BACKGROUND: Uterine leiomyomata (UL), hormone-dependent neoplasms, are a major source of gynecologic morbidity. Metals are hypothesized to influence UL risk through endocrine disruption, and their effects may vary by vitamin D status.
OBJECTIVE: We estimated associations of a metal mixture with incident UL, overall and by vitamin D status.
METHODS: We analyzed data from the Study of Environment, Lifestyle and Fibroids, a Detroit-area prospective cohort study of 1,693 black women 23-35 years of age. We measured concentrations of 17 metals/metalloids in whole blood and 25-hydroxyvitamin D [25(OH)D] in serum collected at baseline (2010-2012). Participants underwent ultrasonography at baseline and after 20 months to detect UL. We used Bayesian kernel machine regression to estimate adjusted associations (β) of the metal mixture with probit of incident UL. We also ran Cox regression models with interaction terms to estimate incidence rate ratios (IRR) by vitamin D status.
RESULTS: Among 1,132 UL-free participants at baseline, 832 (73%) had vitamin D deficiency [25(OH)D <20 ng/mL] and 117 (10%) developed UL within 20 months. Increasing all metals from their 50th to 75th percentiles was weakly positively associated with UL overall [β = 0.06; 95% credible interval (CrI): -0.03, 0.16] and among vitamin D-deficient participants (β=0.13; 95% CrI: 0.01, 0.24), driven by cadmium (overall and vitamin D-deficient) and mercury (vitamin D-deficient only). Increasing cadmium from its 25th to 75th percentile was positively associated with UL overall (β= 0.03; 95% CrI: -0.05, 0.11) and among vitamin D-deficient participants (β = 0.13, 95% CrI: 0.02, 0.24). In Cox models, cadmium [IRR = 1.55; 95% confidence interval (CI): 1.07, 2.24, per 1-unit increase in standardized concentration] and mercury (IRR = 1.38; 95% CI: 0.99, 1.92) were positively associated with UL among vitamin D-deficient participants.
Discussion: The metal mixture was positively associated with incident UL, but the association was weak and imprecise. We observed a stronger association among vitamin D-deficient participants that was driven by cadmium and mercury. https://doi.org/10.1289/EHP15218
Introduction
Uterine leiomyomata (UL) (also called "fibroids") are noncancerous smooth muscle tumors of the uterus. UL are clinically diagnosed in ~ 30% of reproductive-age women1,2 and represent the most common indication for hysterectomy.3 UL can cause symptoms such as heavy menstrual bleeding and pain and sequelae such as iron-deficiency anemia, pregnancy complications, and infertility.5 Research consistently indicates that estrogen and progesterone influence UL development.6,7 Given that UL are hormone-responsive tumors, researchers hypothesize that exposure to endocrine-disrupting chemicals could influence UL risk.8,9
Metals and metalloids (hereafter referred to as "metals") have been implicated in the etiology of UL due to their endocrine-disrupting and tumorigenic properties. Metals including antimony, barium, cadmium, cobalt, copper, lead, mercury, and nickel are classified as metalloestrogens because they can bind to and activate the estrogen receptor.10 Animal studies have shown that metals (e.g., cadmium, mercury) can influence the synthesis and secretion of sex steroids.11-14 Cadmium has been implicated in UL growth in in vitro experiments: Low-dose cadmium exposure is associated with increased proliferation of human UL cells through nongenomic pathways. 15,16
However, few epidemiologic studies have analyzed metal biomarkers in relation to UL.17-21 In cross-sectional studies, positive associations were observed between mercury, 17,21 cadmium,18 and lead18 measured in blood and cobalt measured in urine18 and prevalence of UL. These studies were limited by the analysis of self-reported diagnosis of UL,17,21 clinically diagnosed UL,19,20 or surgical UL cases,18 which represent only a fraction of UL cases due to variability in symptom severity, access to care, and patient and physician preferences regarding disease management. Furthermore, given that UL can take years to develop and come to clinical attention whereas metals have relatively short half-lives in blood,22 blood metal concentrations at the time of UL diagnosis or later may not be etiologically relevant for tumor initiation. Finally, previous research (with the exception of Zhang et al.2) has not used an analytical approach that accounts for collinearity, confounding, and interactions among multiple exposures, which is important because metals often co-occur and biologically interact.23
Micronutrients, including vitamin D, can interact biologically with metals.24 Vitamin D concentrations may influence the absorption and distribution of toxic and essential metals in the body. 25,26 For example, bone loss due to vitamin D deficiency may result in increased release of lead from bones into the bloodstream. Researchers hypothesize that vitamin D may counteract the effects of toxic metals by decreasing oxidative stress.24 Furthermore, vitamin D deficiency-25-hydroxyvitamin D [25(0H)D] <20 ng/mL per Endocrine Society guidelines27'-is a hypothesized risk factor for UL occurrence and growth,28 with evidence from in vitro,29 in vivo,30 and epidemiologic studies.31-34 Given its relevance to UL and its potential ability to influence absorption and distribution of metals in the body, we hypothesized that vitamin D status modifies the association between metals and UL incidence. Specifically, we hypothesized that the effects of toxic metals on UL incidence would be stronger among individuals with vitamin D deficiency. There have been no previous studies of this interaction.
In the US, black individuals are disproportionately affected by toxic metal exposures, vitamin D deficiency, and UL. A conceptual diagram depicting how structural racism could shape disparities in toxic metal exposures and UL incidence is shown in Figure S1.35-38 Residential segregation and the disproportionate siting of roads, waste, and polluting industries near communities of color result in higher exposure to environmental hazards, including toxic metals, in predominantly black neighborhoods.39-42 Reproductive-age black women have higher concentrations of exposure biomarkers for cadmium,43,44 mercury,45,46 lead,44,46 arsenic,46 antimony,44 and multi-metal indices45 compared with those of their white counterparts. These exposures may contribute to racial disparities in reproductive health outcomes, including UL: Black individuals experience higher cumulative incidence of UL,47 more severe symptoms,48,49 lower satisfaction with treatment,50 higher rates of hysterectomy,51 and less access to minimally invasive surgery52,53 compared with white individuals in the US. As discussed above, vitamin D deficiency is a hypothesized risk factor for UL; given that the prevalence of vitamin D deficiency is substantially higher among US black adults (82%) than white adults (31%),54 researchers have hypothesized that vitamin D deficiency may contribute to the racial disparity in UL incidence.28,31 Understanding the influence of metal exposures on UL incidence, and the role of vitamin D deficiency in modifying these effects, could inform strategies to reduce disparities in UL among black individuals. While we do not expect that the biological effects of metals on UL would differ across racial or ethnic groups, black individuals represent both a highly exposed and highly relevant population in which to evaluate this research question.
To address this knowledge gap, we estimated the associations between a mixture of metals and UL incidence and assessed the potential modifying role of vitamin D status using a prospective study design and standardized serial transvaginal ultrasound screening in a cohort of reproductive-age black individuals. As the first prospective epidemiologic study using a mixture approach to investigate metal exposures and incident UL, we chose to analyze a large panel of metals (n= 17) to investigate understudied exposures, control for co-pollutant confounding, and evaluate potential interactions between metals.
Methods
Study Design
Study of Environment, Lifestyle and Fibroids (SELF) is a prospective cohort study designed to identify determinants of UL incidence and growth.55 We enrolled participants (n = 1,693) in 2010-2012. Study design and participants have been described previously.55 Briefly, eligible participants self-identified as black/ African American, had an intact uterus with no prior clinical diagnosis of UL, and were 23-35 years of age and residents of the Detroit, MI area at enrollment. The study was conducted among black/African-American individuals because of their early average onset of and high health burden from UL.55-57 Individuals with a history of cancer or autoimmune disease that required treatment were excluded. The institutional review boards of Henry Ford Health, the National Institute of Environmental Health Sciences, and Boston University Medical Campus approved the study. All participants provided written informed consent.
At the baseline clinic visit, participants completed computerassisted web and telephone interviews, underwent transvaginal ultrasonography for detection of UL, and provided biospecimens including a blood sample. Participants were invited to complete four follow-up visits during ~ 10 years. The four follow-up visits occurred at intervals of ~ 20, 40, 60, and 120 months; hereafter, the first interval is referred to as "20 months." Pregnant participants were asked to delay their visits until 3-4 months postpartum.
Participants
Among 1,693 enrolled participants, we excluded 385 with prevalent UL identified at the baseline ultrasound visit (Figure S2). We excluded 18 participants whose blood sample had insufficient volume for analysis of metals and 55 participants who did not attend at least one follow-up visit. We analyzed incident UL during the first 20 months of follow-up as the primary analysis, to allow for a shorter hypothesized time from exposure to presence of a detectable UL and because metals measured in whole blood generally reflect recent exposures (ranging from days to months).22 For the primary analysis, we excluded 103 participants who did not attend the 20-month visit, resulting in 1,132 participants eligible for analysis. Sixteen participants who attended the 20-month visit but did not have usable ultrasound data were analyzed as noncases because 90% of those with adequate data did not develop UL during the interval. To assess the potential for a longer time from exposure to presence of a detectable UL, we conducted secondary analyses defining the outcome as any incident UL during 10 years of followup and included 1,235 participants who did not have prevalent UL at baseline and attended at least one follow-up visit over 10 years.
Whole Blood Metal Concentrations
At the baseline clinic visit, 1,664 (98%) participants provided nonfasting whole blood samples, including 6 mL of whole blood collected into royal blue Vacutainers and stored at -80°C at Henry Ford Health for batch shipping on dry ice to the National Institute for Environmental Health Sciences as described previously. 58 All samples underwent one freeze-thaw cycle for measurement of a small subset of metals. Subsequently, all samples were shipped on dry ice to the Icahn School of Medicine at Mt. Sinai, where the samples were thawed and assayed for 20 metals: aluminum, antimony, arsenic, barium, cadmium, cesium, chromium, cobalt, copper, lead, magnesium, manganese, mercury, molybdenum, nickel, selenium, thallium, tin, vanadium, and zinc. In the present report, we limited our analysis to 17 metals that were above the limit of detection (LOD) in >60% of samples (excluding aluminum, thallium, and tin).
We assayed all metals except mercury in a multielement panel using an inductively coupled plasma mass spectrometer-triple quadrupole (Agilent 8800-QQQ). We measured total mercury using Direct Mercury Analyzer-80 (Milestone Inc.). We calculated daily LOD values as three times the standard deviation of the concentrations in the method blank (Table 1). Additional details regarding preparation and quality control have been published.58
Vitamin D Assessment with 25(OH)D
Detailed methods regarding the measurement of 25(OH)D concentrations in SELF have been described previously.33 In the present analysis, we analyzed 25(OH)D from samples collected at baseline. Assays were conducted at Heartland Laboratories (Ames, IA) using the LIAISON 25 OH Vitamin D TOTAL Assay, a competitive chemiluminescence immunoassay.59,60 In a subset of samples, we demonstrated high concordance between the immunoassay and liquid chromatography-tandem mass spectrometry, the gold standard.33 We applied seasonal adjustment to 25(OH)D concentrations as described by Harmon et al.33 and categorized concentrations using clinically relevant cut points defined by the Endocrine Society27 as deficient (<20 ng/mL, 73%), insufficient (20 to <30 ng/mL, 20%), or sufficient (>30 ng/mL, 6%). Due to the low prevalence of vitamin D sufficiency, we dichotomized 25(OH)D concentrations as <20 vs. >20 ng/mL.
UL Ascertainment
Experienced sonographers who were trained on the standardized study protocol conducted transvaginal ultrasonography at the baseline and all four follow-up visits to identify UL >0.5 cm in diameter.55 Transvaginal ultrasonography has high sensitivity (99%) and specificity (91%) for detection of UL compared with the gold standard of histologic confirmation.61
Covariates
We identified potential confounders using a causal diagram (Figure S3),62 informed by literature review, subject matter expertise, and previous analyses within SELF.58,63-66 At baseline, participants reported on sociodemographic factors (age, educational attainment, household income, and employment status), behavioral and environmental exposures (smoking status, environmental smoke exposure, and alcohol intake), and reproductive and contraceptive history [parity, breastfeeding history, and history of depot medroxyprogesterone acetate (DMPA) use] through computer-assisted web and telephone interviews. We geocoded participants" residential addresses provided at baseline to ascertain urbanicity (categorized as urban or urban cluster vs. suburban or rural using the National Center for Education Statistics classifications67) and neighborhood disadvantage (operationalized using the 2015 Neighborhood Atlas Area Deprivation Index, a census block group-level measure of within-state disadvantage68). We measured participants' height and weight at the baseline clinic visit and calculated body mass index (BMI) as weight (kg) divided by height (meters) squared. Participants provided data on past-year dietary intakes in terms of frequency and quantity through a Block Food Frequency Questionnaire adapted to a web-based format69; we used these data to calculate past-year average intakes of water and fish (potential sources of metal exposure).58 Fish intake included oyster, shellfish, tuna, fried fish, and other fish (queried separately). Questions on water intake did not differentiate between tap vs. botted or between municipal water supply systems vs. private wells.
To avoid data sparsity, we adjusted for a reduced set of confounders (measured at baseline) that are strongly associated with metal concentrations58,70 and incidence of UL in SELF: age (continuous),71 BMI (<30, 30-34, 35-39, >40 kg/m2, modeled as nominal categorical variables)66, parity (nulliparous vs. parous), use of DMPA within the past 2 years (yes vs. no),64 and current smoking (yes vs. no). Among SELF participants, the incidence of UL increases monotonically with age,71 whereas parity, recent DMPA use,64 and current smoking65 are inversely associated with UL incidence, and BMI is nonlinearly associated with UL (highest risk observed among individuals with a BMI of 30 to <35 kg/m2).66 We did not consider multivitamin use as a potential confounder because it was not appreciably associated with any blood metal concentrations in SELF.58 Our adjustment for parity was informed by previous analyses in SELF that identified meaningfully different metal concentrations according to parity (any births vs. none) but not according to time since last birth (<2 vs. >2 years).58
Descriptive Statistics
We analyzed machine-read values for metal concentrations below the LOD.72 We replaced negative values (resulting from the subtraction of the method blank from the measured sample concentration) with the metal's daily LOD divided by the square root of 2 prior to log transformation.73 As described previously, we replaced mercury values <0.0025 with the LOD divided by the square root of 2 to ensure that the normality assumption for regression modeling was met.58 In order to compare metals on the same scale, we standardized metal concentrations by applying natural log transformation and then subtracting the mean and dividing by the standard deviation of the respective metal concentration on the natural log scale.
We calculated descriptive statistics for participant characteristics and metal concentrations at baseline, overall and by strata of 25(OH) D. We calculated pairwise Spearman correlation coefficients among metal concentrations. To assess the potential for selection bias, we visually assessed boxplots of metal concentrations at baseline according to inclusion in the analysis (primary and secondary analysis, secondary analysis only, excluded due to prevalent UL at baseline, or excluded due to not attending at least one follow-up visit). We also compared metal concentrations according to strata of 25 (OH)D and history of cigarette smoking (ever vs. never) at baseline.
BKMR
We applied Bayesian kernel machine regression (BKMR) with a probit link to estimate the association of the metal mixture with incident UL.74,75 BKMR allows for flexible modeling of multiple exposures and their interactions in relation to a health outcome while adjusting for specified covariates.74 We used component-wise variable selection, which provides a measure of relative importance for every component of the mixture. For the primary analysis, we used the R package bkmrhat76 to run 3 parallel chains of 50,000 iterations each. We assessed convergence by checking the Gelman-Rubin statistics77 and trace plots of the covariates. For secondary analyses, we used the R package bkmr75 to run 40,000 iterations and assessed convergence by checking covariate trace plots. We used the default noninformative prior specifications, used the first 50% of iterations from each chain as burn-in, and selected every SOth iteration for analysis to reduce autocorrelation.
For the primary analysis, we used BKMR to estimate the associations of the metal mixture with probit of incident UL during 20 months of follow-up. We estimated the overall mixture association, defined as the mean difference in probit of UL associated with percentile values of all metals (ranging from 25th to 75th by 5-percentile increments), compared with the median values of all metals. We estimated the posterior inclusion probabilities (PIPs) for each metal, which range from 0 to 1 and are interpreted as the relative posterior probability that a given metal is an important component of the mixture in relation to incident UL.74 We estimated the univariate associations of metals with incident UL, defined as the mean difference in probit of UL associated with increasing an individual metal from its 25th to 75th percentile, holding other metals constant at their medians. We used the approximation βlogit = 1.6 X βprobit to convert probit estimates to the odds ratio (OR) scale."® We visualized the posterior estimates and 95% credible intervals (СП) for associations of standardized metal concentrations with probit of incident UL while holding other metals constant at their median values ("univariate exposure-response plots"). To assess potential pairwise interactions between metals, we plotted the associations of each metal with probit of incident UL while holding a second metal constant at its 25th, 50th, and 75th percentile values and holding all other metals constant at their median values ("bivariate exposure-response plots").
We repeated BKMR analysis restricted to participants with vitamin D deficiency [25(0H)D <20ng/mL]. Due to the high prevalence of vitamin D deficiency (73%), there were too few participants in the 25(OH)D >20ng/mL stratum to conduct BKMR analysis. To further reduce potential confounding by cigarette smoking (which is positively associated with blood cadmium and inversely associated with UL incidence in SELF58,65), we repeated BKMR analysis restricted to participants who reported never smoking cigarettes on a regular basis ("never smokers"; n = 835) at baseline.
Quantile g-Computation
As an alternative statistical mixture method, we used quantile g-computation79 to estimate the overall association of the metal mixture with risk of incident UL and to quantify the relative contribution (i.e., weight) of each metal. We ran quantile g-computation with 500 bootstraps to estimate risk ratios (RR), modeling exposure as quartiles, and repeated the analysis without bootstraps to estimate the weights of metals. Quantile g-computation analyses for the overall mixture association were conducted for incidence of UL during 20 months and 10 years of follow-up. To assess the potential for different associations by vitamin D status, we repeated quantile g-computation for incidence of UL during 20 months of follow-up stratified by 25(OH)D <20 ng/mL vs. >20 ng/mL. We adjusted quantile g-computation models for age (continuous), BMI (<30, 30-34, 35-39, and >40 kg/m"), parity (nulliparous vs. parous), use of DMPA within the past 2 years (yes vs. no), and current smoking (yes vs. no). Due to the smaller number of participants in the >20 ng/mL stratum (п = 300), we removed DMPA use from the model to improve model convergence.
Cox Interaction Models
To further assess potential interaction by vitamin D status, we ran 17 Cox proportional hazards regression models, each with main effect terms for all 17 log-transformed standardized metal concentrations (continuous) and vitamin D status [25(OH)D <20 ng/mL vs. >20 ng/mL], and an interaction (cross product) term for one metal and vitamin D status, to estimate incidence rate ratios (IRR) and 95% confidence intervals (CIs) for associations of metals with incident UL by vitamin D status. We used time on study (months) as the time variable and stratified by age (years) to allow for nonproportional hazards by age. We adjusted for BMI (<30, 30-34, 35-39, >40 kg/m"), parity (nulliparous vs. parous), use of DMPA Within the past 2 years (yes vs. no), and current smoking (yes vs. no). To provide further control of confounding, we conducted a sensitivity analysis in which Cox models were additionally adjusted for alcohol intake (<6, 7-13, or >14 drinks/week), fish intake (<113, 114-226, or >227 g/week), educational attainment (< high school, some college/associate's/technical, or > college graduate), and household income (<$20,000, $20,000 to <$50,000, or >$50,000). Finally, as a post hoc analysis, we ran separate Cox interaction models (operationalized as described above) in subgroups defined by age at baseline, <30 vs. >30 years, to assess the potential for effect measure modification by age. We focused on cadmium and mercury for this post hoc analysis because of their importance across analyses as described in the Results section.
Missing Data
Mercury concentration was missing for one participant due to insufficient sample volume; otherwise, all participants had complete data for metal concentrations. One participant was missing 25(OH)D. There were no missing data for potential confounding variables included in the BKMR analysis (age, BMI, parity, use of DMPA within the past 2 years, and current smoking). Missing data for other potential confounders were low (n missing: educational attainment, n=1; household income, n= 11; urbanicity, n = 19; neighborhood disadvantage, n = 36; remaining covariates, n=0). We multiply imputed missing values for mercury, 25(0H) D, and covariates. We used multiple imputation with fully conditional specification in SAS to create 20 imputed datasets and analyzed the first imputed dataset because methods to combine multiply imputed data across datasets have not yet been developed for BKMR with a probit link. Variables included in the multiple imputation model are listed in Table S1.
Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc.) and R version 4.2.2 (R Foundation for Statistical Computing) software. Consistent with recent statistical guidance,80 we did not use statistical significance to guide our conclusions. Rather, we based our interpretation on the direction, magnitude, and precision of results.
Results
Descriptive Statistics
The cumulative incidence of UL was 117/1,132 (10%) during the first 20 months of follow-up and 389/1,235 (31%) during 10 years of follow-up. At baseline, participants had a median age of 28.8 years and median BMI of 32:2 kg=m2 (Table 2). Twenty-six percent completed a college degree or higher, and 46% had an annual household income of <$20,000. Nineteen percent were current smokers. The majority (64%) of participants were parous, and 12% used DMPA in the past 2 years. Most (73%) participants had vitamin D deficiency [25(OH)D <20 ng=mL]. Compared to those with vitamin D deficiency, participants with 25(OH)D >20ng/mL had a higher median age, lower median BMI, and higher educational attainment and household income. Descriptive statistics for participants who completed 10-years of follow-up were similar to those in the primary analysis (Table S2).
Descriptive statistics for metal concentrations are shown in Table 1. Pairwise Spearman correlations were weak to moderate (ranging from -0.37 to 0.49) (Figure 1) and were generally positive. The strongest correlations were observed for cadmium and lead [Spearman's r (rs) = 0.49], arsenic and mercury (rs = 0.44), chromium and vanadium (rs = 0.42), barium and nickel (rs = 0.41), and chromium and nickel (rs = 0.41). The strongest negative correlation was between molybdenum and nickel (rs = - 0.37). Descriptive statistics and pairwise correlations were similar in the n=1,235 cohort for analysis of cumulative incidence during 10 years (Table S3; Figure S4).
Metal concentrations at baseline were similar among participants included in the primary analysis, those included only in the secondary analysis of 10-year follow-up, those excluded due to prevalent UL at baseline, and those excluded due to not attending at least one follow-up visit (Figure S5). Metal concentrations were generally similar between strata of 25(OH)D; median concentrations of arsenic, cesium, and mercury were slightly lower among participants with a 25(OH)D of <20 ng/mL (Figure S6). As expected, median concentrations of cadmium and lead were higher among ever vs. never smokers (Figure S7).
BKMR
We confirmed that BKMR models converged by assessing convergence diagnostics: The Gelman-Rubin statistics for covariates were <1.05 (Table S4), and the covariate trace plots indicated convergence (Figure S8). The mixture was weakly, monotonically associated with increased probit of incident UL (Figure 2; Table S5). For example, the association with probit of UL when all metals were set to their 75th vs. 50th percentile value was β = 0.06 (95% CrI: -0.03, 0.16), corresponding to an OR of 1.11 (95% Cri: 0.95, 1.29). Cadmium had the strongest contribution to the mixture with respect to probit of UL (PIP = 0.13), followed by barium (PIP - 0.07), cobalt (PIP - 0.04), and zinc (PIP = 0.03) (Table S6). Univariate associations of metals with incident UL were weak. For example, increasing cadmium from its 25th to its 75th percentile value was associated with a mean increase of β=0.03 (95% CrI: -0.05, 0.11; OR = 1.05, 95% CrI: 0.93, 1.20) in probit of incident UL (Table S7). Univariate exposure-response plots of metals with probit of incident UL did not show strong evidence of nonlinear associations (Figure 3). Bivariate exposure- response plots did not show evidence of meaningful interactions between metals (Figure S9).
Among 832 participants with vitamin D deficiency at baseline, we observed a positive monotonic association between the metal mixture and probit of incident UL (Figure 4; Table S8). The overall mixture association in this subgroup appeared stronger than that observed in the primary analysis among all participants. For example, the association with probit of UL when all metals were set to their 75th vs. 50th percentiles was β = 0.13 (95% CrI: 0.01, 0.24; OR = 1.23, 95% CrI: 1.02, 1.47). Cadmium had the strongest contribution to the mixture (PIP=0.43), followed by mercury (PIP = 0.06) and barium (PIP = 0.04) (Table S9). Increasing cadmium from its 25th to its 75th percentile value was associated with a mean increase of β = 0.13 (95% CrI: 0.02, 0.24; OR = 1.23, 95% Crl: 1.02, 1.47) in probit of incident UL (Table S10). Associations of other metals with UL were weak or null (Table S10). Similarly, the univariate exposure-response plot for cadmium showed a strong, positive, monotonic association, whereas associations of other metals with UL were weak or null (Figure 5). We did not observe evidence of meaningful interactions between metals in bivariate exposure-response plots (Figure S10).
The overall association of the metal mixture with probit of incident UL during 10 years of follow-up was close to null (Figure S11; Table S11). The strongest contributors to the association were molybdenum (PIP = 0.15) and nickel (PIP - 0.07) (Table S12). Molybdenum was inversely associated with probit of incident UL, whereas nickel was positively associated with probit of incident UL during 10 years of follow-up (Figure S12; Table S13). We did not observe evidence of meaningful interactions between pairs of metals in bivariate exposure-response plots (Figure S13).
Results were similar to those of the primary analysis when we restricted the study population to never smokers (Figures S14- S16; Tables S14-S16) and after varying the prior specification of the b parameter (Figures S17-S19; Tables S17-S19). Namely, we observed weak, positive, monotonic associations of the overall mixture with probit of incident UL, to which cadmium was the strongest contributor, without meaningful evidence of nonlinear associations or pairwise interactions. The overall mixture associations and univariate associations in these sensitivity analyses were similar in magnitude to the main analysis.
Quantile g-Computation
Using quantile g-computation, we observed an imprecise, positive association of the metal mixture with risk of incident UL during 20 months of follow-up (RR=1.21, 95% CT: 0.81, 1.80 for a 1-quartile increase in metal concentrations). Cobalt, barium, zinc, and mercury had the largest contributions in the positive direction, while molybdenum and arsenic had the largest contributions in the inverse direction (Figure S20A; Table S20). Quantile g-computation results in the full cohort were similar with and without adjustment for DMPA use.
Among participants with vitamin D deficiency at baseline, a 1-quartile increase in metal concentrations was associated with 51% higher risk of incident UL during 20 months of follow-up (RR = 1.51, 95% CI: 0.90, 2.52). Cadmium and mercury had the largest contributions in the positive direction, whereas molybdenum and chromium had the largest contributions in the inverse direction (Figure S20B; Table S21). Among participants with a 25(OH)D of >20ng/mL, a 1-quartile increase in metal concentrations was not appreciably associated with UL incidence during 20 months of follow-up (RR = 1.02, 95% CI: 0.39, 2.68).
The overall association of the metal mixture with risk of incident UL during 10 years of follow-up was close to null (RR = 0.96, 95% CI: 0.78, 1.19, for a 1-quartile increase in metal concentrations). As recommended with respect to weighted quantile sum regression,81 a closely related method, we did not present or interpret weights when the overall mixture association was close to the null [i.e., 25(OH)D >20 ng/mL stratum and analysis of 10-year follow-up].
Cox Interaction Models
We observed positive associations of cadmium and mercury concentrations with incident UL among vitamin D-deficient participants but not among participants with a 25(OH)D of >20 ng/mL (Figure 6; Table S22). Specifically, a 1-unit increase in the log-transformed and standardized cadmium concentration was positively associated with incident UL during 20 months among vitamin D-deficient participants (IRR = 1.55, 95% CI: 1.07, 2.24) but not among participants with a 25(OH)D of >20ng/mL (IRR = 0.76, 95% CI: 0.40, 1.47). Similarly, the IRR for a 1-unit increase in mercury concentration with incident UL was 1.38 (95% CI: 0.99, 1.92) among vitamin D-deficient participants and 0.84 (95% CI: 0.57, 1.25) among participants with a 25(OH)D of >20 ng/mL. We also observed an inverse association of arsenic with incident UL (IRR = 0.63, 95% CI: 0.37, 1.06) and a positive association of barium with incident UL (IRR = 1.48, 95% CI: 0.89, 2.46) only among participants with a 25(OH)D of >20 ng/mL; the respective associations were close to null among vitamin D-deficient participants. Results were similar after further adjustment for alcohol intake, fish intake, educational attainment, and household income (Table S22).
In subgroup analyses according to age at baseline, we observed a strong positive association of cadmium with UL only among vitamin D-deficient participants <30 years of age (IRR - 2.25, 95% CI: 1.29, 3.93) whereas the association of mercury with UL among vitamin D-deficient participants was limited to participants >30 years of age (IRR = 2.57, 95% CI: 1.36, 4.86) (Table S23).
Discussion
Summary
In this prospective ultrasound-based cohort study of reproductiveage black individuals, we observed a weak, positive, monotonic association between a mixture of metals in whole blood and UL incidence during 20 months of follow-up. Cadmium had the strongest positive contribution to the overall mixture association. The association between the metal mixture and incident UL was stronger among vitamin D-deficient participants, and this association was driven by cadmium followed by mercury. Analyses using quantile g-computation and Cox regression confirmed the stronger associations of cadmium and mercury with incident UL over 20 months among vitamin D-deficient participants. When we extended the incident period to 10 years of follow-up, cadmium and mercury were not meaningfully associated with incident UL. Across analyses, we did not observe evidence of meaningful interaction between metals in their associations with incident UL.
Previous Epidemiologic Research
Small case-control studies (number of cases ranging from 22 to 30) identified lower zinc concentrations measured in serum82,83 and subcutaneous fat,19 and lower selenium concentrations measured in plasma84 and subcutaneous fat,19 among UL cases vs. healthy controls but were limited by lack of adjustment for potential confounders, lack of temporality, and reliance on surgical UL cases.
Larger epidemiologic studies used a cross-sectional design; focused on cadmium, lead, and mercury measured in blood; and yielded inconsistent associations with prevalent UL. A crosssectional study of patients undergoing laparoscopy or laparotomy for benign gynecologic indications identified positive associations of cadmium and lead measured in blood and cobalt measured in urine with odds of UL vs. other gynecologic morbidity.19 In an analysis of 1999-2000 National Health and Nutrition Examination Survey (NHANES) data, blood cadmium and lead were not strongly associated with odds of self-reported UL diagnosis, whereas blood mercury was positively but imprecisely associated with odds of UL.17 Likewise, in an analysis of 2001-2006 NHANES data, blood mercury was positively associated with UL across single-chemical regression analyses, weighted quantile sum regression, and BKMR.21 The NHANES research is limited by reliance on self-reported UL diagnosis, which is influenced by factors such as disease severity and health care access. Finally, in a cross-sectional study of a convenience sample of premenopausal women, associations of blood cadmium, lead, and mercury with prevalent UL were close to null and imprecise, but cadmium was positively associated with the volume of largest UL.20 Due to their cross-sectional design, these studies are limited by analysis of blood metal concentrations at or after the time of UL diagnosis, which may be several years after UL incidence.17,18,20,21 Only two studies adjusted for other metals (Jackson et al.17 using conventional regression modeling and Zhang et al.21 using mixture approaches); in both cases, cadmium, lead, and mercury were the only metals included, raising the possibility of confounding by other metals.
Discussion of Present Results
Using BKMR, we observed a weak, positive overall association between the metal mixture and incident UL during 20 months. Cadmium was the strongest contributor to the association. In vivo and in vitro experiments demonstrate that cadmium has estrogenic activity and may also influence synthesis of progesterone by ovarian cells.11,13,85-88 Low-dose cadmium exposure increases the proliferation of UL cells in vitro through nongenomic pathways mediated by the G protein-coupled estrogen receptor, which is more highly expressed in UL tissue than healthy myometrial tissue. 15,16,89 Other pathways by which cadmium could influence UL development include induction of mutagenesis, epigenetic modifications, oxidative stress, and inflammation.90-92 Our results are consistent with epidemiologic studies reporting positive associations of blood cadmium with UL prevalence18 and volume,20 considering that tumors must grow to a detectable size in order to be observed.
Cigarette smoking is a major source of cadmium exposure.93 Current smoking is the strongest identified predictor of blood cadmium concentrations in SELF.58 Other sources of cadmium exposure include dietary intake of leafy green vegetables, potatoes, soybeans, and grains; air and water contaminated by industrial activity; and occupational exposures.93 Cadmium measured in whole blood generally reflects recent exposure (~ 3 months) but can also reflect exposure over the past several years.22 We have previously shown in SELF that cigarette smoking is inversely associated with UL incidence.65 Thus, we found that cigarette smoking is an important confounder of the association of cadmium with UL, and inadequate control may cause downward bias in the observed associations. We conducted an analysis restricted to never smokers, providing stronger control of confounding by smoking, and observed similar results to the main analysis. The inverse association of cigarette smoking with risk of UL, if causal, is likely mediated through components of cigarette smoke other than cadmium.
In secondary BKMR analyses among participants with vitamin D deficiency, we observed a stronger positive overall mixture association, driven by stronger associations of cadmium and to a lesser extent mercury, with incident UL. These findings were confirmed using quantile g-computation and Cox models. Cadmium and mercury concentrations were not starkly different by vitamin D status, so we do not consider the stronger associations among vitamin D-deficient participants to be explained by differences in exposure distributions. We hypothesized a priori that vitamin D status may biologically interact with metals in influencing UL risk. Vitamin D and metals have complex interrelationships in the body. Vitamin D may increase the absorption of essential (e.g., calcium, magnesium, zinc) and toxic (e.g., cadmium, lead) metals through pathways including but not limited to calcium-binding proteins.25 Conversely, toxic metals can impair renal production of 25(OH)D.25,26 Previous work in SELF has shown that sufficient vitamin D is associated with reduced growth of UL.33 In in vitro experiments, treatment of human UL cells with 1,25-dihydroxyvitamin D3 reduced the expression of estrogen and progesterone receptors and inhibited estrogen-induced proliferation.94 Given that cadmium has estrogenic effects, the effects of cadmium on UL cell proliferation may be amplified when vitamin D is deficient. However, the proliferative effects of cadmium on UL cells are mediated through nongenomic pathways rather than through the estrogen receptor, as discussed above. Vitamin D could also mitigate harmful effects of toxic metals through other functions, such as repair of DNA damage.95 Recent work in the Maternal-Infant Research on Environmental Chemicals Study found that the positive association of lead with preterm birth was stronger among participants with insufficient vitamin D [defined as 25(OH)D <50 ng/mL], providing evidence of differential associations of toxic metals with reproductive outcomes by vitamin D status.96
Mercury was not strongly associated with UL incidence in the full cohort BKMR analysis. However, among vitamin D-deficient participants, mercury was the second-largest contributor to the BKMR overall mixture association and the quantile g-computation positive index. While mercury was one of the relatively more important contributors in these analyses, the univariate association of mercury with incident UL in BKMR was small in magnitude. In contrast, mercury was strongly positively associated with UL incidence among vitamin D-deficient participants in the Cox regression analysis. Post hoc analysis identified that this association was driven by the stronger positive association between mercury and UL among participants >30 years of age, which may explain the discrepancy between BKMR and Cox analyses. The Cox analyses adjusted finely for age via stratification while allowing the baseline hazard to vary by age (i.e., a primary time scale for the analysis in addition to "time on study"). We interpret the effect measure modification by age cautiously, but present these results as hypothesis generating for future research. This finding was unexpected, and we are not aware of previous research indicating a stronger effect of mercury on gynecologic outcomes among older reproductive-age women.
Mercury is a metalloestrogen10 and has shown evidence of endocrine disruption and reproductive toxicity in animal studies.11,97 The literature is inconsistent regarding endocrine-disrupting effects of mercury on the female reproductive system, 11,97,98 and the mechanisms by which mercury may influence UL-and the potential role of vitamin D-are unclear. Mercury measured in whole blood generally represents exposure to organic mercury, typically from seafood intake.99 Whole blood mercury was positively associated with UL prevalence in most17,20,21 but not all18 previous epidemiologic studies. Furthermore, previous studies, including research in SELF, have identified positive associations of fish intake100-103 and marine omega-3 fatty acid intake100-101 with incidence or prevalence of UL. Our present results suggest that mercury in fish may contribute to these associations, especially among vitamin D-deficient individuals.
After cadmium, the metals with the greatest relative contributions to the overall mixture association in the main BKMR analysis were barium, cobalt, and zinc. The univariate associations of these metals with incident UL were weak. Barium, cobalt, and zinc also had the largest contributions to the positive index in the quantile g-computation analysis of the full cohort. Barium is a metalloestrogen10,104 and is positively associated with earlier pubic hair development and menarche in peripubertal girls105 and higher odds of polycystic ovary syndrome.106 Barium has previously only been assessed in relation to UL in one cross-sectional study,18 in which urinary barium was not associated with UL prevalence. Cobalt is an essential metal but is toxic in large doses and with low-level chronic occupational exposure.107 Cobalt is also a metalloestrogen10 and may promote tumor initiation through oxidative stress, DNA damage, and activation of hypoxia-inducible factor.107 In a cross-sectional study,18 urinary cobalt was positively associated with prevalent UL. Zinc is an essential element and an antioxidant; its homeostasis in the body is tightly regulated.108 Previous studies have generally observed lower zinc in various biological matrices from UL patients compared with controls, 18,19,82,83 although none of these studies measured zinc in whole blood.
Lead was not identified as an important contributor to UL incidence in our mixture analyses. Lead is a known reproductive toxicant and endocrine disruptor, with evidence from animal studies of ovarian toxicity mediated through alterations to hormone secretion and expression of hormone receptors in the ovary and uterus.11,109-112 Epidemiologic studies among nonoccupationally exposed females have identified associations of blood lead at concentrations <10 pg/dL with alterations in serum reproductive hormone concentrations, although results are inconsistent across studies (reviewed by the Agency for Toxic Substances and Disease Registry in 2020113). In cross-sectional and case-control studies, lead measured in blood18,20 and subcutaneous fat19 was generally positively associated with prevalent UL, but two studies had very imprecise results17,20 and one study did not control for covariates.19 In NHANES, Zhang et al. identified a nonlinear association of blood lead with prevalent UL (with a positive association observed in the second, but not third, tertile), but lead had a small contribution to the mixture associations estimated using weighted quantile sum regression and BKMR.21
Selenium was not meaningfully associated with incident UL in our mixture analyses. Selenium is an essential metal with antioxidant capabilities and has been investigated in relation to cancer prevention, although evidence from randomized controlled trials does not support a protective effect of selenium supplementation.114 Research in Japanese quails found that selenium supplementation (0.2 and 0.4 mg/kg of diet vs. basal diet containing 0.048 mg/kg) was associated with reduced size of oviduct leiomyomata115; however, the doses used in this experiment likely exceed safe levels for humans. In three case-control studies, selenium measured in uterine tissue, 83 serum,83 plasma,84 and subcutaneous fat19 was consistently lower in UL patients compared with controls. Our results do not support a strong association of whole blood selenium with incident UL.
In secondary BKMR analyses evaluating incident UL during 10 years of follow-up, the overall mixture association was close to null, and univariate associations (including that of cadmium) were attenuated. These weaker associations using longer followup may be due to nondifferential measurement error resulting from short half-lives of metals in whole blood.
Limitations and Strengths
Our analysis has several limitations, most notably those related to the measurement of metals. We measured metal concentrations in whole blood, which is a widely accepted matrix in environmental health research for some metals (e.g., cadmium, lead, mercury, and selenium), but not others (e.g., barium, molybdenum, nickel, and zinc).22 Most of the metals that we analyzed have short halflives in whole blood ( ~<2 months), and we measured metals at a single timepoint. Therefore, nondifferential exposure measurement error due to variability in blood metal concentrations is possible. We did not conduct speciation analysis and, therefore, could not address the relative distributions and overall concentrations of metal species. Furthermore, we may not have measured metal concentrations during the etiologically relevant time period. The induction period for effects of metals on UL incidence is unknown and could vary between metals and according to the life stage at which exposure occurs. It is also possible that a given metal may act differentially upon the various biological processes involved in UL development. The latent period from UL initiation to growth to a detectable size is also unknown, contributing further potential for nondifferential measurement error. Finally, given the need to limit the number of covariates for BKMR models to converge, there may be uncontrolled confounding. Reassuringly, Cox model results were similar after further adjustment for alcohol intake, fish intake, educational attainment, and household income.
Our analysis also has several strengths. We identified new UL cases with systematic follow-up ultrasound exams among individuals without UL at baseline, which bolsters temporality and accuracy of outcome ascertainment. We adjusted for strong potential confounders identified using a causal diagram informed by previous research in this cohort. Selection bias is unlikely because we had high cohort retention and observed similar metal concentrations between a) participants with vs. without prevalent UL at baseline and b) participants who were vs. were not lost to followup. We analyzed a large panel of metals, enabling us to investigate understudied exposures and to control for co-pollutant confounding. We used BKMR, a flexible statistical mixture method that allows for nonlinear associations and interactions between mixture components while controlling for confounding by mixture components and specified covariates. We conducted this research in a cohort of black individuals, a population disproportionately affected by vitamin D deficiency, toxic metal exposure, and UL.40,45,47,54 As we reported previously, SELF participants had similar blood metal concentrations compared with same-age nonHispanic black women in NHANES, 2011-2012,58 suggesting that SELF participants" exposure levels are relevant to the general population. If supported by further research, these findings could have large public health significance by identifying an intervention (i.e., prevention and treatment of vitamin D deficiency) to reduce the adverse effects of heavy metal exposure on UL.
Conclusions
This is the first epidemiologic study to use a mixtures approach to examine multiple metals in relation to prospectively identified UL, a major cause of gynecologic morbidity. We observed a weak positive association between metals measured in whole blood and UL incidence during 20 months of follow-up, driven by a positive association of cadmium with incident UL. This work adds to the literature demonstrating harmful associations of cadmium with female reproductive health116 and supports public health efforts to reduce cadmium exposures, e.g., reduction of smoking and environmental tobacco smoke exposure. The associations of cadmium and mercury with incident UL were stronger among participants with vitamin D deficiency, suggesting that vitamin D deficiency may exacerbate the toxicity of these metals for UL. We encourage future research to examine vitamin D-deficient individuals as a hypothesized vulnerable population for reproductive toxicity of environmental contaminants and to investigate vitamin D supplementation as prevention for UL.
Acknowledgments
We thank the SELF participants and staff for their valuable contributions to the study; Robert Wright and Chitra Amarasiriwardena for supervising laboratory analyses; Alexa Friedman, Victoria Fruh, and Samantha Schildroth for discussion of BKMR analysis; and Ashlee Oaks for assistance with geocoding and linkage with spatial variables.
Funding was provided by the National Institute of Environmental Health Sciences (R01-ES028235 and R01-ES024749) and the National Institute of Nursing Research (R00-NR017191) of the National Institutes of Health. The research was supported in part by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences and funds from the American Recovery and Reinvestment Act designated for National Institutes of Health research. The NIH Office of Disease Prevention provided co-funding for baseline visit vitamin D assays.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Address correspondence to Ruth J. Geller, Boston University School of Public Health, 715 Albany St., Boston, MA 02118 USA. Email: rgeller @bu.edu
Supplemental Material is available online (https://doi.org/10.1289/EHP15218).
L.A.W. is a paid consultant for The Gates Foundation (progestin-only contraceptives) and AbbVie, Inc. (fibroids; abnormal uterine bleeding). L.A.W. also receives in-kind donations from Kindara.com (fertility-tracking apps) and Swiss Precision Diagnostics (home pregnancy tests). The other authors declare they have no conflicts of interest related to this work to disclose.
Conclusions and opinions are those of the individual authors and do not necessarily reflect the policies or views of EHP Publishing or the National Institute of Environmental Health Sciences.
Received 23 April 2024; Revised 5 February 2025; Accepted 4 March 2025; Published 25 April 2025.
Note to readers with disabilities: EHP strives to ensure that all journal content is accessible to all readers. However, some figures and Supplemental Material published in EHP articles may not conform to 508 standards due to the complexity of the information being presented. If you need assistance accessing journal content, please contact ehpsubmissions @niehs.nih.gov. Our staff will work with you to assess and meet your accessibility needs within 3 working days.
References
1. Coronado GD, Marshall LM, Schwartz SM. 2000. Complications in pregnancy, labor, and delivery with uterine leiomyomas: a population-based study. Obstet Gynecol 95(5):764-769, PMID: 10775744, https://doi.org/10.1016/s0029-7844(99) 00605-5.
2. Stewart EA, Cookson CL, Gandolfo RA, Schulze-Rath R. 2017. Epidemiology of uterine fibroids: a systematic review. BJOG 124(10):1501-1512, PMID: 28296146, https://doi.org/10.1111/1471-0528.14640.
3. Morgan DM, Kamdar NS, Swenson CW, Kobernik EK, Sammarco AG, Nallamothu B. 2018. Nationwide trends in the utilization of and payments for hysterectomy in the United States among commercially insured women. Am J Obstet Gynecol 218(4):425.е1-425.е18, PMID: 29288067, https://doi.org/10.1016/). ajog.2017.12.218.
4. Wright JD, Herzog TJ, Tsui J, Ananth CV, Lewin SN, Lu YS, et al. 2013. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol 122(2 pt 1):233-241, PMID: 23969789, https://doi.org/10. 1097/A0G.0b013e318299a6cf.
5. Stewart EA, Laughlin-Tommaso SK, Catherino WH, Lalitkumar $, Gupta D, Vollenhoven B. 2016. Uterine fibroids. Nat Rev Dis Primers 2:16043, PMID: 27335259, https://doi.org/10.1038/nrdp.2016.43.
6. Wise LA, Laughlin-Tommaso S. 2013. Uterine leiomyomata. In: Women and Health. Goldman MB, Troisi R, Rexrode KM, eds. 2nd ed. San Diego, CA: Academic Press, 285-306.
7. Bulun SE, Moravek MB, Yin P, Ono M, Coon VJS, Dyson MT, et al. 2015. Uterine leiomyoma stem cells: linking progesterone to growth. Semin Reprod Med 33(5):357-365, PMID: 26251118, https://doi.org/10.1055/s-0035-1558451.
8. Bariani MV, Rangaswamy R, Siblini H, Yang Q, Al-Hendy A, Zota AR. 2020. The role of endocrine-disrupting chemicals in uterine fibroid pathogenesis. Curr Opin Endocrinol Diabetes Obes 27(6):380-387, PMID: 33044243, https://doi.org/ 10.1097/MED.0000000000000578.
9. Wise LA. 2021. Environmental chemicals and risk of uterine leiomyomata. In: Uterine Fibroids. Petrozza JC, ed. Boca Raton, Florida: CRC Press.
10. Darbre PD. 2006. Metalloestrogens: an emerging class of inorganic xenoestrogens with potential to add to the oestrogenic burden of the human breast. J Appl Toxicol 26(3):191-197, PMID: 16489580, https://doi.org/10.1002/jat.1135.
11. Iavicoli I, Fontana L, Bergamaschi A. 2009. The effects of metals as endocrine disruptors. J Toxicol Environ Health B Crit Rev 12(3):206-223, PMID: 19466673, https://doi.org/10.1080/10937400902902062.
12. Lafuente A, Cano P, Esquifino A. 2003. Are cadmium effects on plasma gonadotropins, prolactin, ACTH, GH and TSH levels, dose-dependent? Biometals 16(2):243-250, PMID: 12572682, https://doi.org/10.1023/a:1020658128413.
13. Piasek M, Laskey JW. 1994. Acute cadmium exposure and ovarian steroidogenesis in cycling and pregnant rats. Reprod Toxicol 8(6):495-507, PMID: 7881201, https://doi.org/10.1016/0890-6238(94)90032-9.
14. Drevnick PE, Sandheinrich MB. 2003. Effects of dietary methylmercury on reproductive endocrinology of fathead minnows. Environ Sci Technol 37(19):4390- 4396, PMID: 14572090, https://doi.org/10.1021/es034252m.
15. Gao X, Yu L, Moore AB, Kissling GE, Waalkes MP, Dixon D. 2015. Cadmium and proliferation in human uterine leiomyoma cells: evidence of a role for EGFR/ MAPK pathways but not classical estrogen receptor pathways. Environ Health Perspect 123(4):331-336, PMID: 25343777, https://doi.org/10.1289/ehp.1408234.
16. Liu J, Yu L, Castro L, Yan Y, Sifre MI, Bortner CD, et al. 2019. A nongenomic mechanism for "metalloestrogenic" effects of cadmium in human uterine leiomyoma cells through G protein-coupled estrogen receptor. Arch Toxicol 93(10):2773-2785, PMID: 31468104, https://doi.org/10.1007/s00204-019-02544-0.
17. Jackson LW, Zullo MD, Goldberg JM. 2008. The association between heavy metals, endometriosis and uterine myomas among premenopausal women: National Health and Nutrition Examination Survey 1999-2002. Hum Reprod 23(3):679-687, PMID: 18192673, https://doi.org/10.1093/humrep/dem394.
18. Johnstone EB, Buck Louis GM, Parsons PJ, Steuerwald AJ, Palmer CD, Chen Z, et al. 2014. Increased urinary cobalt and whole blood concentrations of cadmium and lead in women with uterine leiomyomata: findings from the ENDO study. Reprod Toxicol 49:27-32, PMID: 24994689, https://doi.org/10.1016/j. reprotox.2014.06.007.
19. Qin YY, Leung CKM, Leung AOW, Wu SC, Zheng JS, Wong MH. 2010. Persistent organic pollutants and heavy metals in adipose tissues of patients with uterine leiomyomas and the association of these pollutants with seafood diet, BMI, and age. Environ Sci Pollut Res 17(1):229-240, https://doi.org/10.1007/s11356-009-0251-0.
20. Ye S, Chung HW, Jeong K, Sung YA, Lee H, Park SY, et al. 2017. Blood cadmium and volume of uterine fibroids in premenopausal women. Ann Occup Environ Med 29:22, PMID: 28649392, https://doi.org/10.1186/s40557-017-0178-8.
21. Zhang Y, Lu Y, Ma H, Xu Q, Wu X. 2021. Combined exposure to multiple endocrine disruptors and uterine leiomyomata and endometriosis in US women. Front Endocrinol (Lausanne) 12:726876, PMID: 34489871, https://doi.org/10.3389/ fendo.2021.726876.
22. Martinez-Morata I, Sobel M, Tellez-Plaza M, Navas-Acien A, Howe CG, Sanchez TR. 2023. A state-of-the-science review on metal biomarkers. Curr Environ Health Rep 10(3):215-249, PMID: 37337116, https://doi.org/10.1007/ s40572-023-00402-x.
23. Claus Henn B, Coull BA, Wright RO. 2014. Chemical mixtures and children's health. Curr Opin Pediatr 26(2):223-229, PMID: 24535499, https://doi.org/10.1097/ MOP.0000000000000067.
24. Fisher M, Weiler HA, Kuiper JR, Borghese M, Buckley JP, Shutt R, et al. 2024. Vitamin D and toxic metals in pregnancy - a biological perspective. Curr Epidemiol Rep 11(3):153-163, PMID: 39156920, https://doi.org/10.1007/s40471-024-00348-0.
25. Moon J. 1994. The role of vitamin D in toxic metal absorption: a review. J Am Coll Nutr 13(6):559-564, PMID: 7706586, https://doi.org/10.1080/07315724.1994. 10718447.
26. Schwalfenberg GK, Genuis SJ. 2015. Vitamin D, essential minerals, and toxic elements: exploring interactions between nutrients and toxicants in clinical medicine. ScientificWorldJournal 2015:318595, PMID: 26347061, https://doi.org/ 10.1155/2015/318595.
27. Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, et al. 2011. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 96(7):1911-1930, PMID: 21646368, https://doi.org/10.1210/jc.2011-0385.
28. Baird DD. 2008. Hypothesis: vitamin D protects against uterine fibroid development. Ann Epidemiol 18(9):710.
29. Bläuer M, Rovio PH, Ylikomi T, Heinonen PK. 2009. Vitamin D inhibits myometrial and leiomyoma cell proliferation in vitro. Fertil Steril 91(5):1919-1925, PMID: 18423458, https://doi.org/10.1016/j.fertnstert.2008.02.136.
30. Halder SK, Sharan C, Al-Hendy A. 2012. 1,25-dihydroxyvitamin D3 treatment shrinks uterine leiomyoma tumors in the Eker rat model. Biol Reprod 86(4):116, PMID: 22302692, https://doi.org/10.1095/biolreprod.111.098145.
31. Baird DD, Hill MC, Schectman JM, Hollis BW. 2013. Vitamin D and the risk of uterine fibroids. Epidemiology 24(3):447-453, PMID: 23493030, https://doi.org/10. 1097/EDE.0b013e31828acca0.
32. Paffoni A, Somigliana E, Vigano P, Benaglia L, Cardellicchio L, Pagliardini L, et al. 2013. Vitamin D status in women with uterine leiomyomas. J Clin Endocrinol Metab 98(8):E1374-E1378, PMID: 23824422, https://doi.org/10.1210/jc.2013-1777.
33. Harmon QE, Patchel SA, Denslow S, LaPorte F, Cooper T, Wise LA, et al. 2022. Vitamin D and uterine fibroid growth, incidence, and loss: a prospective ultrasound study. Fertil Steril 118(6):1127-1136, PMID: 36150919, https://doi.org/10. 1016/j.fertnstert.2022.08.851.
34. Sabry M, Halder SK, Allah AS, Roshdy E, Rajaratnam V, Al-Hendy A. 2013. Serum vitamin D3 level inversely correlates with uterine fibroid volume in different ethnic groups: a cross-sectional observational study. Int J Womens Health 5:93-100, PMID: 23467803, https://doi.org/10.2147/IJWH.S38800.
35. Bailey ZD, Krieger N, Agénor M, Graves J, Linos N, Bassett MT. 2017. Structural racism and health inequities in the USA: evidence and interventions. Lancet 389(10077):1453-1463, PMID: 28402827, https://doi.org/10.1016/S0140-6736(17) 30569-X.
36. Ross LJ. 2017. Reproductive justice as intersectional feminist activism. Souls 19(3):286-314, https://doi.org/10.1080/10999949.2017.1389634.
37. Zota AR, Shamasunder B. 2017. The environmental injustice of beauty: framing chemical exposures from beauty products as a health disparities concern. Am J Obstet Gynecol 217(4):418.e1-418.e6, PMID: 28822238, https://doi.org/10.1016/ j.ajog.2017.07.020.
38. Katon JG, Plowden TC, Marsh EE. 2023. Racial disparities in uterine fibroids and endometriosis: a systematic review and application of social, structural, and political context. Fertil Steril 119(3):355-363, PMID: 36682686, https://doi.org/10. 1016/j.fertnstert.2023.01.022.
39. Jones DH, Yu X, Guo Q, Duan X, Jia C. 2022. Racial disparities in the heavy metal contamination of urban soil in the southeastern United States. Int J Environ Res Public Health 19(3):1105, PMID: 35162130, https://doi.org/10.3390/ijerph19031105.
40. Trottier BA, NiehoffNM, Keil AP, Jones RR, Levine KE, MacNell NS, et al. 2023. Residential proximity to metal-containing superfund sites and their potential as a source of disparities in metal exposure among U.S. women. Environ Health Perspect 131(3):037701, PMID: 36917478, https://doi.org/10.1289/EHP11045.
41. Davis HT, Aelion CM, Liu J, Burch JB, Cai B, Lawson AB, et al. 2016. Potential sources and racial disparities in the residential distribution of soil arsenic and lead among pregnant women. Sci Total Environ 551-552:622-630, PMID: 26897405, https://doi.org/10.1016/j.scitotenv.2016.02.018.
42. Gee GC, Payne-Sturges DC. 2004. Environmental health disparities: a framework integrating psychosocial and environmental concepts. Environ Health Perspect 112(17):1645-1653, PMID: 15579407, https://doi.org/10.1289/ehp.7074.
43. Mijal RS, Holzman CB. 2010. Blood cadmium levels in women of childbearing age vary by race/ethnicity. Environ Res 110(5):505-512, PMID: 20400068, https://doi.org/10.1016/j.envres.2010.02.007.
44. Geron M, Cowell W, Amarasiriwardena C, Andra SS, Carroll K, Kloog I, et al. 2022. Racial/ethnic and neighborhood disparities in metals exposure during pregnancy in the northeastern United States. Sci Total Environ 820:153249, PMID: 35065119, https://doi.org/10.1016/j.scitotenv.2022.153249.
45. Bulka CM, Bommarito PA, Fry RC. 2019. Predictors of toxic metal exposures among US women of reproductive age. J Expo Sci Environ Epidemiol 29(5):597-612, PMID: 31235790, https://doi.org/10.1038/s41370-019-0152-3.
46. Nguyen VK, Kahana A, Heidt J, Polemi K, Kvasnicka J, Jolliet O, et al. 2020. A comprehensive analysis of racial disparities in chemical biomarker concentrations in United States women, 1999-2014. Environ Int 137:105496, PMID: 32113086, https://doi.org/10.1016/j.envint.2020.105496.
47. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. 2003. High cumulative incidence of uterine leiomyoma in black and white women: ultrasound evidence. Am J Obstet Gynecol 188(1):100-107, PMID: 12548202, https://doi.org/10.1067/ mob.2003.99.
48. Wegienka G, Stewart EA, Nicholson WK, Zhang S, Li F, Thomas L, et al. 2021. Black women are more likely than white women to schedule a uterine-sparing treatment for leiomyomas. J Womens Health (Larchmt) 30(3):355-366, PMID: 33524308, https://doi.org/10.1089/jwh.2020.8634.
49. Stewart EA, Nicholson WK, Bradley L, Borah BJ. 2013. The burden of uterine fibroids for African-American women: results of a national survey. J Womens Health (Larchmt) 22(10):807-816, PMID: 24033092, https://doi.org/10.1089/jwh. 2013.4334.
50. Sengoba KS, Ghant MS, Okeigwe I, Mendoza G, Marsh EE. 2017. Racial/ethnic differences in women's experiences with symptomatic uterine fibroids: a qualitative assessment. J Racial Ethn Health Disparities 4(2):178-183, PMID: 27068661, https://doi.org/10.1007/s40615-016-0216-1.
51. Wechter ME, Stewart EA, Myers ER, Kho RM, Wu JM. 2011. Leiomyoma-related hospitalization and surgery: prevalence and predicted growth based on population trends. Am J Obstet Gynecol 205(5):492.e1-492.e5, PMID: 22035951, https://doi.org/10.1016/j.ajog.2011.07.008.
52. Cohen SL, Vitonis AF, Einarsson JI. 2014. Updated hysterectomy surveillance and factors associated with minimally invasive hysterectomy. JSLS 18(3): e2014.00096, PMID: 25392662, https://doi.org/10.4293/JSLS.2014.00096.
53. Zaritsky E, Le A, Tucker LY, Ojo A, Weintraub MR, Raine-Bennett T. 2022. Minimally invasive myomectomy: practice trends and differences between black and non-black women within a large integrated healthcare system. Am J Obstet Gynecol 226(6):826.e1-826.e11, PMID: 35101407, https://doi.org/10.1016/j. ajog.2022.01.022.
54. Forrest KY, Stuhldreher WL. 2011. Prevalence and correlates of vitamin D deficiency in US adults. Nutr Res 31(1):48-54, PMID: 21310306, https://doi.org/10. 1016/j.nutres.2010.12.001.
55. Baird DD, Harmon QE, Upson K, Moore KR, Barker-Cummings C, Baker S, et al. 2015. A prospective, ultrasound-based study to evaluate risk factors for uterine fibroid incidence and growth: methods and results of recruitment. J Womens Health (Larchmt) 24(11):907-915, PMID: 26334691, https://doi.org/10.1089/jwh. 2015.5277.
56. Laughlin SK, Baird DD, Savitz DA, Herring AH, Hartmann KE. 2009. Prevalence of uterine leiomyomas in the first trimester of pregnancy: an ultrasound-screening study. Obstet Gynecol 113(3):630-635, PMID: 19300327, https://doi.org/10.1097/ AOG.0b013e318197bbaf.
57. Laughlin SK, Schroeder JC, Baird DD. 2010. New directions in the epidemiology of uterine fibroids. Semin Reprod Med 28(3):204-217, PMID: 20414843, https://doi.org/10.1055/s-0030-1251477.
58. Geller RJ, Wesselink AK, Upson K, Claus Henn B, Schildroth S, Wright R, et al. 2023. Correlates of whole blood metal concentrations among reproductive-aged black women. J Expo Sci Environ Epidemiol 33(6):883-894, PMID: 36104525, https://doi.org/10.1038/s41370-022-00477-y.
59. Wagner D, Hanwell HEC, Vieth R. 2009. An evaluation of automated methods for measurement of serum 25-hydroxyvitamin D. Clin Biochem 42(15):1549-1556, PMID: 19631201, https://doi.org/10.1016/j.clinbiochem.2009.07.013.
60. Ersfeld DL, Rao DS, Body J-J, Sackrison JL, Miller AB, Parikh N, et al. 2004. Analytical and clinical validation of the 25 OH vitamin D assay for the LIAISON® automated analyzer. Clin Biochem 37(10):867-874, PMID: 15369717, https://doi.org/10.1016/j.clinbiochem.2004.06.006.
61. Dueholm M, Lundorf E, Hansen ES, Ledertoug S, Olesen F. 2002. Accuracy of magnetic resonance imaging and transvaginal ultrasonography in the diagnosis, mapping, and measurement of uterine myomas. Am J Obstet Gynecol 186(3):409-415, PMID: 11904599, https://doi.org/10.1067/mob.2002.121725.
62. Greenland S, Pearl J, Robins JM. 1999. Causal diagrams for epidemiologic research. Epidemiology 10(1):37-48, PMID: 9888278.
63. Wise LA, Laughlin-Tommaso SK. 2016. Epidemiology of uterine fibroids: from menarche to menopause. Clin Obstet Gynecol 59(1):2-24, PMID: 26744813, https://doi.org/10.1097/GRF.0000000000000164.
64. Harmon QE, Patchel SA, Zhao S, Umbach DM, Cooper TE, Baird DD. 2022. Depot medroxyprogesterone acetate use and the development and progression of uterine leiomyoma. Obstet Gynecol 139(5):797-807, PMID: 35576339, https://doi.org/10.1097/AOG.0000000000004745.
65. Wesselink AK, Wegienka G, Coleman CM, Geller RJ, Harmon QE, Upson K, et al. 2023. A prospective ultrasound study of cigarette smoking and uterine leiomyomata incidence and growth. Am J Obstet Gynecol 229(2):151.e1-151.e8, PMID: 37148957, https://doi.org/10.1016/j.ajog.2023.04.041.
66. Harmon QE, Patchel S, Denslow S, Wegienka G, Baird DD. 2024. Body mass index and uterine fibroid development: a prospective study. J Clin Endocrinol Metab 109(11):e2016-e2023, PMID: 38298165, https://doi.org/10.1210/clinem/dgae036.
67. Geverdt D. 2017. Education Demographic and Geographic Estimates (EDGE) Program. Locale Boundaries, 2015 (NCES 2016-032). Washington, DC: National Center for Education Statistics.
68. Kind AJH, Buckingham WR. 2018. Making neighborhood-disadvantage metrics accessible-the neighborhood atlas. N Engl J Med 378(26):2456-2458, PMID: 29949490, https://doi.org/10.1056/NEJMp1802313.
69. Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J, Gardner L. 1986. A data-based approach to diet questionnaire design and testing. Am J Epidemiol 124(3):453-469, PMID: 3740045, https://doi.org/10.1093/oxfordjournals.aje.a114416.
70. Upson K, Harmon QE, Heffron R, Hall JE, Wise LA, Wegienka G, et al. 2020. Depot medroxyprogesterone acetate use and blood lead levels in a cohort of young women. Environ Health Perspect 128(11):117004, PMID: 33206002, https://doi.org/10.1289/EHP7017.
71. Baird DD, Patchel SA, Saldana TM, Umbach DM, Cooper T, Wegienka G, et al. 2020. Uterine fibroid incidence and growth in an ultrasound-based, prospective study of young African Americans. Am J Obstet Gynecol 223(3):402.e1-402.e18, PMID: 32105679, https://doi.org/10.1016/j.ajog.2020.02.016.
72. Schisterman EF, Vexler A, Whitcomb BW, Liu A. 2006. The limitations due to exposure detection limits for regression models. Am J Epidemiol 163(4):374-383, PMID: 16394206, https://doi.org/10.1093/aje/kwj039.
73. Hornung RW, Reed LD. 1990. Estimation of average concentration in the presence of nondetectable values. Appl Occupation Environ Hygiene 5(1):46-51, https://doi.org/10.1080/1047322X.1990.10389587.
74. Bobb JF, Valeri L, Claus Henn B, Christiani DC, Wright RO, Mazumdar M, et al. 2015. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics 16(3):493-508, PMID: 25532525, https://doi.org/10.1093/biostatistics/kxu058.
75. Bobb JF. 2022. bkmr: Bayesian kernel machine regression. https://CRAN.R-project.org/package=bkmr [accessed 28 October 2022].
76. Keil A. 2022. bkmrhat: parallel chain tools for Bayesian kernel machine regression. https://CRAN.R-project.org/package=bkmrhat [accessed 29 September 2023].
77. Vehtari A, Gelman A, Simpson D, Carpenter B, Bürkner P-C. 2021. Rank-normalization, folding, and localization: an improved R-hat for assessing convergence of MCMC (with discussion). Bayesian Anal 16(2):667-718, https://doi.org/10. 1214/20-BA1221.
78. Bobb JF, Claus Henn B, Valeri L, Coull BA. 2018. Statistical software for analyzing the health effects of multiple concurrent exposures via Bayesian kernel machine regression. Environ Health 17(1):67, PMID: 30126431, https://doi.org/10. 1186/s12940-018-0413-y.
79. Keil AP, Buckley JP, O'Brien KM, Ferguson KK, Zhao S, White AJ. 2020. A quantile-based g-computation approach to addressing the effects of exposure mixtures. Environ Health Perspect 128(4):047004, PMID: 32255670, https://doi.org/10.1289/EHP5838.
80. Wasserstein RL, Lazar NA. 2016. The ASA statement on p-values: context, process, and purpose. Am Stat 70(2):129-133, https://doi.org/10.1080/00031305. 2016.1154108.
81. Carrico C, Gennings C, Wheeler DC, Factor-Litvak P. 2015. Characterization of weighted quantile sum regression for highly correlated data in a risk analysis setting. J Agric Biol Environ Stat 20(1):100-120, PMID: 30505142, https://doi.org/ 10.1007/s13253-014-0180-3.
82. He Q, Ma RH, Tang Y. 2002. [Determination of trace element Cu, Zn, Mg, Cr in serum of women with barrenness and hysteromyoma disease]. Guang Pu Xue Yu Guang Pu Fen Xi 22(4):685-686, PMID: 12938400.
83. Cunzhi H, Jiexian J, Xianwen Z, Jingang G, Shumin Z, Lili D. 2003. Serum and tissue levels of six trace elements and copper/zinc ratio in patients with cervical cancer and uterine myoma. Biol Trace Elem Res 94(2):113-122, PMID: 12958402, https://doi.org/10.1385/BTER:94:2:113.
84. Mutafchiev K, Popov I, Tsafarov M. 2007. [Manganese and selenium content in plasma of women with uterine myoma]. Akush Ginekol (Sofiia) 46(9):43-45, PMID: 18646308.
85. Piasek M, Laskey JW. 1999. Effects of in vitro cadmium exposure on ovarian steroidogenesis in rats. J Appl Toxicol 19:211-217, PMID: 10362273, https://doi.org/10. 1002/(SICI)1099-1263(199905/06)19:3<211:AID-JAT568>3.0.CO;2-4.
86. Paksy K, Rajczy K, Forgács Z, Lázár P, Bernard A, Gáti I, et al. 1997. Effect of cadmium on morphology and steroidogenesis of cultured human ovarian granulosa cells. J Appl Toxicol 17:321-327, PMID: 9339745, https://doi.org/10.1002/(SICI) 1099-1263(199709)17:5<321:AID-JAT443>3.0.CO;2-E.
87. Smida AD, Valderrama XP, Agostini MC, Furlan MA, Chedrese J. 2004. Cadmium stimulates transcription of the cytochrome p450 side chain cleavage gene in genetically modified stable porcine granulosa cells. Biol Reprod 70(1):25-31, PMID: 12954731, https://doi.org/10.1095/biolreprod.103.019000.
88. Johnson MD, Kenney N, Stoica A, Hilakivi-Clarke L, Singh B, Chepko G, et al. 2003. Cadmium mimics the in vivo effects of estrogen in the uterus and mammary gland. Nat Med 9(8):1081-1084, PMID: 12858169, https://doi.org/10.1038/nm902.
89. Yu L, Liu J, Yan Y, Burwell A, Castro L, Shi M, et al. 2021. "Metalloestrogenic" effects of cadmium downstream of G protein-coupled estrogen receptor and mitogen-activated protein kinase pathways in human uterine fibroid cells. Arch Toxicol 95(6):1995-2006, PMID: 33818655, https://doi.org/10.1007/s00204-021-03033-z.
90. Wu X, Cobbina SJ, Mao G, Xu H, Zhang Z, Yang L. 2016. A review of toxicity and mechanisms of individual and mixtures of heavy metals in the environment. Environ Sci Pollut Res Int 23(9):8244-8259, PMID: 26965280, https://doi.org/10. 1007/s11356-016-6333-x.
91. Zhu Y, Costa M. 2020. Metals and molecular carcinogenesis. Carcinogenesis 41(9):1161-1172, PMID: 32674145, https://doi.org/10.1093/carcin/bgaa076.
92. Fan Y, Jiang X, Xiao Y, Li H, Chen J, Bai W. 2024. Natural antioxidants mitigate heavy metal induced reproductive toxicity: prospective mechanisms and biomarkers. Crit Rev Food Sci Nutr 64(31):11530-11542, https://doi.org/10.1080/ 10408398.2023.2240399.
93. Agency for Toxic Substances and Disease Registry. 2012. Toxicological Profile for Cadmium. Atlanta, GA: U.S. Department of Health and Human Services.
94. Al-Hendy A, Diamond MP, El-Sohemy A, Halder SK. 2015. 1,25-Dihydroxyvitamin D3 regulates expression of sex steroid receptors in human uterine fibroid cells. J Clin Endocrinol Metab 100(4):E572-E582, PMID: 25625804, https://doi.org/10. 1210/jc.2014-4011.
95. Elkafas H, Ali M, Elmorsy E, Kamel R, Thompson WE, Badary O, et al. 2020. Vitamin D3 ameliorates DNA damage caused by developmental exposure to endocrine disruptors in the uterine myometrial stem cells of Eker rats. Cells 9(6):1459, PMID: 32545544, https://doi.org/10.3390/cells9061459.
96. Fisher M, Marro L, Arbuckle TE, Potter BK, Little J, Weiler H, et al. 2023. Association between toxic metals, vitamin D and preterm birth in the maternal-infant research on environmental chemicals study. Paediatr Perinat Epidemiol 37(5):447-457, PMID: 36864001, https://doi.org/10.1111/ppe.12962.
97. Liu D, Shi Q, Liu C, Sun Q, Zeng X. 2023. Effects of endocrine-disrupting heavy metals on human health. Toxics 11(4):322, PMID: 37112549, https://doi.org/10. 3390/toxics11040322.
98. Dutta S, Gorain B, Choudhury H, Roychoudhury S, Sengupta P. 2022. Environmental and occupational exposure of metals and female reproductive health. Environ Sci Pollut Res Int 29(41):62067-62092, PMID: 34558053, https://doi.org/10.1007/s11356-021-16581-9.
99. Agency for Toxic Substances and Disease Registry. 1999. Toxicological Profile for Mercury. Atlanta, GA: U.S. Department of Health and Human Services.
100. Wise LA, Radin RG, Kumanyika SK, Ruiz-Narvaez EA, Palmer JR, Rosenberg L. 2014. Prospective study of dietary fat and risk of uterine leiomyomata. Am J Clin Nutr 99(5):1105-1116, PMID: 24598152, https://doi.org/10.3945/ajcn.113.073635.
101. Brasky TM, Bethea TN, Wesselink AK, Wegienka GR, Baird DD, Wise LA. 2020. Dietary fat intake and risk of uterine leiomyomata: a prospective ultrasound study. Am J Epidemiol 189(12):1538-1546, PMID: 32556077, https://doi.org/10. 1093/aje/kwaa097.
102. Lambertino A, Turyk M, Anderson H, Freels S, Persky V. 2011. Uterine leiomyomata in a cohort of Great Lakes sport fish consumers. Environ Res 111(4):565-572, PMID: 21310402, https://doi.org/10.1016/j.envres.2011.01.006.
103. Kim MJ, Kim S, Kim JJ, Kim YS, Song JH, Lee JE, et al. 2024. Dietary intake is associated with the prevalence of uterine leiomyoma in Korean women: a retrospective cohort study. PLoS One 19(2):e0291157, PMID: 38359002, https://doi.org/ 10.1371/journal.pone.0291157.
104. Choe S-Y, Kim S-J, Kim H-G, Lee JH, Choi Y, Lee H, et al. 2003. Evaluation of estrogenicity of major heavy metals. Sci Total Environ 312(1-3):15-21, PMID: 12873394, https://doi.org/10.1016/S0048-9697(03)00190-6.
105. Ashrap P, Sanchez BN, Tellez-Rojo MM, Basu N, Tamayo-Ortiz M, Peterson KE, et al. 2019. In utero and peripubertal metals exposure in relation to reproductive hormones and sexual maturation and progression among girls in Mexico City. Environ Res 177:108630, PMID: 31421446, https://doi.org/10.1016/j.envres.2019. 108630.
106. Liang C, Zhang Z, Cao Y, Wang J, Shen L, Jiang T, et al. 2022. Exposure to multiple toxic metals and polycystic ovary syndrome risk: endocrine disrupting effect from As, Pb and Ba. Sci Total Environ 849:157780, PMID: 35926607, https://doi.org/10.1016/j.scitotenv.2022.157780.
107. Simonsen LO, Harbak H, Bennekou P. 2012. Cobalt metabolism and toxicology-a brief update. Sci Total Environ 432:210-215, PMID: 22732165, https://doi.org/ 10.1016/j.scitotenv.2012.06.009.
108. Michalczyk K, Cymbaluk-Płoska A. 2020. The role of zinc and copper in gynecological malignancies. Nutrients 12(12):3732, PMID: 33287452, https://doi.org/10. 3390/nu12123732.
109. Wide M. 1980. Interference of lead with implantation in the mouse: effect of exogenous oestradiol and progesterone. Teratology 21(2):187-191, PMID: 7394722, https://doi.org/10.1002/tera.1420210209.
110. Junaid M, Chowdhuri DK, Narayan R, Shanker R, Saxena DK. 1997. Lead-induced changes in ovarian follicular development and maturation in mice. J Toxicol Environ Health 50(1):31-40, PMID: 9015130, https://doi.org/10.1080/ 009841097160582.
111. Wiebe JP, Barr KJ. 1988. Effect of prenatal and neonatal exposure to lead on the affinity and number of estradiol receptors in the uterus. J Toxicol Environ Health 24(4):451-460, PMID: 3411631, https://doi.org/10.1080/15287398809531176.
112. Wiebe JP, Barr KJ, Buckingham KD. 1988. Effect of prenatal and neonatal exposure to lead on gonadotropin receptors and steroidogenesis in rat ovaries. J Toxicol Environ Health 24(4):461-476, PMID: 3137361, https://doi.org/10.1080/ 15287398809531177.
113. Agency for Toxic Substances and Disease Registry. 2020. Toxicological Profile for Lead. Atlanta, GA: U.S. Department of Health and Human Services.
114. Vinceti M, Filippini T, Del Giovane C, Dennert G, Zwahlen M, Brinkman M, et al. 2018. Selenium for preventing cancer. Cochrane Database Syst Rev 1(1): CD005195, PMID: 29376219, https://doi.org/10.1002/14651858.CD005195.pub4.
115. Tuzcu M, Sahin N, Ozercan I, Seren S, Sahin K, Kucuk O. 2010. The effects of selenium supplementation on the spontaneously occurring fibroid tumors of oviduct, 8-hydroxy-2'-deoxyguanosine levels, and heat shock protein 70 response in Japanese quail. Nutr Cancer 62(4):495-500, PMID: 20432170, https://doi.org/ 10.1080/01635580903441303.
116. Pollack AZ, Ranasinghe S, Sjaarda LA, Mumford SL. 2014. Cadmium and reproductive health in women: a systematic review of the epidemiologic evidence. Curr Environ Health Rep 1(2):172-184, PMID: 27453808, https://doi.org/10.1007/ s40572-014-0013-0.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025. This work is published under ReproducedfromEnvironmentalHealthPerspectives (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
BACKGROUND: Uterine leiomyomata (UL), hormone-dependent neoplasms, are a major source of gynecologic morbidity. Metals are hypothesized to influence UL risk through endocrine disruption, and their effects may vary by vitamin D status. OBJECTIVE: We estimated associations of a metal mixture with incident UL, overall and by vitamin D status. METHODS: We analyzed data from the Study of Environment, Lifestyle and Fibroids, a Detroit-area prospective cohort study of 1,693 black women 23-35 years of age. We measured concentrations of 17 metals/metalloids in whole blood and 25-hydroxyvitamin D [25(OH)D] in serum collected at baseline (2010-2012). Participants underwent ultrasonography at baseline and after 20 months to detect UL. We used Bayesian kernel machine regression to estimate adjusted associations (β) of the metal mixture with probit of incident UL. We also ran Cox regression models with interaction terms to estimate incidence rate ratios (IRR) by vitamin D status. RESULTS: Among 1,132 UL-free participants at baseline, 832 (73%) had vitamin D deficiency [25(OH)D <20 ng/mL] and 117 (10%) developed UL within 20 months. Increasing all metals from their 50th to 75th percentiles was weakly positively associated with UL overall [β = 0.06; 95% credible interval (CrI): -0.03, 0.16] and among vitamin D-deficient participants (β=0.13; 95% CrI: 0.01, 0.24), driven by cadmium (overall and vitamin D-deficient) and mercury (vitamin D-deficient only). Increasing cadmium from its 25th to 75th percentile was positively associated with UL overall (β= 0.03; 95% CrI: -0.05, 0.11) and among vitamin D-deficient participants (β = 0.13, 95% CrI: 0.02, 0.24). In Cox models, cadmium [IRR = 1.55; 95% confidence interval (CI): 1.07, 2.24, per 1-unit increase in standardized concentration] and mercury (IRR = 1.38; 95% CI: 0.99, 1.92) were positively associated with UL among vitamin D-deficient participants. Discussion: The metal mixture was positively associated with incident UL, but the association was weak and imprecise. We observed a stronger association among vitamin D-deficient participants that was driven by cadmium and mercury. https://doi.org/10.1289/EHP15218
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
1 Department of Epidemiology, Boston University School of Public Health, Boston, Massachusetts, USA
2 Department of Environmental Health, Boston University School of Public Health, Boston, Massachusetts, USA
3 Department of Epidemiology and Biostatistics, College of Human Medicine, Michigan State University, East Lansing, Michigan, USA