Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Aug 15.
Published in final edited form as: Sci Total Environ. 2022 Apr 20;834:155394. doi: 10.1016/j.scitotenv.2022.155394

In-utero exposure to indoor air pollution or tobacco smoke and cognitive development in a South African birth cohort study

Grace M Christensen 1, Claire Rowcliffe 1, Junyu Chen 1, Aneesa Vanker 2, Nastassja Koen 3,4,5, Meaghan J Jones 6, Nicole Gladish 7,8,9, Nadia Hoffman 3,4, Kirsten A Donald 2,3, Catherine J Wedderburn 2,3,10, Michael S Kobor 7,8,9, Heather J Zar 2,5, Dan J Stein 3,4,5, Anke Hüls 1,11
PMCID: PMC9177804  NIHMSID: NIHMS1807601  PMID: 35460774

Abstract

Background and Aims:

There is increasing evidence indicating that air pollution exposure is associated with neuronal damage. Since pregnancy is a critical window of vulnerability, air pollution exposure during this period could have adverse effects on neurodevelopment. This study aims 1) to analyze associations of prenatal exposure to indoor air pollution (particulate matter with diameters ≤10μm, PM10) and tobacco smoke with neurodevelopment and 2) to determine whether these associations are mediated by deviations of epigenetic gestational age from chronological gestational age (ΔGA).

Methods:

Data of 734 children from the South African Drakenstein Child Health Study were analyzed. Prenatal PM10 exposure was measured using devices placed in the families’ homes. Maternal smoking during pregnancy was determined by maternal urine cotinine measures. The Bayley Scales of Infant and Toddler Development III (BSID-III) was used to measure cognition, language and motor development and adaptive behavior at two years of age. Linear regression models adjusted for maternal age, gestational age, sex of child, ancestry, birth weight/length, and socioeconomic status were used to explore associations between air pollutants and BSID-III scores. A mediation analysis was conducted to analyze if these associations were mediated by ΔGA using DNA methylation measurements from cord blood.

Results:

An increase of one interquartile range in natural-log transformed PM10 (lnPM10; 1.58 μg/m3) was significantly associated with lower composite scores in cognition, language, and adaptive behavior sub-scores (composite score β-estimate [95%-confidence interval]: −0.950 [−1.821, −0.120]). Maternal smoking was significantly associated with lower adaptive behavior scores (−3.386 [−5.632, −1.139]). Associations were not significantly mediated by ΔGA (e.g., for PM10 and cognition, proportion mediated [p-value]: 4% [0.52]).

Conclusion:

We found an association of prenatal exposure to indoor air pollution (PM10) and tobacco smoke on neurodevelopment at two years of age, particularly cognition, language, and adaptive behavior. Further research is needed to understand underlying biological mediators.

Keywords: environmental epidemiology, epigenomics, air pollution, neurodevelopment, particulate matter

GRAPHICAL ABSTRACT

graphic file with name nihms-1807601-f0001.jpg

Introduction

There is increasing evidence indicating that air pollution exposure is associated with an impairment of the central nervous system (CNS) and neuronal damage. The assembly of the CNS in utero and early life requires many steps including cell proliferation, differentiation, and migration (Rodier, 2004). Failure of typical cell proliferation or cell migration due to exposure to toxic insults may have profound deleterious effects on the developing brain (Costa et al., 2004). Though neurons are able to make new synapses throughout life, the prenatal period and first few years of life are critical for the formation of the basic circuitry of the nervous system (Costa et al., 2019; Rodier, 1995). Since pregnancy and the first years of life are a critical window of developmental vulnerability for the brain (Costa et al., 2019; Rodier, 1995), few studies have investigated associations between prenatal exposure to air pollution on infants’ cognitive development (Volk et al., 2021). Some studies report significant associations between both prenatal and early life exposure to PM and Autism Spectrum Disorder (ASD) in children (Lam et al., 2016; Raz et al., 2015; Volk et al., 2013). Additional studies have found prenatal particulate matter with a diameter less than 10 microns (PM10) exposure to be associated with lower IQ at ages 4–6 years (Loftus et al., 2019), and decreased mental and psychomotor development scores (Kim et al., 2014).

Even fewer studies focus on indoor air pollutants, as opposed to traffic related or ambient outdoor air pollutants. Indoor air pollution is of particular concern during early life, because infants spend the majority of their time in the home (Pickett & Bell, 2011). A primary source of indoor air pollution is from combustion of gas or other fuel used to cook or heat the home (Balmes, 2019; Levy, 2015). While a few studies have shown associations between combustion of gas or other fuel inside the home and poorer mental development of the child (Vrijheid et al., 2012), developmental delay (Nazif-Muñoz et al., 2020) and delay in reaching gross motor neurodevelopmental milestones (Wei et al., 2018), none of these studies measured the actual indoor air pollution concentrations. Measuring air pollution concentrations directly is more costly than using proxy measures from questionnaires, but is important in determining what pollutant is affecting neurodevelopment. Indoor air pollution is a leading cause of death and disability worldwide (Lim et al., 2012), and is a particularly pertinent area of research for low-to-middle income countries (LMIC) due to the high percentage of households that still use traditional methods for cooking and heating (Balmes, 2019).

Another source of indoor air pollution is environmental tobacco smoke. In addition to high concentrations of indoor air pollution, smoking prevalence is also particularly high in LMICs (Islami et al., 2015). Multiple studies have reported an association between tobacco smoke exposure during pregnancy and lower child performance on measures of intelligence (Huijbregts et al., 2006; Lawlor et al., 2006), behavior problems (Roza et al., 2009), slower reaction times (Mezzacappa et al., 2011), and decreased motor abilities (Polanska et al., 2013) compared to their unexposed peers. However, all of these studies were conducted in high-income countries where exposure to tobacco smoke is typically low (Islami et al., 2015). Tobacco smoke contains PM among other pollutants and no study has tried to disentangle the effects of secondhand smoke exposure and indoor PM for neurodevelopment. In a country where exposure to tobacco smoke is more prevalent, the burden of neurodevelopmental delay associated with exposure to indoor air pollution could be greater than previously thought.

Little is known about the underlying biological pathway of how indoor air pollution affects neurodevelopment. Epigenetics has been discussed as a plausible mechanism by which environmental exposures might regulate the activity of genes relevant to child development. Epigenetic modification, specifically DNA methylation (DNAm), has been linked to a number of neuropsychiatric outcomes such as specific neurodevelopmental delays, ASD, and attention deficit hyperactivity disorder (ADHD) in several cohorts (Caramaschi et al., 2022; Mordaunt et al., 2020; Neumann et al., 2020; Walton et al., 2017) including our South African Drakenstein Child Health Study (DCHS) (Hüls et al., 2021). In addition, newborn DNAm is known to be associated with prenatal air pollution exposure as shown by studies from the Pregnancy And Childhood Epigenetics (PACE) consortium (Gruzieva et al., 2017, 2019). However, to the best of our knowledge, no study has examined the interconnections between prenatal air pollution exposure, DNAm and neurodevelopment in a mediation analysis. One challenge of epigenetic mediation analyses is the high dimension of the mediator (DNAm). One approach to address the dimensionality issue is the use of well-established risk scores that summarize information across the genome into one score (Hüls & Czamara, 2020). The most established methylation risk score is “epigenetic age”, a predictor of biological age that is highly predictive of adverse health outcomes and death (Hannum et al., 2013; Horvath, 2013). Using umbilical cord blood DNAm profiles, similar estimates of epigenetic gestational age have been created (Bohlin et al., 2016; Knight et al., 2016). Similar to epigenetic age in adults, deviations of epigenetic gestational age from chronological gestational age (ΔGA) may be predictive of adverse child development and potentially mediate the association between prenatal risk factors and neurodevelopment. A previous study from Canada found indications that ΔGA might mediate the association between prenatal exposure to traffic-related air pollution and allergic sensitization in a cohort of 145 children (Sbihi et al., 2019), but it is it is currently unknown whether exposure to indoor air pollution and tobacco smoke also affects ΔGA and how ΔGA may be associated with adverse neurodevelopmental outcomes.

In this study, we aim 1) to analyze associations between pre- and postnatal exposure to indoor air pollution (PM10) or tobacco smoke with neurodevelopment at 2 years of age and 2) to determine if these associations are mediated by deviations of epigenetic gestational age from chronological gestational age (ΔGA).

Methods

Study population and participants

The Drakenstein Child Health Study (DCHS) is a longitudinal birth cohort located in the Drakenstein sub-district of the Western Cape, South Africa, a peri-urban area 60 km outside Cape Town (Donald et al., 2018, 2019; Stein et al., 2015; Zar et al., 2015). Consenting pregnant women were enrolled at 20–28 weeks’ gestation at two public primary health clinics from March 1, 2012, to March 31, 2015 (Vanker et al., 2017; Zar et al., 2015). All children were born at Paarl Hospital (Paarl, South Africa) (Vanker et al., 2017; Zar et al., 2015). Mother and infant pairs were followed after birth at 6–10 weeks old, 14 weeks old, and at 6, 9 and 12 months old and then annually thereafter (Vanker et al., 2017). Study questionnaires and clinical data were collected at enrollment and at each follow-up visit. 1137 mothers with 1143 livebirths were originally enrolled in DCHS, but only one livebirth per mother was included in this study’s analysis to ensure independence among observations. The sample included in the present study were 734 children with measurements available for the cognitive Bayley Scales of Infant and Toddler Development, third edition (BSID-III) scores at two years of age.

The DCHS study was approved by the Human Research Ethics Committee of the Faculty of Health Sciences, University of Cape Town, by Stellenbosch University and the Western Cape Provincial Research committee. Written informed consent was provided by the mothers for herself and her infant and is renewed annually.

Assessment of indoor air pollution (IAP) and tobacco smoke

A prenatal (within 4 weeks of enrollment) and postnatal (between 6–10 weeks of the infant’s life) home visit was undertaken to measure IAP; the most common pollutants and by-products of combustion were measured (Vanker et al., 2017). IAP measurements included PM10, carbon monoxide (CO), nitrogen dioxide (NO2), sulfur dioxide (SO2), and the volatile organic compounds (VOCs) benzene and toluene. PM10 was collected using a personal air sampling pump (AirChek 52; SKC, Eighty Four, PA, USA), connected to a styrene filter cassette (37 mm cassette blank; SKC) with a gravimetrically pre-weighed filter (PVC filter 37 mm×5 μm with support pad; SKC) left in the common area of the home for 24 hours (Vanker et al., 2015). Filters were weighed post-sampling and analyzed using National Institute for Occupational Safety and Health method 0600 to obtain an average concentration over 24 h (Ashley, 2017). These average concentrations over 24 h were used as exposures in the analyses. Additionally, the source of heating and type of stove in each participant’s house was noted. More details on home characteristics and IAP have been provided elsewhere (Vanker et al., 2015). During the postnatal home visit between 6–10 weeks of the infant’s life, these same measurements were repeated.

Mothers also provided urine within four weeks of enrollment for analysis of cotinine levels as a measure of tobacco smoke exposure (Vanker et al., 2015). Urine cotinine levels were classified as <10 ng/ml (non-smoker), 10–499 ng/ml (passive smoker), or ≥500 ng/ml (active smoker) (Vanker et al., 2015). In addition, we used a dichotomized smoking variable indicating if the pregnant women were exposed to any (≥10 ng/ml) versus no (<10 ng/ml) tobacco smoke; these participants were categorized as exposed or non-exposed, respectively. For the postnatal assessment of tobacco smoke exposure, mothers filled out a questionnaire approximately 6–10 weeks after giving birth asking current smoking status. If the mothers indicated they were a current smoker, a secondary question asked about how many cigarettes they smoked on average per day.

Assessment of neurodevelopment

At two years of age, children in the DCHS underwent a neurodevelopment assessment using the BSID-III (Bayley, 2006).The BSID-III is one of the most frequently used directly observed tests in infant developmental assessment, and has been reported in both clinical and research settings (Albuquerque et al., 2018; Anderson & Burnett, 2017; Johnson et al., 2014; Kaya-Kara et al., 2019). Cognitive development, language development, motor development, and adaptive behavior were measured using the BSID-III. Trained assessors administered the assessment to generate scores for the cognitive, language, and motor domains via direct observation of the children, and using a caregiver-completed questionnaire for adaptive behavior (Donald et al., 2018). Composite scores were generated which have been used in previous research and validated for use in a South African setting (Ballot et al., 2017; Rademeyer & Jacklin, 2013). Additionally, for this study, an overall composite score was calculated as the average of the four domain sub-scores. If a child did not complete an assessment for each domain, the overall composite score was calculated as the average of the composite sub-scores available. Of the 1143 children originally enrolled, 734 completed at least one BSID-III domain assessment and were subsequently included in the analysis. The results of the BSID-III composite sub-scores and the overall composite score followed a normal distribution (Supplemental Figures S1A-E).

Epigenetic age predictors

DNA was isolated from cord blood samples that were collected at time of delivery (Morin et al., 2017). DNAm data was quantified using the Illumina HumanMethylation450K BeadChip (450K) or Infinium MethylationEPIC BeadChip (EPIC) which were subsequently combined, followed by quality control and normalization. Pre-processing and statistics were done using R 3.5.1 (R Core Team, 2018). Raw iDat files were imported to RStudio where intensity values were converted into beta values. The 450K and EPIC datasets were then combined using the minfi package (Aryee et al., 2014) resulting in 316 samples and 453,093 probes that were available on both arrays. Background subtraction, color correction and normalization were performed using the preprocessFunnorm function (Fortin et al., 2014). After sample and probe filtering, 273 samples and 409,033 probes remained for downstream analyses (Drzymalla et al., 2021). Of these, 191 children had data available on the Bayley Scales scores at two years of age. To assess differences in PM10 distribution between the full BSID-III sample and this subsample, we conducted Wilcoxon Sum Rank tests.

The Bohlin-London gestational age predictor (Bohlin et al., 2016) for each participant was calculated using predictGA function (predictGA package) in R. The Bohlin-London predictor is a weighted sum of beta values of 96 CpG sites with weights as the coefficients of these CpG sites obtained from prediction models trained in 1068 cord blood DNA samples (Bohlin et al., 2016). Among the 96 CpG sites needed to compute Bohlin-London predictor, 8 were missing among EPIC array samples and 3 were missing among both 450K and EPIC array samples. Median beta values from the 450K array were used to impute the beta values of CpG sites missing only among the EPIC array samples and publicly available DNAm data from cord blood (GSE69176 and GSE97628) were used to impute missing values of CpG sites that were missing data from both arrays.

The Knight-Clock gestational age predictor (Knight et al., 2016) was calculated using R code available on github (https://github.com/akknight/PredictGestationalAge). This predictor is a weighted sum of beta values of 142 CpG sites, 6 of which were missing and were imputed by k-nearest neighbors method implemented in the tool.

In our study, the Bohlin-London predictor showed a higher correlation with chronological gestational age (Figures S2A-B) and was therefore used in the subsequent analyses. The Pearson correlations of the Knight-Clock and Bohlin-London predictors with chronological gestational age were 0.387 and 0.641, respectively, and there was no difference between children of Black African or mixed ancestry (Figures S2A-B).

Statistical Analysis

Descriptive statistics were calculated to characterize the study population, summarizing categorical data as N (%) and continuous data as median (IQR); and Pearson correlations between all indoor air pollutants. Linear regression models were used to explore associations of indoor air pollution and maternal smoke exposure on neurodevelopment, adjusting for demographic and socioeconomic characteristics (maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and prenatal parental socioeconomic status); results are presented as β-estimates and 95% confidence intervals (CI). To test if gestational age, birth weight and birth length were intermediates of the association between PM10/tobacco smoke and neurodevelopment and not confounders, sensitivity models were run with and without those covariates. An additional sensitivity analysis was conducted, in which we further adjusted our association analyses with PM10 for season. The distribution of PM10 was skewed in both pre- and postnatal samples, therefore the variable was transformed using the natural log (lnPM10) (Figure S3). Correlations of prenatal IAPs were calculated using a Pearson correlation and correlations of prenatal IAPs and maternal smoking were calculated using Point Biserial correlation. Results showing the association between lnPM10 and the BSID-III overall composite score are standardized by the IQR. In main analyses, we estimated the association of prenatal PM10 and maternal tobacco smoke exposure with the overall composite BSID-III score as the primary outcome. Secondary analyses with each subscale composite score were conducted to assess if specific neurodevelopmental domains are more affected by indoor air pollution or tobacco smoke exposure than others. In addition to the single pollutant models, we conducted two- and multi-pollutant models. In the two-pollutant models, PM10 and smoking were both included in the same linear regression; which can be interpreted as the associations between PM10 and the BSID-III scores additionally adjusted for smoking and the associations between smoking and the BSID-III scores additionally adjusted for PM10. In the multi-pollutant models, linear regression models for the associations were additionally adjusted for benzene, toluene, sulfur dioxide (SO2), nitrogen dioxide (NO2), and carbon monoxide (CO). To assess interaction of PM10 and smoking, crude, adjusted, and multi-pollutant models were run with an interaction term for PM10*smoking.

To identify susceptible time windows of exposure, we analyzed associations of postnatal PM10 and maternal tobacco smoke exposure on the overall composite BSID-III score as well as each composite sub-score and compared the associations to the ones with the prenatal exposures. All data analysis was conducted in R (version 4.0.2).

A mediation analysis using the mediation R package (Tingley et al., 2014) was conducted to analyze if the associations with developmental outcomes were mediated by deviations of epigenetic gestational age from chronological gestational age (ΔGA). ΔGA was calculated as residuals that result from regressing epigenetic gestational age on chronological gestational age. Since the Bohlin-London was determined to be a more accurate predictor of gestational age, this predictor was used for the mediation analysis. A mediator model was fitted where the measure of ΔGA was modeled as a linear regression function of the exposure variable (PM10 or maternal tobacco smoke) while adjusting for the same variables as discussed above. Next, the outcome variable (BSID-III overall composite score, cognition subscale language subscale, motor subscale, or the adaptive behavior subscale) was modeled as a linear regression function of the mediator, exposure variable, and confounding variables as discussed above. Then the mediate function (mediation package) was used to estimate the average causal mediation effects (Indirect), the average direct effects (Direct), the total effects of the mediator and outcome models (Total), and the proportion of the outcome mediated by the mediator. Results of the mediation analysis determined the mediated, direct, and total effects of the exposure on the outcome with the mediated effect accounting for the influence of the mediator, (ΔGA). Sample size for mediation analyses was n=132 for analyses with PM10 as exposure, and n=172 for analyses with smoking status as the exposure, considering complete data for the exposure, mediator, and outcome.

Results

Description of Study Participants

The analysis sample included 734 children with complete information on neurodevelopment at two years of age (Table 1). Overall, about half of the children were male (51.9%). Over half of the children were of Black African ancestry (52.6%) and the rest were of mixed ancestry (47.4%). Participants were categorized into quartiles as lowest, low-to-moderate, moderate-to-high, or highest socioeconomic status based on a composite socioeconomic status score (Vanker et al., 2015; Zar et al., 2015). The percentage of participants using paraffin or fossil fuels in their stove and heating were slightly higher in the subset for multi-pollutant models compared to the full analysis sample (Table 1).

Table 1.

Characteristics of birth cohort and prenatal IAP measurements for sample population subgroups.

Full Cohort Final Analysis Sample Subset for multi-pollutant models Subset for mediation analyses with ΔGA
N 1143 734 347 191
Maternal Age (yrs) (median [IQR]) 26.07 [22.29, 31.09] 26.59 [22.49, 31.35] 26.96 [22.39, 27.47] 26.61 [22.50, 31.96]
Infant Sex = Male (%) 589 (51.5) 379 (51.6) 191 (55.04) 109 (57.07)
Gestational Age (wks) (median [IQR]) 39.00 [37.00, 40.00] 39.00 [38.00, 40.00] 39.00 [38.00, 40.00] 39.00 [38.00, 40.00]
Birth Weight (kg) (median [IQR]) 3.09 [2.71, 3.42] 3.10 [2.71, 3.41] 3.12 [2.82, 3.44] 3.12 [2.76, 3.44]
Birth Length (cm) (median [IQR]) 50.00 [48.00, 52.00] 50.00 [48.00, 52.00] 51.00 [49.00, 53.00] 51.00 [48.00, 58.00]
Ancestry
Mixed ancestry (%) 511 (44.7) 348 (47.4) 167 (48.13) 92 (48.17)
Black African ancestry (%) 632 (55.29) 386 (52.59) 180 (51.87) 99 (51.83)
Parental Socioeconomic status (%)
Lowest SES 274 (24.16) 180 (24.5) 89 (25.65) 48 (25.13)
Low-moderate SES 296 (26.1) 196 (26.7) 94 (27.09) 52 (27.23)
Moderate-high SES 290 (25.57) 193 (26.3) 94 (27.09) 49 (25.65)
Highest SES 283 (24.96) 165 (22.5) 70 (20.17) 42 (21.99)
Paraffin Stove (%) 128 (17.44) 128 (17.44) 76 (21.90) 21 (10.99)
Fossil Stove (%) 176 (23.98) 176 (23.98) 107 (30.84) 28 (14.66)
Paraffin Heating (%) 129 (17.57) 129 (17.57) 77 (22.19) 20 (10.470
Fossil Heating (%) 141 (19.21) 141 (19.21) 87 (25.07) 21 (10.99)
Prenatal Maternal Smoking = Exposed (%) 831 (76.03) 559 (78.1) 261 (75.22) 147 (76.96)
Prenatal Maternal Smoking (%)
Non-Smoker 262 (23.99) 157 (21.9) 86 (24.78) 44 (23.04)
Passive Smoker 479 (43.86) 308 (43.0) 143 (41.21) 83 (43.46)
Active Smoker 352 (32.23) 251 (35.1) 118 (34.01) 64 (33.51)
Postnatal Maternal Smoking = Exposed (%) 252 (27.97) 187 (30.11) 91 (30.23) 49 (28.65)
Prenatal Benzene Concentration (μg/m3) (median [IQR]) 4.28 [1.75, 11.29] 3.67 [1.45, 9.71] 4.28 [1.51, 10.48] 4.34 [2.41, 9.52]
Prenatal Toluene Concentration (μg/m3) (median [IQR]) 16.79 [7.04, 44.24] 16.57 [6.79, 46.24] 17.63 [8.12, 52.67] 17.63 [8.79, 46.50]
Prenatal SO2 Concentration (μg/m3) (median [IQR]) 0.00 [0.00, 0.28] 0.00 [0.00, 0.28] 0.00 [0.00, 0.34] 0.00 [0.00, 0.29]
Prenatal NO2 Concentration (μg/m3) (median [IQR]) 7.13 [3.33, 12.69] 6.61 [3.36, 12.59] 6.92 [3.49, 10.57] 8.58 [4.59, 13.50]
Prenatal PM10 Concentration (μg/m3) (median [IQR]) 33.37 [12.49, 64.80] 35.51 [13.17, 67.57] 33.45 [14.63, 64.41] 38.89 [14.53, 73.88]
Prenatal CO Concentration (mg/m3) (median [IQR]) 0.00 [0.00, 6.02] 0.00 [0.00, 2.86] 0.00 [0.00, 7.88] 0.00 [0.00, 6.58]
Postnatal Benzene Concentration (μg/m3) (median [IQR]) 3.08 [1.06, 9.46] 2.61 [0.90, 7.34] 3.23 [0.83, 11.05] 3.69 [1.30, 14.08]
Postnatal Toluene Concentration (μg/m3) (median [IQR]) 15.50 [5.90, 48.97] 13.26 [5.42, 47.15] 17.04 [6.66, 63.73] 18.68 [6.14, 52.98]
Postnatal SO2 Concentration (μg/m3) (median [IQR]) 0.00 [0.00, 0.00] 0.00 [0.00, 0.00] 0.00 [0.00, 0.00] 0.00 [0.00, 0.00]
Postnatal NO2 Concentration (μg/m3) (median [IQR]) 5.83 [2.58, 12.55] 5.28 [2.57, 11.81] 6.15 [2.57, 12.24] 6.36 [2.59, 13.90]
Postnatal PM10 Concentration (μg/m3) (median [IQR]) 29.47 [12.66, 52.47] 29.28 [12.48, 53.13] 25.86 [9.86, 49.17] 24.38 [12.41, 48.69]
Postnatal CO Concentration (mg/m3) (median [IQR])* 0.00 [0.00, 0.00] 0.00 [0.00, 2.09] 0.00 [0.00, 4.65] 0.00 [0.00, 15.94]
Bayley Scales-Cognitive (median [IQR]) 85.00 [80.00, 90.00] 85.00 [80.00, 90.00] 85.00 [80.0, 90.0] 85.00 [80.00, 90.00]
Bayley Scales-Language (median [IQR]) 83.00 [77.00, 91.00] 83.00 [77.00, 91.00] 83.00 [77.0, 93.25] 85.00 [74.00, 91.00]
Bayley Scales-Motor (median [IQR]) 92.50 [85.00, 100.00] 92.50 [85.00, 100.00] 94.00 [85.00, 100.00] 91.00 [85.00, 100.00]
Bayley Scales-Adaptive Behavior (median [IQR]) 84.00 [72.25, 92.00] 84.00 [72.25, 92.00] 84.00 [74.99, 92.00] 84.00 [71.00, 92.00]
Bayley Scales-Overall Composite (median [IQR]) 86.25 [80.25, 92.25] 86.25 [80.25, 92.25] 87.33 [81.83, 93.67] 86.67 [81.33, 93.67]
Open in a new tab
*

A majority of participants (76%) have a postnatal CO measurement of 0 mg/m3

Participant demographics did not change substantially when the full cohort (N=1143) was subset into those with BSID-III data (n=734), those with full IAP measurements included in the multi-pollutant models (n=347), and those with ΔGA (n=191) (Table 1). There was low correlation between PM10 and the other IAPs, and low correlation between all IAPs and maternal smoking (Figure S4).

PM10 and neurodevelopment

There was a significant association between an increase of one IQR (1.58 μg/m3) in lnPM10 and a lower overall composite BSID-III score (adjusted β-estimate [95% confidence interval]: −0.950 [−1.800, 0.1004]) (Figure 1A, Table S1A). When maternal smoking was added in the two-pollutant model, the effect estimate for PM10 on BSID-III composite score increased and remained statistically significant (−1.040 [−1.8942, −0.186]). lnPM10 was significantly associated with the cognition sub-score (−0.887 [−1.743, −0.032]), the language sub-score (−1.062 [−2.158, −0.062]), and the adaptive behavior sub-score (−1.331 [−2.464, −0.197]), but not with the motor sub-score (Figure 2A, Table S1B). In the sensitivity analysis removing potential intermediates from the model, effect estimates were similar as in the main models. Association between PM10 and BSID-III composite score were robust to additional adjustment for season. (Table S2-3).

Figure 1.

Figure 1.

Open in a new tab

Association of lnPM10 and smoking on Bayley Scales overall composite score. A. lnPM10 and Bayley Scales overall composite score (single and two-pollutant models, standardized by IQR of lnPM10 [1.58 μg/m3]). Crude: crude association between lnPM10 and Bayley score; Adjusted: adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status; Two-pollutant: additionally adjusted for maternal smoking. B. lnPM10 and Bayley Scales overall composite score (single and multi-pollutant models, standardized by IQR of lnPM10 [1.58 μg/m3]). Crude: crude association between PM10 and Bayley score in the subsample, in which all IAPs (PM10, benzene, toluene, SO2, NO2, CO) were available; Adjusted: adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status; Multi-Pollutant: additionally, adjusted for maternal smoking and all other IAPs. C. Smoking and Bayley Scales overall composite score (single and two-pollutant models). Crude: crude association between maternal tobacco smoke and Bayley score; Adjusted: adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status; Two-pollutant: additionally adjusted for PM10. D. Smoking and Bayley Scales overall composite score (single and multi-pollutant models). Crude: crude association between smoking and Bayley score in the subsample, in which all IAPs were available; Adjusted: additionally, adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status; Multi-Pollutant: additionally, adjusted for PM10 and IAPs.

Figure 2. Association of lnPM10 and smoking on Bayley Scales sub-scores.

Figure 2.

Open in a new tab

A. Single pollutant analysis of lnPM10 exposure (per IQR of PM10 [1.58 μg/m3]) on the Bayley Scales four composite sub-scores adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status. B. Single pollutant analysis of maternal tobacco smoke exposure on the Bayley Scales four composite sub-scores adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status.

When the sample was limited to only participants with full IAP measurements (N=347), an IQR increase in lnPM10 in the crude model showed a significant decrease in the BSID III overall composite score (−1.216 [−2.231, −0.200]). This association was strengthened after adjusting for confounders (−1.304 [−2.305, −0.303]). The association between lnPM10 and the BSID-III overall composite score was robust to further adjustment of other IAPs −1.261 [−2.269, −0.253]) (Figure 1B, Table S1C).

Tobacco smoke and neurodevelopment

Exposure to maternal smoking was significantly associated with a lower overall composite score in the crude model (−1.672 [−3.332, −0.012]), but not after adjusting for confounding variables (−1.349 [−3.066, 0.367]). Associations were similar after adjusting for lnPM10 exposure in the two-pollutant model (N=496) (−1.818 [−3.867, 0.231]) (Figure 1C, Table S4A). Maternal smoking during pregnancy was also significantly associated with a lower adaptive behavior scores (−3.386 [−5.632, −1.1389]), but not with the other developmental domains when adjusted for confounding variables (Figure 2B, Table S4B). There was no significant interaction between PM10 and smoking for overall composite score (Table S3A).

When the sample was limited to only participants with all IAP measurements, maternal smoking was associated with the BSID-III overall composite score in the crude model (−2.404 [−4.603, −0.271]), but not after adjusting for confounders (−1.976 [−4.365, 0.414]) or other IAPs (−1.777 [−4.197, 0.642]) (Figure 2D, Table S4C). In models with the PM10*smoking interaction term, there was not a significant effect of the interaction term (Table S5B).

Identification of susceptible time window of exposure

In order to investigate if postnatal exposure to indoor air pollution or tobacco smoke was also impacting the results of the BSID-III scores, a sensitivity analysis was conducted. Due to the longitudinal and prospective nature of the study, it was possible to explore the time window of exposure at which cognitive development appears most susceptible to indoor PM10 and tobacco smoke. This question may be investigated by comparing linear regression models assessing the association of the BSID-III overall composite score with prenatal and postnatal exposure to the same IAP. Pre- and postnatal PM10 exposure was not correlated (r = −0.003) with each other. In contrast to prenatal exposure discussed previously, postnatal exposure to lnPM10 (−0.935 [−1.910, 0.040])(Figure 4A and 4B, Table S6) was not associated with BSID-III composite score. Postnatal tobacco smoke (−1.062 [−2.013, −0.110]) (Figure 4C and 4D, Table S7) was associated with the BSID-III overall composite score. Postnatal lnPM10 was associated with significantly lower cognitive score (−1.062 [−2.013, −0.110]), but not with the other domains. Postnatal tobacco smoke exposure was not significantly associated with performance in any of the developmental domains.

Figure 4. Mediation analysis for deviations of epigenetic gestational age from chronological gestational age (ΔGA).

Figure 4.

Open in a new tab

A mediation analysis was conducted to analyze if the associations of lnPM10 and smoking on the Bayley scores were mediated by ΔGA. All figures show the average causal mediation effects (Indirect), the average direct effects (Direct), the total effects of the mediator and outcome models (Total), and the proportion of the outcome mediated by the mediator (in % with p-value). A. Mediation analysis of ΔGA (mediator) and lnPM10 exposure on Bayley Scales overall composite score B-E. Mediation analysis of ΔGA (mediator) and lnPM10 exposure on the four Bayley Scales sub-scores. F. Mediation analysis of ΔGA and maternal tobacco smoke exposure on Bayley Scales overall composite score. G-J. Mediation analysis of ΔGA and maternal tobacco smoke exposure on the four Bayley Scales sub-scores. All models were adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status.

Mediation by ΔGA

Distribution of PM10 was not significantly different in this subsample compared to the full BSID-III sample (Table S8). Our subsequent mediation analysis did not support the role of ΔGA as mediating pathway for the association between indoor air pollution or smoking and neurodevelopment. Only 1% [p-value = 0.95], 4% [0.52], 4% [0.74], −5% [0.70] of the association between PM10 and the overall composite score, cognition, language, and motor function were mediated by ΔGA, respectively (Figures 4A–D). While 20% [0.13] of the association between PM10 and adaptative behavior was mediated by ΔGA, this finding needs to be interpreted with caution due to the lack of a significant total effect (Figure 4E). Similarly, only −4% [0.79], 1% [0.95], 2% [0.868], −4% [0.91], 7% [0.544] of the association between maternal smoking and the overall composite score, cognition, language, motor, and adaptive behavior were mediated by ΔGA, respectively (Figure 4F–J).

Discussion

In this prospective longitudinal cohort study of pregnant women in South Africa whose children underwent a neurodevelopmental assessment at two years of age, our findings suggest that in-utero exposure to indoor air pollution and tobacco smoke may be negatively associated with neurodevelopment at two years of age. Exposure to indoor PM10 during pregnancy was associated with poorer developmental outcomes particularly for cognitive and language domains, as well as for adaptive behavior Even after adjusting for confounding variables, gestational age, smoking and other IAPs (CO, NO2, SO2, VOCs), PM10 remained significantly associated with the lower overall BSID-III composite score. The increased effect estimate in the full IAP subsample may be due to uncertainty in the model, as wider 95% CIs are observed. Though, the significant association indicates that PM10 is an important neurotoxicant independent of other IAPs. There were no significant interaction between PM10 and smoking in any models assessing effect on BSID-III overall composite score. The association between PM10 and neurodevelopment was only significant for prenatal exposure (not postnatal exposure), indicating that pregnancy is a critical period for neurodevelopment and vulnerability to IAP. Postnatal PM10 was significantly associated with the cognitive domain performance, the postnatal early life period may be a sensitive window for PM10 exposure. We did not find any indication that ΔGA may be a mediating pathway for the association between IAP and BSID-III scores. Further studies are needed to better understand the biological mechanisms of how exposure to air pollutants causes neurological changes and the role of epigenetics.

Our study addresses an important gap in the current literature on air pollution and neurodevelopment by providing the first evidence for an association between measured indoor PM10 concentrations and neurodevelopmental outcomes in very young children. While no other studies researched the specific effect of prenatal PM10 exposure on neurodevelopment using BSID-III, PM10 is thought to be neurotoxic and has been associated with reduced IQ, poorer mental and psychomotor development, and ASD diagnosis (Costa et al., 2019; Kim et al., 2014; Loftus et al., 2019; Volk et al., 2013). In addition, one study found outdoor PM2.5 exposure during pregnancy was associated with lower motor scores in the BSID (Lertxundi et al., 2015). Lertxundi et al. did not measure PM10 exposure, but PM10 and PM2.5 are typically highly correlated (Eeftens et al., 2012). Conversely, a limitation of our study is that PM2.5 was not measured, this highlights the need for broad pollutant monitoring in epidemiology studies, Indoor PM10 can come from a variety of sources including smoking, cooking using kerosene and biomass fuels, woodstoves and furnaces (Vardoulakis et al., 2020). These sources, as opposed to sources of outdoor pollutants, can be easily modified by individuals to reduce PM10 concentrations and exposure in the home. Identifying specific pollutants that affect neurodevelopment is important for identifying and eliminating the source of those pollutants to prevent exposure and subsequent health outcomes. Future studies can incorporate outdoor exposure measurements for a holistic assessment of air pollution exposure.

Exposure to maternal tobacco smoke was not associated with lower overall neurodevelopment in adjusted models, except with respect to BSID-III adaptive behavior outcomes. These findings are in line with previous studies that have found self-reported environmental tobacco smoke exposure during pregnancy to be associated with specific domains of development as measured on the BSID-III , specifically cognition and language (He et al., 2018) (B.-E. Lee et al., 2011; M. Lee et al., 2019). Another study that also used maternal urinary cotinine levels to assess maternal tobacco smoke exposure found cotinine levels were associated with mental development index scores, in a study using the Korean version of the Bayley Scales of Infant Development version II (K-BSID-II) (B.-E. Lee et al., 2011; M. Lee et al., 2019).

The two-pollutant models that included both PM10 and smoking status reached statistical significance for PM10 and not smoking. Though, effect estimates did not differ between models that were or were not adjusted for co-exposure to PM10 and smoking. This suggests that there may be independent effects of PM10 and smoking on neurodevelopment. While smoking cigarettes does emit PM10 into the home, cigarettes also expose the smoker to other toxicants that may harm the developing fetus (Avino et al., 2018).

Subsequently, our findings indicated that the observed significance of indoor air pollution exposure on neurodevelopment is primarily the result of the prenatal, in-utero exposure, rather than postnatal early-life exposure. Previous studies have reported that the second trimester was the most sensitive time window for behavioral and active muscle tone development (Chen et al., 2020), indicating that this trimester could also be highly sensitive to air pollution exposure. However, this study only investigated exposure during the second trimester, so future studies would be needed to assess sensitive windows in other stages of pregnancy. In contrast to the present study, other studies have found both pre- and postnatal secondhand smoke exposure associated with poorer neurodevelopmental outcomes at two years (Mohamed et al., 2018). This may, however, be due to the use of different outcome measures of neurodevelopment, the use of self-reported smoking status in the postnatal period as opposed to measured cotinine levels, or due to limited statistical power of our analyses. A limitation of the sensitivity analyses comparing pre- and postnatal exposure periods is the differential measurement of exposure to tobacco smoke. These differences could potentially explain stronger associations with prenatal tobacco smoke exposure than with postnatal exposures

In our study, the association of prenatal exposure to PM10 and tobacco smoke on neurodevelopment were not significantly mediated by epigenetic gestational age (measured by DNAm in cord blood). So far, only one study has investigated ΔGA as a potential mediating pathway for the adverse effects of prenatal air pollution exposure (Sbihi et al., 2019). Sbihi et al. found indications that ΔGA might mediate the association between prenatal exposure to traffic-related air pollution and allergic sensitization in a cohort of 145 children (Sbihi et al., 2019). Future studies should investigate the role of other epigenetic markers in addition to ΔGA as potential mediators for the association between air pollution and neurodevelopment using high-dimensional mediation analysis or pre-selected CpG sites based on their known association with air pollution and/or neurodevelopment.

The DCHS is a unique birth cohort that enables the analysis of various exposure assessments over time to investigate potential associations between in-utero air pollution and neurodevelopment due to prenatal recruitment and postnatal follow-ups. Prenatal recruitment allows assessment of exposure during gestation and after birth, use of a population-based sample helps to eliminate selection biases inherent in case-based approaches, and the cohort design ensures clear identification of the time-order of associations (Zar et al., 2015). Additional strengths include a large sample size that evaluated high-risk infants using the BSID-III, a well- established assessment tool. Though, not all children in the cohort completed the BSID-III, limiting sample size in this study (Wedderburn et al., 2019). The limitations of this study include the short PM10 exposure record time, and missing values for IAPs and DNAm. Although PM10 was recorded prenatally and postnatally and therefore, highlighted multiple time points, it was only recorded for 24 hours at each assessment. Therefore, this measurement could potentially incorrectly reflect actual indoor air pollution exposure. This may also explain low correlation between pre- and postnatal PM10 measurements. Additionally, IAP measurements were only available for a subsample, reducing the sample size in multi pollutant models. There was also reduced sample size for analyses using DNAm, because only a subsample of participants had DNAm data. Another limitation was the one-time measurement of the neurodevelopment assessment. While neurodevelopment occurs over multiple years, the focus of this study was to examine the impact of IAP on early cognitive development. Future studies should look at the association between prenatal exposure to air pollution and the trajectory of neurodevelopmental outcomes over the early years of life. Additionally, future studies can introduce interventions aimed at preventing IAP exposure to reduce neurodevelopmental outcomes.

To our knowledge, this is the first study to assess prenatal exposure to measured indoor air pollutants on neurodevelopment in a large cohort study, and therefore, addresses an important gap in epidemiologic and environmental health research. Moreover, this is also the first study to include research on the epigenetic mechanisms that may be influencing neurodevelopment in the context of exposure to indoor air pollutants. Although our findings determined these associations are not significantly mediated by ΔGA, other epigenetic mechanisms could be influencing neurodevelopment upon in-utero exposure to air pollution. Additionally, our study supports the view that pregnancy is a vulnerable time window of exposure for neurodevelopment, emphasizing the importance of pregnancy as a time to implement integrated approaches to support healthy lifestyles and reduce substance use and other factors associated with neurodevelopmental risk in the unborn child.

Conclusion

In summary, our findings show a significant association between prenatal exposure to indoor air pollution (PM10) and tobacco smoke on neurodevelopment at two years of age. Further research to understand underlying biological mediators of toxic exposure and neurodevelopment is needed.

Supplementary Material

1

Figure 3. Identification of the most vulnerable time window of exposure.

Figure 3.

Open in a new tab

A. lnPM10 (per IQR) and Bayley Scales overall composite score. B. lnPM10 (per IQR) and Bayley Scales composite sub-scores. C. Smoking and Bayley Scales overall composite score. D. Smoking and Bayley Scales composite sub-scores. Prenatal: Prenatal lnPM10 or tobacco smoke exposure (within 4 weeks of enrollment), Postnatal: Postnatal lnPM10 or tobacco smoke exposure (between 6–10 weeks of the infant’s life); IQR prenatal lnPM10 exposure: 1.58 μg/m3; IQR postnatal lnPM10 exposure: 1.39 μg/m3. All associations were adjusted for maternal age, gestational age, sex of child, ancestry, birth weight, birth length, and parental socioeconomic status

Highlights.

  • Well-characterized longitudinal birth cohort from South Africa (N=734)

  • Measured indoor air pollution data (PM10) from pre- and postnatal visits

  • Exposure to indoor PM10 was associated with poorer cognitive development

  • Investigation of epigenetic mechanisms of the PM10-neurodevelopment association

Acknowledgments:

The authors thank the study and clinical staff at Paarl Hospital, Mbekweni and TC Newman clinics, as well as the CEO of Paarl Hospital, and the Western Cape Health Department for their support of the study. The authors thank the families and children who participated in this study.

Funding:

The Drakenstein Child Health Study was funded by the Bill & Melinda Gates Foundation (OPP 1017641), Discovery Foundation, South African Medical Research Council, National Research Foundation South Africa, CIDRI Clinical Fellowship and Wellcome Trust (204755/2/16/z). Additional support for the DNA methylation work was by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health (NICHD) under Award Number R21HD085849, and the Fogarty International Center (FIC). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. AH is supported the HERCULES Center (NIEHS P30ES019776). GMC is supported by the NIEHS T32 Training Program in Environmental Health and Toxicology (5T32ES12870). CJW was supported by a Wellcome Trust fellowship (203525/Z/16/Z). DJS and HJZ are supported by the South African Medical Research Council (SAMRC). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of manuscript.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

  1. Albuquerque P. L. de, Guerra M. Q. de F., Lima M. de C., & Eickmann SH (2018). Concurrent validity of the Alberta Infant Motor Scale to detect delayed gross motor development in preterm infants: A comparative study with the Bayley III. Developmental Neurorehabilitation, 21(6), 408–414. 10.1080/17518423.2017.1323974 [DOI] [PubMed] [Google Scholar]
  2. Anderson PJ, & Burnett A (2017). Assessing developmental delay in early childhood—Concerns with the Bayley-III scales. The Clinical Neuropsychologist, 31(2), 371–381. 10.1080/13854046.2016.1216518 [DOI] [PubMed] [Google Scholar]
  3. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, & Irizarry RA (2014). Minfi: A flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics, 30(10), 1363–1369. 10.1093/bioinformatics/btu049 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Ashley K (2017). NIOSH Manual of Analytical Methods (NMAM), 5th Edition. 860. [PMC free article] [PubMed] [Google Scholar]
  5. Avino P, Scungio M, Stabile L, Cortellessa G, Buonanno G, & Manigrasso M (2018). Second-hand aerosol from tobacco and electronic cigarettes: Evaluation of the smoker emission rates and doses and lung cancer risk of passive smokers and vapers. Science of The Total Environment, 642, 137–147. 10.1016/j.scitotenv.2018.06.059 [DOI] [PubMed] [Google Scholar]
  6. Ballot DE, Ramdin T, Rakotsoane D, Agaba F, Davies VA, Chirwa T, & Cooper PA (2017). Use of the Bayley Scales of Infant and Toddler Development, Third Edition, to Assess Developmental Outcome in Infants and Young Children in an Urban Setting in South Africa. International Scholarly Research Notices, 2017, 1631760. 10.1155/2017/1631760 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Balmes JR (2019). Household air pollution from domestic combustion of solid fuels and health. The Journal of Allergy and Clinical Immunology, 143(6), 1979–1987. 10.1016/j.jaci.2019.04.016 [DOI] [PubMed] [Google Scholar]
  8. Bayley N (2006). No Title. Bayley Scales of Infant and Toddler Development, Technical Manual. 2006; 3rd Ed. [Google Scholar]
  9. Bohlin J, Håberg SE, Magnus P, Reese SE, Gjessing HK, Magnus MC, Parr CL, Page CM, London SJ, & Nystad W (2016). Prediction of gestational age based on genome-wide differentially methylated regions. Genome Biology, 17(1), 207. 10.1186/s13059-016-1063-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Caramaschi D, Neumann A, Cardenas A, Tindula G, Alemany S, Zillich L, Pesce G, Lahti JMT, Havdahl A, Mulder R, Felix JF, Tiemeier H, Sirignano L, Frank J, Witt SH, Rietschel M, Deuschle M, Huen K, Eskenazi B, … Sharp G (2022). Meta-analysis of epigenome-wide associations between DNA methylation at birth and childhood cognitive skills. Molecular Psychiatry, 1–10. 10.1038/s41380-022-01441-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen B, Huang S, He J, He Q, Chen S, Liu X, Peng S, Luo D, & Duan Y (2020). Sex-specific influence of prenatal air pollutant exposure on neonatal neurobehavioral development and the sensitive window. Chemosphere, 254, 126824. 10.1016/j.chemosphere.2020.126824 [DOI] [PubMed] [Google Scholar]
  12. Costa LG, Aschner M, Vitalone A, Syversen T, & Soldin OP (2004). Developmental neuropathology of environmental agents. Annual Review of Pharmacology and Toxicology, 44, 87–110. 10.1146/annurev.pharmtox.44.101802.121424 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Costa LG, Cole TB, Dao K, Chang Y-C, & Garrick JM (2019). Developmental impact of air pollution on brain function. Neurochemistry International, 131(1), 104580. 10.1016/j.neuint.2019.104580 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Donald KA, Hoogenhout M, du Plooy CP, Wedderburn CJ, Nhapi RT, Barnett W, Hoffman N, Malcolm-Smith S, Zar HJ, & Stein DJ (2018). Drakenstein Child Health Study (DCHS): Investigating determinants of early child development and cognition. BMJ Paediatrics Open, 2(1), e000282. 10.1136/bmjpo-2018-000282 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Donald KA, Wedderburn CJ, Barnett W, Nhapi RT, Rehman AM, Stadler JAM, Hoffman N, Koen N, Zar HJ, & Stein DJ (2019). Risk and protective factors for child development: An observational South African birth cohort. PLoS Medicine, 16(9), e1002920. 10.1371/journal.pmed.1002920 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Drzymalla E, Gladish N, Koen N, Epstein MP, Kobor MS, Zar HJ, Stein DJ, & Hüls A (2021). Association between maternal depression during pregnancy and newborn DNA methylation. Translational Psychiatry, 11(1), 1–8. 10.1038/s41398-021-01697-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Eeftens M, Tsai M-Y, Ampe C, Anwander B, Beelen R, Bellander T, Cesaroni G, Cirach M, Cyrys J, de Hoogh K, De Nazelle A, de Vocht F, Declercq C, Dėdelė A, Eriksen K, Galassi C, Gražulevičienė R, Grivas G, Heinrich J, … Hoek G (2012). Spatial variation of PM2.5, PM10, PM2.5 absorbance and PMcoarse concentrations between and within 20 European study areas and the relationship with NO2 – Results of the ESCAPE project. Atmospheric Environment, 62, 303–317. 10.1016/j.atmosenv.2012.08.038 [DOI] [Google Scholar]
  18. Fortin J-P, Labbe A, Lemire M, Zanke BW, Hudson TJ, Fertig EJ, Greenwood CM, & Hansen KD (2014). Functional normalization of 450k methylation array data improves replication in large cancer studies. Genome Biology, 15(12), 503. 10.1186/s13059-014-0503-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gruzieva O, Xu C-J, Breton CV, Annesi -Maesano Isabella, Ant ó J. M., Auffray C, Ballereau S, Bellander T, Bousquet J, Bustamante M, Charles M-A, de KY, den DHT, Duijts, Felix JF, Gehring U, Guxens M, Jaddoe VVW, Jankipersadsing SA, … Mel én E. (2017). Epigenome-Wide Meta-Analysis of Methylation in Children Related to Prenatal NO2 Air Pollution Exposure. Environmental Health Perspectives, 125(1), 104–110. 10.1289/EHP36 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gruzieva O, Xu C-J, Yousefi P, Relton C, Merid SK, Breton CV, Gao L, Volk HE, Feinberg JI, Ladd -Acosta Christine, Bakulski K, Auffray C, Lemonnier N, Plusquin M, Ghantous A, Herceg Z, Nawrot TS, Pizzi C, Richiardi L, … Mel én E. (2019). Prenatal Particulate Air Pollution and DNA Methylation in Newborns: An Epigenome-Wide Meta-Analysis. Environmental Health Perspectives, 127(5), 057012. 10.1289/EHP4522 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hannum G, Guinney J, Zhao L, Zhang L, Hughes G, Sadda S, Klotzle B, Bibikova M, Fan J-B, Gao Y, Deconde R, Chen M, Rajapakse I, Friend S, Ideker T, & Zhang K (2013). Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular Cell, 49(2), 359–367. 10.1016/j.molcel.2012.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. He Y, Luo R, Wang T, Gao J, & Liu C (2018). Prenatal Exposure to Environmental Tobacco Smoke and Early Development of Children in Rural Guizhou Province, China. International Journal of Environmental Research and Public Health, 15(12), 2866. 10.3390/ijerph15122866 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Horvath S (2013). DNA methylation age of human tissues and cell types. Genome Biology, 14(10), 3156. 10.1186/gb-2013-14-10-r115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Huijbregts SCJ, Séguin JR, Zelazo PD, Parent S, Japel C, & Tremblay RE (2006). Interrelations between maternal smoking during pregnancy, birth weight and sociodemographic factors in the prediction of early cognitive abilities. Infant and Child Development, 15(6), 593–607. 10.1002/icd.480 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hüls A, & Czamara D (2020). Methodological challenges in constructing DNA methylation risk scores. Epigenetics, 15(1–2), 1–11. 10.1080/15592294.2019.1644879 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Hüls A, Wedderburn CJ, Groenewold NA, Gladish N, Jones MJ, Koen N, MacIsaac JL, Lin DTS, Ramadori KE, Epstein MP, Donald KA, Kobor MS, Zar HJ, & Stein DJ (2021). Newborn differential DNA methylation and subcortical brain volumes as early signs of severe neurodevelopmental delay in a South African Birth Cohort Study. The World Journal of Biological Psychiatry, 0(0), 1–12. 10.1080/15622975.2021.2016955 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Islami F, Torre LA, & Jemal A (2015). Global trends of lung cancer mortality and smoking prevalence. Translational Lung Cancer Research, 4(4), 327–338. 10.3978/j.issn.2218-6751.2015.08.04 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Johnson S, Moore T, & Marlow N (2014). Using the Bayley-III to assess neurodevelopmental delay: Which cut-off should be used? Pediatric Research, 75(5), 670–674. 10.1038/pr.2014.10 [DOI] [PubMed] [Google Scholar]
  29. Kaya-Kara Ö, Kerem-Günel M, & Yiğit Ş (2019). Correlation of the Bayley scales of infant-toddler development-3rd edition and neuro-sensory motor assessment in preterm infants during the first year of life. The Turkish Journal of Pediatrics, 61(3), 399–406. 10.24953/turkjped.2019.03.012 [DOI] [PubMed] [Google Scholar]
  30. Kim E, Park H, Hong Y-C, Ha M, Kim Y, Kim B-N, Kim Y, Roh Y-M, Lee B-E, Ryu J-M, Kim B-M, & Ha E-H (2014). Prenatal exposure to PM₁₀ and NO₂ and children’s neurodevelopment from birth to 24 months of age: Mothers and Children’s Environmental Health (MOCEH) study. The Science of the Total Environment, 481, 439–445. 10.1016/j.scitotenv.2014.01.107 [DOI] [PubMed] [Google Scholar]
  31. Knight AK, Craig JM, Theda C, Bækvad-Hansen M, Bybjerg-Grauholm J, Hansen CS, Hollegaard MV, Hougaard DM, Mortensen PB, Weinsheimer SM, Werge TM, Brennan PA, Cubells JF, Newport DJ, Stowe ZN, Cheong JLY, Dalach P, Doyle LW, Loke YJ, … Smith AK (2016). An epigenetic clock for gestational age at birth based on blood methylation data. Genome Biology, 17(1), 206. 10.1186/s13059-016-1068-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lam J, Sutton P, Kalkbrenner A, Windham G, Halladay A, Koustas E, Lawler C, Davidson L, Daniels N, Newschaffer C, & Woodruff T (2016). A Systematic Review and Meta-Analysis of Multiple Airborne Pollutants and Autism Spectrum Disorder. PLoS ONE, 11(9), e0161851. 10.1371/journal.pone.0161851 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Lawlor DA, Najman JM, Batty GD, O’Callaghan MJ, Williams GM, & Bor W (2006). Early life predictors of childhood intelligence: Findings from the Mater-University study of pregnancy and its outcomes. Paediatric and Perinatal Epidemiology, 20(2), 148–162. 10.1111/j.1365-3016.2006.00704.x [DOI] [PubMed] [Google Scholar]
  34. Lee B-E, Hong Y-C, Park H, Ha M, Hyeong Kim J, Chang N, Roh Y-M, Kim B-N, Kim Y, Oh S, Ju Kim Y, & Ha E-H (2011). Secondhand smoke exposure during pregnancy and infantile neurodevelopment. Environmental Research, 111(4), 539–544. 10.1016/j.envres.2011.02.014 [DOI] [PubMed] [Google Scholar]
  35. Lee M, Ha M, Hong Y-C, Park H, Kim Y, Kim E-J, Kim Y, & Ha E (2019). Exposure to prenatal secondhand smoke and early neurodevelopment: Mothers and Children’s Environmental Health (MOCEH) study. Environmental Health, 18, 22. 10.1186/s12940-019-0463-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Lertxundi A, Baccini M, Lertxundi N, Fano E, Aranbarri A, Martínez MD, Ayerdi M, Álvarez J, Santa-Marina L, Dorronsoro M, & Ibarluzea J (2015). Exposure to fine particle matter, nitrogen dioxide and benzene during pregnancy and cognitive and psychomotor developments in children at 15months of age. Environment International, 80, 33–40. 10.1016/j.envint.2015.03.007 [DOI] [PubMed] [Google Scholar]
  37. Levy RJ (2015). Carbon monoxide pollution and neurodevelopment: A public health concern. Neurotoxicology and Teratology, 49, 31–40. 10.1016/j.ntt.2015.03.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lim SS, Vos T, Flaxman AD, Danaei G, Shibuya K, Adair-Rohani H, AlMazroa MA, Amann M, Anderson HR, Andrews KG, Aryee M, Atkinson C, Bacchus LJ, Bahalim AN, Balakrishnan K, Balmes J, Barker-Collo S, Baxter A, Bell ML, … Ezzati M (2012). A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: A systematic analysis for the Global Burden of Disease Study 2010. The Lancet, 380(9859), 2224–2260. 10.1016/S0140-6736(12)61766-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Loftus CT, Hazlehurst MF, Szpiro AA, Ni Y, Tylavsky FA, Bush NR, Sathyanarayana S, Carroll KN, Karr CJ, & LeWinn KZ (2019). Prenatal air pollution and childhood IQ: Preliminary evidence of effect modification by folate. Environmental Research, 176, 108505. 10.1016/j.envres.2019.05.036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Mezzacappa E, Buckner JC, & Earls F (2011). Prenatal cigarette exposure and infant learning stimulation as predictors of cognitive control in childhood. Developmental Science, 14(4), 881–891. 10.1111/j.1467-7687.2011.01038.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Mohamed NN, Loy SL, Lim PY, Al Mamun A, & Jan Mohamed HJ (2018). Early life secondhand smoke exposure assessed by hair nicotine biomarker may reduce children’s neurodevelopment at 2years of age. The Science of the Total Environment, 610–611, 147–153. 10.1016/j.scitotenv.2017.08.030 [DOI] [PubMed] [Google Scholar]
  42. Mordaunt CE, Jianu JM, Laufer BI, Zhu Y, Hwang H, Dunaway KW, Bakulski KM, Feinberg JI, Volk HE, Lyall K, Croen LA, Newschaffer CJ, Ozonoff S, Hertz-Picciotto I, Fallin MD, Schmidt RJ, & LaSalle JM (2020). Cord blood DNA methylome in newborns later diagnosed with autism spectrum disorder reflects early dysregulation of neurodevelopmental and X-linked genes. Genome Medicine, 12(1), 88. 10.1186/s13073-020-00785-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Morin AM, Gatev E, McEwen LM, MacIsaac JL, Lin DTS, Koen N, Czamara D, Räikkönen K, Zar HJ, Koenen K, Stein DJ, Kobor MS, & Jones MJ (2017). Maternal blood contamination of collected cord blood can be identified using DNA methylation at three CpGs. Clinical Epigenetics, 9(1), 75. 10.1186/s13148-017-0370-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Nazif-Muñoz JI, Spengler JD, Arku RE, & Oulhote Y (2020). Solid fuel use and early child development disparities in Ghana: Analyses by gender and urbanicity. Journal of Exposure Science & Environmental Epidemiology, 30(4), 698–706. 10.1038/s41370-020-0224-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Neumann A, Walton E, Alemany S, Cecil C, González JR, Jima DD, Lahti J, Tuominen ST, Barker ED, Binder E, Caramaschi D, Carracedo Á, Czamara D, Evandt J, Felix JF, Fuemmeler BF, Gutzkow KB, Hoyo C, Julvez J, … Tiemeier H (2020). Association between DNA methylation and ADHD symptoms from birth to school age: A prospective meta-analysis. Translational Psychiatry, 10(1), 1–11. 10.1038/s41398-020-01058-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Pickett AR, & Bell ML (2011). Assessment of Indoor Air Pollution in Homes with Infants. International Journal of Environmental Research and Public Health, 8(12), 4502–4520. 10.3390/ijerph8124502 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Polanska K, Hanke W, Sobala W, Trzcinka-Ochocka M, Ligocka D, Brzeznicki S, Strugala-Stawik H, & Magnus P (2013). Developmental Effects of Exposures to Environmental Factors: The Polish Mother and Child Cohort Study. BioMed Research International, 2013, 1–11. 10.1155/2013/629716 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. R Core Team. (2018). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna. [Google Scholar]
  49. Rademeyer V, & Jacklin L (2013). A study to evaluate the performance of black South African urban infants on the Bayley Scales of Infant Development III. South African Journal of Child Health, 7(2), 54. 10.7196/sajch.547 [DOI] [Google Scholar]
  50. Raz R, Roberts AL, Lyall K, Hart JE, Just AC, Laden F, & Weisskopf MG (2015). Autism spectrum disorder and particulate matter air pollution before, during, and after pregnancy: A nested case-control analysis within the Nurses’ Health Study II Cohort. Environmental Health Perspectives, 123(3), 264–270. 10.1289/ehp.1408133 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Rodier PM (1995). Developing brain as a target of toxicity. Environmental Health Perspectives, 103 Suppl, 73–76. 10.1289/ehp.95103s673 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Rodier PM (2004). Environmental Causes of Central Nervous System Maldevelopment. Pediatrics, 113(Supplement_3), 1076–1083. 10.1542/peds.113.S3.1076 [DOI] [PubMed] [Google Scholar]
  53. Roza SJ, Verhulst FC, Jaddoe VW, Steegers EA, Mackenbach JP, Hofman A, & Tiemeier H (2009). Maternal smoking during pregnancy and child behaviour problems: The Generation R Study. International Journal of Epidemiology, 38(3), 680–689. 10.1093/ije/dyn163 [DOI] [PubMed] [Google Scholar]
  54. Sbihi H, Jones MJ, MacIsaac JL, Brauer M, Allen RW, Sears MR, Subbarao P, Mandhane PJ, Moraes TJ, Azad MB, Becker AB, Brook JR, Kobor MS, & Turvey SE (2019). Prenatal exposure to traffic-related air pollution, the gestational epigenetic clock, and risk of early-life allergic sensitization. Journal of Allergy and Clinical Immunology, 144(6), 1729–1731.e5. 10.1016/j.jaci.2019.07.047 [DOI] [PubMed] [Google Scholar]
  55. Stein DJ, Koen N, Donald KA, Adnams CM, Koopowitz S, Lund C, Marais A, Myers B, Roos A, Sorsdahl K, Stern M, Tomlinson M, van der Westhuizen C, Vythilingum B, Myer L, Barnett W, Brittain K, & Zar HJ (2015). Investigating the psychosocial determinants of child health in Africa: The Drakenstein Child Health Study. Journal of Neuroscience Methods, 252, 27–35. 10.1016/j.jneumeth.2015.03.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Tingley D, Yamamoto T, Hirose K, Keele L, & Imai K (2014). mediation: R Package for Causal Mediation Analysis. Journal of Statistical Software, 59(5). 10.18637/jss.v059.i05 [DOI] [Google Scholar]
  57. Vanker A, Barnett W, Nduru PM, Gie RP, Sly PD, & Zar HJ (2015). Home environment and indoor air pollution exposure in an African birth cohort study. The Science of the Total Environment, 536, 362–367. 10.1016/j.scitotenv.2015.06.136 [DOI] [PubMed] [Google Scholar]
  58. Vanker A, Barnett W, Workman L, Nduru PM, Sly PD, Gie RP, & Zar HJ (2017). Early-life exposure to indoor air pollution or tobacco smoke and lower respiratory tract illness and wheezing in African infants: A longitudinal birth cohort study. The Lancet. Planetary Health, 1(8), e328–e336. 10.1016/S2542-5196(17)30134-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Vardoulakis S, Giagloglou E, Steinle S, Davis A, Sleeuwenhoek A, Galea KS, Dixon K, & Crawford JO (2020). Indoor Exposure to Selected Air Pollutants in the Home Environment: A Systematic Review. International Journal of Environmental Research and Public Health, 17(23), 8972. 10.3390/ijerph17238972 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Volk HE, Lurmann F, Penfold B, Hertz-Picciotto I, & McConnell R (2013). Traffic Related Air Pollution, Particulate Matter, and Autism. JAMA Psychiatry, 70(1), 71–77. 10.1001/jamapsychiatry.2013.266 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Volk HE, Perera F, Braun JM, Kingsley SL, Gray K, Buckley J, Clougherty JE, Croen LA, Eskenazi B, Herting M, Just AC, Kloog I, Margolis A, McClure LA, Miller R, Levine S, & Wright R (2021). Prenatal air pollution exposure and neurodevelopment: A review and blueprint for a harmonized approach within ECHO. Environmental Research, 196, 110320. 10.1016/j.envres.2020.110320 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Vrijheid M, Martinez D, Aguilera I, Bustamante M, Ballester F, Estarlich M, Fernandez-Somoano A, Guxens M, Lertxundi N, Martinez MD, Tardon A, Sunyer J, & INMA Project. (2012). Indoor air pollution from gas cooking and infant neurodevelopment. Epidemiology (Cambridge, Mass.), 23(1), 23–32. 10.1097/EDE.0b013e31823a4023 [DOI] [PubMed] [Google Scholar]
  63. Walton E, Pingault J-B, Cecil C. a. M., Gaunt TR, Relton CL, Mill J, & Barker ED (2017). Epigenetic profiling of ADHD symptoms trajectories: A prospective, methylome-wide study. Molecular Psychiatry, 22(2), 250–256. 10.1038/mp.2016.85 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Wedderburn CJ, Yeung S, Rehman AM, Stadler JAM, Nhapi RT, Barnett W, Myer L, Gibb DM, Zar HJ, Stein DJ, & Donald KA (2019). Neurodevelopment of HIV-exposed uninfected children in South Africa: Outcomes from an observational birth cohort study. The Lancet Child & Adolescent Health, 3(11), 803–813. 10.1016/S2352-4642(19)30250-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Wei C-F, Chen M-H, Lin C-C, Guo YL, Lin S-J, Hsieh W-S, & Chen P-C (2018). Household incense burning and infant gross motor development: Results from the Taiwan Birth Cohort Study. Environment International, 115, 110–116. 10.1016/j.envint.2018.03.005 [DOI] [PubMed] [Google Scholar]
  66. Zar HJ, Barnett W, Myer L, Stein DJ, & Nicol MP (2015). Investigating the early-life determinants of illness in Africa: The Drakenstein Child Health Study. Thorax, 70(6), 592–594. 10.1136/thoraxjnl-2014-206242 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

1
NIHMS1807601-supplement-1.docx (90.1KB, docx)

RESOURCES