Ambient PM2.5 exposure and salivary cortisol output during pregnancy in a multi-ethnic urban sample

Abstract Objectives Evidence from murine research supports that fine particulate matter (PM2.5) may stimulate the hypothalamic-pituitary-adrenal axis, leading to elevated circulating glucocorticoid levels. Epidemiologic research examining parallel associations document similar associations. We examined these associations among a diverse sample of pregnant individuals exposed to lower levels of ambient PM2.5. Materials and Methods Participants included pregnant individuals enrolled in the PRogramming of Intergenerational Stress Mechanisms (PRISM) pre-birth cohort. Daily residential PM2.5 exposure was estimated using a satellite-based spatial-temporal hybrid model. Maternal 3rd trimester salivary cortisol levels were used to calculate several features of the diurnal cortisol rhythm. We used multivariable linear regression to examine PM2.5 during the pre-conception period and during each trimester in relation to cortisol awakening rise (CAR), slope, and area under the curve relative to ground (AUCG). Results and Discussion The average PM2.5 exposure level across pregnancy was 8.13 µg/m3. PM2.5 in each exposure period was positively associated with AUCG, a measure of total cortisol output across the day. We also observed an inverse association between PM2.5 in the 3rd trimester and diurnal slope, indicating a steeper decline in cortisol throughout the day with increasing exposure. We did not detect strong associations between PM2.5 and slope for the other exposure periods or between PM2.5 and CAR for any exposure period. Conclusions In this sample, PM2.5 exposure across the preconception and pregnancy periods was associated with increased cortisol output, even at levels below the U.S. National Ambient Air Quality Annual Standard for PM2.5 of 12.0 µg/m3.

Particulate matter is a mixture of solid particles and liquid droplets present in the air that arise from a range of sources, including traffic, construction activity, commercial kitchens, and smokestacks. Fine particles (PM 2.5 ) have a diameter approximately 1/30th that of a strand of hair and are small enough to penetrate deep lung tissue, where they can stimulate a local immune response, produce reactive oxygen species, and trigger more systemic inflammatory reactions (Calderon-Garciduenas et al. 2015;Feng et al. 2016). Given ubiquitous exposure and links to numerous adverse health outcomes, PM 2.5 is widely considered one of the most dangerous air pollutants, especially in urban areas (Braithwaite et al. 2019;Bekkar et al. 2020;Lederer et al. 2021).
The hypothalamic-pituitary-adrenal (HPA) axis is an adaptive stress response system under endocrine control that responds to a range of environmental perturbations, including those of emotional, physical, and immunological origin (Itoi et al. 1994). The axis originates in the paraventricular nucleus (PVN) of the hypothalamus and includes a cascade of events regulated via negative feedback. Briefly, following neurotransmitter stimulation, corticotrophin releasing hormone (CRH) is released from the PVN and enters the portal circulation where it acts on the pituitary to trigger the secretion of adrenocorticotropin hormone (ACTH). ACTH then stimulates glucocorticoid (cortisol in humans, corticosterone in rodents) secretion from the adrenal glands. Glucocorticoids bind receptors (GRs) located on nearly all cell types to regulate a range of processes, including glucose metabolism, inflammation, and hormone activity.
Evidence from studies conducted using murine models supports that PM 2.5 exposure may stimulate the HPA axis (Snow et al. 2018;Thomson 2019). In rats, repeated studies have shown that short-term exposure to concentrated fine air particles (500 mg/m 3 ) increases levels of both the neurotransmitter norepinephrine, a key activator of CRH neurons (Itoi et al. 1994), and CRH in the hypothalamus (Sirivelu et al. 2006;Balasubramanian et al. 2013). Research in mice has also shown that long term exposure to unfiltered (mean PM 2.5 : 91 mg/m 3 ) versus filtered (18 mg/m 3 ) air results in a significant decrease in the expression of glucocorticoid receptors (GR) in the brain that are responsible for HPA axis negative feedback responsivity, leading to sustained HPA activation (Jia et al. 2018). These studies and others also consistently report increased levels of corticosterone in rodent serum following exposure to high levels of particulates (Sirivelu et al. 2006;Balasubramanian et al. 2013;Thomson et al. 2013;Jia et al. 2018). Gene profiling research has further demonstrated that rats exposed to high dose urban air particles display transient changes in gene expression, including an upregulation of stress-responsive genes in the pituitary, as well as an upregulation of genes that mediate the anti-inflammatory action of glucocorticoids in multiple organs (Thomson et al. 2013). Chronic activation of the HPA axis is associated with a number of disease states, including those that affect the psychiatric, neurological, cardiovascular, and reproductive systems, and could be one mechanism underlying the broad negative health effects associated with PM 2.5 exposure. Research conducted in humans has also linked PM 2.5 exposure with increased cortisol levels in a range of matrices (blood, saliva, urine); however, studies conducted to date are generally limited by small sample sizes and high exposure levels (Li et al. 2017;Jia et al. 2018;Niu et al. 2018;Boll et al. 2020;Khamirchi et al. 2020;Ahlers and Weiss 2021;Verheyen et al. 2021;Wei et al. 2021). Further, only a subset of these studies have considered the pregnancy period (Boll et al. 2020;Khamirchi et al. 2020;Ahlers and Weiss 2021;Verheyen et al. 2021). In the present analysis, we examined the prospective and cross-sectional relation between lower level PM 2.5 exposure and indices of the diurnal cortisol rhythm among a sample of pregnant individuals. Whether findings from animal and human studies investigating higher dose exposure remain at exposure levels relevant to the general U.S. population is important for understanding the public health implications of current PM 2.5 National Ambient Air Quality Standards (NAAQS). Glucocorticoids play a critical role in fetal development and during pregnancy, the fetal stress response relies heavily on inputs from the maternal and placental system (Itoi et al. 1994;Moisiadis and Matthews 2014). Moreover, dynamic physiologic changes that occur to support the developing fetus, including shifts to neuroendocrine and immune system functioning, may increase maternal susceptibility to PM 2.5 (Varshavsky et al. 2020). These shifts, including placental incorporation with the maternal HPA axis, may additionally reduce the generalizability of findings from studies of non-pregnant adults to pregnant persons, supporting the need to specifically examine these associations in a pregnant sample.

Study sample
The study sample includes a subset of pregnant individuals enrolled in the PRogramming of Intergenerational Stress Mechanisms (PRISM) pre-birth cohort. The study recruited n ¼ 1110 participants receiving prenatal care from the Beth Israel Deaconess Medical Center and East Boston Neighborhood Health Center in Boston, MA (from March 2011-December 2013) and Mount Sinai Hospital in New York City, NY (from April 2013-ongoing). Eligibility criteria include singleton gestation, age 18 years or older, fluency in English or Spanish, intake of seven or fewer alcoholic drinks per week before pregnancy and no alcohol after pregnancy recognition, and HIV negative status. Participants who deliver a live newborn with no significant congenital anomalies noted during pregnancy or at birth that would impact ongoing participation remain eligible for study follow-up. The present analysis is restricted to 987 participants for whom residential address during pregnancy was available and geocoded to obtain PM 2.5 estimates. Of these, 289 provided saliva samples during the 3rd trimester for measurement of cortisol levels. We additionally excluded three participants missing key covariate data and three participants with excessive cortisol levels, resulting in a final analytic sample of 283 pregnant individuals. Study procedures were approved by the institutional review boards at the Icahn School of Medicine at Mount Sinai in New York City and the Brigham and Women's Hospital (BWH) in Boston. Written informed consent was obtained from all participants in their primary language.

PM 2.5 exposure
We used the ArcGIS software framework to geocode maternal residential address during pregnancy, accounting for any moves. Geocoding has been robustly validated as previously described (Brunst et al. 2018). We predicted daily residential PM 2.5 exposure for each study participant using an adaptation of a previously described satellite based hybrid model. Briefly, we used extreme gradient boosting (XGboost) to predict PM 2.5 based on spectral Aerosol Optical Depth (AOD) estimates from the MODIS sensor of the Aqua and Terra satellites in combination with PM 2.5 monitoring data and spatiotemporal predictors such as height of the planetary boundary layer, percentage of developed area, air temperature, specific humidity and others (Just et al. 2020). We implemented a recursive feature selection to develop a parsimonious model and demonstrate excellent predictions of withheld observations (RMSE of 2.10 lg/m 3 and RMSE of 3.11 lg/m 3 in our spatial cross-validation). We considered average exposure over the following periods: 3 months preconception, 1st trimester, 2nd trimester, and 3rd trimester (cross-sectional).

Salivary cortisol
Study participants provided five passive drool saliva samples at the following defined time windows across three days within a 1-week period during the 3rd trimester of pregnancy (mean ± standard deviation: 33.0 ± 3.6 weeks, range: 27-39 weeks): (1) upon awakening, (2) 45 minutes (range: 30-90 minutes) after awakening and before the consumption of caffeine or food, (3) 4 hours (range: 4-6.5 hours) after awakening, (4) 10 hours (range: 7.5-11.5 hours) after awakening, and (5) before sleep (>11.5 hours after awakening) (Cowell et al. 2021). Samples were collected at home and stored in the participant's personal refrigerator until collection by study staff. After receipt by our laboratory, samples were processed and stored at -80 c until cortisol analysis, which was conducted by the Kirschbaum laboratory (Dresden, Germany) using a commercial chemiluminescence assay (IBL International). The assay sensitivity was 0.47 nmol/L, and the limit of detection was 0.08 nmol/L; intra-and inter-assay coefficients of variation were both below 8%.
Cortisol output follows a diurnal rhythm characterized by a rapid rise after awakening followed by a slower decline throughout the remainder of the day (Edwards et al. 2001). To capture this rhythm, we calculated three features of the daily cortisol curve as previously described in detail (Cowell et al. 2021). Briefly, we calculated cortisol awakening rise (CAR), which provides information on the immediate postawakening increase, as the difference in cortisol between the first and second time points. We calculated diurnal slope across the day (slope), which reflects the rate of decline in cortisol from awakening to bedtime, by regressing cortisol on time of sample collection within a person, after removing the sample collected 45 minutes after awakening to minimize the impact of the morning rise (Matthews et al. 2006). Finally, we calculated area under the curve relative to ground (AUC G ), which provides a measure of total cortisol output throughout the day, according to the trapezoid approach described by Pruessner (Pruessner et al. 2003). For each participant, we calculated the three measures for each of the three sampling days before averaging across days.

Covariates
Information on sociodemographic and lifestyle characteristics was collected during a structured in-person interview conducted during pregnancy. Participant age, race/ethnicity, and level of education were self-reported. Information on cigarette smoking during pregnancy and exposure to environmental tobacco smoke (ETS) was ascertained by questionnaire during pregnancy and again during the post-partum period. Participants were considered smoke exposed if they reported active smoking or exposure to ETS for 1-hour or more during pregnancy. Maternal pre-pregnancy body mass index (BMI) was calculated as self-reported weight (kg) divided by height squared (m 2 ); because we relied on participant recall of weight prior to pregnancy, we acknowledge the potential for measurement error in this covariate. Similar to PM 2.5 AOD products, daily surface temperature was obtained from the MODIS instruments of the NASA Terra and Aqua satellites (Carri on et al. 2021). These measures were calibrated to air temperature at the reference height (2 meters above ground) using ground monitoring data derived from the National Climate Data Center, the Meteorological Assimilation Data Ingest System of the National Oceanic and Atmospheric Administration, and a large number of aggregated nongovernmental meteorologic stations. The calibration also included a temporal smoothing algorithm to account for location, season, year, and land-use regression terms for greenness, elevation, and land use.

Statistical analysis
We first calculated descriptive statistics for: PM 2.5 at each exposure period, each cortisol parameter, and each covariate. We visualized the distributions of exposure and outcome variables using histograms and boxplots and examined the bivariate relationship using scatter plots with loess curves to visualize any non-linear patterns. Given an apparent linear relationship and normal residuals, we used multivariable linear regression to examine associations between average PM 2.5 during four exposure periods (3 months preconception, 1st trimester, 2nd trimester, 3rd trimester) and each feature of the diurnal cortisol curve (CAR, slope, AUC G ) in separate models. Based on the evaluation of Cook's distance, we excluded three highly influential data points from all analyses (Cook's D > 1.00). In secondary analyses, we further examined PM 2.5 averaged between pregnancy onset and the time of 3rd trimester saliva sample collection in relation to salivary cortisol levels at each of the five collection time points in models adjusting for the tier three set of covariates as described below. To visualize the relationship between PM 2.5 and the overall diurnal cortisol rhythm, we plotted least square mean cortisol values at each time point across the day stratified by participants with higher versus lower PM 2.5 exposure defined by dichotomization at the median.
We took a tiered approach to covariate adjustment. First, we adjusted all models for gestational week at saliva collection, which we considered an important precision variable that could help to minimize the influence of gestational timing on cortisol levels. We next considered a set of models additionally adjusting for maternal age (continuous years), race/ethnicity (White-Hispanic vs. Black/Black-Hispanic vs. White, non-Hispanic), education (<high school vs. high school vs. >high school), smoke exposure (active smoking or ETS as described vs. neither), and temperature (continuous in degrees Celsius). Finally, we ran a third set of models additionally adjusting for maternal pre-pregnancy BMI, which has been linked to cortisol output and may be associated with PM 2.5 exposure, although that literature to date has been inconsistent. Given our focus on biological indicators of HPA axis functioning, we did not adjust models for potential upstream determinants of cortisol output, such as objective or perceived measures of maternal stress.
Exogenous steroids may influence HPA axis activity, therefore, we performed a sensitivity analysis excluding participants who used inhaled (n ¼ 10), nasal (n ¼ 4), topical (n ¼ 10) and/or oral (n ¼ 9) steroids during pregnancy or in the immediate post-partum period (n ¼ 29 excluded). Finally, because cigarette smoking provides an additional source of particle exposure independent of ambient air pollution, we explored models excluding participants who reported smoking during pregnancy (n ¼ 66). Table 1 provides sociodemographic and lifestyle characteristics of the included sample. On average, participants were 30 years old at enrollment with the majority self-identifying as White, non-Hispanic (25%), White-Hispanic (33%) or Black/Black-Hispanic (34%). Approximately 15% of the sample had less than a high school education and 23% reported actively smoking or exposure to ETS during pregnancy. The distribution of PM 2.5 was approximately normal for each exposure period considered, with an average ± SD level across pregnancy of 8.13 ± 1.03 mg/m 3 (range: 5.62 À 10.68). Likewise, the distributions of AUC G and slope appeared normal, whereas the distribution of CAR showed a slight right tail. Despite this skew, linear regression was considered acceptable based on normal residuals. Notably, compared to participants enrolled in PRISM, but excluded from the current analysis, the included sample had fewer racially minoritized participants, was more highly educated, was on average 1.5 years older at enrollment, had lower average BMI, was less likely to be exposed to smoke during pregnancy and had slightly higher (average 0.2 mg/m 3 ) PM 2.5 exposure (Supplemental Table 1). The included sample was also more likely to be recruited from Boston, MA (56%) compared to the excluded sample (32%), which likely underlies these demographic differences and reflects a timedependent shift in recruitment from Boston to New York City beginning in 2013. There were no statistically significant differences in cortisol features between those included and those excluded (all p-value > 0.05).

Associations between PM 2.5 and salivary cortisol features
In our primary analyses adjusting for week of saliva collection, temperature, age, race/ethnicity, education, and smoke exposure, we found positive associations between PM 2.5 in each exposure period and AUC G , a measure of total cortisol output across the day (Figure 1 and Supplemental Table 3). The effect sizes ranged from 8.87 to 13.35 nmol/L increase per 1 mg/m 3 increase (approximately  Abbreviations: AUCg: cortisol area under the curve relative to ground in nmol/L Â hours; CAR: cortisal awakening rise in nmol/L; slope: nmol/L change in cortisol per hour; PM 2.5 : fine particulate matter; Pre: 3-months preconception; T1: first trimester; T2: second trimester; T3: third trimester. Points indicate beta coefficients for the change in cortisol feature for a 1 mg/m 3 increase in PM 2.5 and lines indicate 95% confidence intervals.
1 SD) in PM 2.5 exposure, depending on the period considered. We also observed a positive trend between increasing PM 2.5 and CAR, although associations did not reach statistical significance. Finally, we detected an inverse association between PM 2.5 in the 3rd trimester and diurnal slope, indicating a steeper decline in cortisol throughout the day with increasing exposure. Results from models adjusting only for gestational week of saliva collection (Supplemental Table 2), as well as the fully adjusted model that also controlled for maternal pre-pregnancy BMI (Supplemental Table 4), were not meaningfully different. When considering PM 2.5 averaged between pregnancy onset and saliva collection in relation to cortisol at each of the five collection times across the day, we observed the following positive associations: awakening (b for a nmol/L change in cortisol per a 1 mg/m 3 increase in PM 2 . This is consistent with our finding of higher AUC G and suggests the inverse association with slope may be driven by higher cortisol at awakening. To help visualize these associations, Figure 2 provides diurnal cortisol rhythms derived from least square mean values at each time point among participants with PM 2.5 levels below and above the sample median.

Sensitivity analyses
Results from models excluding participants who used steroids (Supplemental Table 5) or smoked (Supplemental Table 6) during pregnancy were similar to the primary results in direction and magnitude

Discussion
In this study, we found that PM 2.5 exposure during the preconception period and across the course of pregnancy, even at levels below the U.S. NAAQS PM 2.5 annual (12 mg/m 3 ) and 24-hour (35 mg/m 3 ) standards, was associated with higher total daily cortisol output during the 3rd trimester.
We are aware of two prior studies conducted in pregnant persons that have examined similar PM 2.5 exposure levels in relation to cortisol output. Consistent with our findings, among a sample 50 participants in California, PM 2.5 exposure (average 8.1 mg/m 3 ) during late pregnancy was associated with higher salivary cortisol AUC G (mg/dL change per 1 mg/m 3 increase in PM 2.5 : 15.93, p ¼ 0.005), but not with CAR or diurnal slope (Ahlers and Weiss 2021). Similarly, a cross-sectional study of 81 pregnant persons based in Belgium observed a positive association between PM 2.5 exposure (geometric mean: 11.61 mg/m 3 ) and higher levels of cortisol in 3rd trimester hair, a measure of longer-term cortisol output, but only after adjusting for season of sampling (Verheyen et al. 2021). Parallel associations during the second trimester (n ¼ 133) were attenuated and positive only in unadjusted models. A third study conducted in Iran (n ¼ 150) found positive associations between PM 2.5 exposure during pregnancy (mean 47.0 mg/m 3 ) and cord blood cortisol, however, they did not examine maternal cortisol output and exposure levels were considerably higher (Boll et al. 2020;Khamirchi et al. 2020). Studies of non-pregnant adults have largely documented similar relationships (Li et al. 2017;Jia et al. 2018;Niu et al. 2018;Wei et al. 2021). Most recently, a large study (n ¼ 6223) of adults living in rural China found that longterm PM 2.5 exposure (median across a 3-year period: 68 mg/ m 3 ) was positively associated with cortisone, but not cortisol, measured in fasting serum samples (Wei et al. 2021). In a much smaller natural experiment of 12 adult men in China, plasma cortisol levels were higher following a week of exposure to high ambient PM 2.5 (277 mg/m 3 ) compared to levels following a control week when ambient levels were lower (48 mg/m 3 ) (Jia et al. 2018). Similarly, an air purification crossover trial of 60 college students in China found that higher PM 2.5 exposure during periods of sham (mean: 53 mg/m 3 ) versus real (mean: 24 mg/m 3 ) purification was associated with higher levels of cortisol, cortisone, CRH and ACTH measured via serum metabolomics, suggesting that PM 2.5 may interfere with feedback processes along the HPA axis (Li et al. 2017). However, we note that findings have not always been consistent. For example, among a sample of older adults enrolled in the Multi-Ethnic Study of Atherosclerosis (MESA), no statistically significant cross-sectional associations were detected between ambient PM 2.5 and several features of the diurnal cortisol curve, including AUC (Hajat et al. 2019). Comparisons across these studies are also challenging given the various methods used to derive pollution measures. Salivary cortisol levels at each sampling time for participants with lower ( 8.18 mg/m 3 ) compared to higher (>8.18 mg/m 3 ) PM 2.5 exposure averaged across pregnancy onset and 3rd trimester cortisol sampling date. Values are least square mean cortisol levels from models adjusted for gestational week of saliva collection, temperature, race/ethnicity, age, education, smoke exposure and maternal pre-pregnancy body mass index. Smoothing between points is shown using a loess curve.
As described previously, experimental animal research has demonstrated that PM 2.5 exposure activates central stress centers and increases plasma ACTH and corticosterone levels (Sirivelu et al. 2006;Thomson et al. 2013;Jia et al. 2018). In addition, research conducted using murine models has identified associations with altered expression of genes involved in stress responding. For example, in a gene profiling study designed to understand initial pollutant effects, a single 4-hour exposure to PM 2.5 was associated with gene expression changes in nearly every organ studied, including the pituitary (Thomson et al. 2007(Thomson et al. , 2013. Other research has shown that PM 2.5 down regulates glucocorticoid receptor expression in the brain, which play a key role in inhibiting HPA axis activity via negative feedback (Asaba et al. 2004;Miller et al. 2009;Jia et al. 2018). Hypothesized mechanisms through which PM 2.5 acts on the HPA axis and other extrapulmonary systems include direct effects of absorbed nano-sized particulate components, local injury and subsequent stimulation of a systemic pro-inflammatory response, and activation of the autonomic nervous system leading to systemic alterations (Snow et al. 2018). Future research, including during pregnancy when significant physiological shifts occur, that is designed to investigate these pathways will help inform our understanding of the mechanisms through which PM 2.5 acts as stressful trigger capable of stimulating the HPA axis.
Our finding of an association between PM 2.5 and cortisol at each sampling time point is consistent with our findings of a positive association with AUC G . Further, our finding of stronger associations with earlier time points is consistent with the positive associations observed with CAR and negative associations observed with slope, suggesting the PM 2.5 may act most prominently on early-day features of the cortisol rhythm. Importantly, while a range of disease states and adverse health outcomes have been associated with specific features of the diurnal cortisol rhythm, a definitive biological basis for inter and intra-individual variation in these features remains poorly understood. Nonetheless, increased cortisol output during pregnancy is particularly notable as fetal development and programming of key physiological systems is dependent on maternal cortisol supplies, with excessive levels linked to increased stress reactivity (Coe and Lubach 2005), impaired brain development (Coe and Lubach 2005), and emotional disturbances (French et al. 1999), among other adverse outcomes. Fetal exposure to excessive maternal cortisol is typically prevented through the activity of placental 11-beta-dehydrogenase-2 (11bHSD2), which metabolizes cortisol into inactive cortisone (Brown et al. 1996). Research has shown that under optimal conditions, approximately 20% of maternal cortisol crosses the placenta and that attenuation of placental 11bHSD2 activity exposes the fetus to higher levels of cortisol (Jansson and Powell 2007). While we are not aware of prior research that has examined associations between PM 2.5 and placental HSD11b2, this is an important future direction as PM 2.5 has been associated with changes in the placenta (Liu et al. 2016;Maghbooli et al. 2018;Naav et al. 2020), as well as placenta-related diseases, such as abruption (Ananth et al. 2018;Huang et al. 2021).
Strengths of this study include the prospective design and our relatively large, multiethnic and socioeconomically diverse sample, which should reasonably generalize to the broader U.S. population of pregnant individuals. We were able to control for a number of potential confounding factors and conduct sensitivity analyses to understand the influence of glucocorticoid use. We estimated residential PM 2.5 exposure using a state-of-the-art satellite-based hybrid model that incorporates data from an extensive network of ground monitors and has high predictive accuracy at a 1 Â 1 km 2 and higher resolution, which should minimize measurement error. We were also able to predict temperature at the same temporal and spatial resolution, allowing us to account for the influence of this important variable in our models. Importantly, fine particulate pollution is a complex mix of particles and gases that depends on local source contributions. Thus, the specific mixture of fine particles that our participants were exposed to may not translate to populations in regions with different exposure levels, sources and atmospheric conditions. Additionally, PM 2.5 exposure was estimated at the residential address and did not account for time spent away from home. However, we note that Americans are estimated to spend an average of 67% of time at home (Leech et al. 2002) and residential exposure levels have been linked to a range of health outcomes in prior studies. With regard to our measures of salivary cortisol, we were unable to control for certain lifestyle factors that could impact levels, such as sleep, differences in arousal and food intake. Interparticipant variability in wake-sleep cycles and the exact timing of cortisol sampling could also have reduced the precision of cortisol measurements, however, we minimized noise in the data by excluding samples outside of specified temporal collection windows and by repeating measurements across three days within a 1-week period. While we trained participants to record the exact time of sample collection in a cortisol daily diary, we did not have an objective measure of sample collection timing.
In conclusion, our findings are consistent with prior animal and human research that support an association between PM 2.5 exposure and HPA axis activation. This is notable as dysregulation of neuroendocrine stress response systems is a common feature of many chronic diseases previously linked to air pollution exposure, including those with reproductive (Bekkar et al. 2020), psychiatric (Braithwaite et al. 2019), cardiovascular (Lederer et al. 2021), and neurodegenerative (Calderon-Garciduenas et al. 2016) features and those that have been linked to prenatal programming. Future research that further elucidates the proximal events, including those related to immune system activity, linking inhaled pollutants with effects on the neuroendocrine system and broader brain regions will contribute to our understanding of how this ubiquitous environmental exposure contributes to pregnancy health as well as prenatal programming of adverse health outcomes for children.

Disclosure statement
No potential conflict of interest was reported by the authors.