Secondary Logo

Journal Logo

Original Research Article

Outdoor air pollution and anti-Müllerian hormone concentrations in the Sister Study

Gregoire, Allyson M.a; Upson, Kristenb; Niehoff, Nicole M.a; Chin, Helen B.c; Kaufman, Joel D.d; Weinberg, Clarice R.e; Sandler, Dale P.a; Nichols, Hazel B.f; White, Alexandra J.a,*

Author Information
Environmental Epidemiology: October 2021 - Volume 5 - Issue 5 - p e163
doi: 10.1097/EE9.0000000000000163

Abstract

What this study adds

The aim of this study was to evaluate the association between exposure to residential outdoor criteria air pollutants and anti-Müllerian hormone (AMH) concentrations. This research is important as outdoor air pollution is ubiquitous, and lower age-specific AMH concentrations are associated with earlier menopause, which has been associated with adverse health effects. Overall, our results support that outdoor air pollution is not strongly related to AMH in our population of older reproductive-aged women. Our paper will be of interest to the readers of Environmental Epidemiology as these findings are relevant for understanding the role of air pollution in ovarian reserve.

Introduction

Anti-Müllerian hormone (AMH) is an established marker of ovarian reserve.1 Ovarian reserve refers to the primordial follicle count in the ovaries and is indicative of a woman’s reproductive life span.2 AMH is produced by the granulosa cells of developing small antral and preantral follicles and reflects the size of the remaining primordial follicle pool in reproductive-age women.3,4 The relationship between AMH concentrations and ovarian reserve remains evident until about 5 years before menopause, at which point AMH becomes undetectable in most women.5 Lower age-specific AMH concentrations are associated with shorter time to menopause.6 Younger age of menopause has been related to higher risk of adverse health outcomes including cardiovascular disease7 and osteoporosis,8 but reduced risk of other health outcomes, such as breast cancer.9

Indoor and outdoor air pollution, as well as smoke from tobacco products, exposes individuals to a complex mixture of compounds including polycyclic aromatic hydrocarbons (PAHs).10–12 Previous studies have demonstrated that AMH is lower in smokers compared with nonsmokers.13,14 Similarly, we previously reported lower AMH concentrations among women exposed to indoor air pollution from burning wood in indoor stoves/fireplaces.14 PAH-DNA adducts have been detected in ovarian cells, suggesting that PAHs are able to reach the ovaries.15 Likewise, some previous research has supported a possible association between outdoor air pollution concentrations and both earlier age at menopause16 and lower ovarian reserve in younger reproductive-aged populations.17–19 It is unclear if this association extends to women of older reproductive age.

Our study evaluated adult residential exposure to three criteria air pollutants whose concentrations are regulated by the Environmental Protection Agency (EPA): particulate matter less than 2.5 µm in diameter (PM2.5), particulate matter less than 10 µm in diameter (PM10), and nitrogen dioxide (NO2), in relation to AMH concentrations.20 We also investigated the association between residential exposure to vehicular traffic at both the childhood and adult enrollment address in relation to adult AMH concentrations measured in blood samples collected at enrollment.

Methods

Study population

The Sister Study enrolled 50,884 women from all 50 states and Puerto Rico between 2003 and 2009.21 Briefly, women ages 35–74 who had at least one sister who had been previously diagnosed with breast cancer, but had no history of breast cancer themselves, were recruited into the study. Participants completed extensive computer-assisted telephone interviews to collect information on demographics, residential history, dietary and lifestyle factors, and existing health conditions. Baseline activities included a home visit in which a trained examiner collected a fasting blood sample and measured weight and height.21 The Sister Study was approved by the Institutional Review Boards of the National Institute of Health. Written informed consent was obtained from all participants.

AMH concentrations were measured for a subset of premenopausal women selected for a case-control study of AMH and breast cancer risk.22 Cases included all women who developed breast cancer between study enrollment and December 31, 2012 (N = 458), and a sample of participants who had not developed breast cancer by that date, matched to cases at a 2:1 ratio on age at enrollment and enrollment year (N = 916).22 All women in the case-control study were ages 35–54 at enrollment and premenopausal (defined as having at least one menstrual cycle in the past 12 months or having had a hysterectomy without a bilateral oophorectomy).22 Women selected as controls (N = 916) in the prior case-control study were eligible for inclusion in the current cross-sectional analysis; we excluded breast cancer cases given the prior association found between breast cancer case status and AMH concentrations.22 From this sample, we also excluded women with a history of polycystic ovarian syndrome (N = 27) or who were missing information on polycystic ovarian syndrome (N = 3) and those missing information on radiation or chemotherapy that caused cessation in menses before enrollment (N = 1).

Exposure assessment

Annual average concentrations of PM2.5 and NO2 in 2006 and PM10 in 2000 were estimated for each woman’s enrollment residence from a validated universal kriging regression model with spatial smoothing using air pollution regulatory monitoring data.23 These air pollution models have been previously described.23,24 Briefly, annual averages were estimated from data obtained from the air quality system (AQS) and interagency monitoring of protected visual environments (IMPROVE) networks for estimates of PM2.5 and PM10.23 Satellite information from the ozone monitoring instrument (OMI) and data from AQS were incorporated into estimates of NO2.24 Models employed partial least squares regressions using the air monitoring data and GIS-based geographic covariates to account for collinearity between covariates and to estimate pollution levels at each geocoded enrollment address.23,24 Estimates for PM2.5 and NO2 from 2006 were selected as it was the middle of the enrollment period, for PM10 there were only available estimates for the years 2000 and 2010 therefore estimates for 2000 were used as it was before the blood draw. Distance to a nearest major road in meters (A1, A2, or A3 corresponding to interstates, highways, and major roads, respectively) was determined for both the geocoded enrollment and longest-lived childhood (before age 14 years) residential addresses using roadway maps from the year 2000.25

Outcome assessment

Measurement of AMH concentrations has been previously described.22 Briefly, fasting blood serum samples collected at baseline were measured at the Reproductive Endocrinology laboratory at the University of Southern California Keck School of Medicine for AMH with the Ultrasensitive AMH ELISA kits (LOD = 0.07 ng/mL). Samples that did not have detectable levels by the Ultrasensitive ELISA were assayed using picoAMH ELISA kits (LOD = 0.003 ng/mL). We excluded 1 woman with a low-quality blood sample. For women with values of AMH below the limit of detection by the picoAMH assay (N = 245), a single value (0.0015 ng/mL) equal to half of the LOD was used.22 The distribution of AMH levels were right skewed and therefore were natural log-transformed to approximate a normal distribution.

Statistical analysis

For the primary analysis, we used linear regression to estimate the association between air pollution and natural log-transformed AMH levels. Air pollutant concentrations were examined both continuously (per interquartile range [IQR] increase) and categorically based on quartile cutpoints using the distribution of pollutant measures among women included in this analysis. Distance to roadway variables for adult and childhood residences were assessed categorically using a priori cut points (<50 m, 50–100 m, 100–200 m, ≥200 m). Percent changes per IQR increase in the exposure of interest were calculated by first exponentiating the product of the IQR and the relevant β values. Values were then transformed into percentages by subtracting 1 and multiplying by 100. The same calculation was used for percent change in categorical exposures except that the β value alone was exponentiated.26 We estimated the correlation between distance to major roadway and NO2 exposure by calculating a Pearson correlation coefficient, as these are both considered to be indicators of traffic-related air pollution exposure.

Potential confounding variables were identified a priori from the construction of directed acyclic graphs (DAGs).27 Information for covariates was obtained at baseline through a home visit conducted by a trained examiner and a computer-assisted telephone interviews. We selected two confounder adjustment sets, the first for air pollution exposure at enrollment during adulthood and the second for residential distance to major road during childhood. For all exposures, we first adjusted for age at baseline blood draw (continuous) as our “crude” model due to the high correlation between age and AMH levels in adult women. In our fully adjusted model for adulthood exposures, we further adjusted for education (≤high school, associate’s degree/technical degree/some college, bachelor’s degree, graduate degree), body mass index (BMI) calculated from baseline height and weight measurements taken by a trained examiner at baseline (continuous), and self-reported race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other). In our fully adjusted model for adulthood exposures, we further adjusted for education (≤high school, associate’s degree/technical degree/some college, bachelor’s degree, graduate degree), body mass index (BMI) calculated from baseline height and weight measurements taken by a trained examiner at baseline (continuous), and self-reported race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other). We conducted sensitivity analyses to consider confounding by area deprivation index (a continuous measure of neighborhood-level socioeconomic status),28 marital status (never married, married/living as married, widowed/divorced/separated), parity (nulliparous, 1 child, 2–3 children, 4+ children), hormonal contraceptive use at baseline (currently taking a hormonal contraceptive, not currently taking a hormonal contraceptive), smoking status (never smoker, past smoker, current smoker), and residential location type (suburban, urban, rural/small town). Potential spatial confounding was also considered with adjustment for census division of residence (New England, middle Atlantic, east north central, west north central, south Atlantic, east south central, west south central, Mountain, Pacific). We conducted a complete-case analysis and excluded women with missing covariate information (N = 1). In our fully adjusted model for childhood exposures, we included highest education of any adult in the household at age 13 (≤high school, associate’s degree/technical degree/some college, bachelor’s degree, graduate degree), relative weight at age 10 (lighter, same, heavier than peers), and race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other) and a complete-case analysis approach excluded individuals missing any covariate information (N = 4). After these exclusions, N = 883 and N = 880 women were eligible for statistical analyses.

We evaluated whether residential location type (suburban, urban, rural/small town) or distance to nearest road (<50 m, 50–100 m, 100–200 m, ≥200 m) was an effect measure modifier of the associations between adult exposure to criteria pollutants (PM2.5, PM10, NO2) and AMH. Effect measure modification was tested by using cross-product terms between the exposure and the modifier and assessing whether the interaction term significantly impacted the variance of the residuals using an F-test. We excluded individuals from the analysis who reported their residential location type as “other” as well as those who were missing information for residential location type.

We conducted sensitivity analyses excluding women who were current users of or were missing information for hormonal contraceptives (N = 99) or who were current smokers (N = 73) for ambient air pollution metrics for the adult enrollment address. Women who are current users of hormonal contraceptives may have lower AMH concentrations than women who are not current users.29,30 Smoking has been shown to increase the rate of ovarian aging and exposure to tobacco smoke has been hypothesized to overwhelm any health effects from exposure to air pollution.13

We also conducted a sensitivity analysis that included all women selected for the case-control study (N = 1,374) as well as Sister Study participants who were 35–54 at enrollment but postmenopausal (N = 5,911). Postmenopausal women were included in this study to account for the possibility of selection bias, which could arise if women who were more susceptible to the effects of air pollution on ovarian reserve experienced an earlier menopause making them ineligible for selection into the case-control study. This approach has been previously applied to an analysis of farm exposures and AMH concentrations in the Sister Study.31 Using this approach, we again excluded women with polycystic ovarian syndrome (N = 291), bilateral oophorectomy or unknown oophorectomy status (N = 2,270), radiation or chemotherapy that caused a cessation of menses (N = 24), low-quality AMH samples (N = 2), and missing covariate data (N = 2). This resulted in a sample of 4,696 women. In this sensitivity analysis, we utilized a reverse Cox regression approach32 to handle the large amount of missing AMH values (79.3% overall) arising from samples with values below the LOD (7.4%) and the inclusion of the postmenopausal women without measured AMH (71.8%). Reverse Cox regression treats these missing data points as left-censored data, presuming them to be below the LOD. Using this method, we estimated a hazard ratio (HR) and 95% confidence interval (CI) for the associations between air pollution and proximity to roadways in relation to AMH concentrations. With the reverse Cox regression, an HR > 1.0 is interpreted as an increase in AMH concentrations per one IQR increase in air pollution measures. An HR < 1.0 is interpreted as a decrease in AMH concentrations per IQR decrease in air pollution measures. We applied inverse probability weighting and included a robust variance estimator in our models to weight the selected participants back to the original cohort. The weights for controls were devised with a logistic regression model in which age at baseline, year of enrollment into the study, age at baseline squared, and the cross-products of both age variables and year of enrollment were regressed on whether an individual was selected as a control into the case-control study. The inverse of the predicted probabilities from the regression models was then applied as the weights for participants that were selected as controls. The cases from the case-control study as well as the control women who were postmenopausal at baseline were assigned a weight equal to 1.0 because all available premenopausal cases and young postmenopausal women meeting the study criteria were selected. When incorporating the sampling weights, 40.9% of participants had AMH below the LOD (79.3% in unweighted population). This sensitivity analysis allowed us to evaluate the impact of both the potential selection bias that may arise from exclusion of postmenopausal women in the original study population and to accommodate the 27.7% of nondetects in our control sample with measured AMH.

Results

In our primary controls-only study population, the median AMH concentration was 0.134 ng/mL (IQR = 0.0015–0.74 ng/mL) and the mean age was 47 years. Women tended to be highly educated (58.2% with at least a bachelor’s degree), non-Hispanic White (87.1%) and nonsmokers (91.7% never or past smoker) (Table 1). The average concentration and IQR of pollutants at the study participants’ enrollment residences were 10.62 µg/m3 (IQR = 3.5 µg/m3), 22.26 µg/m3 (IQR = 5.4 µg/m3), and 10.48 ppb (IQR = 6.1 ppb) for PM2.5, PM10, and NO2, respectively. Most women (64.9%) lived more than 200 m from the nearest major road at enrollment.

Table 1. - Study participant characteristics, Sister Study, 2003–2009
Characteristic N Mean (SD)/count (%)
AMH level (ng/mL) [median (IQR)] 883 0.134 (0.0015–0.74)
Age at blood draw 883 47.19 (4.49)
BMI (kg/m2) 883 27.22 (6.54)
NO2 (ppb) 874 10.48 (4.82)
PM25 (µg/m3) 875 10.62 (2.32)
PM10 (µg/m3) 875 22.26 (5.61)
Distance from primary adult residence at baseline  to major (A1, A2, or A3) road 883
 ≥200 m 573 (64.9%)
 100 m–≤200 m 134 (15.2%)
 50 m–≤100 m 65 (7.4%)
 <50 m 111 (12.6%)
Education 883
 ≤High school 104 (11.8%)
 Associate’s degree, technical degree or some college 265 (30%)
 Bachelor’s degree 279 (31.6%)
 Graduate degree 235 (26.6%)
Race/Ethnicity 883
 Non-Hispanic White 769 (87.1%)
 Non-Hispanic Black 57 (6.5%)
 Hispanic 34 (3.9%)
 Other 23 (2.6%)
Smoking status 883
 Never smoked 547 (61.9%)
 Past smoker 263 (29.8%)
 Current smoker 73 (8.3%)
Residential Location Type 883
 Urban 397 (45.0%)
 Suburban 152 (17.2%)
Rural/small town 329 (37.3%)
 Missing 5 (0.6%)
Hormonal contraceptive use 883
 Not currently using 784 (88.8%)
 Currently using 77 (8.7%)
 Missing 22 (2.5%)

Overall, air pollution measures were not consistently associated with AMH concentrations (Table 2). An IQR increase in PM2.5 showed little difference in AMH concentrations (1.9%; 95% CI = –17.9, 26.3). IQR increases in PM10 and NO2 were associated with estimated positive percent changes in AMH concentrations (7.4% and 12.2%, respectively) but were not statistically significant. Conversely, living closer to a major road in adulthood was inversely associated with AMH concentration (<50 m from nearest roadway vs. ≥200 m = –32.9%; 95% CI = –56.1, 2.6) (Table 3). No consistent pattern was observed for proximity of childhood residence to major roads and AMH concentrations. NO2 is a traffic-related air pollutant; however, the correlation between proximity to a major roadway and NO2 was modest and inverse (Pearson correlation coefficient = –0.32, P < 0.001).

Table 2. - Associations between adult residential air pollution exposures and AMH concentrations, Sister Study 2003–2009
Age-adjusted model Fully adjusted modela
Air pollution exposure N Percent change (%) 95% CI Percent change (%) 95% CI
PM2.5
Q1b 219f 0.0 (reference) 0.0 (reference)
Q2 219 –23.2 (–48.0, 13.4) –22.6 (–47.7, 14.4)
Q3 218 –9.4 (–38.7, 33.9) –6.8 (–37.0, 37.9)
Q4 219 –7.2 (–37.1, 37.0) 0.9 (–32.2, 50.0)
IQRe 875 –2.8 (–21.3, 20.0) 1.9 (–17.9, 26.3)
PM10
Q1c 219f 0.0 (reference) 0.0 (reference)
Q2 219 –2.3 (–33.9, 44.4) –1.1 (–33.1, 46.2)
Q3 218 15.9 (–21.5, 71.2) 16.5 (–21.2, 72.2)
Q4 219 13.5 (–23.1, 67.6) 17.0 (–21.0, 73.4)
IQRe 875 7.0 (–6.2, 22.0) 7.4 (–6.0, 22.7)
NO2
Q1d 219f 0.0 (reference) 0.0 (reference)
Q2 218 28.4 (–13.0, 89.5) 28.1 (–13.2, 89.0)
Q3 218 52.5 (3.4, 125.1) 53.8 (4.2, 126.8)
Q4 219 41.4 (–4.1, 108.5) 42.9 (–3.4, 111.4)
IQRe 874 11.3 (–6.6, 32.6) 12.2 (–6.0, 33.9)
aAdjusted for age, education, BMI, and race/ethnicity.
bQuartile cutpoints: PM2.5 (µg/m3): Q1 PM2.5 ≤ 8.8, Q2: 8.8 < PM2.5 ≤ 10.9, Q3: 10.9 < PM2.5 ≤ 12.4, Q4: PM2.5 > 12.4.
cQuartile cutpoints: PM10 (µg/m3): Q1 PM10 ≤ 19.0, Q2: 19.0 < PM10 ≤ 21.9, Q3: 21.9 < PM10 ≤ 24.3, Q4: PM10 > 24.3.
dQuartile cutpoints: NO2 (ppb): Q1 NO2 ≤ 7.0, Q2: 7.0 < NO2 ≤ 9.6, Q3: 9.6 < NO2 ≤ 13.1, Q4: NO2 > 13.1.
eIQRs: PM2.5 = 3.5 µg/m3, PM10 = 5.4 µg/m3, and NO2 = 6.1 ppb.
fNumbers after exclusion for missing exposure information, for PM2.5 and PM10 (N = 8) and NO2 (N = 9).

Table 3. - Associations between distance from residence to nearest a major road and adult AMH concentrations, Sister Study, 2003–2009
Age-adjusted Fully adjusted modela
Distance lived from major road N Percent change (%) 95% CI Percent change (%) 95% CI
Adult residence
 ≥200 m 573 0.0 (reference) 0.0 (reference)
 100 m–200 m 134 3.3 (–30.1, 52.5) 6.9 (–27.7, 58.1)
 50 m–100 m 65 –9.6 (–47.0, 54.0) –9.8 (–47.1, 53.6)
 <50 m 111 –37.7 (–59.1, –5.0) –32.9 (–56.1, 2.6)
Childhood residence
 ≥200 m 443b 0.0 (reference) 0.0 (reference)
 100 m–200 m 161 30.8 (–9.9, 89.9) 32.0 (–9.1, 91.7)
 50 m–100 m 70 –20.2 (–52.8, 34.9) –21.7 (–53.8, 32.6)
 <50 m 165 –6.4 (–35.8, 36.4) –7.2 (–36.5, 35.5)
aAdult residence estimates adjusted for age, education, BMI, and race/ethnicity; childhood residence estimates adjusted for age, highest education in household at age 13, relative weight at age 10, and race/ethnicity.
bNumbers after exclusion for missing exposure information, for childhood residence (N = 41).

The associations between adulthood enrollment air pollution measures and AMH did not notably change when excluding current users of hormonal contraceptives or current smokers (eTable 1; https://links.lww.com/EE/A149). We did see a suggestive 12.4% (95% CI = –30.1, 9.9) decrease in AMH levels for every IQR increase in PM2.5 among women who were noncurrent users of hormonal contraceptives; however, these results did not reach statistical significance. Additionally, we saw little evidence of modification of the adulthood air pollution and AMH associations by either residential location type or by distance to nearest major roadway (eTable 2; https://links.lww.com/EE/A149).

In the sensitivity analysis in the expanded study sample (including all case-control study participants and postmenopausal women between the ages of 35–54), patterns of association between the assessed pollutants and AMH levels were similar to those in our main analysis limited to controls from the original case-control study (eTable 3; https://links.lww.com/EE/A149).

Adjustment for area deprivation index, marital status, parity, hormonal contraception use at baseline, smoking status, and residential location type did not materially change results and were therefore not included in the final fully adjusted models (data not shown). There was no change in our results after adjustment for the census region of residence (data not shown).

Discussion

Overall, we observed little evidence to support an association between air pollution and AMH concentrations in our population of older reproductive-aged women. In our analysis, exposure to both estimated criteria air pollution levels at the adult residence or proximity to roadways at both the adult and childhood residence were not strongly or consistently associated with AMH concentrations. This study is one of the largest to date to evaluate both ambient air pollution and proximity to roadways in relation to AMH levels in a population of women not seeking fertility treatments. Given the widespread nature of both air pollution and traffic, achieving a better understanding of the relationship between air pollution and AMH levels is important for women’s health.

Laboratory evidence supports a role for air pollution in diminished ovarian reserve; proposed mechanisms have included increased oxidative stress as well as induction of the aryl hydrocarbon receptor.33,34 For example, exposure to PM2.5 reconstituted from particles collected in Beijing, China, led to decreased levels of AMH in adult female mice33 and maternal exposure to polycyclic aromatic hydrocarbons (PAHs) was associated with a decreased follicular pool in offspring in mice.34 Similarly, Ogliari et al. observed an inverse association between maternal exposure to diesel exhaust and markers of ovarian reserve in mice.35

There have been few epidemiologic studies examining the associations between air pollution and ovarian reserve. In a population of women undergoing treatment at a fertility clinic, Gaskins et al. (2019) found that an IQR increase (2 µg/m3) increase in residential PM2.5 was associated with a 7.2% decrease (95% CI = –10.4, –3.8) in antral follicle count.18 In our study population overall, we observed a 1.9% nonsignificant decrease in adult AMH levels (95% CI = –17.9, 26.3) for an IQR increase in PM2.5 (3.5 µg/m3). However, in an analysis limited to women not currently using hormonal contraceptives, we observed a 12.4% (95% CI = –30.1, 9.9) decrease in AMH. A recent cross-sectional study of 67 women in Iran suggested that higher levels of PM less than 1 µm in aerodynamic diameter (PM1) and PM2.5 were both associated with lower levels of AMH.17 Our study population was exposed to a much lower concentration of PM2.5 on average (10.6 µg/m3) than the women in Iran (42.4 µg/m3), which may explain, in part, some discrepancies in our findings.

We observed a suggestive, but unexpected, positive association between AMH and NO2 levels, which is a traffic-related air pollutant. Conversely, we observed living closer to a major road during adulthood was associated with lower AMH levels. A recent Italian study reported an inverse correlation (Rho = –0.111, P <0.001) between AMH concentrations and residential NO2 levels, in contrast to our findings with estimated NO2.19 In our data, there was a negative correlation between distance from adult residence to major roadway and NO2 levels at the residence, suggesting that our measured variables are capturing different patterns in NO2 exposure.25 Further studies are needed to help to clarify whether there is an impact of distance from residence to a major roadway and traffic-related air pollution on adult AMH levels.

Important strengths of this study included the ability to assess the associations between both early life and adult air pollution exposures or proxies. Our study used a validated AMH assay and validated air pollution exposure models to estimate annual average exposure at each individuals’ enrollment residential location. Our study population was not affiliated with seeking infertility care and thus may have more generalizable AMH concentrations. Although we were able to consider residential proximity to traffic during childhood, we did not have any information on participants’ mothers’ exposure during their gestation, which may be a critical window of exposure when the primordial follicle pool is established.36 As with most epidemiologic studies of residential air pollution, there is potential for misclassification of exposures due to the inability to account for air pollution exposure while women are at work, indoors, or during other large periods of time spent away from the home. Additionally, our estimates for distance lived to roadway were determined based on roadway maps from the year 2000, which may not be an accurate representation of childhood exposure in this population.

We also considered the potential for selection bias by including postmenopausal women in a sensitivity analysis using an inverse probability weighting scheme and reverse Cox proportional hazards regression. Our overall results were similar when including these additional women, suggesting that there was little selection bias resulting from women most susceptible to air pollution exposure undergoing an earlier age at menopause and thus are not eligible for inclusion in the original case-control population. This lack of selection bias is not particularly surprising given that air pollution appeared to have little effect on AMH concentrations in this population. This sensitivity analysis also gave us the opportunity to address the proportion <LOD among our main analysis study population by using another method to accommodate this missing data. The consistency in our findings suggests that the single imputation for values below the LOD used in our main statistical models did not substantially affect results.

We measured AMH concentrations in a sample of women of older reproductive age. Therefore, median AMH concentrations were lower, which may have made it more difficult to detect any possible associations with air pollution. Additionally, as AMH declines across the life course from age 25 on, this may not be the most suitable population from which to draw conclusions about the relationship between air pollution and reproductive health.5 However, as our participants are largely past their prime reproductive years, the results observed here may be more interpretable as informing the relationship between air pollution, AMH concentrations, and menopause.

We were unable to consider smaller windows of time closer to the blood draw, which may more closely align with follicular development37 as seen in prior studies of ovarian reserve and air pollution.18,19 However, as we were more interested in chronic exposure to air pollution on ovarian reserve depletion as opposed to the acute effects on a single event, annual averages serve as an appropriate measure of pollution exposure. Additionally, women in this population have relatively moderate exposures to air pollution and it is possible that any effects were too modest to detect.

In our study population, both air pollution exposure measures and proximity to roadways in childhood and adulthood were not consistently associated AMH levels in older reproductive-age US women. More research is needed to understand whether air pollution may play a role in women’s ovarian reserve, particularly considering exposure during other developmental life stages or at higher exposure levels.

References

1. Visser JA, de Jong FH, Laven JS, Themmen AP. Anti-Müllerian hormone: a new marker for ovarian function. Reproduction. 2006;131:1–9.
2. Broekmans FJ, Visser JA, Laven JS, Broer SL, Themmen AP, Fauser BC. Anti-Müllerian hormone and ovarian dysfunction. Trends Endocrinol Metab. 2008;19:340–347.
3. Dewailly D, Andersen CY, Balen A, et al. The physiology and clinical utility of anti-Mullerian hormone in women. Hum Reprod Update. 2014;20:370–385.
4. Hansen KR, Hodnett GM, Knowlton N, Craig LB. Correlation of ovarian reserve tests with histologically determined primordial follicle number. Fertil Steril. 2011;95:170–175.
5. Kelsey TW, Wright P, Nelson SM, Anderson RA, Wallace WH. A validated model of serum anti-müllerian hormone from conception to menopause. PLoS One. 2011;6:e22024.
6. Broer SL, Eijkemans MJ, Scheffer GJ, et al. Anti-Mullerian hormone predicts menopause: a long-term follow-up study in normoovulatory women. J Clin Endocrinol Metab. 2011;96:2532–2539.
7. Quinn MM, Cedars MI. Cardiovascular health and ovarian aging. Fertil Steril. 2018;110:790–793.
8. Lane NE. Epidemiology, etiology, and diagnosis of osteoporosis. Am J Obstet Gynecol. 2006;194(2 suppl):S3–11.
9. Rojas K, Stuckey A. Breast cancer epidemiology and risk factors. Clin Obstet Gynecol. 2016;59:651–672.
10. Cincinelli A, Martellini T. Indoor air quality and health. Int J Environ Res Public Health. 2017;14:E1286.
11. Fromme H, Lahrz T, Piloty M, Gebhardt H, Oddoy A, Rüden H. Polycyclic aromatic hydrocarbons inside and outside of apartments in an urban area. Sci Total Environ. 2004;326:143–149.
12. Turner MC, Andersen ZJ, Baccarelli A, et al. Outdoor air pollution and cancer: an overview of the current evidence and public health recommendations. CA Cancer J Clin. 2020;70:460–479.
13. Plante BJ, Cooper GS, Baird DD, Steiner AZ. The impact of smoking on antimüllerian hormone levels in women aged 38 to 50 years. Menopause. 2010;17:571–576.
14. White AJ, Sandler DP, D’Aloisio AA, et al. Antimüllerian hormone in relation to tobacco and marijuana use and sources of indoor heating/cooking. Fertil Steril. 2016;106:723–730.
15. Zenzes MT, Puy LA, Bielecki R. Immunodetection of benzo[a]pyrene adducts in ovarian cells of women exposed to cigarette smoke. Mol Hum Reprod. 1998;4:159–165.
16. Li H, Hart JE, Mahalingaiah S, Nethery RC, Bertone-Johnson E, Laden F. Long-term exposure to particulate matter and roadway proximity with age at natural menopause in the Nurses’ Health Study II Cohort. Environ Pollut. 2021;269:116216.
17. Abareshi F, Sharifi Z, Hekmatshoar R, et al. Association of exposure to air pollution and green space with ovarian reserve hormones levels. Environ Res. 2020;184:109342.
18. Gaskins AJ, Fong KC, Abu Awad Y, et al. Time-varying exposure to air pollution and outcomes of in vitro fertilization among couples from a fertility clinic. Environ Health Perspect. 2019;127:77002.
19. La Marca A, Spaggiari G, Domenici D, et al. Elevated levels of nitrous dioxide are associated with lower AMH levels: a real-world analysis. Hum Reprod. 2020;35:2589–2597.
20. Environmental Protection Agency. Criteria Air Pollutants. Available at: https://www.epa.gov/criteria-air-pollutants. Accessed March 21, 2021.
21. Sandler DP, Hodgson ME, Deming-Halverson SL, et al.; Sister Study Research Team. The sister study cohort: baseline methods and participant characteristics. Environ Health Perspect. 2017;125: 127003.
22. Nichols HB, Baird DD, Stanczyk FZ, et al. Anti-Mullerian hormone concentrations in premenopausal women and breast cancer risk. Cancer Prev Res (Phila Pa). 2015;8:528–534.
23. Sampson PD, Richards M, Szpiro AA, et al. A regionalized national universal Kriging model using Partial Least Squares regression for estimating annual PM2.5 concentrations in epidemiology. Atmos Environ (1994). 2013;75:383–392.
24. Young MT, Bechle MJ, Sampson PD, et al. Satellite-based NO2 and model validation in a national prediction model based on Universal Kriging and land-use regression. Environ Sci Technol. 2016;50: 3686–3694.
25. Allen RW, Davies H, Cohen MA, Mallach G, Kaufman JD, Adar SD. The spatial relationship between traffic-generated air pollution and noise in 2 US cities. Environ Res. 2009;109:334–342.
26. DuPré NC, Hart JE, Holmes MD, et al. Particulate matter and traffic-related exposures in relation to breast cancer survival. Cancer Epidemiol Biomarkers Prev. 2019;28:751–759.
27. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008;8:70.
28. Kind AJ, Jencks S, Brock J, et al. Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med. 2014;161:765–774.
29. Bentzen JG, Forman JL, Pinborg A, et al. Ovarian reserve parameters: a comparison between users and non-users of hormonal contraception. Reprod Biomed Online. 2012;25:612–619.
30. Johnson LN, Sammel MD, Dillon KE, Lechtenberg L, Schanne A, Gracia CR. Antimüllerian hormone and antral follicle count are lower in female cancer survivors and healthy women taking hormonal contraception. Fertil Steril. 2014;102:774–781.e3.
31. Upson K, Weinberg CR, Nichols HB, et al. Early-life farm exposure and ovarian reserve in a U.S. cohort of women. Epidemiology. 2021;32:672–680.
32. Dinse GE, Jusko TA, Ho LA, et al. Accommodating measurements below a limit of detection: a novel application of Cox regression. Am J Epidemiol. 2014;179:1018–1024.
33. Gai HF, An JX, Qian XY, Wei YJ, Williams JP, Gao GL. Ovarian damages produced by aerosolized fine particulate matter (PM2.5) pollution in mice: possible protective medications and mechanisms. Chin Med J (Engl). 2017;130:1400–1410.
34. Jurisicova A, Taniuchi A, Li H, et al. Maternal exposure to polycyclic aromatic hydrocarbons diminishes murine ovarian reserve via induction of Harakiri. J Clin Invest. 2007;117:3971–3978.
35. Ogliari KS, Lichtenfels AJ, de Marchi MR, Ferreira AT, Dolhnikoff M, Saldiva PH. Intrauterine exposure to diesel exhaust diminishes adult ovarian reserve. Fertil Steril. 2013;99:1681–1688.
36. Broekmans FJ, Soules MR, Fauser BC. Ovarian aging: mechanisms and clinical consequences. Endocr Rev. 2009;30:465–493.
37. Fleming R, Seifer DB, Frattarelli JL, Ruman J. Assessing ovarian response: antral follicle count versus anti-Müllerian hormone. Reprod Biomed Online. 2015;31:486–496.
Keywords:

anti-Müllerian hormone; air pollution; ovarian reserve; cohort studies; traffic-related air pollutants

Supplemental Digital Content

Copyright © 2021 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of The Environmental Epidemiology. All rights reserved.