Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Mar 1.
Published in final edited form as: Environ Res. 2021 Sep 16;204(Pt B):112062. doi: 10.1016/j.envres.2021.112062

Critical windows of perinatal particulate matter (PM2.5) exposure and preadolescent kidney function

Maria José Rosa 1,†, Maria D Politis 1,†, Marcela Tamayo-Ortiz 2, Elena Colicino 1, Ivan Pantic 3, Guadalupe Estrada-Gutierrez 3, Mari Cruz Tolentino 3, Aurora Espejel-Nuñez 3, Maritsa Solano-Gonzalez 4, Itai Kloog 5, Nadya Rivera Rivera 1, Andrea A Baccarelli 6, Martha M Tellez-Rojo 4, Robert O Wright 1,7, Allan C Just 1,*, Alison P Sanders 7,8,*
PMCID: PMC8678189  NIHMSID: NIHMS1743060  PMID: 34537199

Abstract

Air pollution exposure, especially particulate matter ≤ 2.5 μm in diameter (PM2.5), is associated with poorer kidney function in adults and children. Perinatal exposure may occur during susceptible periods of nephron development. We used distributed lag nonlinear models (DLNMs) to examine time-varying associations between early life daily PM2.5 exposure (periconceptional through age 8 years) and kidney parameters in preadolescent children aged 8 to 10 years. Participants included 427 mother-child dyads enrolled in the PROGRESS birth cohort study based in Mexico City. Daily PM2.5 exposure was estimated at each participant’s residence using a validated satellite-based spatio-temporal model. Kidney function parameters included estimated glomerular filtration rate (eGFR), serum cystatin C, and blood urea nitrogen (BUN). Models were adjusted for child’s age, sex and body mass index (BMI) z-score, as well as maternal education, indoor smoking report, seasonality and average first year of life PM2.5 exposure. We also tested for sex-specific effects. Average perinatal PM2.5 was 22.7 μg/m3 and ranged 16.4–29.3 μg/m3. Early pregnancy PM2.5 exposures were associated with higher eGFR in preadolescence. Specifically, we found that PM2.5 exposure between weeks 1 to 18 of gestation was associated with increased preadolescent eGFR, whereas exposure in the first 14 months of life after birth were associated with decreased eGFR. Specifically, a 5 μg/m3 increase in PM2.5 during the detected prenatal window was associated with a cumulative increase in eGFR of 4.44 mL/min/1.732 (95%CI: 1.37, 7.52), and during the postnatal window we report a cumulative eGFR decrease of −10.36 mL/min/1.732 (95%CI: −17.68, −3.04). We identified perinatal windows of susceptibility to PM2.5 exposure with preadolescent kidney function parameters. Follow-up investigating PM2.5 exposure with peripubertal kidney function trajectories and risk of kidney disease in adulthood will be critical.

Keywords: particulate matter, kidney function, distributed lag models, prenatal exposure, Bayesian distributed lag interaction models

Introduction

Chronic kidney disease (CKD) affects 10%–15% of the population worldwide, with rising prevalence due to global epidemics of major risk factors for CKD including hypertension, obesity, and diabetes (Hill et al., 2016). However, the increasing prevalence of CKD cannot be fully attributed to these risk factors, and exposure to environmental pollutants is increasingly recognized as a potential contributor to the rising prevalence (Xu et al., 2018). Moreover, rates of CKD and hypertension among children and young adults are increasing (Collaboration, 2020; Kim et al., 2006) furthering the need to examine early life determinants of disease.

Exposure to particulate matter (PM) air pollution has been associated with adverse cardiorenal outcomes including decreased kidney function and elevated blood pressure (BP) in children and adolescents (Rosa et al., 2020; Sanders et al., 2018). Long-term PM exposure contributes to the incidence and progression of chronic kidney disease (CKD) and hypertension (HTN) in adults (Blum et al., 2020; Bowe et al., 2018). Inhaled particles can rapidly trigger changes in BP and stimulate an inflammatory response of activated immune cells and cytokines leading to imbalance of vascular homeostatic responses. Increased demand on the kidney, may in turn lead to compromised glomerular function or acute kidney injury (Matsumura et al., 2018). Given the growing body of evidence that air pollution (i.e. PM2.5) may cross the placenta to disrupt normal homeostatic responses, it is possible that perinatal exposure to air pollution may occur during a particularly vulnerable period of nephron development and maturation, priming the kidney for subsequent CKD (Sanders et al., 2018; Zheng et al., 2017).

Indeed, the perinatal period comprises key susceptible windows of kidney development. In the developing fetus, metanephric kidney development begins around week 4 gestational age (GA). By the second trimester (11–14 weeks GA) the majority of branching morphogenesis is complete (Sampogna et al., 2015), with 90% of nephron formation accomplished in the third trimester (26–36 weeks GA) (Hinchliffe et al., 1991). After birth, renal blood flow, glomerular filtration rate (GFR) and renal homeostatic processes continue to mature and are also processes susceptible to environmental insults (Gubhaju et al., 2014; Veille et al., 1998). Environmental insults during these critical developmental periods may program adverse health outcomes in adulthood such as CKD or HTN (Solhaug et al., 2004). These events experienced in utero can lead to a higher risk for cardiovascular diseases and metabolic anomalies observed in children with CKD who have GFR decline (Furth et al., 2011; Luyckx et al., 2013). Even minor changes in GFR within a normal or mildly reduced range can predict CKD or mortality later in life (Matsushita et al., 2009; Turin et al., 2013). As Mexico has one of the highest prevalence of CKD worldwide (Collaboration, 2020) as well as elevated ambient PM2.5 concentrations in Mexico City (Just et al., 2015), we set out to examine perinatal windows of susceptibility to PM2.5 exposure on preadolescent eGFR and kidney function parameters among a high risk population. We additionally examined potential effect modification by child sex, given evidence of sex-specific differences in studies of PM exposure and children’s eGFR as well as CKD risk (Carrero et al., 2018; Liu et al., 2020). Our a priori hypothesis was that developmental PM2.5 exposure was associated with declines in eGFR, and an increase in serum cystatin C, serum creatinine, and blood urea nitrogen (BUN).

Methods

Study population

Between July 2007 and February 2011, pregnant women who were receiving prenatal care through the Mexican Social Security System (Instituto Mexicano del Seguro Social –IMSS) were recruited into the Programming Research in Obesity, Growth, Environment and Social Stressors (PROGRESS) study. Participants’ eligibility criteria included: less than 20 weeks’ gestation, at least 18 years of age, had completed primary education, planned to stay in Mexico City for the next 3 years, had access to a telephone, had no medical history of heart or kidney disease, did not consume alcohol daily, and did not use any steroid or anti-epilepsy medications. Procedures were approved by institutional review boards at the Brigham and Women’s Hospital, Icahn School of Medicine at Mount Sinai (protocol number (−00751), and the Mexican National Institute of Public Health (project #560). After a detailed explanation of the project protocols, during each study visit women provided written informed consent and at the preadolescent visit children provided assent. Of 948 mothers who delivered a live birth, a total of 571 children were followed longitudinally and presented at the 8–10 year-old visit. Our analysis excluded children born preterm (<37 weeks gestation; n=64) and post term (>42 weeks gestation; n=3), missing kidney parameter measure (n=73), missing daily PM2.5 estimates (n=1), and missing child characteristics (n=3). Derivation of gestational age was previously reported elsewhere (Sanders et al., 2015). Thus, study participants for this analysis were restricted to the 427 dyads who had at least one kidney parameter measure, daily PM2.5 estimates, and were born term.

PM2.5 levels

Daily exposure to PM2.5 at each participant’s residence was estimated during pregnancy through age 8 years using a previously described spatio-temporal model that incorporates Moderate Resolution Imaging Spectroradiometer (MODIS) satellite-derived Aerosol Optical Depth (AOD) measurements at a 1×1 km spatial resolution (Just et al., 2015). Briefly, remote sensing data were calibrated with municipal ground level monitors of PM2.5, meteorological data and land use regression variables to generate estimates of daily ambient PM2.5 concentrations for each participant at their residential location as recorded by research staff using handheld GPS devices. Mixed effect models with spatial and temporal predictors and day-specific random effects were used to account for temporal variations in the PM2.5–AOD relationship. The model was fit with a seasonal smooth function of latitude and longitude and time-varying average incorporating local monitoring for days without AOD data. We refer the reader to (Just et al., 2015) for additional details of the model methods and performance. In trimester-specific analyses, PM2.5 levels were averaged over clinically defined trimesters (in gestational weeks). In these trimester-specific analyses, we calculated the average PM2.5 over the first trimester defined as weeks 1–13 gestation, the second trimester defined as weeks 14–27 gestation, and the third trimester defined as 28 weeks’ gestation to delivery. Average postnatal PM2.5 was also estimated for the first year after delivery.

Serum collection details

At children’s arrival to the study clinic, fasting venous blood samples were collected in red top vacutainer tubes during the morning between 8 and 11 AM. Samples were centrifuged and serum was separated, aliquoted and stored at −70 ° C until subsequent analyses.

Kidney Function Parameters

Serum cystatin C was measured using the Quantikine Enzyme-Linked Immunosorbent Assay (ELISA) Human Cystatin C Immunoassay (R&D Systems, Minneapolis, MN) with intra- and inter-assay coefficient of variation below 5%. Optical density was measured by a microplate reader (Synergy HT, BioTek Instruments, Inc., Winooski, VT) set to 450 nm with a limit of detection of 31 pg/ml. Serum creatinine was measured using creatinine FS reagent and respons910 (DiaSys, Holzheim, Germany), based on Jaffe’s kinetic method without deproteinization, with a detection limit of 0.2 mg/dL and coefficient of variation <5%. BUN was calculated with the following formula: Serum urea (mg/dL)/2.14; serum urea was determined through the Urease-GLDH enzymatic UV test (Sarkar, 2013).

Estimated Glomerular Filtration Rate (eGFR)

eGFR was calculated according to cystatin C- based formula: and eGFRCysC2012 = 70.69 * (Cystatin C)−0.931 where serum cystatin C is in mg/L (Schwartz et al., 2012). The metabolite cystatin C has been proposed as an improved assessment that more closely estimates measured GFR in pediatric populations (Aggarwal et al., 2012; Ng et al., 2018). Cystatin C is produced by all mononucleated cells and freely filtered at the glomerulus. Compared to serum creatinine, cystatin C is not as subject to variation with muscle mass, BMI and delayed sensitivity to kidney damage (Gerchman et al., 2009). For sensitivity analysis, eGFR was also calculate using the validated serum creatinine eGFRSchwartz=k*height in cmSCr in mg/dL where k= 0.55 for children < 13 years (the preferred equation for serum creatinine measured by the Jaffe method) and creatinine-cystatin C-based CKID formula eGFR =39.8 × [height/serum creatinine]0.456 × [1.8/cystatin C]0.418 × [30/BUN]0.079 × [1.076male] [1.00female] × [height/1.4]0.179 (Schwartz et al., 2009).

Covariates

Covariates were selected a priori based on previous literature (Aris et al., 2019) and included child’s sex, age, and body mass index (BMI) z-score at blood draw, average first postnatal year PM2.5, seasonality of conception, maternal educational attainment at enrollment (categorized as <high school, some high school or high school graduate, >high school), and prenatal environmental tobacco smoke exposure. Children were weighed with the least amount of clothing possible and without footwear using an InBody230, and height was measured with a SECA stadiometer without footwear. BMI z-score was estimated according to WHO guidelines (Group, 2006): Underweight, normal weight, overweight, and obesity were defined as <−2SD, >−2SD to ≤+1SD, >+1SD to +2SD, and >+2SD, respectively. As there was one underweight child, this observation was excluded from the analyses. Prevalence of maternal smoking was very low in the cohort; in our sample, only 2 mothers reported smoking in pregnancy, therefore only prenatal exposure to environmental tobacco smoke (ETS) was considered in the models. Prenatal exposure to ETS was defined as report of any smoker in the home during the second or third trimester of pregnancy. Seasonality of PM2.5 exposure was defined as sine and cosine of time of year (Stolwijk et al., 1999; van Rossem et al., 2015). In linear regression models, where no time series data were used, season of conception was used to account for seasonality and was defined according to weather patterns in Mexico City as dry cold (January-February; November-December), dry warm (March-April) and rainy (May-October). Postnatal models included adjustment for season of kidney function assessment.

Statistical Analyses

Kidney outcomes of interest included eGFRCysC2012, cystatin C, creatinine and BUN. First, traditional linear regression models with adjustment for covariates listed above were run using clinically defined trimesters and average PM2.5 over pregnancy in separate models. In order to avoid bias in estimates (Wilson et al., 2017b), average exposure from all three trimesters were included in a single model.

To estimate the time-varying association between estimated daily PM2.5 levels and each kidney parameter, we then fitted distributed lag nonlinear models (DLNMs) using (a) prenatal and (b) postnatal exposures. The prenatal models included a 390-day exposure period starting 60 days prior to last menstrual period (LMP) and ending 50 days after 40 weeks gestation. A postnatal model included an exposure period starting one month prior to birth and ending at 8 years. The DLNMs were based on a generalized additive model with linear terms for the association of exposure and outcome and a penalized spline basis for the lag structure with penalties for overall smoothness. Additionally, for the prenatal models, ridge penalties (Gasparrini et al., 2017) were applied under the assumption that the associations at earliest lags (preconception period) approach the null value. A sensitive window was identified when the pointwise 95% confidence bands did not contain zero. In sensitivity analyses, we additionally examined the consistency of our detected windows to the use of alternative smoothers over the lag period and compared our results using b-splines, natural splines and penalized cubic regression splines. We also additionally adjusted for temperature by including a separate cross-basis for daily mean temperature (for prenatal models) and monthly mean temperatures (for postnatal models) covering the same lags. DLNMs were implemented using the dlnm package version 2.4.2 (Gasparrini, 2011) in R Version 3.5.1 (R Development Core Team) and other analyses were performed in SPSS version 24 (Chicago, IL).

Exploratory Bayesian distributed lag interaction models (BDLIM) were employed to examine effect modification by child sex using weekly average PM2.5. In addition to sensitive windows, BDLIM estimates the cumulative effect of PM2.5 exposure over the entire pregnancy for each sex-specific subgroup, accounting for identified sensitive windows and within-window effects (Wilson et al., 2017a). In these models, each stratum can have either the same or different sensitive windows or the same or different within-window effects (Wilson et al., 2017a). The model quantifies the likelihood of each pattern of heterogeneity and estimates the association between exposure and outcome under the effect modification pattern that is best supported by the data.

Results

Children in our study were an average of 9 years of age, equally distributed between males and females, and over a third experienced prenatal exposure to ETS (Table 1). More than half (53%) were normal weight, and 24% and 22% were overweight or obese, respectively. Nearly all eGFR measures were within normal range, only 4 participants had an eGFR< 60 mL/min/1.732, a level associated with adult CKD (Levin et al., 2013).

Table 1.

Demographics for mother-child dyads participating in the PROGRESS study at age 8 to 10 years.

Characteristic N=427
n (%)
Child sex
 Female 213 (49.9)
 Male 214 (50.1)
Prenatal environmental tobacco smoke (ETS) exposure (yes) 152 (35.6)
Maternal education at enrollment
 < High school 170 (39.8)
 Some high school or high school graduate 161 (37.7)
 > High school 96 (22.5)
Mean ± std
Maternal age at enrollment (years) 27.8 ± 5.71
Gestational age (weeks) 38.8 ± 1.13
Average prenatal PM2.5 (μg/m3) 22.7 ± 2.51
Average postnatal PM2.5 (μg/m3) year 1 22.7 ± 2.21
Average postnatal PM2.5 (μg/m3) year 2 22.0 ± 2.33
Average postnatal PM2.5 (μg/m3) year 3 23.0 ± 1.85
Average postnatal PM2.5 (μg/m3) year 4 22.2 ± 1.93
Average postnatal PM2.5 (μg/m3) year 5 21.6 ± 2.00
Average postnatal PM2.5 (μg/m3) year 6 21.2 ± 1.91
Average postnatal PM2.5 (μg/m3) year 7 20.8 ± 1.78
Average postnatal PM2.5 (μg/m3) year 8 20.8 ± 1.82
At preadolescence study visit (8–10 years)
Child age (years) 9.65 ± 0.65
Child BMI z-score 0.92 ± 1.24
Normal weight, n (%) 228 (53.2)
Overweight, n (%) 104 (24.4)
Obese, n (%) 95 (22.2)
Child kidney markers
 eGFRcysC2012 (mL/min/1.732) 99.4 ± 22.4
 Serum Cystatin C (mg/L) 0.73 ± 0.17
 Serum Creatinine (mg/dL) 0.43 ± 0.09
 Missing 80
 BUN (mg/dL) 12.3 ± 3.05
 Missing 81
Open in a new tab

First, we assessed associations between averaged PM2.5 across pregnancy, each trimester, and one year postnatally with each outcome in generalized linear models (Table 2). We found that each 5 μg/m3 increase in PM2.5 exposure averaged across the pregnancy period was associated with a statistically significant increase of 4.57 mL/min/1.73m2 in eGFRcysC2012 (95% CI: 0.13, 9.00). Correspondingly, each 5 μg/m3 increase in average PM2.5 exposure during pregnancy was associated with a decrease of −0.05 (95% CI: −0.08, −0.01) mg/dL and −0.02 (95% CI: −0.05, −0.002) mg/dL of serum cystatin C and serum creatinine, respectively. In models using the average PM2.5 exposure in each trimester, we observed that 5 μg/m3 increase in first trimester PM2.5 exposure was associated with a 3.44 mL/min/1.732 increase in eGFRcysC2012 (95% CI: 0.35, 6.52). Additionally, average PM2.5 exposure in the first postnatal year was associated with a 5 min/mL (95% CI: −9.87, −0.19) decrease in eGFRcysC2012, and corresponding increased serum cystatin C (β: 0.05; 95% CI: 0.01, 0.08) and serum creatinine (β: 0.04; 95% CI: 0.02, 0.06). No significant relationships were observed with BUN (Table 2). Sensitivity analyses using the creatinine-cystatin C-based equation showed similar results (Table S1, Supplemental Figure S1), whereas results derived with the “old” Schwartz equation showed a similar shape, but did not reach statistical significance (results not shown).

Table 2.

Association of average PM2.5 levels calculated during across pregnancy, each trimester, and the first postnatal year with kidney parameters assessed at preadolescence (ages 8–10 years)

eGFRCysC2012
(mL/min/1.732)		 Serum Cystatin C
(mg/L)		 Serum Creatinine
(mg/dL)		 BUN
(mg/dL)
β (95% CI) β (95% CI) β (95% CI) β (95% CI)
Model 1
Average pregnancy PM2.5 4.57*
(0.13, 9.00)		 −0.05*
(−0.08, −0.01)		 −0.02*
(−0.05, −0.002)		 0.04
(−0.70, 0.77)
Model 2
First trimester PM2.5 3.44*
(0.35, 6.52)		 −0.03*
(−0.05, −0.01)		 −0.15
(−0.03, 0.00)		 0.18
(−0.33, 0.68)
Second trimester PM2.5 −0.09
(−3.16, 2.97)		 −0.01
(−0.03, 0.01)		 −0.01
(−0.02, 0.00)		 −0.41
(−0.89, 0.08)
Third trimester PM2.5 1.46
(−1.50, 4.43)		 −0.01
(−0.03, 0.1)		 0.00
(−0.02, 0.01)		 0.41
(−0.07, 0.89)
First Postnatal year −5.00*
(−9.87, −0.19)		 0.05*
(0.01, 0.08)		 0.04†
(0.02, 0.06)		 −0.49
(−1.23, 0.25)
Open in a new tab

Estimates shown for a 5 μg/m3 increase in PM2.5. Linear models adjusted for child’s age, sex and BMI z-score, maternal education, prenatal ETS exposure, season of conception, other trimester PM2.5 concentrations (Model 1) and first postnatal year average PM2.5 (Models 1 and 2).

*

p<0.05,

†

p<0.01

In DLNMs with daily resolution PM2.5 estimates, we observed several non-linear relationships in the timing of the association between PM2.5 exposure and kidney parameters (Figure 1 a–d). We identified a window of susceptibility to PM2.5 exposure early during gestation with higher eGFRcysC2012. Specifically, we found that first trimester PM2.5 exposure between weeks 1 to 18 of gestation (days 1 to 129) was associated with increased preadolescent eGFRcysC2012. In context, a 5 μg/m3 increase in PM2.5 throughout this critical window would predict a cumulative increase in eGFRcysC2012 of 4.44 mL/min/1.732 (95% CI: 1.37, 7.52). In DLNMs examining monthly postnatal PM2.5 averages, we observed that PM2.5 exposure in the first 14 months of life was associated with a cumulative decrease in eGFRcysC2012 (β:−10.36; 95% CI: −17.68, −3.04) (Figure 2).

Figure 1.

Figure 1.

Open in a new tab

Associations between daily prenatal PM2.5 and a) eGFR, b) cystatin C, c) creatinine, and d) BUN at 8–10 years. Models adjusted for child’s age, sex and BMI z-score, maternal education, prenatal ETS exposure, seasonality and postnatal year 1 average PM2.5. Dotted lines represent conception and 40 weeks gestation.

Figure 2.

Figure 2.

Open in a new tab

Associations between monthly average postnatal PM2.5 and a) eGFR, b) cystatin C, c) creatinine, and d) BUN at 8–10 years. Models adjusted for child’s age, sex and BMI z-score, maternal education, prenatal ETS exposure, and seasonality at time of visit of kidney assessment. Dotted line represents birth.

Correspondingly, we also report that PM2.5 exposure between weeks 2 to 15 of gestation (days 15 to 108) was associated with decreased (β: −0.04; 95% CI: −0.07, −0.01) cystatin C. Among the monthly postnatal PM2.5 averages, we found that PM2.5 exposure in the first 8 months of life was associated with a slight cumulative increase in cystatin C (β: 0.038; 95% CI: 0.005, 0.071). For serum creatinine, we detected a susceptibility window at 5–19 weeks of gestation in which PM2.5 was associated with lower creatinine (β: −0.008; 95% CI: −0.002, −0.0007), although the observed magnitude of effect was close to zero. There was a small cumulative increase (β: 0.05; 95% CI: 0.01, 0.09) in creatinine among the first 18 months of life. For BUN the susceptibility window was estimated later (approximately third trimester), from gestational weeks 25–34 (days 176 to 240) (β: 0.70; 95% CI: 0.23, 1.18). The postnatal analysis examining BUN did not find a susceptibility window. Sensitivity analyses using of alternative smoothers over the lag period (Table S2, Figures S2–S5) or additionally adjusting for temperature showed similar results (Table S3). In BDLIM analysis, there was no evidence of sex-modification between PM2.5 and kidney parameters (data not shown).

Discussion

In this study, we found that children exposed perinatally to PM2.5 had altered kidney parameters at approximately 9 years of age. Exposure to PM2.5 early in pregnancy was associated with higher eGFR, lower cystatin C and lower creatinine. We also report that PM2.5 in late pregnancy was associated with higher BUN. We observed that early postnatal exposure was associated with lower eGFR and higher cystatin C and creatinine. Our findings suggest that PM2.5 exposure in the first trimester and first 14 months of life may have implications for kidney health in adolescence. While several prior studies have examined time-specific PM exposure and children’s kidney function or BP (Liu et al., 2020; Liu et al., 2021; Sanders et al., 2018), time-varying relationships of perinatal PM exposure with preadolescent eGFR have not yet been reported. As adolescent kidney function is a predictor of kidney health across the life course, the findings may have important public health implications for mitigating the risk of CKD and HTN.

There are several important observations from this work. First, one important interpretation is that in utero PM2.5 exposure may account for some portion of later-life kidney dysfunction, observable as early as 9 years of age. Our data suggest that the first trimester is an important window of susceptibility for increased preadolescent eGFR. Indeed, this overlaps with a critical developmental window of nephron progenitor development prior to branching morphogenesis (Sampogna et al., 2015). Importantly, loss of nephron progenitor cells affects uteric branching and nephron endowment (Cebrian et al., 2014). Our data also suggest that postnatal PM2.5 exposure is associated with lower eGFR, as suggested in prior studies (Liu et al., 2020; Liu et al., 2021). Postnatal kidney development and maturation is characterized by increased renal blood flow and GFR as renal homeostatic processes continue to mature (Veille et al., 1998). The opposite directionality of the two observed windows may seem counterintuitive, however, this is in-line with the sequalae of nephron damage and initially presenting as overcompensation by hyperfiltration followed by eGFR decline (Orr and Bridges, 2017). Additionally, several mechanisms have hypothesized potential developmental or teratogenic effects of PM exposure, including placental inflammation, oxidative stress and coagulation changes, which may occur during embryogenesis in the first trimester, leading to potential DNA damage in cells and future disruptions in cell activity in adolescence and adulthood (Kannan et al., 2006; Vrijheid et al., 2011). Additionally, experimental animal studies suggest that exposure to diesel and deep exhaust particles may lead to changes in renal hemodynamics, DNA damage in renal tissue, and further worsens acute kidney injury and chronic renal injury (Nemmar et al., 2010; Nemmar et al., 2016). Future research is needed to examine the opposite directionally of the two observed windows, including unmeasured or residual confounding, when examining the prenatal versus postnatal time periods.

Evidence from animal models supports in utero biological programming of kidney dysfunction and disease. There are several biologically plausible mechanisms by which PM2.5 could result in lower nephron endowment coupled with alterations in the renin-angiotensin-aldosterone system (RAS). For example, prenatal PM exposure can affect the RAS pathways in animal models, interfering with placental angiogenesis and nutrient transfer at the maternal-fetal interface (Soto et al., 2017). Oxidative stress and inflammation are the predominant unifying mechanisms that underlie PM2.5-associated kidney damage in animal models (Zhang et al., 2018). Fetal nephrogenesis is subject to modification according to the energy made available to the fetus (Chevalier, 2020). Prenatal PM2.5 exposure can disrupt mitochondrial energy production potentially resulting in decreased nephron number at birth, which carries an increased risk for CKD later in life (Chevalier, 2020). Notably, our prior work showed that third trimester PM2.5 levels were associated with decreased mitochondrial copy number in cord blood (Rosa et al., 2017). Longitudinal assessment of cumulative mitochondrial DNA damage is needed in future studies to better understand effects of renal-accumulated oxidative stress across the life course.

Prior studies of time-varying PM exposure and children’s kidney function have omitted the perinatal exposure period. In a study of 105 children aged 4–13-year-old in China, real-time PM2.5 was measured at three time points in childhood, prior to serum creatinine collection (Liu et al., 2020). The authors showed that personal exposure to PM2.5 was associated with a decrease in serum creatinine-based eGFR with a potential 2-day lag, showing acute changes in kidney function (Liu et al., 2020). In a subsequent study, the same group reported on real-time size-fractionated particle number counts (PNC) and 13 PM constituents, reporting greater reductions in eGFR in association with PNC1 and PNC2.5 (with an aerodynamic diameter ≤ 1 μm and ≤ 2.5 μm, respectively) than PNC0.5 (with an aerodynamic diameter ≤ 0.5 μm) in this panel of children (Liu et al., 2021).

While we did not find any sex-specific effects with the kidney biomarkers assessed herein, Liu et al. reported greater reductions in eGFR with short-term exposure to PM2.5 in boys when compared to girls (Liu et al., 2020). Recent data support male sex as an important risk factor for primary HTN in children (Theodore et al., 2015; Wang et al., 2006). Moreover, boys have been previously shown to be more susceptible to particulate matter exposure in utero, including higher risk of adverse fetal outcomes (Lakshmanan et al., 2015), differences in body composition (Chiu et al., 2017) and measures of oxidative stress (Song et al., 2019). As gender-associated risk disparities exist in CKD, sex-specific models will be important at later life stages (Carrero et al., 2018).

The PROGRESS cohort is an established prospective birth cohort with well-characterized demographic and covariate data. Strengths of fine-resolution satellite models include the ability to reconstruct ambient exposure estimates for the early pregnancy period based on residential locations and prospectively collected ground stations and remote sensing data. Furthermore, these modelled ambient exposure metrics are less susceptible to biases than personal exposure measurements, which risk confounding by individual behaviors (Weisskopf and Webster, 2017). We assessed multiple kidney parameters at age 9, prior to adolescence - a stage with hormone-associated growth spurts which affect kidney function. Our primary statistical approach included traditional linear regression models using averaged PM2.5 across pregnancy, each trimester, and one year postnatally as well as fine-resolution DLNMs using daily PM2.5 estimates of exposure. Advantages of our approach using different modeling methods (simple linear regression, DLNMs with different smoother specifications) enables comparison of the consistency of identified critical windows as well as overall shape of time-varying associations. Standard techniques wherein average PM during prenatal periods are assessed by linear regression models are unable to capture the distinct windows of susceptibility that may lie within or across trimesters.

PROGRESS is an urban disadvantaged population and our results may translate to other disadvantaged populations who face similar urban pollution, stressors and poverty. Similar studies conducted in other regions may help elucidate whether these results generalize to other populations. In our study, daily PM2.5 measures were used in the prenatal analysis, with a standard deviation between 7–8.5 μg/m3, and monthly measures were used in the postnatal analysis with a standard deviation of 1.67–7.81 ug/m3. In a recent study, a 1 μg/m3 interquartile range increase in PM2.5 concentration was associated with a 2% reduction in eGFR (Kuźma et al., 2021). Therefore, we believe there was substantial exposure variability in our study area to support the observed associations with kidney function. Our reliance on ambient exposure estimates do not capture indoor sources (although the ventilation patterns in Mexican homes let outdoor pollution indoors), nor did we include estimates of humidity, and thus may not completely reflect personal exposure. PM2.5 is attributed as the largest component of the global burden of disease due to air pollution (Cohen et al., 2017) and has an established association with adult kidney disease (Bowe et al., 2018). However, air pollution is a complex mixture and our analysis may be leaving out the potential contribution of other toxicants found in the study region. We cannot rule out residual confounding due to unmeasured factors or incomplete adjustment for measured factors related to PM2.5 that may also influence kidney function in childhood (e.g., socioeconomic factors related to residential location). Nevertheless, our findings were limited to specific perinatal time periods, thus unmeasured confounders would need to co-vary with PM2.5 and time-invariant characteristics, such as diet, would not explain these associations (Weisskopf et al., 2015). It is possible that BDLIM analyses of sex differences with kidney parameters were underpowered. Kidney function was assessed at a single time point and future analyses will include longitudinal follow-up. We assessed eGFR using both cystatin C and creatinine-based equations. Decrements in kidney function are traditionally assessed by biomarkers such as serum creatinine and urea (Khan et al., 2010); however, changes in these biomarkers may not manifest until substantial kidney impairment is present or clinical kidney disease is apparent. Thus, future work will examine biomarkers such as validated urinary proteins (i.e. beta-2-microglobulin, kidney injury molecule-1, neutrophil gelatinase-associated lipocalin) that capture subclinical kidney damage including potential tubular dysfunction. We note that comparison of serum creatinine vs. cystatin C findings showed greater variance and wider confidence intervals observed with creatinine and supports prior observations that cystatin C may be an improved biomarker for estimating GFR in children (Schwartz et al., 2012). We acknowledge that debate remains in the ideal equation for estimating GFR. Use of existing formulae requires careful selection considering the biomarker analyzed, type of molecular assay employed, as well as demographics of the study population (e.g., children vs. adult). Given the inequities in CKD across different racial/ethnic groups which are compounded by socioeconomic disparities, examining potential health disparities will be important considerations in future studies (Norris and Nissenson, 2008).

Mexico has one of the highest prevalence of CKD worldwide (Collaboration, 2020) and our research in PROGRESS children is directly relevant for understanding the role that early life environment plays in HTN and CKD etiology. Understanding regional causes for CKD and HTN are key factors for developing and maintaining effective disease care models (Stanifer et al., 2018) and furthermore can help inform public health strategies for disease prevention (Joshi et al., 2017; Luyckx et al., 2021). Longitudinal assessments of at-risk populations during adolescence and early adulthood will be an important next step.

Supplementary Material

1

Acknowledgements:

This work was supported by NIEHS grants R00ES027508 (Sanders AP, PI), R00ES027496 (Rosa MJ, PI), and R00ES023450 (Just AC, PI). The PROGRESS cohort has been supported by grants R01ES014930, R01ES013744, R24ES028522, P42ES016454, P30ES023515 (Wright RO, PI), R01ES021357 and R01ES020268 (Baccarelli A and Wright RO, MPI). This study was supported by the National Institute of Public Health/Ministry of Health of Mexico, and the National Institute of Perinatology. We thank the ABC (American British Cowdray Medical Center) in Mexico for providing some of the needed research facilities.

Abbreviations:

BDLIM

Bayesian distributed lag interaction model

BP

Blood pressure

BUN

Blood urea nitrogen

CI

confidence interval

DLNM

Distributed lag nonlinear models

eGFR

Estimated glomerular filtration rate

GPS

global positioning system

PM2.5

Particulate matter less than or equal 2.5 microns in diameter

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.

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. Aggarwal N, et al. , 2012. Creatinine-based estimations of kidney function are unreliable in obese kidney donors. J Transplant. 2012, 872894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aris IM, et al. , 2019. Early Life Predictors of Systolic Blood Pressure Trajectories from Infancy to Adolescence: Findings from Project Viva. Am J Epidemiol. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Blum MF, et al. , 2020. Particulate Matter and Albuminuria, Glomerular Filtration Rate, and Incident CKD. Clin J Am Soc Nephrol. 15, 311–319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bowe B, et al. , 2018. Particulate Matter Air Pollution and the Risk of Incident CKD and Progression to ESRD. J Am Soc Nephrol. 29, 218–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carrero JJ, et al. , 2018. Sex and gender disparities in the epidemiology and outcomes of chronic kidney disease. Nat Rev Nephrol. 14, 151–164. [DOI] [PubMed] [Google Scholar]
  6. Cebrian C, et al. , 2014. The number of fetal nephron progenitor cells limits ureteric branching and adult nephron endowment. Cell Rep. 7, 127–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chevalier RL, 2020. Bioenergetic Evolution Explains Prevalence of Low Nephron Number at Birth: Risk Factor for CKD. Kidney360. 1, 863–879. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chiu YM, et al. , 2017. Prenatal particulate air pollution exposure and body composition in urban preschool children: Examining sensitive windows and sex-specific associations. Environ Res. 158, 798–805. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cohen AJ, et al. , 2017. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: an analysis of data from the Global Burden of Diseases Study 2015. Lancet. 389, 1907–1918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Collaboration, G. B. D. C. K. D., 2020. Global, regional, and national burden of chronic kidney disease, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 395, 709–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Furth SL, et al. , 2011. Metabolic abnormalities, cardiovascular disease risk factors, and GFR decline in children with chronic kidney disease. Clinical journal of the American Society of Nephrology : CJASN. 6, 2132–2140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gasparrini A, 2011. Distributed Lag Linear and Non-Linear Models in R: The Package dlnm. Journal of Statistical Software. 43, 1–20. [PMC free article] [PubMed] [Google Scholar]
  13. Gasparrini A, et al. , 2017. A penalized framework for distributed lag non-linear models. Biometrics. 73, 938–948. [DOI] [PubMed] [Google Scholar]
  14. Gerchman F, et al. , 2009. Body mass index is associated with increased creatinine clearance by a mechanism independent of body fat distribution. J Clin Endocrinol Metab. 94, 3781–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Group, W. H. O. M. G. R. S., 2006. WHO Child Growth Standards based on length/height, weight and age. Acta Paediatr Suppl. 450, 76–85. [DOI] [PubMed] [Google Scholar]
  16. Gubhaju L, et al. , 2014. Assessment of renal functional maturation and injury in preterm neonates during the first month of life. Am J Physiol Renal Physiol. 307, F149–58. [DOI] [PubMed] [Google Scholar]
  17. Hill NR, et al. , 2016. Global Prevalence of Chronic Kidney Disease - A Systematic Review and Meta-Analysis. PLoS One. 11, e0158765. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hinchliffe SA, et al. , 1991. Human intrauterine renal growth expressed in absolute number of glomeruli assessed by the disector method and Cavalieri principle. Lab Invest. 64, 777–84. [PubMed] [Google Scholar]
  19. Joshi R, et al. , 2017. The Potential Impact of Public Health Interventions in Preventing Kidney Disease. Semin Nephrol. 37, 234–244. [DOI] [PubMed] [Google Scholar]
  20. Just AC, et al. , 2015. Using High-Resolution Satellite Aerosol Optical Depth To Estimate Daily PM2.5 Geographical Distribution in Mexico City. Environ Sci Technol. 49, 8576–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kannan S, et al. , 2006. Exposures to airborne particulate matter and adverse perinatal outcomes: a biologically plausible mechanistic framework for exploring potential effect modification by nutrition. Environmental health perspectives. 114, 1636–1642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Khan E, et al. , 2010. Emergence of biomarkers in nephropharmacology. Biomark Med. 4, 805–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kim J, et al. , 2006. Trends in overweight from 1980 through 2001 among preschool-aged children enrolled in a health maintenance organization. Obesity (Silver Spring). 14, 1107–12. [DOI] [PubMed] [Google Scholar]
  24. Kuźma Ł, et al. , 2021. Exposure to air pollution and renal function. Scientific Reports. 11, 11419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lakshmanan A, et al. , 2015. Associations between prenatal traffic-related air pollution exposure and birth weight: Modification by sex and maternal pre-pregnancy body mass index. Environmental Research. 137, 268–277. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Levin A, et al. , 2013. Kidney disease: Improving global outcomes (KDIGO) CKD work group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney International Supplements. 3, 1–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Liu M, et al. , 2020. Personal exposure to fine particulate matter and renal function in children: A panel study. Environ Pollut. 266, 115129. [DOI] [PubMed] [Google Scholar]
  28. Liu M, et al. , 2021. Short-term effects of size-fractionated particulate matters and their constituents on renal function in children: A panel study. Ecotoxicol Environ Saf. 209, 111809. [DOI] [PubMed] [Google Scholar]
  29. Luyckx VA, et al. , 2021. Sustainable Development Goals relevant to kidney health: an update on progress. Nat Rev Nephrol. 17, 15–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Luyckx VA, et al. , 2013. Effect of fetal and child health on kidney development and long-term risk of hypertension and kidney disease. Lancet. 382, 273–83. [DOI] [PubMed] [Google Scholar]
  31. Matsumura K, et al. , 2018. Trajectory of Estimated Glomerular Filtration Rate Predicts Renal Injury in Children with Multicystic Dysplastic Kidney. Nephron. 140, 18–23. [DOI] [PubMed] [Google Scholar]
  32. Matsushita K, et al. , 2009. Change in estimated GFR associates with coronary heart disease and mortality. Journal of the American Society of Nephrology : JASN. 20, 2617–2624. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Nemmar A, et al. , 2010. Diesel exhaust particles in the lung aggravate experimental acute renal failure. Toxicol Sci. 113, 267–77. [DOI] [PubMed] [Google Scholar]
  34. Nemmar A, et al. , 2016. Prolonged Pulmonary Exposure to Diesel Exhaust Particles Exacerbates Renal Oxidative Stress, Inflammation and DNA Damage in Mice with Adenine-Induced Chronic Renal Failure. Cell Physiol Biochem. 38, 1703–13. [DOI] [PubMed] [Google Scholar]
  35. Ng DK, et al. , 2018. Combination of pediatric and adult formulas yield valid glomerular filtration rate estimates in young adults with a history of pediatric chronic kidney disease. Kidney Int. 94, 170–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Norris K, Nissenson AR, 2008. Race, gender, and socioeconomic disparities in CKD in the United States. J Am Soc Nephrol. 19, 1261–70. [DOI] [PubMed] [Google Scholar]
  37. Orr SE, Bridges CC, 2017. Chronic Kidney Disease and Exposure to Nephrotoxic Metals. Int J Mol Sci. 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Praga M, 2005. Synergy of low nephron number and obesity: a new focus on hyperfiltration nephropathy. Nephrol Dial Transplant. 20, 2594–7. [DOI] [PubMed] [Google Scholar]
  39. Rosa MJ, et al. , 2020. Identifying critical windows of prenatal particulate matter (PM2.5) exposure and early childhood blood pressure. Environ Res. 182, 109073. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Rosa MJ, et al. , 2017. Identifying sensitive windows for prenatal particulate air pollution exposure and mitochondrial DNA content in cord blood. Environ Int. 98, 198–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sampogna RV, et al. , 2015. Developmental Programming of Branching Morphogenesis in the Kidney. J Am Soc Nephrol. 26, 2414–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sanders AP, et al. , 2015. microRNA expression in the cervix during pregnancy is associated with length of gestation. Epigenetics. 10, 221–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sanders AP, et al. , 2018. Perinatal and childhood exposure to environmental chemicals and blood pressure in children: a review of literature 2007–2017. Pediatr Res. 84, 165–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Sarkar R, 2013. Establishment of Biological Reference Intervals and Reference Curve for Urea by Exploratory Parametric and Non-Parametric Quantile Regression Models. Ejifcc. 24, 61–7. [PMC free article] [PubMed] [Google Scholar]
  45. Schwartz GJ, et al. , 2009. New equations to estimate GFR in children with CKD. J Am Soc Nephrol. 20, 629–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Schwartz GJ, et al. , 2012. Improved equations estimating GFR in children with chronic kidney disease using an immunonephelometric determination of cystatin C. Kidney Int. 82, 445–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Solhaug MJ, et al. , 2004. The developing kidney and environmental toxins. Pediatrics. 113, 1084–91. [PubMed] [Google Scholar]
  48. Song LL, et al. , 2019. Effects of maternal exposure to ambient air pollution on newborn telomere length. Environment International. 128, 254–260. [DOI] [PubMed] [Google Scholar]
  49. Soto SF, et al. , 2017. Exposure to fine particulate matter in the air alters placental structure and the renin-angiotensin system. PLoS One. 12, e0183314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Stanifer JW, et al. , 2018. Chronic kidney disease care models in low- and middle-income countries: a systematic review. BMJ Glob Health. 3, e000728. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Stolwijk AM, et al. , 1999. Studying seasonality by using sine and cosine functions in regression analysis. Journal of Epidemiology and Community Health. 53, 235–238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Theodore RF, et al. , 2015. Childhood to Early-Midlife Systolic Blood Pressure Trajectories: Early-Life Predictors, Effect Modifiers, and Adult Cardiovascular Outcomes. Hypertension. 66, 1108–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Turin TC, et al. , 2013. Change in the estimated glomerular filtration rate over time and risk of all-cause mortality. Kidney Int. 83, 684–91. [DOI] [PubMed] [Google Scholar]
  54. van Rossem L, et al. , 2015. Prenatal Air Pollution Exposure and Newborn Blood Pressure. Environmental Health Perspectives. 123, 353–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Veille JC, et al. , 1998. Renal Hemodynamics: Longitudinal Study from the Late Fetal Life to One Year of Age. J Matern Fetal Investig. 8, 6–10. [PubMed] [Google Scholar]
  56. Vrijheid M, et al. , 2011. Ambient air pollution and risk of congenital anomalies: a systematic review and meta-analysis. Environmental health perspectives. 119, 598–606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Wang X, et al. , 2006. Ethnic and gender differences in ambulatory blood pressure trajectories: results from a 15-year longitudinal study in youth and young adults. Circulation. 114, 2780–7. [DOI] [PubMed] [Google Scholar]
  58. Weisskopf MG, et al. , 2015. Air Pollution and Autism Spectrum Disorders: Causal or Confounded? Curr Environ Health Rep. 2, 430–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Weisskopf MG, Webster TF, 2017. Trade-offs of Personal Versus More Proxy Exposure Measures in Environmental Epidemiology. Epidemiology. 28, 635–643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Wilson A, et al. , 2017a. Bayesian distributed lag interaction models to identify perinatal windows of vulnerability in children’s health. Biostatistics. 18, 537–552. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Wilson A, et al. , 2017b. Potential for Bias When Estimating Critical Windows for Air Pollution in Children’s Health. Am J Epidemiol. 186, 1281–1289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Xu X, et al. , 2018. Environmental pollution and kidney diseases. Nat Rev Nephrol. 14, 313–324. [DOI] [PubMed] [Google Scholar]
  63. Zhang Y, et al. , 2018. The Kidney Injury Induced by Short-Term PM2.5 Exposure and the Prophylactic Treatment of Essential Oils in BALB/c Mice. Oxid Med Cell Longev. 2018, 9098627. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Zheng LY, et al. , 2017. Environmental exposures and pediatric kidney function and disease: A systematic review. Environ Res. 158, 625–648. [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
NIHMS1743060-supplement-1.docx (722.9KB, docx)

RESOURCES