Portal de Periódicos da CAPES

Prenatal exposure to traffic-related air pollution, the gestational epigenetic clock, and risk of early-life allergic sensitization

2019; Elsevier BV; Volume: 144; Issue: 6 Linguagem: Inglês

10.1016/j.jaci.2019.07.047

1097-6825

Hind Sbihi, Meaghan J. Jones, Julie L. MacIsaac, Michael Bräuer, Ryan W. Allen, Malcolm R. Sears, Padmaja Subbarao, Piush J. Mandhane, Theo J. Moraes, Meghan B. Azad, Allan B. Becker, Jeffrey R. Brook, Michael S. Kobor, Stuart E. Turvey,

Noise Effects and Management

The Developmental Origins of Health and Disease paradigm posits that an adverse early-life environment induces gradients at birth that can alter lifetime health trajectories, including the development of atopic diseases. An important adverse environmental factor that has received considerable scrutiny is air pollution. Meta-analyses of birth cohort studies suggest that early-life exposure to traffic-related air pollution (TRAP) is associated with an increased risk of asthma and allergic sensitization, with a caveat that there is substantial variability across studies.1Bowatte G. Lodge C. Lowe A.J. Erbas B. Perret J. Abramson M.J. et al.The influence of childhood traffic-related air pollution exposure on asthma, allergy and sensitization: a systematic review and a meta-analysis of birth cohort studies.Allergy. 2015; 70: 245-256Crossref PubMed Scopus (293) Google Scholar In seeking to understand the underlying biology, the study of epigenetics offers a plausible mechanism by which such environmental exposures in early life can get "under the skin" to regulate the activity of genes relevant to allergy and asthma.2Feinberg A.P. The key role of epigenetics in human disease prevention and mitigation.N Engl J Med. 2018; 378: 1323-1334Crossref PubMed Scopus (303) Google Scholar We used a recently developed biomarker of fetal development—epigenetic gestational age (eGA)—to further characterize the relationship between in utero exposure to TRAP and allergic sensitization. As a biomarker, eGA holds great promise because it may identify infants who deviate from the typical developmental profile, allowing targeted early preventative interventions. In adults, patterns of DNA methylation (DNAm) at specific CpG sites are used to estimate "epigenetic age" and this is highly correlated with chronological age.3Jones M.J. Goodman S.J. Kobor M.S. DNA methylation and healthy human aging.Aging Cell. 2015; 14: 924-932Crossref PubMed Scopus (473) Google Scholar Moreover, deviations between epigenetic and chronological age have been associated with many age-related health outcomes.4Marioni RE, Harris SE, Shah S, McRae AF, von Zglinicki T, Martin-Ruiz C, et al. The epigenetic clock and telomere length are independently associated with chronological age and mortality. Int J Epidemiol. published online ahead of print April 13, 2016, pii: dyw041.Google Scholar Using umbilical cord blood DNAm profiles, similar estimates of eGA have been created.5Knight A.K. Craig J.M. Theda C. Bækvad-Hansen M. Bybjerg-Grauholm J. Hansen C.S. et al.An epigenetic clock for gestational age at birth based on blood methylation data.Genome Biol. 2016; 17: 206Crossref PubMed Scopus (52) Google Scholar, 6Bohlin J. Håberg S.E. Magnus P. Reese S.E. Gjessing H.K. Magnus M.C. et al.Prediction of gestational age based on genome-wide differentially methylated regions.Genome Biol. 2016; 17: 207Crossref PubMed Scopus (24) Google Scholar Although deviations in epigenetic age as quantified by DNAm in adults are typically associated with morbidity, the direction of the change in newborns in relation to various health outcomes is only beginning to be defined. In this context, we developed our hypotheses that deviation in eGA from chronological age (a variable referred to as "delta gestational age" or dGA) is associated with both prenatal TRAP exposures and early-life allergic sensitization, and that the relationship between prenatal TRAP exposure and sensitization risk is mediated in part by dGA. In an atopy-enriched subset of the Canadian Healthy Infant Longitudinal Development (CHILD) cohort (n = 145, 29% having a positive skin prick test result to at least 1 of 10 standardized allergens at age 1 year), the mean gestational age (GA) was 276 ± 7.9 days or 39.4 weeks (see this article's Methods section in the Online Repository at www.jacionline.org). The mean eGA was 243 ± 8.2 and 279 ± 5.8 days as calculated by the 2 available gestational epigenetic clock algorithms developed by Knight et al5Knight A.K. Craig J.M. Theda C. Bækvad-Hansen M. Bybjerg-Grauholm J. Hansen C.S. et al.An epigenetic clock for gestational age at birth based on blood methylation data.Genome Biol. 2016; 17: 206Crossref PubMed Scopus (52) Google Scholar and Bohlin et al.6Bohlin J. Håberg S.E. Magnus P. Reese S.E. Gjessing H.K. Magnus M.C. et al.Prediction of gestational age based on genome-wide differentially methylated regions.Genome Biol. 2016; 17: 207Crossref PubMed Scopus (24) Google Scholar Both algorithms resulted in eGAs that were statistically associated with chronological GA (Fig 1, A and B). Given the stronger positive correlation using the method of Bohlin et al (r = 0.66; P < .001) compared with that of Knight et al (r = 0.44; P = .001), all further analyses were conducted on the basis of Bohlin et al's eGA. All subsequent analyses did not control for the following features because no statistical differences were observed: child's sex, maternal factors (age, health status, parity, gestational diabetes, ethnicity, and antibiotic use), and child's birth weight (GA- and sex-standardized z score). Mode of delivery, which was strongly associated with dGA (see Table E1 in this article's Online Repository at www.jacionline.org), was included in our analyses. Children born by elective cesarian section had a statistically larger deviation between their eGA and chronological GA (5.74 days; 95% CI, 0.35-8.23). Children of mothers exposed to higher levels of TRAP during the first trimester of pregnancy were at increased risk of allergic sensitization at age 1 year (P = .06; see Table E1). These children also had a larger deviation, meaning their eGA was larger than their chronological age (2.1 days; 95% CI, 0.11-4.08, for each 10 μg/m3 increase in NO2 exposure, which approximates an interquartile increase). Proximity to green space (as quantified from satellite imagery using the normalized difference vegetation index) was not associated with a change in dGA (see Table E1). Building on this observation, we further established that infants with a 1-day deviation in dGA were 8% more likely to be sensitized to allergens at age 1 year (odds ratio, 1.07; 95% CI, 1.00-1.14; P = .05) (Fig 1, C). Finally, when estimating the direct effect of air pollution (ie, adjusting for dGA), an in utero exposure increase of 10 μg/m3 NO2 was associated with a 34% increased risk of allergic sensitization (odds ratio, 1.34; 95% CI, 0.97-1.87), compared with the 42% increase (odds ratio, 1.42; 95% CI, 1.02-1.99) without dGA adjustment (see Table E2 in this article's Online Repository at www.jacionline.org). Our hypothesis that dGA might indicate a role for epigenetics on the causal pathway between prenatal exposures and later development of allergic sensitization was then formally tested using the counterfactual framework-based mediation analysis (details are provided in this article's Online Repository at www.jacionline.org). Based on 1000 bootstrapped samples and after controlling for mode of delivery, the average causal mediation effect was 0.03 (P = .08), whereas the average direct effect was 0.06 (P = .4)—results compatible with mediation for the prespecified α = 0.10 given the limited sample size (Fig 1; see also Fig E1 in this article's Online Repository at www.jacionline.org). Our results suggest that, in our modest sample of 145 children, exposure to TRAP during the first trimester of pregnancy increases the risk of allergic sensitization at age 1 year, and that the impact of exposure to air pollution can be detected through epigenetic patterns present at birth. Specifically, children whose mothers were exposed to higher TRAP during the first trimester had larger dGA. Importantly, a formal statistical test of mediation showed that 31% of the association between TRAP and the risk of allergic sensitization at age 1 year is mediated by modification of DNAm-based eGA (Fig 1, C). Studies have only recently begun to examine the relationship between eGA deviations and exposures or health outcomes.7Suarez A. Lahti J. Czamara D. Lahti-Pulkkinen M. Knight A.K. Girchenko P. et al.The epigenetic clock at birth: associations with maternal antenatal depression and child psychiatric problems.J Am Acad Child Adolesc Psychiatry. 2018; 57: 321-328.e2Abstract Full Text Full Text PDF PubMed Scopus (54) Google Scholar, 8Knight A.K. Smith A.K. Conneely K.N. Dalach P. Loke Y.J. Cheong J.L. et al.Relationship between epigenetic maturity and respiratory morbidity in preterm infants.J Pediatr. 2018; 198: 168-173.e2Abstract Full Text Full Text PDF PubMed Scopus (13) Google Scholar, 9Khouja J.N. Simpkin A.J. O'Keeffe L.M. Wade K.H. Houtepen L.C. Relton C.L. et al.Epigenetic gestational age acceleration: a prospective cohort study investigating associations with familial, sociodemographic and birth characteristics.Clin Epigenet. 2018; 10: 86Crossref PubMed Scopus (24) Google Scholar Although more studies are needed, currently available data indicate that various adverse prenatal environmental exposures (including TRAP, maternal depression, extreme prematurity, and gestational diabetes mellitus) are associated with alterations in eGA. The results of this investigation are based on several strengths. We used novel and comprehensive approaches to determine epigenetic age. In addition, the availability of extensive data from the CHILD cohort enabled the assessment of potential confounders such as maternal age, GA, ethnicity, and child's sex. Finally, the environmental exposures were rigorously assessed at the individual's reported home addresses throughout pregnancy up to the time of birth. Although greenness exposure was averaged across the whole pregnancy, TRAP prediction models at a fine spatiotemporal scale allowed us to report that the first trimester of pregnancy is likely a critical time window when the methylome is particularly vulnerable to environmental insults such as traffic air pollution. In conclusion, we report data that indicate an important trend where prenatal exposure to TRAP is linked to an increased risk of allergic sensitization, which is potentially mediated by altering the gestational epigenetic clock. Our "proof-of-concept" investigation strongly suggests that this relationship deserves reexamination in larger populations to confirm the plausible mechanism of the impact of prenatal TRAP exposure on allergic sensitization via epigenetic modifications. The human ethics board at all universities directly involved with the Canadian Healthy Infant Longitudinal Development (CHILD) cohort (McMaster University, University of British Columbia, University of Toronto, University of Manitoba, and University of Alberta) approved this study. The 145 study participants (n = 145) constitute an atopy-enriched subset of the CHILD cohort.E1Takaro T.K. Scott J.A. Allen R.W. Anand S.S. Becker A.B. Befus A.D. et al.The Canadian Healthy Infant Longitudinal Development (CHILD) birth cohort study: assessment of environmental exposures.J Expo Sci Environ Epidemiol. 2015; 25: 580Crossref PubMed Scopus (44) Google Scholar, E2Subbarao P. Anand S.S. Becker A.B. Befus A.D. Brauer M. Brook J.R. et al.The Canadian Healthy Infant Longitudinal Development (CHILD) study: examining developmental origins of allergy and asthma.Thorax. 2015; 70: 998-1000Crossref PubMed Scopus (120) Google Scholar, E3Moraes T. Lefebvre D. Chooniedass R. Becker A. Brook J. Denburg J. et al.The Canadian Healthy Infant Longitudinal Development birth cohort study: biological samples and biobanking.Paediatr Perinat Epidemiol. 2015; 29: 84-92Crossref PubMed Scopus (38) Google Scholar Allergic sensitization was assessed by standardized epicutaneous skin tests administered to each infant at age 1 year using a panel of 6 inhalant (Alternaria alternata, cat hair, dog epithelium, house-dust mites [Dermatophagoides pteronyssinus and Dermatophagoides farinae], and German cockroach) and 4 food (whole cow's milk, egg white, soybean, and peanut) allergens. Histamine and glycerin were used as positive and negative controls, respectively. The wheal responses were measured at 10 (histamine) and 15 (allergen) minutes, and the maximum diameter and its orthogonal were averaged. A child was diagnosed with allergic sensitization (or atopy) if they developed a wheal greater or equal to 2 mm in diameter for any of the 10 allergens tested. Subjects testing negative to histamine were not included in this cohort unless they tested positive (with a wheal diameter of ≥2 mm) for 1 of the 10 allergens. If a subject tested positive to glycerin, the wheal size for glycerin was subtracted from the wheal size of any positive allergen response. Umbilical cord blood samples were obtained in the delivery room, processed to isolate cord blood mononuclear cells (CBMCs), and frozen as previously described.E3Moraes T. Lefebvre D. Chooniedass R. Becker A. Brook J. Denburg J. et al.The Canadian Healthy Infant Longitudinal Development birth cohort study: biological samples and biobanking.Paediatr Perinat Epidemiol. 2015; 29: 84-92Crossref PubMed Scopus (38) Google Scholar DNAm data were generated on frozen CBMCs using the Illumina HumanMethylation450 array, as previously described.E4Esposito E.A. Jones M.J. Doom J.R. MacIsaac J.L. Gunnar M.R. Kobor M.S. Differential DNA methylation in peripheral blood mononuclear cells in adolescents exposed to significant early but not later childhood adversity.Dev Psychopathol. 2016; 28: 1385-1399Crossref PubMed Scopus (48) Google Scholar Briefly, DNA was extracted from CBMC pellet using the Qiagen DNAeasy kit, per manufacturer's instructions. The Zymo EZ DNAm gold kit was used to bisulfite convert 750 ng of this DNA, and 160 ng of bisulfite-converted genomic DNA was applied to the 450k arrays as recommended by the manufacturer. Next, we used the Illumina GenomeStudio software to perform background subtraction and color correction. Two published methods were used for deriving eGA. Bohlin eGA calculation required unfiltered but normalized data, and so we BMIQ normalized the unfiltered data,E5Teschendorff A.E. Marabita F. Lechner M. Bartlett T. Tegner J. Gomez-Cabrero D. et al.A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data.Bioinformatics. 2013; 29: 189-196Crossref PubMed Scopus (911) Google Scholar then applied the eGA predictor using the 96 sites as previously reported.E6Bohlin J. Håberg S.E. Magnus P. Reese S.E. Gjessing H.K. Magnus M.C. et al.Prediction of gestational age based on genome-wide differentially methylated regions.Genome Biol. 2016; 17: 207Crossref PubMed Scopus (77) Google Scholar Knight eGA was calculated according to previous reports, using filtered but unnormalized data.E7Knight A.K. Craig J.M. Theda C. Bækvad-Hansen M. Bybjerg-Grauholm J. Hansen C.S. et al.An epigenetic clock for gestational age at birth based on blood methylation data.Genome Biol. 2016; 17: 206Crossref PubMed Scopus (129) Google Scholar To filter, we removed X and Y chromosome probes, probes where the CpG is a single nucleotide polymorphism or the probe binds in silico to the X or Y chromosome.E8Price E.M. Cotton A.M. Lam L.L. Farré P. Emberly E. Brown C.J. et al.Additional annotation enhances potential for biologically-relevant analysis of the Illumina Infinium HumanMethylation450 BeadChip array.Epigenet Chrom. 2013; 6: 4Crossref PubMed Scopus (338) Google Scholar Next, values where fewer than 3 beads contributed to signal or with a detection P value of less than .01 were set to not available (NA), and any probe with more than 3 NA values was removed. Eight of the required 149 CpGs had been filtered out, which were added back to the data with all values set as NA. Individual exposure estimates to greenness and TRAP were assessed at the home address(es) throughout pregnancy. Green space exposure was quantified from satellite imagery using the normalized difference vegetation index (NDVI), which is a metric of overall greenness, as previously described.E9Cusack L. Sbihi H. Larkin A. Chow A. Brook J.R. Moraes T. et al.Residential green space and pathways to term birth weight in the Canadian Healthy Infant Longitudinal Development (CHILD) study.Int J Health Geogr. 2018; 17: 43Crossref PubMed Scopus (25) Google Scholar NDVI was based on satellite imagery (Landsat 5) of land surface reflectance of visible and near-infrared parts of spectrum, with values ranging from −1 to 1, with the higher numbers indicating more greenness. These values were computed at mothers' residential addresses during the 9 months of pregnancy. A time-weighted average from all addresses during pregnancy (11.5% of participants moved) was calculated for different buffer distances (ie, 100, 250, and 500 m) around addresses to capture various representations of green space exposures. Air pollution levels were estimated using highly spatially resolved surfaces of nitrogen dioxide (NO2). NO2 is of primary interest because this pollutant demonstrates fine-scale spatial variability. City-specific land use regression models to estimate NO2 concentrations were used to assign exposures at the residential locations of all participants. The methodology to adjust both temporally and spatially all geo-referenced residences reported by participants is described elsewhere.E10Sbihi H. Allen R.W. Becker A. Brook J.R. Mandhane P. Scott J.A. et al.Perinatal exposure to traffic-related air pollution and atopy at 1 year of age in a multi-center Canadian birth cohort study.Environ Health Perspect. 2015; 123: 902-908Crossref PubMed Scopus (44) Google Scholar Briefly, exposure to land use regression–derived NO2 was estimated using a time-weighted average exposure based on the time each participant resided in each reported address and temporally adjusted on the local fixed-site ambient monitoring data on a biweekly basis. Our exposure assessment approach enabled the computation of trimester-specific individual estimates of TRAP, whereas the greenness measure was derived for the entire pregnancy, which could potentially dilute the signal for an association (compared with that for TRAP), should one exist because NDVI varies by season. For both Bohlin and Knight methods, deviations in eGA from chronological age (referred to as dGA) were calculated by taking the difference between eGA in days and chronological GA in days. To estimate the total, direct, and indirect effects of environmental exposures on sensitization through dGA, we first examined the risk of having allergic sensitization associated with prenatal exposures to TRAP and greenness. Next, we tested whether these environmental factors affected dGA. Finally, dGA was tested against allergic sensitization. Effect estimates are presented for a 10 μg/m3 increase in NO2 and for 0.1 increase in NDVI, which approximate the interquartile range for each of these exposures. With 145 participants, the statistical testing of differences in sensitization status had 80% power to detect differences with 90% confidence.E11Faul F. Erdfelder E. Buchner A. Lang A.G. Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses.Behav Res Methods. 2009; 41: 1149-1160Crossref PubMed Scopus (14646) Google Scholar Before investigating the association between TRAP, dGA, and atopy, we assessed bivariate associations between potential confounders and each of dGA and atopy. In light of the sequential pathway between exposures, cord blood DNAm, and 1-year outcome, the selection of covariates pertained to the prenatal period up to the time of birth and included maternal age and gestational diabetes as well as the mode of delivery, which was defined as vaginal birth or cesarian section with and without labor, respectively. In addition, maternal health, in particular atopic status, has been shown to be predictive of offspring atopy; thus, maternal self-reported asthma and atopy (via skin prick test) as well as antibiotic use during pregnancy were also examined. In line with putative risk factors for allergic disorders shown in the literature, we also considered parity and education, the latter serving as a surrogate for socioeconomic status. DNAm is highly associated with genotype, so in the absence of genotyping data, we used self-reported ethnicity. We tested the associations of dGA against TRAP and greenness independently on the risk of sensitization to allergens in this population by examining the effect of (1) each exposure on atopic outcomes with and without adjustment for dGA, (2) the dGA on the risk of atopy, and (3) each exposure on the dGA (Table E2). Counterfactual-based causal mediation analysisE12Imai K. Keele L. Tingley D. A general approach to causal mediation analysis.Psychol Methods. 2010; 15: 309-334Crossref PubMed Scopus (1697) Google Scholar was used to dissect the total effect of a treatment (here TRAP exposure, a continuous variable) into direct and indirect effects, where the latter is transmitted via a mediator (here dGA, a continuous variable) to the outcome (atopy, a binary variable). To complete this analysis, the mediation package,E13Tingley D. Yamamoto T. Hirose K. Keele L. Imai K. mediation: R Package for Causal Mediation Analysis.J Stat Software. 2014; 59: 38Crossref Scopus (1472) Google Scholar designed to perform causal mediation analysis under the assumption of sequential ignorability, was used. Nonparametric bootstrapping was used for variance estimation with 1000 resamples.Table E1Selected characteristics of the study population by atopic status∗Categorical variables: n and % with χ2 P value; continuous variables: mean and SD with t test P value. and by dGA†Categorical variable: mean and SD, P value of t test and ANOVA for dichotomous and multiple categories, respectively; continuous variable: P value of simple linear regression; arrow indicates direction of effect. Statistically significant associations are shown in boldface.CharacteristicNonatopic (n = 102)Atopic (n = 42)dGABohlin dGA (d)2.26 ± 5.334.58 ± 7.16P = .06Study center (n, %)P = .33P = .09 Vancouver (21, 14%)67%33%5.31 ± 8.3 Edmonton (38, 26%)66%34%3.23 ± 5.2 Winnipeg (43, 30%)81%19%1.31 ± 5.6 Toronto (43, 30%)67%33%3.04 ± 5.3Sex (n, %)P = .19P = .27 Male (84, 58%)67%33%3.4 ± 5.3 Female (60, 42%)77%23%2.2 ± 6.8Mother ethnicity (n, %)P = .11P = .83 White (n = 81)75%25%2.35 ± 5.3 Asian (n = 11)45%54%3.09 ± 4.1 Other (n = 18)56%44%2.54 ± 6.35Mother educationP = .61P = .9 No university degree (63, 44%)73%27%3.0 ± 6.5 University degree (80, 56%)68%32%2.9 ± 5.5Mother atopyP = .77P= .17E12Imai K. Keele L. Tingley D. A general approach to causal mediation analysis.Psychol Methods. 2010; 15: 309-334Crossref PubMed Scopus (1697) Google Scholar Nonsensitized (51, 35%)73%27%2.0 ± 5.1 Sensitized (94, 65%)70%30%3.4 ± 6.4Mother asthmaP = .22P = .44 No (99, 69%)74%26%2.72 ± 6.2 Yes (44, 31%)64%36%3.53 ± 5.6Mother antibiotic useP = .7P = .56 No (75, 82%)71%29%3.22 ± 6.6 Yes (16, 18%)75%25%4.25 ± 6.2Birth weight (Z score)0.18 ± 1.000.27 ± 0.95P = .30P = .6 ↓Maternal age (y)31.97 ± 4.332.55 ± 4.5P = .24P = .7 ↑Mode of delivery (n, %)30.8%P = .31P = .02 Vaginal (n=107)69.2%10%2.31 ± 6.1 C-section with labor (n = 10)90%28.6%1.79 ± 6.3 C-section without labor (n = 28)71.4%5.74 ± 5.0Greenness0.36 ± 0.160.32 (0.19)P = .25P = .15 ↑NO2 first trimester (μg/m3)20.3 ± 12.925.4 ± 12.6P = .06P = .04 ↑NO2 second trimester (μg/m3)23.3 ± 13.724.6 ± 13.6P = .56P = .20 ↑NO2 third trimester (μg/m3)22.6 ± 14.223.7 ± 13.0P = .68P = .30 ↑∗ Categorical variables: n and % with χ2 P value; continuous variables: mean and SD with t test P value.† Categorical variable: mean and SD, P value of t test and ANOVA for dichotomous and multiple categories, respectively; continuous variable: P value of simple linear regression; arrow indicates direction of effect. Statistically significant associations are shown in boldface. Open table in a new tab Table E2Effect estimates of TRAP and greenness on both allergic sensitization at age 1 year (atopy) (adjusted and nonadjusted for dGA) and deviation of cord blood DNAm from GA or dGAObserved outcomes∗All models are adjusted for mode of delivery.Atopy, OR (95% CI)P valueDGA, b (95% CI)P valueNO2 (first trimester of pregnancy, 10 μg/m3 increase) Unadjusted1.42 (1.02 to 1.99).0372.1 (0.11 to 4.08).038 Adjusted1.34 (0.97 to 1.87).08Greenness in pregnancy Unadjusted2.42 (0.05 to 110.5).651.22 (−0.32 to 2.76).12 Adjusted1.47 (0.05 to 59.4).92OR, Odds ratio.∗ All models are adjusted for mode of delivery. Open table in a new tab OR, Odds ratio.