Secondary Logo

Journal Logo

Original Research Article

Ambient air pollution during pregnancy and cardiometabolic biomarkers in cord blood

Friedman, Chloea,b,*; Dabelea, Danaa–c; Bloemsma, Lizan D.a,b; Thomas, Deborah S. K.d; Peel, Jennifer L.e; Adgate, John L.f; Magzamen, Sherylf,g; Martenies, Sheena E.e; Allshouse, William B.f; Starling, Anne P.a,b,h

Author Information
Environmental Epidemiology: April 2022 - Volume 6 - Issue 2 - p e203
doi: 10.1097/EE9.0000000000000203

Abstract

What this study adds

Ambient air pollution exposure during gestation has been associated with adverse childhood cardiometabolic outcomes, although it is unknown whether markers of exposure-related altered cardiometabolic profile are identifiable at birth. In the present study, ozone during several periods of pregnancy was consistently associated with higher insulin and lower glucose/insulin ratio in cord blood. Our work highlights the potential for in utero exposure to air pollution to impact markers of neonate cardiometabolic status at birth.

Introduction

In the United States, the prevalence of cardiometabolic diseases in childhood, including obesity,1 type 2 diabetes,2 and other cardiovascular disease risk factors,3,4 has risen markedly in recent decades. While dietary factors and increasingly sedentary lifestyle have played a major role in the rising prevalence of childhood obesity, exposure to higher levels of air pollution early in life has also been posited as a potential explanatory factor.5,6

The prenatal period may represent a critical time for environmental chemical exposures to impact health later in life.7,8 For example, maternal exposure to ambient air pollution during pregnancy has been associated with offspring obesity and adverse cardiometabolic outcomes during childhood and beyond, in both animal9,10 and human studies.6,11–14

Despite these suggestive findings, understanding of the potential pathophysiological mechanisms of these observed associations between prenatal air pollution and childhood cardiometabolic outcomes remains limited. In utero exposure to air pollutants may adversely affect offspring’s metabolic function and promote the development of childhood obesity.15,16 Further, disturbances in leptin and other adipokines in cord blood have been associated with cardiometabolic disease markers in infancy and childhood,17–21 although there are sparse data describing this relationship with other cardiometabolic biomarkers.22 Alterations in cord blood cardiometabolic profiles may indicate both short-term, and potentially long-term, consequences of in utero air pollution exposure for offspring metabolic health.

A small number of epidemiologic studies have investigated air pollution-related changes in biomarkers of cardiometabolic function at birth in umbilical cord blood.23–26 Both Alderete et al23 and Lavigne et al24 found associations between maternal air pollution exposure and higher levels of cord blood leptin and adiponectin, two adipocyte-secreted hormones associated with adiposity at birth and in childhood.18,27 Notably, Lavigne et al24 observed a significant association between maternal air pollution exposure and cord blood leptin only after adjustment for birthweight z-score, suggesting that for any given birth weight, the relative size of the fat compartment may be altered by exposure. Further, in a cross-sectional study of 150 mother-infant pairs, maternal exposures to particulate matter of aerodynamic diameter <1 µm (PM1), 2.5 µm (PM2.5), and 10 µm (PM10) were associated with lipid biomarkers in cord blood, including triglycerides and cholesterol.26 Finally, Madhloum et al25 reported that higher maternal PM2.5 and PM10 exposures during pregnancy were associated with higher cord blood insulin. However, no study has examined multiple markers of cardiometabolic function concomitantly, nor have previous studies examined the role of ozone (O3), recently linked to diabetes risk in adults.28

The US Environmental Protection Agency (US EPA) sets the National Ambient Air Quality Standards (NAAQS) for PM2.5 (12 µg/m3 annual average) and O3 (0.070 parts per million annual fourth-highest daily maximum 8-hour concentration, averaged over 3 years), due to the known harmful effects of these pollutants on human health.29 PM2.5 levels in the Denver-metropolitan area are relatively low compared to the levels in some of the other previously studied urban locations, including the United States, Iran, and Belgium.23,25,26 However, O3 levels in Denver have been above the NAAQS standard since 2008.30 In this study, we investigated the relationships between exposure to ambient PM2.5 and O3 during pregnancy and cardiometabolic biomarkers in umbilical cord blood in a birth cohort based in Denver, Colorado. Specifically, we examined associations between period-specific average concentrations of air pollution at the maternal residence and cord blood glucose, insulin, leptin, non-high-density lipoprotein (non-HDL) cholesterol, HDL cholesterol, triglycerides, and free fatty acids (FFAs). We hypothesized that higher exposure to ambient air pollutants during pregnancy would be associated with higher levels of cord blood cardiometabolic biomarkers.

Methods

Study sample

Participants included in this analysis were enrolled in the Healthy Start study, an ongoing, prebirth cohort study of ethnically diverse mother-infant pairs. Between 2009 and 2014, pregnant women ≥16 years old who had not yet reached 24 weeks of gestation were recruited from the obstetrics clinics at the University of Colorado Hospital in Aurora, Colorado. Additional inclusion criteria were singleton pregnancy, no previous stillbirth or birth <25 weeks, and no preexisting chronic diseases, including diabetes, cancer, asthma managed with steroids, or serious psychiatric illness. Participants were invited to attend three in-person visits: in early pregnancy (median 17 weeks gestation), in mid-pregnancy (median 27 weeks gestation), and at delivery (median 1 day after birth). All participants provided written informed consent, and all study protocols were approved by the Colorado Multiple Institutional Review Board.

Of the 1,410 women enrolled in the Healthy Start study, participants were excluded from this analysis if they withdrew from the study prior to delivery (n = 11), experienced fetal demise (n = 17), did not have cord blood analyzed for cardiometabolic biomarkers (n = 527), and/or did not reside in the Denver-metropolitan area or had no O3 and/or PM2.5 monitors within 50 km of the residence with sufficient nonmissing data, as described below (n = 43). This led to a sample of 812 eligible mother-infant pairs (Supplementary Figure 1; https://links.lww.com/EE/A181).

Assessment of exposures

Participants self-reported their residential address at enrollment. Average PM2.5 and 8-hour maximum O3 levels during the study period (2009–2014) were estimated using data collected at stationary monitors located within 50 km from each participant’s address via the US EPA Air Quality System. The methods used to estimate air pollution exposure for the Healthy Start cohort have been published previously31 and are briefly described below.

Average ambient PM2.5 and average 8-hour maximum O3 concentrations outside the maternal residences were estimated for the following periods during pregnancy: first, second, and third trimesters, full pregnancy, and the 30 and 7 days prior to delivery. There were 10 stationary monitors that measured 24-hour average PM2.5 and 19 stationary monitors that reported hourly O3 during the study period that were within 50 km of at least one participant’s residence. The frequency of reporting varied by monitor for PM2.5, ranging from daily to every 6 days, with most reports occurring every 3 days. All O3 monitors reported hourly values. We used all monitors within 50 km of the participant’s residence at enrollment and for which ≥75% of the expected concentrations were available to estimate period-specific averages for each participant. Inverse-distance-weighting (1/distance2) was used to calculate the average exposure during the specified period for each participant.32

Assessment of outcomes

Umbilical cord blood was collected at delivery, and samples were analyzed by the University of Colorado Clinical and Translational Sciences Institute Core Laboratories. Serum leptin (ng/ml) was measured by ELISA (Alpco, Salem, NH). Insulin (µIU/ml) was measured using radioimmunoassay according to manufacturer’s instructions (EMD Millipore Corporation, Billerica, MA). Glucose (mg/dl), total cholesterol (mg/dl), HDL (mg/dl), triglycerides (mg/dl), and free fatty acids (µIU/ml) were measured using an AU400e Chemistry Analyzer (Olympus America, Center Valley, PA). We calculated non-HDL cholesterol by subtracting HDL cholesterol from total cholesterol in cord blood. Glucose/insulin ratio (GIR) was used as an indirect measure of insulin sensitivity.33,34

Other variables

Maternal age, race, ethnicity, education, and parity were self-reported at the time of study enrollment. Smoking status and exposure to secondhand smoke were self-reported at all three study visits during pregnancy. Maternal prepregnancy body mass index (BMI, kg/m2) was calculated using height measured by research staff at the first visit and prepregnancy weight, which was obtained through medical record abstraction (87.9%) or when unavailable through self-report (12.1%) at the first visit. Gestational diabetes (GDM) status was determined by a positive oral glucose tolerance test and/or a diagnosis in the medical record. Mode of delivery, season of birth, and infant sex were collected from medical records. Median income in the Census tract of the mother’s residence was obtained by linkage of the geocoded address reported at enrollment to 2012–2016 American Community Survey data and used an indicator of neighborhood-level socioeconomic status.

Statistical analysis

Spearman correlation coefficients were calculated for each pair of cardiometabolic markers. Insulin, GIR, leptin, and triglycerides in cord blood had right-skewed distributions and were natural log-transformed prior to analysis to better approximate normality and minimize the influence of outliers.

Multivariable linear regression models were used to investigate the relationships between prenatal PM2.5 and O3 and cardiometabolic biomarkers in cord blood. Associations were expressed as beta-coefficients and 95% confidence intervals (CIs) per interquartile range (IQR) increase in air pollutant exposure. Potential confounders were identified via directed acyclic graphs based on published literature, and included maternal age at delivery (years), prepregnancy BMI (kg/m2), race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, all others), maternal smoking during pregnancy (any, none), exposure to secondhand smoke during pregnancy (>1 hour per week at any point during pregnancy), median annual income in the census tract (per $1,000), maternal education completed (<12th grade, high school degree or General Education Development [GED], some college or Associate’s degree, four years of college, graduate degree), mode of delivery (cesarean, vaginal), previous pregnancy (any, none), season of birth (winter, spring, summer, fall), and the corresponding copollutant during the specified pregnancy period (PM2.5 or O3). Infant sex (male, female) was included as a precision covariate of the outcome. Birth weight and gestational age at birth were considered potential causal intermediates and accordingly were not included in the models. Because there is some evidence that fetal metabolic response to maternal environmental chemical exposures may differ by sex,35 potential effect modification by infant sex was assessed by including product interaction terms between each air pollutant exposure and infant sex in fully adjusted models. We hypothesized that the corresponding pollutant (PM2.5 or O3) confounded the relationships between the other pollutant and the cardiometabolic biomarkers; therefore, we included both pollutants in all primary models. However, in a sensitivity analysis, we investigated the results of single pollutant models. To evaluate the potential for nonlinear relationships between the air pollutants and the cardiometabolic markers, adjusted generalized additive models with penalized regression splines were run. The fit of models with the linear air pollutant exposure term was compared to that of models with the spline-transformed exposure using Akaike Information Criterion (AIC). We additionally assessed associations with quartiles of the exposure. Finally, because the levels of cardiometabolic biomarkers of infants born to mothers with GDM may differ,36 we conducted a sensitivity analysis excluding 34 participants with gestational diabetes.

We conducted all analyses in SAS (Version 9.4; The SAS Institute, Cary, NC) and R (Version 4.0.2; R Foundation for Statistical Computing, Vienna, Austria).

Results

The mean ± SD of maternal age at delivery was 28 ± 6 years, and the mean prepregnancy BMI was 26 ± 7 kg/m2. There were 34 participants (4%) with GDM. The sample was racially and ethnically diverse (Hispanic, 25%; non-Hispanic White, 53%; non-Hispanic Black, 16%; and all others, 7%). Approximately 9% of participants smoked during pregnancy, and 23% of participants were exposed to secondhand smoke (>1 hour per week at any point during pregnancy) (Table 1). There were 432 (53%) male infants. Characteristics of participants in this subsample were comparable to those of the excluded participants and to the full potentially eligible cohort (Supplemental Table 1; https://links.lww.com/EE/A181).

Table 1. - Characteristics of 812 eligible mother-infant pairs in the Healthy Start Study
Characteristic Mean ± SD, median (25th–75th percentiles), or N (%)
Covariates
 Maternal age at delivery (years) 27.5 ± 6.2
 Prepregnancy body mass index (kg/m2) 25.9 ± 6.5
 Race/ethnicity
  Hispanic 201 (25)
  Non-Hispanic White 428 (53)
  Non-Hispanic Black 127 (16)
  All others 56 (7)
 Maternal smoking during pregnancy (any) 77 (9)
 Exposure to secondhand smoke during pregnancy 186 (23)
 Median annual income in the census tract ($) 63,500 ± 26,500
 Maternal education completed
  Less than 12th grade 122 (15)
  High school degree or GED/high school equivalency diploma 163 (20)
  Some college or associate’s degree 178 (22)
  Four years of college (BA, BS) 174 (21)
  Graduate degree (Master’s, PhD) 175 (22)
 Mode of delivery = cesarean 172 (21)
 Previous pregnancy 510 (63)
 Gestational diabetesa 34 (4)
 Season of birth
  Winter 173 (21)
  Spring 199 (25)
  Summer 250 (31)
  Fall 190 (23)
 Infant sex = male 432 (53)
Exposures
 Average PM2.5 (µ/m3)a
  Trimester 1 7.5 (6.9–8.2)
  Trimester 2 7.4 (6.8–8.1)
  Trimester 3 7.3 (6.8–8.1)
  Full pregnancy 7.5 (7.0–7.9)
  Prior 30 days to delivery 7.2 (6.4–8.3)
  Prior 7 days to delivery 6.7 (5.6–8.2)
 Average 8-hour maximum O3 (ppb)
  Trimester 1 41.8 (33.0–53.0)
  Trimester 2 42.6 (33.9–52.4)
  Trimester 3 46.6 (35.5–54.0)
  Full pregnancy 43.9 (40.7–47.0)
  Prior 30 days to delivery 48.0 (34.2–55.9)
  Prior 7 days to delivery 47.0 (35.5–55.6)
aSample sizes differ for the following variables due to missing data: gestational diabetes, n = 774; average PM2.5 trimester 1, n = 749; average PM2.5 trimester 2, n = 749; average PM2.5 trimester 3, n = 787; average PM2.5 full pregnancy, n = 806; average PM2.5 prior 30 days to delivery, n = 765; and average PM2.5 prior 7 days to delivery, n = 771.

The median full pregnancy average PM2.5 exposure was 7.5 µg/m3, the 5th percentile was 6.6 µg/m3, and the 95th percentile was 8.4 µg/m3. Similar median and range values were observed for trimester-average PM2.5 exposures and the 30- and 7- day exposure windows. The median full pregnancy average 8-hour maximum O3 concentration was 43.9 parts per billion (ppb), the 5th percentile was 36.6 ppb, and the 95th percentile was 49.9 ppb (Table 1 and Supplemental Table 2; https://links.lww.com/EE/A181).

Glucose, insulin, GIR, and leptin in cord blood were generally weakly correlated with each other, except for a strong inverse correlation between insulin and GIR (rs = –0.92). Most cord blood lipid biomarkers were weakly to moderately correlated with each other, with the strongest correlation observed between triglycerides and non-HDL cholesterol (rs = 0.44) (Table 2).

Table 2. - Levels and pairwise spearman correlations of cardiometabolic biomarkers in cord blood among eligible mother-infant pairs in the Healthy Start Study
Cardiometabolic biomarker Median (25th–75th percentiles) Glucose (mg/dl)a Insulin (µIU/ml)a Glucose/insulin ratioa Leptin (ng/ml)a Non-HDL cholesterol (mg/dl)a HDL (mg/dl)a Free fatty acids (µIU/ml)a Triglycerides (mg/dl)a
Glucose (mg/dl)a 79 (70–90) 1.00 0.19 0.17 0.10 –0.03 –0.07 0.10 0.08
Insulin (µIU/ml)a 7 (5–11) 1.00 –0.92 0.27 –0.09 –0.02 –0.23 –0.26
Glucose/insulin ratioa 11 (7–15) 1.00 –0.23 0.07 –0.02 0.26 0.28
Leptin (ng/ml)a 11 (6–21) 1.00 –0.06 –0.01 0.09 0.02
Non-HDL cholesterol (mg/dl)a 33 (26–41) 1.00 0.40 0.13 0.44
HDL (mg/dl)a 25 (21–30) 1.00 0.06 –0.13
Free fatty acids (µIU/ml)a 260 (178–366) 1.00 0.36
Triglycerides (mg/dl)a 39 (29–55) 1.00
aMissing data for cord blood: glucose, n = 19; insulin, n = 50; glucose/insulin ratio, n = 52; leptin, n = 144; non-HDL cholesterol, n = 109; HDL, n = 109; free fatty acids, n = 108; and triglycerides, n = 50.

Higher average 8-hour maximum O3 during trimesters 2 and 3, full pregnancy, and the prior 7 and 30 days to delivery was associated with higher insulin (Table 3). For example, an IQR increase in prior 30-day O3 (21.7 ppb) was associated with a 0.21 log-unit (95% CI = 0.07, 0.35) increase in cord blood insulin, after adjusting for covariates. Additionally, higher O3 during all exposure windows, except for trimester 1, was associated with lower GIR. For example, an IQR increase in prior 30-day O3 (21.7 ppb) was associated with a 0.22 log-unit (95% CI = –0.36, –0.09) decrease in cord blood GIR, after adjusting for covariates. Higher full pregnancy PM2.5 was associated with higher leptin, and average PM2.5 in each trimester was nonsignificantly positively associated with leptin. We did not observe consistent associations between O3 and leptin, nor between PM2.5 or O3 and glucose (Table 3). Associations in single pollutant models were similar to those observed in copollutant adjusted models (Supplemental Table 3; https://links.lww.com/EE/A181).

Table 3. - Associations between average PM2.5 and 8-hour maximum O3 at the residential address during pregnancy and glucose, insulin, glucose/insulin ratio, and leptin in cord blood
Air pollution exposure n Glucose (mg/dl)β (95% CI)a per IQR Sex interactionP n Log-transformed insulin (log-unit)β (95% CI)a per IQR Sex interactionP n Log-transformed glucose/insulin ratioβ (95% CI)a per IQR Sex interactionP n Log-transformed Leptin (log-unit)β (95% CI)a per IQR Sex interactionP
Trimester 1 731 706 704 614
 Average PM2.5 (µg/m3) –0.91 (–2.64, 0.82) 0.80 0.00 (–0.05, 0.06) 0.54 –0.02 (–0.08, 0.03) 0.58 0.08 (–0.01, 0.18) 0.72
 Average 8-hour maximum O3 (ppb) 2.91 (–1.91, 7.72) 0.56 –0.02 (–0.17, 0.14) 0.11 0.05 (–0.11, 0.20) 0.18 –0.12 (–0.38, 0.14) 0.32
Trimester 2 731 700 698 642
 Average PM2.5 (µg/m3) 0.96 (–0.82, 2.74) 0.28 –0.01 (–0.06, 0.04) 0.04 0.03 (–0.03, 0.08) 0.02 0.05 (–0.03, 0.14) 0.42
 Average 8-hour maximum O3 (ppb) –0.30 (–4.99, 4.40) 0.36 0.15 (0.01, 0.30) 0.32 –0.15 (–0.30, –0.01) 0.25 –0.04 (–0.26, 0.18) 0.47
Trimester 3 768 737 735 662
 Average PM2.5 (µg/m3) 0.94 (–0.93, 2.81) 0.64 0.00 (–0.05, 0.06) 0.61 0.00 (–0.06, 0.06) 0.68 0.01 (–0.09, 0.10) 0.29
 Average 8-hour maximum O3 (ppb) 0.04 (–4.62, 4.70) 0.31 0.23 (0.09, 0.37) 0.56 –0.24 (–0.38, –0.10) 0.43 0.08 (–0.16, 0.31) 0.59
Full pregnancy 787 756 754 665
 Average PM2.5 (µg/m3) 0.39 (–1.64, 2.42) 0.66 0.02 (–0.04, 0.08) 0.13 –0.02 (–0.08, 0.04) 0.08 0.11 (0.01, 0.21) 0.86
 Average 8-hour maximum O3 (ppb) 0.73 (–2.28, 3.73) 0.96 0.13 (0.04, 0.22) 0.28 –0.13 (–0.22, –0.04) 0.34 0.01 (–0.14, 0.16) 0.94
Prior 30 days to delivery 747 718 716 643
 Average PM2.5 (µg/m3) 0.76 (–0.95, 2.48) 0.91 –0.03 (–0.08, 0.03) 0.24 0.03 (–0.02, 0.09) 0.24 –0.01 (–0.10, 0.09) 0.70
 Average 8-hour maximum O3 (ppb) –0.19 (–4.65, 4.27) 0.17 0.21 (0.07, 0.35) 0.14 –0.22 (–0.36, –0.09) 0.07 0.06 (–0.16, 0.29) 0.83
Prior 7 days to delivery 752 723 721 642
 Average PM2.5 (µg/m3) 0.79 (–0.35, 1.93) 0.95 0.02 (–0.01, 0.06) 0.19 –0.01 (–0.05, 0.02) 0.21 0.02 (–0.04, 0.08) 0.90
 Average 8-hour maximum O3 (ppb) –1.79 (–5.66, 2.08) 0.24 0.12 (0.00, 0.24) 0.15 –0.16 (–0.28, –0.04) 0.08 0.10 (–0.10, 0.30) 0.95
aAdjusted for race, maternal age, prepregnancy BMI, smoking status, exposure to secondhand smoke during pregnancy, median income in the census tract, education, mode of delivery, parity, season of birth, infant sex, and the corresponding copollutant during the specified pregnancy period.

We observed significant exposure-by-infant sex interactions in models for associations between trimester 2 PM2.5 and insulin (P = 0.04), and trimester 2 PM2.5 and GIR (P = 0.02) (Table 3). Sex-specific analyses indicated different directions of association in male and female infants (Supplemental Tables 5 and 6; https://links.lww.com/EE/A181). There was a nonsignificant, positive association between trimester 2 PM2.5 and insulin among males and an inverse association among females (Supplemental Table 5; https://links.lww.com/EE/A181). There was a nonsignificant inverse association between trimester 2 PM2.5 and GIR observed among males and a positive association among females (β = 0.09 per IQR; 95% CI = 0.02, 0.17) (Supplemental Table 6; https://links.lww.com/EE/A181). We did not observe significant exposure-by-infant sex interactions for glucose, leptin, or lipids (Table 4 and Supplemental Tables 5–8; https://links.lww.com/EE/A181).

Table 4. - Associations between average PM2.5 and 8-hour maximum O3 at the residential address during pregnancy and lipid biomarkers in cord blood
Air pollution exposure n Non-HDL Cholesterol (mg/dl)β (95% CI)a per IQR Sex interactionP n HDL (mg/dl)β (95% CI)a per IQR Sex interactionP n Free fatty acids (µIU/ml)β (95% CI)a per IQR Sex interactionP n Log-transformed triglycerides (log-unit)β (95% CI)a per IQR Sex interactionP
Trimester 1 647 647 657 703
 Average PM2.5 (µg/m3) 0.12 (–1.63, 1.86) 0.41 0.18 (–0.64, 1.00) 0.69 –8.98 (–24.13, 6.16) 0.15 –0.02 (–0.07, 0.03) 0.24
 Average 8-hour maximum O3 (ppb) 3.20 (–1.69, 8.09) 0.91 1.54 (–0.77, 3.85) 0.39 –32.34 (–74.76, 10.07) 0.64 –0.08 (–0.21, 0.06) 0.77
Trimester 2 647 647 644 701
 Average PM2.5 (µg/m3) –0.71 (–2.38, 0.96) 0.70 –0.55 (–1.35, 0.25) 0.12 –2.04 (–16.85, 12.76) 0.28 –0.02 (–0.07, 0.03) 0.29
 Average 8-hour maximum O3 (ppb) 6.19 (1.71, 10.68) 0.32 0.69 (–1.46, 2.84) 0.65 25.44 (–12.70, 63.57) 0.52 0.07 (–0.05, 0.20) 0.45
Trimester 3 684 684 682 737
 Average PM2.5 (µg/m3) –0.91 (–2.60, 0.79) 0.59 0.42 (–0.39, 1.24) 0.74 –12.65 (–28.43, 3.14) 0.97 –0.05 (–0.10, 0.00) 0.47
 Average 8-hour maximum O3 (ppb) 0.24 (–4.10, 4.58) 0.76 0.62 (–1.46, 2.70) 0.20 31.63 (–8.44, 71.70) 0.94 0.08 (–0.04, 0.20) 0.28
Full pregnancy 698 698 700 756
 Average PM2.5 (µg/m3) –0.74 (–2.61, 1.12) 0.55 0.13 (–0.77, 1.03) 0.84 –11.38 (–27.83, 5.07) 0.54 –0.05 (–0.11, 0.00) 0.47
 Average 8-hour maximum O3 (ppb) 2.97 (0.12, 5.81) 0.54 1.54 (0.17, 2.91) 0.77 3.28 (–21.02, 27.58) 0.90 –0.02 (–0.10, 0.06) 0.86
Prior 30 days to delivery 663 663 661 716
 Average PM2.5 (µg/m3) –1.42 (–3.03, 0.18) 0.99 –0.28 (–1.05, 0.49) 0.92 –15.22 (–30.12, –0.33) 0.59 0.00 (–0.04, 0.05) 0.93
 Average 8-hour maximum O3 (ppb) –0.08 (–4.30, 4.15) 0.80 1.14 (–0.88, 3.16) 0.24 15.83 (–22.58, 54.23) 0.73 0.01 (–0.11, 0.12) 0.39
Prior 7 days to delivery 667 667 666 721
 Average PM2.5 (µg/m3) –0.61 (–1.68, 0.45) 0.39 –0.57 (–1.07, –0.06) 0.88 –2.95 (–12.27, 6.38) 0.24 0.01 (–0.02, 0.04) 0.10
 Average 8-hour maximum O3 (ppb) –1.33 (–4.99, 2.33) 0.77 0.95 (–0.79, 2.68) 0.16 5.98 (–26.75, 38.70) 0.68 –0.03 (–0.13, 0.07) 0.38
aAdjusted for race, maternal age, prepregnancy BMI, smoking status, exposure to secondhand smoke during pregnancy, median income in the census tract, education, mode of delivery, parity, season of birth, infant sex, and the corresponding copollutant during the specified pregnancy period.

Trimester 1, 2 and full-pregnancy average 8-hour maximum O3 were positively associated with non-HDL cholesterol in cord blood, although the 95% CI for trimester 1 included the null. We observed inverse associations of third trimester and full pregnancy average PM2.5 with triglycerides in cord blood. Generally, there were nonsignificant, inverse associations between PM2.5 and FFA, and nonsignificant, positive associations between O3 during trimesters 2 and 3 and the prior 30 and 7 days to delivery and FFA. We did not observe consistent patterns of associations between either PM2.5 or O3 and HDL cholesterol (Table 4). Results from single pollutant models were similar (Supplemental Table 4; https://links.lww.com/EE/A181).

In a sensitivity analysis excluding participants with gestational diabetes (n = 34), results did not change appreciably (Supplemental Tables 9 and 10; https://links.lww.com/EE/A181). Evidence for nonlinearity, defined as lower AIC in the model with spline transformation of the exposure, was observed for some associations; therefore, we ran additional models in which cord blood biomarkers were regressed on quartiles of exposure (Supplemental Tables 11 and 12; https://links.lww.com/EE/A181). We observed a nonsignificant, U-shaped relationship between O3 in the 7 days prior to delivery and cord blood glucose (Supplemental Table 11; https://links.lww.com/EE/A181). Several quartile models showed nonmonotonic responses between both pollutants and non-HDL cholesterol, HDL cholesterol, FFA, and triglycerides in cord blood, but results were generally nonsignificant (Supplemental Table 12; https://links.lww.com/EE/A181).

Discussion

In this prospective birth cohort study, greater average 8-hour maximum O3 during trimesters 2, 3, full-pregnancy and the 30 and 7 days prior to delivery was associated with higher levels of cord blood insulin and lower GIR. Additionally, average PM2.5 during certain periods of pregnancy was positively associated with cord blood leptin, and O3 was positively associated with non-HDL cholesterol and inversely associated with triglycerides and FFA. Our findings provide suggestive evidence that prenatal air pollution exposure may be associated with dysregulation of metabolism at birth. Alterations in some metabolic biomarkers in cord blood have been associated with cardiometabolic health outcomes later in infancy and childhood, including weight gain, BMI, and body size and composition.17–22 Notably, the Denver-metropolitan area represents a unique exposure profile with regard to these two criteria pollutants. While PM2.5 levels are below the NAAQS standard, the area has been out of attainment for O3 since 2008.30 Although the changes observed in the present study are small at the individual level, given the widespread nature of the exposure, on a population-scale these changes may be associated with an increased prevalence of obesity and adverse metabolic outcomes.

We noted consistently positive associations between 8-hour maximum O3 concentrations during multiple periods of pregnancy with cord blood insulin. One previous study investigated air pollution exposure during pregnancy and cord blood insulin.25 Among 590 mother-infant pairs enrolled in a cohort study in Belgium, cord blood insulin was positively associated with full pregnancy PM2.5 and PM10, but not nitrogen dioxide (NO2).25 The study did not examine associations with O3, nor did they find any significant pollutant-by-sex interactions. While there is a dearth of literature examining this relationship in the perinatal period, chronic exposure to PM10 and NO2 and proximity to a major roadway at the birth address has been associated with insulin resistance in children 10 years old.37 Importantly, cord blood collection in the Healthy Start study did not discriminate between the umbilical vein (from the mother) and artery (from the fetus) sources, and thus, the origin of insulin in our samples is unknown. However, given that the fetal pancreas produces measurable levels of insulin by mid-gestation, and maternal insulin does not typically cross the placenta (with a few exceptions),38,39 we can assume that most of the insulin in our cord blood samples collected at delivery is from fetal production. In our study, we used GIR as a measure of insulin sensitivity; alterations in GIR in cord blood have been identified in infants whose mothers were overweight or obese33 and had impaired glucose tolerance,40 suggesting it is a useful marker in cord blood of metabolic homeostasis.

Studies from animal models also support an association between prenatal air pollution exposure and impaired metabolic homeostasis later in life.10,41 Among mice exposed prenatally to diesel exhaust and fed a high-fat diet postnatally, males developed hyperinsulinemia and insulin resistance, and excess weight gain was observed in both sexes.10,41 We also noted sex-specific effects for some of the PM2.5 exposure windows and cord blood insulin. Moreover, our finding of an inverse association between O3 and GIR is consistent with animal studies of insulin resistance. There are several proposed mechanisms through which particulate matter and ozone may induce insulin resistance, including oxidative stress and inflammation.42,43

We found positive associations between trimester-long and full-pregnancy PM2.5 and cord blood leptin. However, we did not find consistent associations between PM2.5 during any of the shorter exposure windows and cord blood leptin. In infants, cord blood leptin is a known marker for neonatal fat mass,44 however, the nature of the relationship between cord blood leptin and adiposity across infancy and childhood is complex.18,21 In a Canadian cohort study, authors reported a positive association between full pregnancy PM2.5 and NO2 and cord blood leptin, when adjusting for birthweight z-score.24 More recently, results from a California-based birth cohort suggest positive associations between full pregnancy nonfreeway Nitrogen oxide (NOx) exposure and proximity to major roadways and cord blood leptin.23 Further, an epigenetic study found PM2.5 exposure during trimester 2 was associated with lower deoxyribonucleic acid methylation of the leptin gene in the placenta.45

We did not observe associations between exposure to PM2.5 and O3 during pregnancy and glucose in cord blood. To our knowledge, no previous study has examined this relationship in the perinatal period. During gestation, placental transport of maternal glucose serves as the primary source of fuel for the fetus,46 and it is therefore difficult to compare our findings with existing literature investigating this relationship in childhood. Nonetheless, in one prior study, among overweight and obese African American and Hispanic children in Los Angeles, CA, PM2.5, NO2, and nonfreeway NOx, but not O3, exposure was associated with higher fasting glucose.47

Our findings on lipid biomarkers showed little consistency across lipids or across periods of pregnancy. We observed positive associations between O3 and non-HDL cholesterol in cord blood during some periods of pregnancy and inverse associations between trimester-long PM2.5 exposure and triglycerides. Although we observed limited evidence of associations between PM2.5 (inverse) and O3 (positive) during some periods of pregnancy and cord blood FFA, patterns were inconsistent and generally nonsignificant. In a cross-sectional analysis of Iranian mother-infant pairs, higher exposure to PM1, PM2.5, and PM10 was associated with higher total cholesterol levels and triglycerides, but not HDL, in cord blood.26 It is notable that our study had a much smaller range of variability in average full pregnancy PM2.5 (IQR, 0.9 µg/m3) compared to that of the Heydari et al26 study (IQR, 33.2 µg/m3). A recent investigation in a Mexico city-based birth cohort found that exposure to PM2.5 during trimester 3 of pregnancy was associated with higher levels of low-density lipoprotein (LDL), non-HDL, and total cholesterol levels in childhood (mean 4.8 years), suggesting that these alterations may also be identifiable in childhood.48 Authors also found that prenatal exposure to late-pregnancy PM2.5 was associated with lower triglyceride levels in children.48 While we observed only isolated associations between prenatal air pollution exposure and lipid biomarkers in cord blood, the abovementioned studies suggest that prenatal air pollution exposure remains an important potentially modifiable early-life exposure with evidence of enduring disruptions to cardiovascular risk later in life.49

There are several strengths of this study. We used data from a prospective, prebirth cohort study, which allowed for the longitudinal assessment of the relationship between in utero air pollution exposure and cardiometabolic biomarker outcomes at delivery. The study population was racially and ethnically diverse, with well-characterized covariate data. We found novel associations between period-specific exposures to PM2.5 and O3 during pregnancy and some cardiometabolic biomarkers in cord blood.

Our study also has some limitations. Residential address information was only collected at enrollment. Because we did not have information on residential mobility during the study period, there is the possibility for misclassification of air pollution exposure. A previous study in New York showed limited residential mobility during pregnancy.50 While it is unknown if these findings are generalizable to women living in Colorado, it provides evidence that the degree of exposure misclassification due to local moves may be minimal. Further, we were not able to account for differences in human behavior that may affect personal air pollution exposure, including time spent indoors and in nonresidential settings. We also did not account for potential differences in exposure levels due to small-scale spatial variability, such as proximity to roadways or other concentrated sources of air pollution. The measurement error in air pollution due to the relatively sparse monitoring network is likely nondifferential with respect to the cord blood metabolic markers, and thus, may have resulted in an underestimation of associations.51 While we included several relevant covariates in our models, there is still the possibility for residual confounding due to unmeasured factors related to air pollution, including green space and traffic noise, or due to the use of imperfect proxies of socioeconomic status. There is the possibility of residual confounding due to the use of self-reported smoking data, as it may have resulted in an underestimation of smoking status, in our study. We performed a large number of statistical analyses and did not adjust for multiple testing, because our emphasis was on the magnitude and precision of the associations rather than their statistical significance. It is possible that some of the findings are statistically significant due to chance. However, the main findings of our study were consistent across several exposure time periods, and thus, are less likely to be spurious results. Independent replication of these results in another cohort would strengthen the interpretation.

Conclusions

In conclusion, we found higher average 8-hour maximum O3 concentrations associated with higher cord blood insulin and lower GIR. We also found limited evidence of associations of ambient air pollution with lipids and the adipocyte-derived hormone leptin. These findings highlight the potential for air pollution exposure during pregnancy to disrupt the metabolic and lipid homeostasis of the offspring in the perinatal period. These findings were noted in an urban area with a unique air pollution profile, with levels of PM2.5 below the NAAQS standard, but O3 regularly out of attainment. Future studies from Healthy Start and other birth cohorts should examine whether air pollution-associated changes in biomarkers persist into childhood, and therefore whether maternal exposure to air pollutants during pregnancy may contribute to insulin resistance and other adverse child cardiometabolic outcomes.

Conflicts of interest statement

The authors declare that they have no conflicts of interest with regard to the content of this report.

This work was supported in part by grants from the National Institute of Environmental Health Sciences (R00ES025817), the National Institute of Diabetes and Digestive and Kidney Diseases (R01DK076648), and the National Institutes of Health Office of the Director (UH3OD023248). The University of Colorado Cancer Center laboratory received support from the National Cancer Institute (UL1 TR001082).

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

C.F., D.D., D.S.K.T., J.L.P., and A.P.S. contributed to the conception and design of this study. All authors contributed to the acquisition, analysis, and interpretation of data. C.F. and A.P.S. participated in drafting the article. D.D., L.D.B., D.S.K.T., J.L.P., J.L.A., S.M., S.E.M., and W.B.A. participated in revising the article critically for important intellectual content. All authors gave approval of the final article.

References

1. Ogden CL, Carroll MD, Lawman HG, et al. Trends in obesity prevalence among children and adolescents in the United States, 1988-1994 through 2013-2014. JAMA. 2016;315:2292–2299.
2. Mayer-Davis EJ, Lawrence JM, Dabelea D, et al. Incidence trends of type 1 and type 2 diabetes among youths, 2002–2012. N Engl J Med. 2017;376:1419–1429.
3. Muntner P, He J, Cutler JA, Wildman RP, Whelton PK. Trends in blood pressure among children and adolescents. JAMA. 2004;291:2107–2113.
4. Rosner B, Cook NR, Daniels S, Falkner B. Childhood blood pressure trends and risk factors for high blood pressure: the NHANES experience 1988-2008. Hypertension. 2013;62:247–254.
5. Heindel JJ, vom Saal FS. Role of nutrition and environmental endocrine disrupting chemicals during the perinatal period on the aetiology of obesity. Mol Cell Endocrinol. 2009;304:90–96.
6. Jerrett M, McConnell R, Wolch J, et al. Traffic-related air pollution and obesity formation in children: a longitudinal, multilevel analysis. Environ Health. 2014;13:49.
7. Lawlor DA, Chaturvedi N. Treatment and prevention of obesity–are there critical periods for intervention? Int J Epidemiol. 2006;35:3–9.
8. Grün F, Blumberg B. Minireview: the case for obesogens. Mol Endocrinol. 2009;23:1127–1134.
9. Wei Y, Zhang JJ, Li Z, et al. Chronic exposure to air pollution particles increases the risk of obesity and metabolic syndrome: findings from a natural experiment in Beijing. FASEB J. 2016;30:2115–2122.
10. Bolton JL, Smith SH, Huff NC, et al. Prenatal air pollution exposure induces neuroinflammation and predisposes offspring to weight gain in adulthood in a sex-specific manner. FASEB J. 2012;26:4743–4754.
11. Rundle A, Hoepner L, Hassoun A, et al. Association of childhood obesity with maternal exposure to ambient air polycyclic aromatic hydrocarbons during pregnancy. Am J Epidemiol. 2012;175:1163–1172.
12. Fleisch AF, Luttmann-Gibson H, Perng W, et al. Prenatal and early life exposure to traffic pollution and cardiometabolic health in childhood. Pediatr Obes. 2017;12:48–57.
13. Huang JV, Leung GM, Schooling CM. The association of air pollution with body mass index: evidence from Hong Kong’s “Children of 1997” birth cohort. Int J Obes (Lond). 2019;43:62–72.
14. Mao G, Nachman RM, Sun Q, et al. Individual and joint effects of early-life ambient exposure and maternal prepregnancy obesity on childhood overweight or obesity. Environ Health Perspect. 2017;125:067005.
15. Grandjean P, Barouki R, Bellinger DC, et al. Life-long implications of developmental exposure to environmental stressors: new perspectives. Endocrinology. 2015;156:3408–3415.
16. Heindel JJ, Vandenberg LN. Developmental origins of health and disease: a paradigm for understanding disease cause and prevention. Curr Opin Pediatr. 2015;27:248–253.
17. Ong KKL, Ahmed ML, Sherriff A, et al. Cord blood leptin is associated with size at birth and predicts infancy weight gain in humans. J Clin Endocrinol Metab. 1999;84:1145–1148.
18. Mantzoros CS, Rifas-Shiman SL, Williams CJ, Fargnoli JL, Kelesidis T, Gillman MW. Cord blood leptin and adiponectin as predictors of adiposity in children at 3 years of age: a prospective cohort study. Pediatrics. 2009;123:682–689.
19. Volberg V, Heggeseth B, Harley K, et al. Adiponectin and leptin trajectories in Mexican-American children from birth to 9 years of age. PLoS One. 2013;8:e77964.
20. Isganaitis E, Rifas-Shiman SL, Oken E, et al. Associations of cord blood metabolites with early childhood obesity risk. Int J Obes (Lond). 2015;39:1041–1048.
21. Boeke CE, Mantzoros CS, Hughes MD, et al. Differential associations of leptin with adiposity across early childhood. Obesity (Silver Spring). 2013;21:1430–1437.
22. Lee IL, Barr ELM, Longmore D, et al.; PANDORA study team. Cord blood metabolic markers are strong mediators of the effect of maternal adiposity on fetal growth in pregnancies across the glucose tolerance spectrum: the PANDORA study. Diabetologia. 2020;63:497–507.
23. Alderete TL, Song AY, Bastain T, et al. Prenatal traffic-related air pollution exposures, cord blood adipokines and infant weight. Pediatr Obes. 2018;13:348–356.
24. Lavigne E, Ashley-Martin J, Dodds L, et al. Air pollution exposure during pregnancy and fetal markers of metabolic function: the MIREC study. Am J Epidemiol. 2016;183:842–851.
25. Madhloum N, Janssen BG, Martens DS, et al. Cord plasma insulin and in utero exposure to ambient air pollution. Environ Int. 2017;105:126–132.
26. Heydari H, Abroudi M, Adli A, et al. Maternal exposure to ambient air pollution during pregnancy and lipid profile in umbilical cord blood samples; a cross-sectional study. Environ Pollut. 2020;261:114195.
27. Tsai PJ, Yu CH, Hsu SP, et al. Cord plasma concentrations of adiponectin and leptin in healthy term neonates: positive correlation with birthweight and neonatal adiposity. Clin Endocrinol (Oxf). 2004;61:88–93.
28. Li YL, Chuang TW, Chang PY, et al. Long-term exposure to ozone and sulfur dioxide increases the incidence of type 2 diabetes mellitus among aged 30 to 50 adult population. Environ Res. 2021;194:110624.
29. EPA United States Environmental Protection Agency. NAAQS Table. Available at: https://www.epa.gov/criteria-air-pollutants/naaqs-table. [Accessed July 2021]
30. EPA United States Environmental Protection Agency. 8-Hour Ozone (2008) Designated Area/State Information With Design Values. 2008. Updated 31 May 2021. Available at: https://www3.epa.gov/airquality/greenbook/hbtcw.html. [Accessed July 2021]
31. Starling AP, Moore BF, Thomas DSK, et al. Prenatal exposure to traffic and ambient air pollution and infant weight and adiposity: the Healthy Start study. Environ Res. 2020;182:109130.
32. Jerrett M, Arain A, Kanaroglou P, et al. A review and evaluation of intraurban air pollution exposure models. J Expo Anal Environ Epidemiol. 2005;15:185–204.
33. Lemas DJ, Brinton JT, Shapiro AL, Glueck DH, Friedman JE, Dabelea D. Associations of maternal weight status prior and during pregnancy with neonatal cardiometabolic markers at birth: the Healthy Start study. Int J Obes (Lond). 2015;39:1437–1442.
34. Setia S, Sridhar MG, Bhat V, Chaturvedula L, Vinayagamoorti R, John M. Insulin sensitivity and insulin secretion at birth in intrauterine growth retarded infants. Pathology. 2006;38:236–238.
35. Ashley-Martin J, Dodds L, Arbuckle TE, et al. Maternal blood metal levels and fetal markers of metabolic function. Environ Res. 2015;136:27–34.
36. Westgate JA, Lindsay RS, Beattie J, et al. Hyperinsulinemia in cord blood in mothers with type 2 diabetes and gestational diabetes mellitus in New Zealand. Diabetes Care. 2006;29:1345–1350.
37. Thiering E, Cyrys J, Kratzsch J, et al. Long-term exposure to traffic-related air pollution and insulin resistance in children: results from the GINIplus and LISAplus birth cohorts. Diabetologia. 2013;56:1696–1704.
38. Jovanovic L, Pettitt DJ. Treatment with insulin and its analogs in pregnancies complicated by diabetes. Diabetes Care. 2007;30(2 suppl):S220–S224.
39. Schwartz R. Hyperinsulinemia and Macrosomia. Mass Medical Soc; 1990.
40. Luo ZC, Delvin E, Fraser WD, et al. Maternal glucose tolerance in pregnancy affects fetal insulin sensitivity. Diabetes Care. 2010;33:2055–2061.
41. Bolton JL, Auten RL, Bilbo SD. Prenatal air pollution exposure induces sexually dimorphic fetal programming of metabolic and neuroinflammatory outcomes in adult offspring. Brain Behav Immun. 2014;37:30–44.
42. Haberzettl P, O’Toole TE, Bhatnagar A, Conklin DJ. Exposure to fine particulate air pollution causes vascular insulin resistance by inducing pulmonary oxidative stress. Environ Health Perspect. 2016;124:1830–1839.
43. Liu C, Xu X, Bai Y, et al. Air pollution-mediated susceptibility to inflammation and insulin resistance: influence of CCR2 pathways in mice. Environ Health Perspect. 2014;122:17–26.
44. Schubring C, Siebler T, Kratzsch J, et al. Leptin serum concentrations in healthy neonates within the first week of life: relation to insulin and growth hormone levels, skinfold thickness, body mass index and weight. Clin Endocrinol (Oxf). 1999;51:199–204.
45. Saenen ND, Vrijens K, Janssen BG, et al. Lower placental leptin promoter methylation in association with fine particulate matter air pollution during pregnancy and placental nitrosative stress at birth in the ENVIRONAGE cohort. Environ Health Perspect. 2017;125:262–268.
46. Castillo-Castrejon M, Powell TL. Placental nutrient transport in gestational diabetic pregnancies. Front Endocrinol (Lausanne). 2017;8:306.
47. Toledo-Corral CM, Alderete TL, Habre R, et al. Effects of air pollution exposure on glucose metabolism in Los Angeles minority children. Pediatr Obes. 2018;13:54–62.
48. McGuinn LA, Coull BA, Kloog I, et al. Fine particulate matter exposure and lipid levels among children in Mexico city. Environ Epidemiol. 2020;4:e088.
49. Breton CV, Mack WJ, Yao J, et al. Prenatal air pollution exposure and early cardiovascular phenotypes in young adults. PLoS One. 2016;11:e0150825.
50. Chen L, Bell EM, Caton AR, Druschel CM, Lin S. Residential mobility during pregnancy and the potential for ambient air pollution exposure misclassification. Environ Res. 2010;110:162–168.
51. Armstrong BG. Effect of measurement error on epidemiological studies of environmental and occupational exposures. Occup Environ Med. 1998;55:651–656.
Keywords:

Air pollution; Lipids; Metabolism; Neonate; Prenatal exposure

Supplemental Digital Content

Copyright © 2022 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of The Environmental Epidemiology. All rights reserved.