, we observed associations between urinary concentrations of essential trace metals and oxidative stress biomarkers. Urinary Cu, Mn, Mo, and Se, all essential trace metals or elements, were associated with higher 8-isoprostane, 8-OHdG, or both. In addition, the principal component representing essential metals was associated with higher levels of both oxidative stress biomarkers. Findings from BKMR confirmed our primary findings, highlighting Cu and Se as the strongest predictors for each outcome, and indicated an overall effect of increasing trace metal exposure associations with higher oxidative stress biomarker concentrations. Generally, we did not observe associations between toxic trace metals and oxidative stress biomarkers. These findings were contrary to our a priori hypotheses; however, in previous analyses with adverse birth outcomes in this study population, we observed stronger associations with the essential metals and adverse outcomes (Kim et al. 2018). Additionally, findings from other pregnant populations suggest that higher exposure to essential metals is not always beneficial (Hao et al. 2019; Li et al. 2018; Neggers et al. 2000).

Essential metals are often crucial antioxidants and deficiency has been associated with increased markers of oxidative damage, including DNA oxidation, protein oxidation, and lipid peroxidation (Eide 2011; Jomova and Valko 2012; Powell 2000; Valko et al. 2016). However, depending on dose, essential metals increase oxidative stress as well. Urinary Se, an essential element, had the strongest association with oxidative stress biomarkers in our study, with 109% increase in 8-isoprostane and 71% increase in 8-OHdG in linear models and similar results in BKMR analyses. The women in our study population had higher urinary concentrations of Se compared to a nationally representative population of Canadian women and compared to women from two other pregnancy cohorts from Spain and Australia (Callan et al. 2013; Fort et al. 2014; Health Canada 2010). However, our population had lower levels of urinary Se compared to a study of pregnant women in Michigan (Goodrich et al. 2019). Se is involved in several redox reactions which are essential for human function but may cause oxidative stress and genotoxicity at higher exposure levels (Jablonska and Vinceti 2015; Shamberger 1985). Se reactions with thiols produce superoxide and hydrogen peroxide ROS, and if antioxidant capacity cannot handle this production then oxidative damage could occur (Jablonska and Vinceti 2015; Spallholz 1997). The few human studies examining the effect of Se supplementation on oxidative stress have shown a beneficial effect of supplementation (Asemi et al. 2015; Razavi et al. 2016). However, our findings are in line with one previous study of men and women aged 18 to 85 years that observed higher levels of plasma 8-oxo-7,8-dihydroguanine, another biomarker of oxidative stress, with increasing quintiles of plasma Se (Galan-Chilet et al. 2014).

As with Se, we observed associations between higher levels of Cu and increases in both oxidative stress biomarkers. The women in our population had higher urinary Cu concentrations compared to the nationally representative population in Canada, but similar levels to pregnant women from Michigan and Australia (Callan et al. 2013; Goodrich et al. 2019; Health Canada 2010). Cu can initiate oxidative stress by catalyzing the formation of ROS through a Fenton-like reaction (Ercal et al. 2001; Valko et al. 2016). Also, elevated levels of Cu can decrease the activity of the antioxidant glutathione (Gaetke and Chow 2003; Jomova and Valko 2011). In animal studies, moderate dietary intake of Cu reduces oxidative stress; however, high levels of Cu can lead to an increase (Gaetke and Chow 2003; Powell 2000). While we did not observe a U-shaped curve for the relationship between Cu and oxidative stress in this study population, we might expect that since this is a highly educated population most participants have sufficient dietary intake of Cu, so we may be unable to observe effects at low levels. Alternatively, it is possible that the positive association between Cu and oxidative stress could be attributed to reverse-causality. Pregnant women with higher levels of oxidative stress could have higher uptake of antioxidants like Cu (Mistry et al. 2015), resulting in higher exposure biomarker concentrations.

It is known that toxic metals (e.g., Pb, Cd, As, Hg) can increase oxidative stress by multiple pathways, including damage to the antioxidant defense system by depletion of glutathione and affinity for sulfhydryl groups on other proteins (Ercal et al. 2001; Valko et al. 2005). Further, metals with a 2+ ion state, such as Cd and Pb, can interfere with intracellular calcium ion (Ca2+) concentrations and Ca2+ dependent enzymatic activity, ultimately disrupting intracellular redox conditions and gene expression (Choong et al. 2014; Gorkhali et al. 2016). Several human studies have found associations between higher levels of these metals and oxidative stress biomarkers (Costa et al. 1997; Dobrakowski et al. 2017; Ercal et al. 2001; Gurer-Orhan et al. 2004), however most of these studies are in adult occupational settings. One study from adults in the general US population observed increases in serum γ-glutamyltransferase, another biomarker of oxidative stress, with increasing blood Pb and urinary Cd (Lee et al. 2006). We did not observe increases in oxidative stress biomarkers with the toxic metals in our study possibly because the concentrations of our urinary metals were not as high compared to other populations where these associations have been previously studied (CDC 2018; Health Canada 2010; Kim et al. 2018). For As in particular, our null results could be attributed to the fact that we did not speciate during analysis so we have the total measurements of the organic and inorganic forms, but not the specific types. Inorganic As is highly toxic but organic As is less so, which may have diluted associations in our study (ATSDR 2007a).

We had several limitations to our study. First, we measured all metals in urine when blood or other biomarkers are preferred for certain metals, such as Pb (ATSDR 2007b). Second, as with other environmental studies, there is potential confounding by other environmental exposures that were not measured. This may be of particular concern for Se, which is present in air pollution and has been identified as a contaminant in several Superfund sites (ATSDR 2003; Vinceti et al. 2018). Third, though we adjusted for co-exposures because the metals were not highly correlated in this sample, this may have unintentionally amplified bias (Weisskopf et al. 2018). Fourth, we did not collect information on the mother’s diet during pregnancy, which could have been a confounder of the associations between urinary trace metals and oxidative stress. Finally, while we wanted to explore how correlated groups of exposure were associated with oxidative stress with PCA, we acknowledge that fitting linear regression models with PCs may give unstable effect estimates (Kioumourtzoglou et al. 2014).

In addition to these limitations, this is a cross-sectional analysis. We would expect that this is appropriate for the present analysis, since we hypothesize that recent exposure to metals would be associate with current oxidative stress levels. However, the half-lives of the metals measured in this study vary from a few hours to years. Essential metals tend have shorter half-lives that range between days to weeks, but the more toxic metals have longer half-lives depending on where they are stored in the body (ATSDR 2003, 2007b, 2015). In contrast, the oxidative stress biomarkers have shorter half-lives (Kaviarasan et al. 2009; Lai et al. 2005; Roberts and Morrow 2000), although we have previously shown they have moderate reliability over the course of pregnancy (Ferguson et al. 2015). While this analysis efficiently assayed a large number of metals in a single specimen, future studies addressing this research question should consider measuring metals in the matrix (i.e., urine, blood, or other tissue) that best captures recent exposure.

Limitations notwithstanding, our study had several strengths. Our study measured many metals in a large population of pregnant women during pregnancy, thus reducing concerns about confounding across metals and small sample bias. There are only a small number of studies measuring multiple metals in association with oxidative stress in pregnant women and our paper can add insight into these associations. Additionally, we used two biomarkers of oxidative stress to measure oxidative lipid damage (8-isoprostane) as well as DNA damage (8-OHdG) which reflect different biological processes. Finally, we utilized two mixtures methods, PCA and BKMR, which supported the findings from our multipollutant linear models, indicating that the essential metals and specifically Se and Cu were the most important predictors of 8-isoprostane and 8-OHdG levels from within the mixture. These methods also showed a suggestive interactive effect between Cu and the overall metals mixture, and an association between cumulative trace metals and oxidative stress.

5. Conclusion

We report an increase in oxidative stress biomarkers in association with urinary essential trace metals measured in the beginning of the third trimester of pregnancy, particularly Se and Cu. Elevated oxidative stress levels in pregnancy have been associated with adverse birth outcomes, including preterm birth (Aouache et al. 2018; Biri et al. 2007; Jauniaux et al. 2000; Takagi et al. 2004). Thus, the beneficial effects of these compounds should be carefully questioned.

Supplementary Material

Supplemental materials

Acknowledgments

Funding:

This research was supported in part by the Intramural Research Program of the National Institute of Environmental Health Sciences (NIEHS), National Institute of Health (Z1AES103321). Additional funding was provided by NIEHS (R01ES018872). Trace metal analysis was provided by the Children’s Health Exposure Analysis Research (CHEAR) Program (U2CES026555).

Footnotes

Human Studies Research:

This study was deemed exempt by the Institutional Review Boards (IRB) at the University of Michigan and the National Institute of Environmental Health Sciences. The protocol for the LIFECODES birth cohort was approved by the IRB at Brigham and Women’s Hospital.

Conflicts of interests: Authors declare no conflicts of interest.

References

  1. Aouache R, Biquard L, Vaiman D, Miralles F. 2018. Oxidative stress in preeclampsia and placental diseases. Int J Mol Sci 19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Asemi Z, Jamilian M, Mesdaghinia E, Esmaillzadeh A. 2015. Effects of selenium supplementation on glucose homeostasis, inflammation, and oxidative stress in gestational diabetes: Randomized, double-blind, placebo-controlled trial. Nutrition 31:1235–1242. [DOI] [PubMed] [Google Scholar]
  3. ATSDR. 2003. Toxicological profile for selenium. Atlanta, GA:U.S. Department of Health and Human Services, Public Health Service. [Google Scholar]
  4. ATSDR. 2007a. Toxicological profile for arsenic. Atlanta, GA. [PubMed] [Google Scholar]
  5. ATSDR. 2007b. Toxicological profile for lead. Atlanta, GA. [Google Scholar]
  6. ATSDR. 2015. Public health statement: Cadmium. [Google Scholar]
  7. Barr DB, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, Pirkle JL. 2004. Urinary creatinine concentrations in the us population: Implications for urinary biologic monitoring measurements. Environmental health perspectives 113:192–200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Betteridge DJ. 2000. What is oxidative stress? Metabolism 49:3–8. [DOI] [PubMed] [Google Scholar]
  9. Biri A, Bozkurt N, Turp A, Kavutcu M, Himmetoglu O, Durak I. 2007. Role of oxidative stress in intrauterine growth restriction. Gynecol Obstet Invest 64:187–192. [DOI] [PubMed] [Google Scholar]
  10. Bobb JF, Valeri L, Claus Henn B, Christiani DC, Wright RO, Mazumdar M, et al. 2015. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics 16:493–508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bobb JF. 2017. Bkmr: Bayesian kernel machine regression. Vol. R package version 0.2.0. [Google Scholar]
  12. Bobb JF, Claus Henn B, Valeri L, Coull BA. 2018. Statistical software for analyzing the health effects of multiple concurrent exposures via bayesian kernel machine regression. Environ Health 17:67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Callan AC, Hinwood AL, Ramalingam M, Boyce M, Heyworth J, McCafferty P, et al. 2013. Maternal exposure to metals--concentrations and predictors of exposure. Environ Res 126:111–117. [DOI] [PubMed] [Google Scholar]
  14. Cantonwine D, Hu H, Sanchez BN, Lamadrid-Figueroa H, Smith D, Ettinger AS, et al. 2010. Critical windows of fetal lead exposure: Adverse impacts on length of gestation and risk of premature delivery. J Occup Environ Med 52:1106–1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. CDC. 2018. Fourth report on human exposure to environmental chemicals, updated tables, (march 2018). Available: https://www.cdc.gov/exposurereport/ [accessed 04/01/2018.
  16. Choong G, Liu Y, Templeton DM. 2014. Interplay of calcium and cadmium in mediating cadmium toxicity. Chem Biol Interact 211:54–65. [DOI] [PubMed] [Google Scholar]
  17. Costa CA, Trivelato GC, Pinto AM, Bechara EJ. 1997. Correlation between plasma 5-aminolevulinic acid concentrations and indicators of oxidative stress in lead-exposed workers. Clin Chem 43:1196–1202. [PubMed] [Google Scholar]
  18. Cuffe JS, Xu ZC, Perkins AV. 2017. Biomarkers of oxidative stress in pregnancy complications. Biomark Med 11:295–306. [DOI] [PubMed] [Google Scholar]
  19. Dobrakowski M, Pawlas N, Kasperczyk A, Kozlowska A, Olewinska E, Machon-Grecka A, et al. 2017. Oxidative DNA damage and oxidative stress in lead-exposed workers. Hum Exp Toxicol 36:744–754. [DOI] [PubMed] [Google Scholar]
  20. Eide DJ. 2011. The oxidative stress of zinc deficiency. Metallomics 3:1124–1129. [DOI] [PubMed] [Google Scholar]
  21. Ercal N, Gurer-Orhan H, Aykin-Burns N. 2001. Toxic metals and oxidative stress part i: Mechanisms involved in metal-induced oxidative damage. Curr Top Med Chem 1:529–539. [DOI] [PubMed] [Google Scholar]
  22. Ferguson KK, McElrath TF, Meeker JD. 2014. Environmental phthalate exposure and preterm birth. JAMA Pediatr 168:61–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Ferguson KK, McElrath TF, Chen YH, Loch-Caruso R, Mukherjee B, Meeker JD. 2015. Repeated measures of urinary oxidative stress biomarkers during pregnancy and preterm birth. Am J Obstet Gynecol 212:208 e201–208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ferguson KK, Chin HB. 2017. Environmental chemicals and preterm birth: Biological mechanisms and the state of the science. Curr Epidemiol Rep 4:56–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Fort M, Cosin-Tomas M, Grimalt JO, Querol X, Casas M, Sunyer J. 2014. Assessment of exposure to trace metals in a cohort of pregnant women from an urban center by urine analysis in the first and third trimesters of pregnancy. Environ Sci Pollut Res Int 21:9234–9241. [DOI] [PubMed] [Google Scholar]
  26. Gaetke LM, Chow CK. 2003. Copper toxicity, oxidative stress, and antioxidant nutrients. Toxicology 189:147–163. [DOI] [PubMed] [Google Scholar]
  27. Galan-Chilet I, Tellez-Plaza M, Guallar E, De Marco G, Lopez-Izquierdo R, Gonzalez-Manzano I, et al. 2014. Plasma selenium levels and oxidative stress biomarkers: A gene-environment interaction population-based study. Free Radic Biol Med 74:229–236. [DOI] [PubMed] [Google Scholar]
  28. Gelman A, Carlin J, Stern H, Dunson D, Vehtari A, Rubin D. 2013. Bayesian data analysis, 3rd edition. Boca Raton, FL:CRC Press. [Google Scholar]
  29. Goodrich JM, Ingle ME, Domino SE, Treadwell MC, Dolinoy DC, Burant C, et al. 2019. First trimester maternal exposures to endocrine disrupting chemicals and metals and fetal size in the michigan mother-infant pairs study. J Dev Orig Health Dis:1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Gorkhali R, Huang K, Kirberger M, Yang JJ. 2016. Defining potential roles of pb(2+) in neurotoxicity from a calciomics approach. Metallomics 8:563–578. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Gurer-Orhan H, Sabir HU, Ozgunes H. 2004. Correlation between clinical indicators of lead poisoning and oxidative stress parameters in controls and lead-exposed workers. Toxicology 195:147–154. [DOI] [PubMed] [Google Scholar]
  32. Halliwell B, Cross CE. 1994. Oxygen-derived species: Their relation to human disease and environmental stress. Environ Health Perspect 102 Suppl 10:5–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Hao Y, Pang Y, Yan H, Zhang Y, Liu J, Jin L, et al. 2019. Association of maternal serum copper during early pregnancy with the risk of spontaneous preterm birth: A nested case-control study in china. Environ Int 122:237–243. [DOI] [PubMed] [Google Scholar]
  34. Health Canada. 2010. Report on human biomonitoring of environmental chemicals in canada: Results of the canadian health measures survey cycle 1 (2007–2009). Ottawa, Canada. [Google Scholar]
  35. Hurst R, Siyame EW, Young SD, Chilimba AD, Joy EJ, Black CR, et al. 2013. Soil-type influences human selenium status and underlies widespread selenium deficiency risks in malawi. Sci Rep 3:1425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Jablonska E, Vinceti M. 2015. Selenium and human health: Witnessing a copernican revolution? J Environ Sci Health C Environ Carcinog Ecotoxicol Rev 33:328–368. [DOI] [PubMed] [Google Scholar]
  37. Jauniaux E, Watson AL, Hempstock J, Bao YP, Skepper JN, Burton GJ. 2000. Onset of maternal arterial blood flow and placental oxidative stress. A possible factor in human early pregnancy failure. Am J Pathol 157:2111–2122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Jelliffe-Pawlowski LL, Miles SQ, Courtney JG, Materna B, Charlton V. 2006. Effect of magnitude and timing of maternal pregnancy blood lead (pb) levels on birth outcomes. J Perinatol 26:154–162. [DOI] [PubMed] [Google Scholar]
  39. Jomova K, Valko M. 2011. Advances in metal-induced oxidative stress and human disease. Toxicology 283:65–87. [DOI] [PubMed] [Google Scholar]
  40. Jomova K, Valko M. 2012. Redox cycling mechanisms in the colon. Med Hypotheses 79:418–419. [DOI] [PubMed] [Google Scholar]
  41. Kaviarasan S, Muniandy S, Qvist R, Ismail IS. 2009. F(2)-isoprostanes as novel biomarkers for type 2 diabetes: A review. J Clin Biochem Nutr 45:1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Kim SS, Meeker JD, Carroll R, Zhao S, Mourgas MJ, Richards MJ, et al. 2018. Urinary trace metals individually and in mixtures in association with preterm birth. Environ Int 121:582–590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Kioumourtzoglou MA, Coull BA, Dominici F, Koutrakis P, Schwartz J, Suh H. 2014. The impact of source contribution uncertainty on the effects of source-specific pm2.5 on hospital admissions: A case study in boston, ma. J Expo Sci Environ Epidemiol 24:365–371. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Kupsco A, Kioumourtzoglou MA, Just AC, Amarasiriwardena C, Estrada-Gutierrez G, Cantoral A, et al. 2019. Prenatal metal concentrations and childhood cardiometabolic risk using bayesian kernel machine regression to assess mixture and interaction effects. Epidemiology 30:263–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Lai CH, Liou SH, Lin HC, Shih TS, Tsai PJ, Chen JS, et al. 2005. Exposure to traffic exhausts and oxidative DNA damage. Occup Environ Med 62:216–222. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Lee DH, Lim JS, Song K, Boo Y, Jacobs DR Jr., 2006. Graded associations of blood lead and urinary cadmium concentrations with oxidative-stress-related markers in the u.S. Population: Results from the third national health and nutrition examination survey. Environ Health Perspect 114:350–354. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Li Z, Liang C, Huang K, Yan S, Tao R, Sheng J, et al. 2018. Umbilical serum copper status and neonatal birth outcomes: A prospective cohort study. Biol Trace Elem Res 183:200–208. [DOI] [PubMed] [Google Scholar]
  48. Mistry HD, Gill CA, Kurlak LO, Seed PT, Hesketh JE, Méplan C, et al. 2015. Association between maternal micronutrient status, oxidative stress, and common genetic variants in antioxidant enzymes at 15 weeks׳ gestation in nulliparous women who subsequently develop preeclampsia. Free radical biology and medicine 78:147–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Munzel T, Daiber A. 2018. Environmental stressors and their impact on health and disease with focus on oxidative stress. Antioxid Redox Signal 28:735–740. [DOI] [PubMed] [Google Scholar]
  50. Munzel T, Sorensen M, Schmidt F, Schmidt E, Steven S, Kroller-Schon S, et al. 2018. The adverse effects of environmental noise exposure on oxidative stress and cardiovascular risk. Antioxid Redox Signal 28:873–908. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Myers SL, Lobdell DT, Liu Z, Xia Y, Ren H, Li Y, et al. 2010. Maternal drinking water arsenic exposure and perinatal outcomes in inner mongolia, china. J Epidemiol Community Health 64:325–329. [DOI] [PubMed] [Google Scholar]
  52. Neggers YH, Goldenberg RL, Dubard MB, Cliver SP. 2000. Increased risk of preterm delivery with elevated maternal alpha-fetoprotein and plasma zinc levels in african-american women. Acta Obstetricia et Gynecologica Scandinavica 79:160–164. [PubMed] [Google Scholar]
  53. O’Rourke N, Hatcher L. 2013. A step-by-step approach for using sas for factor analysis and structural equation modeling, second edition. Cary, North Carolina:SAS Institute Inc.. [Google Scholar]
  54. Perkins M, Wright RO, Amarasiriwardena CJ, Jayawardene I, Rifas-Shiman SL, Oken E. 2014. Very low maternal lead level in pregnancy and birth outcomes in an eastern massachusetts population. Ann Epidemiol 24:915–919. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Powell SR. 2000. The antioxidant properties of zinc. J Nutr 130:1447S-1454S. [DOI] [PubMed] [Google Scholar]
  56. Rayman MP, Wijnen H, Vader H, Kooistra L, Pop V. 2011. Maternal selenium status during early gestation and risk for preterm birth. CMAJ 183:549–555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Razavi M, Jamilian M, Kashan ZF, Heidar Z, Mohseni M, Ghandi Y, et al. 2016. Selenium supplementation and the effects on reproductive outcomes, biomarkers of inflammation, and oxidative stress in women with polycystic ovary syndrome. Horm Metab Res 48:185–190. [DOI] [PubMed] [Google Scholar]
  58. Roberts LJ, Morrow JD. 2000. Measurement of f(2)-isoprostanes as an index of oxidative stress in vivo. Free Radic Biol Med 28:505–513. [DOI] [PubMed] [Google Scholar]
  59. Shamberger RJ. 1985. The genotoxicity of selenium. Mutation Research/Reviews in Genetic Toxicology 154:29–48. [DOI] [PubMed] [Google Scholar]
  60. Spallholz JE. 1997. Free radical generation by selenium compounds and their prooxidant toxicity. Biomed Environ Sci 10:260–270. [PubMed] [Google Scholar]
  61. Takagi Y, Nikaido T, Toki T, Kita N, Kanai M, Ashida T, et al. 2004. Levels of oxidative stress and redox-related molecules in the placenta in preeclampsia and fetal growth restriction. Virchows Arch 444:49–55. [DOI] [PubMed] [Google Scholar]
  62. Torres-Sanchez LE, Berkowitz G, Lopez-Carrillo L, Torres-Arreola L, Rios C, Lopez-Cervantes M. 1999. Intrauterine lead exposure and preterm birth. Environ Res 81:297–301. [DOI] [PubMed] [Google Scholar]
  63. Valeri L, Mazumdar MM, Bobb JF, Claus Henn B, Rodrigues E, Sharif OIA, et al. 2017. The joint effect of prenatal exposure to metal mixtures on neurodevelopmental outcomes at 20–40 months of age: Evidence from rural bangladesh. Environ Health Perspect 125:067015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Valko M, Morris H, Cronin MT. 2005. Metals, toxicity and oxidative stress. Curr Med Chem 12:1161–1208. [DOI] [PubMed] [Google Scholar]
  65. Valko M, Jomova K, Rhodes CJ, Kuca K, Musilek K. 2016. Redox- and non-redox-metal-induced formation of free radicals and their role in human disease. Arch Toxicol 90:1–37. [DOI] [PubMed] [Google Scholar]
  66. Vinceti M, Filippini T, Wise LA. 2018. Environmental selenium and human health: An update. Curr Environ Health Rep 5:464–485. [DOI] [PubMed] [Google Scholar]
  67. Wang H, Liu L, Hu YF, Hao JH, Chen YH, Su PY, et al. 2016. Association of maternal serum cadmium level during pregnancy with risk of preterm birth in a chinese population. Environ Pollut 216:851–857. [DOI] [PubMed] [Google Scholar]
  68. Weisskopf MG, Seals RM, Webster TF. 2018. Bias amplification in epidemiologic analysis of exposure to mixtures. Environ Health Perspect 126:047003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Wilson RL, Grieger JA, Bianco-Miotto T, Roberts CT. 2016. Association between maternal zinc status, dietary zinc intake and pregnancy complications: A systematic review. Nutrients 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Wu LL, Chiou CC, Chang PY, Wu JT. 2004. Urinary 8-ohdg: A marker of oxidative stress to DNA and a risk factor for cancer, atherosclerosis and diabetics. Clin Chim Acta 339:1–9. [DOI] [PubMed] [Google Scholar]
  71. Yang J, Huo W, Zhang B, Zheng T, Li Y, Pan X, et al. 2016. Maternal urinary cadmium concentrations in relation to preterm birth in the healthy baby cohort study in china. Environ Int 94:300–306. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental materials
NIHMS1058731-supplement-Supplemental_materials.docx (86.2KB, docx)

RESOURCES