The Imaging Innovations for Placental Assessment in Response to Environmental Exposures (PARENTs) study recruited a cohort of 199 women early in pregnancy from antenatal clinics at the University of California Los Angeles during 2016–2019 [18]. Women were enrolled as early as the 10th week of gestation and were asked to participate in a once-per-trimester and at-birth study visit that included urinary sample collections we used for oxidative stress biomarkers and metabolites of hydroxyl polycyclic aromatic hydrocarbons (OH-PAHs) assessment. Phone interviews were conducted at three timepoints during pregnancy and at birth to collect environmental and behavioral risk factor data.
Our study population consists of 159 women enrolled in the PARENTs study for whom at least one useable urine sample was available at the time of laboratory analysis supported by the Emory Children’s Health Exposure Analysis Resource (CHEAR) program.
2.2 Biomarkers assessment
We focused on two oxidative stress biomarkers, malondialdehyde (MDA) and 8-hydroxy-2’-deoxyguanosine (8-OHdG), and a total of 7 hydroxyl PAH metabolites (OH-PAHs). Specifically, we examined measures for the combined 2-hydroxyfluorene + 3-hydroxyfluorene (2&3-FLUO) metabolites, the single metabolites 2-hydroxynaphthalene (2-NAP), 1-hydroxyphenanthrene (1-PHEN), 2-hydroxyphenanthrene (2-PHEN), 3-hydroxyphenanthrene (3-PHEN), 4-hydroxyphenanthrene (4-PHEN), and 1-hydroxypyrene (1-PYR), and also the sum of 1-, 2-, 3-, and 4- hydroxyphenanthrene (Σ4OH-PHEN), and of all 7 analytes (Σ7OH-PAH), respectively.
Study visits were timed to be optimal for Magnetic resonance imaging (MRI) evaluations (1st MRI 14th -18th gestational week, 2nd MRI 19th -24th gestational week) in the PARENTs cohort study, and sample collection followed this schedule with the 1st sample being collected in the 10-17th gestational week, the 2nd sample collection in the 18-29th gestational week, and the 3rd sample collection after the 30th gestational week and prior to delivery. Maternal urinary samples were collected at each study visit, and we collected at least one and at most three urine samples from all participants during pregnancy. Specifically, multiple urine samples were collected among 146 out of 159 participants (92%).
The urine samples were stored at -80°C after collection at UCLA and were shipped on dry ice to the CHEAR Laboratory Hub to measure both the OH-PAHs and the oxidative stress biomarkers. All samples were randomized using a Fisher-Yates shuffling algorithm prior to laboratory analysis to reduce any potential batch effects [19, 20]. The OH-PAHs were measured by tandem mass spectrometry (MS/MS) [21], and the oxidative stress biomarkers (MDA and 8-OHdG) were measured by liquid chromatography-mass spectrometry (LC-MS) [22].
Samples with measures below the limit of detection (LOD) values were replaced with the LOD/√2 [23]. Concentrations were further corrected for urine dilution by adjusting for specific gravity (SG) measured with a Reichert AR200 refractometer. We excluded samples with an invalid SG value below 1 (N = 14, 18 and 16 samples during the 1st, 2nd and 3rd sample collection interval, respectively), resulting in a total of 391 samples available for analysis. To correct for the hydration status of pregnant women, SG-standardized biomarker concentrations were calculated using the following formula [24]:
$${CHEM}_{SG\_Adj }={CHEM}_{i}*\left[\right({SG}_{m}-1)/({SG}_{i}-1)]$$
where CHEMSG_Adj is the specific gravity-standardized biomarker concentration (nmol/L for MDA, ng/mL for 8-OHdG, ng/L for OH-PAHs), CHEMi is the observed biomarker concentration, SGi is the specific gravity of the urine sample and SGm is the median specific gravity for the total samples with valid SG values.
2.3 Covariates
Information on maternal age (years), parity (continuous), maternal pre-pregnancy body mass index (BMI) (< 18.5, 18.5–24.9, 25.0–30.0 and ≥ 30.0), maternal race/ethnicity (White, non-White), smoking (yes, no), maternal educational attainment (bachelor’s degree or less, master’s degree, doctoral/professional degree) were collected in interviews. Gestational age (based on the best obstetric estimate obtained during a 1st trimester ultrasound exam), as well as information about pregnancy complications including gestational diabetes, gestational hypertension, and pre-eclampsia were obtained from medical records. Season of sample collection was categorized based on the month of sample collection, as spring (March, April, May), summer (June, July, August), fall (September, October, November) and winter (December, January, February).
2.4 Statistical analysis
First, we estimated effects for different time intervals during pregnancy by first conducting multiple linear regression analyses and calculated the expected percentage of change in each oxidative stress biomarker concentration according to PAH metabolite levels in each sample collection interval, separately. The oxidative stress biomarker concentrations were treated as continuous variables and log-transformed for statistical analyses. For all OH-PAHs, we log-transformed (base 2) the values such that in statistical model the exposure effect estimate represents an increase per doubling of the OH-PAHs concentration (ng/L). Second, we also used linear mixed models with a random intercept for participant to take repeated measurements into account, i.e., we relied on up to 3 samples collected across pregnancy for each exposure and biomarker and assessed associations between OH-PAHs exposure urine measures and oxidative stress concentrations across multiple time points in pregnancy.
In one-point-in-time linear and in longitudinal linear mixed regression models, we adjusted for maternal age, maternal race/ethnicity, maternal education, parity, pre-pregnancy BMI, and season of sampling. We also conducted stratified one-point-in-time linear regression analyses by season of sampling, fetal sex, and maternal race/ethnicity to evaluate potential effect measure modification. Sensitivity analyses were conducted by additionally adjusting for gestational day at sample collection. Furthermore, as women experiencing pregnancy complications are likely to have higher oxidative stress levels due to these conditions [25], we also restricted some analyses to women without pregnancy complications specifically gestational diabetes, gestational hypertension, or pre-eclampsia. All statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).