Introduction

Hypertension (HTN), an important risk factor for cardiovascular disease, affects ~30% of adults 18 years of age and older in the United States.1 The prevalence of HTN in children, in contrast, is 3.5%, although it ranges from 3.8 to 24.8% in overweight or obese youth populations2 (reviewed in3). Its prevalence is thus rising as child obesity rates increase.2,4 Of concern, later life HTN is strongly correlated with early life blood pressure (BP) elevation and impaired renal development.5,6

The etiology of HTN is complex, and its programming is considered to reflect a complex interplay between genetic predisposition and environment, mediated in part by epigenetic factors. Although it is generally accepted that overweight/obesity,7 a sedentary lifestyle,8,9 and salt intake10 contribute to HTN, environmental chemical exposures to heavy metals, phthalates, and arsenic are also established risk factors for development of HTN in adult populations, estimated to account for 3–19% of the population attributable risk for high BP.11 Despite the predictive value of childhood BP on HTN later in life, there exist a limited number of studies that examine the effect of environmental chemical exposures on childhood BP.

The developmental origins of health and disease (DOHaD) hypothesis posits that an adverse intrauterine environment programs chronic disease including elevated BP and kidney disease later in life.12 Metanephric kidney development typically begins at 5 weeks’ gestation with the full complement of ~1 million nephrons per kidney largely achieved by ~35 weeks gestation.13,14 Renal blood flow and glomerular filtration, which increase significantly during fetal life, continue to mature after birth, reaching adult levels by 2 years of age.15 Overall, these prenatal and postnatal periods are likely to be critical windows that are susceptible to nephrotoxic environmental chemical exposures predisposing the developing kidney for later life HTN (Fig. 1).16,17 Although evidence supports this relationship for in utero tobacco exposure and child BP,18,19 the effects of other highly prevalent environmental nephrotoxicants on renal development and BP are poorly understood. We previously reviewed non-BP effects of environmental chemicals on pediatric kidney function and disease.20

Fig. 1
figure 1

The timing of exposures to environmental chemicals during periods of renal development from embryogenesis to birth and through childhood can inform critical windows of kidney susceptibility. Printed with permission of ©Mount Sinai Health System

Full size image

To summarize the state of the literature on the role of environmental chemicals as predictors of childhood BP, we conducted a review of human studies examining exposures during prenatal and postnatal susceptibility windows. We focused on exposure to air pollution, and included a wider range of publication dates and pollutants than previously reported.21,22 We also included metals and organic contaminants as pervasive and/or emerging contaminants, which have not been reviewed with respect to childhood BP. We present conclusions based on this body of evidence and provide recommendations for future studies.

Methods

We conducted a literature search to identify human studies that examined the effects of air pollutants [particulate matter (PM10 and PM2.5), nitric oxides (NOx and NO2), ozone (O3), sulfur dioxide (SO2), carbon monoxide (CO), black carbon (BC), and polycyclic aromatic hydrocarbons (PAHs)], nephrotoxic metals [arsenic (As), cadmium (Cd), lead (Pb), or mercury (Hg)], phthalates and other organic contaminants with respect to children’s BP using the following search terms (child* OR infan* OR school OR postnatal OR post-natal OR prenatal OR pre-natal OR fetal OR pregnan* OR in utero) AND (blood pressure OR systolic OR diastolic OR pulse pressure) AND (metal OR “cadmium” [MeSH Terms] OR “arsenic” [MeSH Terms] OR “lead” [MeSH Terms] OR “mercury” [MeSH Terms] OR “BPA” [MeSH Terms] OR “Bisphenol A” [MeSH Terms] OR perfluoroalkyl OR pesticides OR PAH OR polycyclic aromatic hydrocarbon OR phthalate). Among additional contaminants included in the initial search terms including cannabis/marijuana, brominated flame retardants, parabens, dioxins, or furans, we identified no published studies of children’s BP. We used PubMed and Web of Science search engines and restricted the search to peer-reviewed human studies published in English between January 2007 and July 2017. We summarized a total of 14, 10, and 12 studies on air pollution, metals, or phthalates and other organic contaminants, respectively. We note that this manuscript is not a systematic review, which was not possible due to the paucity of peer-reviewed studies for some exposures as well as the heterogeneity of reported exposure and outcome measures.

This review uses updated terminology for pediatric BP as defined in the 2017 report by the American Academy of Pediatrics (AAP) Subcommittee On Screening Management Of High Blood Pressure In Children and Adolescents2 wherein normal BP, elevated BP (formerly called “preHTN”), Stage 1 HTN and Stage 2 HTN are defined and we refer the reader to the AAP report for details.2 The term “high BP” may refer to both HTN and “elevated BP” together. To further clarify BP nomenclature, we use the term “increased” BP to describe associations with continuous BP (formerly termed “elevated” in some studies). In cases where the former terminology was used in reports herein, we have indicated the authors’ intended definition of “HTN”, “preHTN”, and “elevated BP” and corrected linear regression interpretations to “increased BP” where appropriate.

Epidemiologic evidence of association between air pollution and childhood BP

We identified 14 studies examining the effects of air pollutants on child BP (Table 1). Exposures assessed included PM10 (n = 10), PM2.5 (n = 8), NOx or NO2 (n = 8), O3 (n = 5), SO2 (n = 3), CO (n = 2), BC (n = 2), PAHs (n = 1), distance to roadway/traffic/point source (n = 3), nano-sized ultrafine particles (n = 1); eleven studies assessed more than one pollutant. Outdoor air quality was assessed via ambient monitoring, estimates derived with geographic information system (GIS) models (distance to traffic, land-use regression), or satellite/dispersion or other national database estimates. Two studies assessed indoor air quality. Three of the study designs were prospective and 12 were cross-sectional. Among 10 studies of PM10 and 8 studies of PM2.5, four and two reported significant associations with childhood BP, respectively. Five of seven studies of NO2, four of five studies of O3, two of three studies of SO2, one of two studies of CO, and one of two studies of BC reported associations with childhood BP; one study reported evidence of a relationship with nano-sized ultrafine particles (UFP) (particles with a diameter ≤100 nm). Two studies examined indoor air pollution and reported no association with children’s BP.23,24 A single study using school location as proxy for PAH exposure reported an association with childhood BP, but not with urine PAH biomarkers.25 A study using distance to major road as a proxy for ambient PM10 reported higher mean carotid arterial stiffness (derived from BP and ultrasound parameters) among children living closer to the major road.26 A study of road traffic at schools near an airport found no association with child BP.27

Table 1 Studies assessing exposure to air pollutants and blood pressure outcomes (n = 14)
Full size table

Among the three prospective cohort studies, all identified significant associations with exposure to air pollution (PM2.5 and other pollutants) and child BP. In a prospective study of 1131 infants born in the United States, higher mean PM2.5 and BC in the third trimester (90 days before birth) were associated with higher SBP in newborns ~30 h after birth.28 Ozone (O3) was associated with lower SBP, and no significant associations were observed for NOx, NO2, or CO averages during the third trimester. The observed significant third trimester findings with PM2.5, BC, and O3 were bolstered by estimates of “moving averages” calculated from 2 to 7, 14, 30, 60, and 90 days before birth. No significant relationships were observed with first trimester pollutant averages; however, putative relationships in the second trimester were observed for NOx and CO with lower SBP. The study did not report associations with DBP. Two studies conducted in the PIAMA birth cohort in the Netherlands29,30 reported associations between PM2.5 and increased DBP at 12 years. In a subset of 471 children who did not move since birth, long-term NO2 and PM2.5 (reported as annual averages at participants’ home and school addresses estimated by land use regression) were associated with increased DBP.30 No associations were observed with short-term (7-day average preceding BP measure) exposure. In a concurrent study, the composition of PM2.5 and PM10 (estimated as annual averages at participants’ home addresses), reported that increased DBP was associated with levels of iron, silicon, and potassium in PM10, and showed marginal associations (p < 0.1) with iron and silicon in PM2.5.29 The authors attributed part of the observed effect to NO2 (a marker of traffic exhaust emissions), and noted that the possible association with iron suggested non-exhaust emissions were present.

Among the 10 cross-sectional studies, 8 reported significant relationships between exposure to air pollution and children’s BP. Three of these studies were based in a large cross-sectional study of 9354 children ages 5–17 years old in China, the Seven Northeastern Cities Chinese Children’s Study (SNECCS) cohort.31,32,33 The studies assessed the effects of long-term32,33 or short-term exposure.31 Four-year average concentrations of PM10, SO2, NO2, O3, and CO were calculated from monitoring stations. One analysis reported the increased odds of childhood HTN with O3 and PM10. Increased DBP was associated with O3, PM10, and CO; increased SBP was associated with O3 and PM10.32 These relationships were attenuated when stratified by children who were breastfed versus those who were not. Another analysis conducted in the SNECCS cohort reported the interaction effects of obesity and air pollutant exposure on BP outcomes, and found significant relationships between NO2, O3, and PM10 with the odds of HTN that were consistently larger effect sizes among overweight/obese children compared to normal weight children.33 The findings reported evidence of an interactive effect with child weight and exposure to PM10, SO2, and O3 for both SBP and DBP. A third study assessed short-term effects using 5-day mean exposures to PM10, SO2, NO2, and O3, and reported increased odds of elevated BP (defined as either SBP or DBP >95th percentile) for both O3 and PM10.31 SBP and DBP were positively associated with all four pollutants at various lag times.31 When stratified by sex the relationship between elevated BP and NO2 was observed only in males. Another large cross-sectional study of 2368 German children at 10 years of age exposed to air and noise pollution due to traffic34,35 found no association between NO2, PM2.5, or PM10 and BP overall, however, when restricted to the 605 participants with noise information a significant relationship with NO2 and increased DBP was observed. 34

In a secondary analysis of a subsample of 276 children age 9 and 10 years in the UK attending school near a major airport, no association was observed between NO2 levels and child BP with or without adjustment for noise,27 and the study replicated previous findings that aircraft and road traffic noise were not associated with child BP.36 In a cross-sectional study of 184 adolescent males ages 10–14 years in Saudi Arabia, increased SBP and DBP and odds of “preHTN” were associated with attendance at schools located closest to an oil refinery, compared to a school location further from the refinery.25 Concentrations of PM10 and PAH metabolites were measured including total hydroxyphenanthrenes and 1-hydroxypyrene, however no relationships were identified. In a longitudinal study of 130 children in Belgium ages 6–12 years, increased total ultrafine particle fraction as well as fraction of nano-sized (20–30 nm) ultrafine particles were associated with increased SBP. No effects were identified for nano-sized fractions with diameter >100 nm (PM0.1–2.5), nor PM2.5, PMcourse, or PM10.37 In a cross-sectional study of 81 Mexican children 6–13 years of age, 7-day average PM2.5 and O3 were measured/estimated based on residence in Polotitlan (n = 22, control), Northeast (n = 19) or Southwest Mexico City (n = 40). Doppler echocardiography was used to measure systolic pulmonary arterial pressure (PAP) and diastolic PAP, and calculate mean pulmonary arterial pressure (MPAP).38 MPAP was increased in NE and SW Mexico City locations compared to Polotitlan, and the 7-day mean PM2.5 was associated with MPAP. The assessment of MPAP, rather than systemic BP, provides insight into potential pulmonary effects of PM2.5, but may be less informative for developmental renal effects.

In addition to two studies discussed above,25,27 three of the cross-sectional studies of indoor air quality or pollution assessed by proxy measures (e.g., distance, location) reported no association between exposure to air pollution and children’s BP. A cross-sectional study of 240 Chinese study children 10 years of age, identified no association between 24-h indoor PM2.5 or BC exposure,24 using previously reported assessment techniques.39 A cross-sectional study of 179 Pakistani children 8–12 years of age measured indoor and outdoor PM10 and PM2.5 at schools located in geographic regions with low (n = 79) and high (n = 93) air pollution.23 Although direct associations were not reported for indoor or outdoor PM measures, the authors observed higher SBP and DBP in areas with high pollution compared to controls (both indoor and outdoor levels were increased in “high” pollution area). Finally, a cross-sectional analysis of 52 Italian children, which assessed distance to major road as a proxy for PM2.5 exposure, reported no significant association with MAP, or BP, however the authors could not adjust for covariates.26

Epidemiologic evidence of association between metals and childhood BP

We identified 10 studies that examined the effects of selected nephrotoxic metals comprised of As (n = 2), Pb (n = 2) Cd (n = 4), and Hg (n = 3) on child BP (Table 2). One study examined both Cd and As and one study examined Cd and Pb.

Table 2 Studies assessing exposure to nephrotoxic metals and children’s blood pressure outcomes (n = 10)
Full size table

Of the two studies examining early life As exposure, one identified a significant relationship with childhood BP, for both prenatal and postnatal As exposure.40 In a prospective cohort study of 1887 mother–baby pairs in Bangladesh, an increase in pregnancy urine-As (U-As) (average measure at weeks 8 and 30 of gestation) was associated with increased SBP and DBP at 4.5 years of age. Additionally, children’s U-As at 18 months was associated with an increase in SBP at 4.5 years. In a subsequent cross-sectional study of 1356 children aged 4.5 years from the same population, no association was observed with U-As levels (reported as the sum of inorganic, methylarsonic acid, and dimethylarsonic metabolites) measured approximately 8 months following BP assessment.41 Although U-As typically reflects ongoing exposure, the study aimed to capture spot urine measures as a proxy for consistent As exposure through drinking water, as well as chronic Cd exposure (discussed below).

Of the two studies examining early life Pb exposure, both were conducted in prospective birth cohorts; however, only one identified a significant relationship with childhood BP.42 In a prospective birth cohort of 457 mother–child pairs in Mexico City, an increase in maternal tibia Pb was associated with increased SBP and DBP in girls, but not boys. No associations were observed with patella Pb or cord blood Pb levels. In a prospective cohort of 1511 dyads in Bangladesh, no association was observed with SBP or DBP at 4.5 years with maternal blood Pb measured at 14 and 30 weeks gestation.43

Among the four studies examining early life Cd exposure, none reported a significant relationship with childhood BP.41,44,45,46 Among these studies, two measured urine Cd (U-Cd), in either maternal41 or children’s samples,46 and two measured blood Cd in children’s samples.44,45 Two studies were prospective (including a secondary analysis of a randomized control trial), and two were cross-sectional (including a case–control design measured at two time points). The prospective study of 1356 Bangladeshi children reported no association with maternal U-Cd levels in pregnancy.41 In a prospective cohort (secondary analysis of randomized control trial data) of 223 control and 218 US children who received succimer treatment, blood Cd measured at 2 years of age was not associated with BP at 2, 5, or 7 years.44 A decrease in SBP was observed 1-week following succimer treatment, although the authors note the fall in BP may be attributable to reduction in Pb burden and there was no clear association with Cd and BP. In a cross-sectional study of 594 Thai school-age children randomly selected from 10 Cd-contaminated schools and 3 outside the contamination area, no significant association was found for U-Cd when analyzed as a continuous variable or categorized by tertile with SBP or DBP.46 Blood Pb was measured in this study but the relationship with BP was not reported. Lastly, in a case–control study of 320 children from Iran (160 normal weight, 160 cases with metabolic syndrome), blood Cd measured at 10 and 18 years showed a positive, but not significant association with DBP. The analysis did not include adjustment for covariates, and did not report the method of BP measurement. Notably, the Cd levels were relatively high among the control group (mean was 10.09 μg/L for cases and 9.97 μg/L for controls), and the analysis included linear regression across all cases and controls and reported null associations for the relationship between blood Cd and DBP or SBP.

We identified three studies examining early life Hg exposure, and only one reported an association with childhood BP. All three studies were prospective cohorts. A prospective cohort study of 645 and 561 children in the Seychelles that assessed maternal hair Hg and followed children to 12 and 15 years, respectively, reported that prenatal maternal Hg levels were associated with increased DBP among 15-year-old boys but not girls.47 No associations were observed among 12-year-old children. A large prospective cohort study of 1103 mother–child pairs in the United States, found no association with maternal blood erythrocyte Hg measured during the second trimester and SBP at 3.2 or 7.7 years.48 Another prospective cohort study of 226 Nunavik Inuit children, measured cord blood Hg, as well as blood and hair Hg at 11 years of age, and reported no association with Hg exposure at any time point and BP at 11 years.49 Notably, the studies above reported lower Hg levels than a 1999 study (outside the range of years reviewed herein) conducted of 917 children in the Faroe Islands.50 The study identified a significant relationship between cord blood Hg and SBP at 7 years, and we note that Hg levels were 2–3× higher in cord or maternal hair levels than those reported in refs. 49,47 which may account for the lack of statistically significant findings in the later studies. Additionally, the selection of appropriate biomarker matrices for nephrotoxicity may have affected the results (i.e., hair vs. urine).

Epidemiologic evidence of association between phthalates and other organic compounds and childhood BP

We identified 12 studies that examined the effects of other potential environmental nephrotoxicants on child BP, comprised of phthalates (n = 4), Bisphenol A (PBA) (n = 3), polychlorinated biphenols (PCBs) (n = 2), pesticides (n = 4, two also measured PCBs) and flame retardants (n = 1) (Table 3). The studies were either cross-sectional (n = 6) or prospective (n = 6) in design. Studies of phthalates, pesticides, and BPA used urinary levels of the parent compound or metabolites, whereas PCBs and flame retardants were measured in blood.

Table 3 Studies assessing exposure to phthalates (n = 4), bisphenol A (n = 3), polycholinated byphenols (n = 2), pesticides (n = 3, one also measured PCBs), and flame retardants/perfluoroalkyl chemicals (n = 1) and children’s blood pressure outcomes (n = 12)
Full size table

Of the four studies examining early life phthalate exposure, all reported a significant relationship with subsequent childhood BP, for both prenatal and postnatal exposure. In a prospective study of 379 mother–child pairs in Spain, maternal phthalate exposure (averaged from first and third trimester urine measures, grouped by high or low molecular weight), showed significantly lower SBP z-score among girls comparing the 2nd and 3rd tertiles to the first tertile of summed molar concentrations of high molecular weight phthalates (HMWP) and comparing only the 3rd tertile to first for summed molar concentrations of low molecular weight phthalates (LMWP).51 In another prospective cohort study of 258 children in the United States, prenatal levels of mono(2-ethyl-5-oxohexyl) phthalate (MEOHP) and mono(2-ethyl-5- hydroxyhexyl) phthalate (MEHHP) measured in maternal urine (at 12 and 20 weeks gestation) were associated with DBP at 14 years of age, but not at 5 or 9 years of age.52 This study did not directly report estimates of the association between phthalates and SBP or DBP. Associations were also noted in two large cross-sectional studies in the United States.53,54 Among 2447 adolescents participating in NHANES 2003–2008, grouped urine phthalates and metabolite molar sums (HMWP, LMWP, and diethylhexyl phthalate [DEHP]) were assessed in relation to SBP z-score and “preHTN”.54 An increase in DEHP metabolites was associated with a significantly higher SBP z-score. Individual metabolites including MEHP, MBP, MEHHP, and MEOHP were associated with increased SBP z-score and MEP was associated with increased odds of preHTN. No association was observed with the LMWP or HMWP groups. In a subsequent study of 1329 adolescents participating in NHANES 2009–2012, molar sums of HMWP and urinary DEHP metabolites as well as the parent compound for its replacements, di-isononyl phthalate (DINP) and di-isodecyl phthalate (DIDP), were each associated with increased SBP z-score.53 No associations were observed with the LMWP group. An increase in summed DEHP metabolites was also associated with increased DBP z-score. This study further identified the significant metabolites within each group and we refer the reader to Table 3 and the original article for details.

Among three studies examining early life BPA exposure, two were conducted in prospective cohorts and one was a cross-sectional study. Two reported significant associations with childhood BP. In a prospective cohort of 486 mother–child pairs in Korea, higher prenatal BPA exposure (measured at 20 weeks gestation) was associated with a significant increase in DBP at levels above an identified threshold of 4.5 μg BPA per g creatinine.55 Additionally, for prenatal BPA levels above 5.0 μg/g creatinine, a relationship with decreased pulse pressure was observed. Among girls, a U-shaped relationship with SBP was identified above and below the level of 0.7 μg BPA/g creatinine. However, in a prospective cohort study of 500 mother–child pairs in Greece, urine BPA (measured during the first trimester, as well as at 2.5 and 4 years) was not associated with child BP at 4 years.56 Lastly, a limited cross-sectional study of 39 obese or overweight children ages 3–8 years in the United States, urinary BPA adjusted for creatinine was associated with elevated DBP among males, but not girls.57 No associations were observed with SBP.

Among two studies examining early life PCB exposure (a persistent organic pollutant banned in the 1970s), one reported evidence of association with BP. In a prospective cohort study of 427 Greek children, the persistent organic pollutant hexachlorobenzene (HCB) was associated with higher SBP, and p,p′-DDE, a metabolite of DDT, was associated with higher DBP at 4 years of age.58 No association was observed for a group of summed PCBs comprised of six congeners. In a cross-sectional study of 158 Korean children aged 7–9 years at baseline, the sum of 32 measured PCBs and 2 individual PCBs (congeners 138 and 153) were associated with significantly higher DBP at 1-year follow-up, after adjustment for sex, age, household income, and BMI.59 No associations were identified with SBP or DBP and 19 organochlorines measured.

Among four studies examining early life organochlorine compound exposure, three reported significant associations with BP. The prospective cohort study by Lee et al. (described above) reported no association with organocholorine pesticides,59 whereas the study by Vefaidi et al. (described above),58 and two cross-sectional studies from the ESPINA cohort in Ecuador reported associations.60,61 In a cross-sectional examination of 87 children, prenatal pesticide exposure assessed by parental occupation interview were marginally associated with higher SBP at 6–8 years among a subset of 6960; no relationships were observed with urine organophosphate metabolites. In a follow-up study of 271 children, duration of living with a flower worker as a proxy for exposure as well as number of “bad practices” likely to result in home contamination were significantly associated with lower SBP.61

Finally, we identified a single study examining early life exposure to perfluoroalkyl flame retardants with childhood BP. A cross-sectional analysis of 1655 children participating in NHANES in years 1999–2000 and 2003–2008, reported no association with serum PFOA or PFOS and continuous BP measure or HTN.62

Discussion

Summary and recommendations for future directions

There is a growing body of literature supporting adverse cardiorenal effects of air pollution exposure. For metals, the literature is sparse in children, but suggests that As, Hg, and Pb may affect childhood BP. Among the persistent organic pollutants there is limited but consistent evidence that phthalates and pesticide exposure may alter childhood BP. There is mixed evidence for BPA and PCB effects, and there is no current evidence to support an association between childhood PFC exposure and BP. There remain many gaps in our knowledge, including the importance of developmental susceptibility windows (i.e., the role of exposure timing in predicting effects), appropriate study design such as prospective studies that can avoid the issue of reverse causality (in which renal damage alters biomarker levels—a potential issue in cross-sectional studies), more standardized measurement and definition of BP outcomes, need for more complex statistical modeling approaches that factor in developmental trajectories or mixed exposure, and consideration of genetic susceptibility and epigenetic mediation as a way to define susceptible subpopulations and mechanisms of action. Future studies should also consider possible effects of emerging contaminants given that over 100,000 chemicals are registered for industrial use in the United States yet only a handful have been studied with regards to renal toxicity. Additional attention is also needed to examine potential maternal BP effects from exposures in pregnancy. Incorporating these considerations into new studies will enable the research community to address existing limitations and enhance our understanding of early life environmental exposures on BP. Given the heterogeneity of reported exposures and outcomes in the peer-reviewed studies summarized herein, we also recommend that exposure-specific systematic reviews and meta-analyses are needed in the advancing field of developmental exposures and childhood BP.

Conclusions

This review of the recent literature suggests that early life exposure to air pollutants, metals, and organic pollutants may contribute to changes in childhood BP. Limited evidence suggests that prenatal exposure may be more deleterious to BP than postnatal exposures and requires additional investigation. Because childhood BP tracks with adult BP and HTN63,64, it is imperative that hypertensive children are identified early for intervention and potential environmental, behavioral, and biological risk factors.