Environmental phenols are ubiquitous endocrine disruptors and putatively diabetogenic. However, data during pregnancy are scant. We investigated the prospective associations between pregnancy phenol concentrations and gestational diabetes mellitus (GDM) risk. In a nested matched case-control study of 111 individuals with GDM and 222 individuals without GDM within the prospective PETALS cohort, urinary bisphenol A (BPA), BPA substitutes (bisphenol F and bisphenol S [BPS]), benzophenone-3, and triclosan were quantified during the first and second trimesters. Cumulative concentrations across the two times were calculated using the area under the curve (AUC). Multivariable conditional logistic regression examined the association of individual phenols with GDM risk. We conducted mixture analysis using Bayesian kernel machine regression. We a priori examined effect modification by Asian/Pacific Islander (A/PI) race/ethnicity resulting from the case-control matching and highest GDM prevalence among A/PIs. Overall, first-trimester urinary BPS was positively associated with increased risk of GDM (adjusted odds ratio comparing highest vs. lowest tertile [aORT3 vs. T1] 2.12 [95% CI 1.00–4.50]). We identified associations among non-A/Ps, who had higher phenol concentrations than A/PIs. Among non-A/PIs, first-trimester BPA, BPS, and triclosan were positively associated with GDM risk (aORT3 vs. T1 2.91 [95% CI 1.05–8.02], 4.60 [1.55–13.70], and 2.88 [1.11–7.45], respectively). Triclosan in the second trimester and AUC were positively associated with GDM risk among non-A/PIs (P < 0.05). In mixture analysis, triclosan was significantly associated with GDM risk. Urinary BPS among all and BPA, BPS, and triclosan among non-A/PIs were associated with GDM risk. Pregnant individuals should be aware of these phenols’ potential adverse health effects.

As a common metabolic complication during pregnancy, gestational diabetes mellitus (GDM) has increased in prevalence by 33–90% over the last decades and is currently affecting 6–12% of pregnant individuals worldwide (1,2). Burgeoning evidence illustrates the plethora of wide-ranging adverse effects of GDM on pregnant individuals and their offspring, including perinatal complications and long-term diabetes and cardiovascular diseases, representing a salient public health issue (3). Therefore, it is imperative to identify risk factors for GDM, especially overlooked or understudied factors, given the rapidly evolving landscape of environmental exposures.

The increase in GDM prevalence is concomitant with the increasing use of endocrine-disrupting chemicals (EDCs). Accumulating evidence has linked EDCs to obesity and altered glucose homeostasis via mechanisms of interfering with the synthesis, activity, and elimination of natural hormones regulating glucose metabolism (4,5). One such group of EDCs, with ubiquitous human exposure, comprises phenols used in the manufacture of food packaging, thermal receipts, and other consumer products (bisphenol A [BPA]) and personal care products (e.g., triclosan as an antiseptic agent and benzophenone-3 [BP-3] as a sunscreen agent). Despite previous investigations of BPA in relation to obesity and type 2 diabetes among nonpregnant individuals (6,7), data on exposure to BPA and other phenols during pregnancy in relation to risk of GDM are scarce. Moreover, underlying these limited data are important limitations that further undermine the evidence: small sample sizes, skewed racial/ethnic distributions, assessment of phenol concentrations at or after GDM diagnosis, single measure of phenols, and use of serum rather than urine to quantify phenol concentrations, which may result in external contamination and exposure misclassification, because phenol concentrations are several orders of magnitude lower in serum than in urine (812). Recognizing the knowledge gaps, several societies and groups have called for timely action to prioritize research on health effects of EDCs including phenols, especially among vulnerable subgroups such as pregnant individuals (1315). Furthermore, the raised concern of potential harmful health effects of BPA has prompted a gradual shift to products labeled BPA free and increased use of BPA substitutes, such as bisphenol F (BPF) and bisphenol S (BPS) (16). The changing landscape of phenols in commerce highlights the need to examine the as-yet understudied potential health effects of BPA substitutes in a timely manner. Of note, substantial racial/ethnic differences were observed in concentrations of urinary phenols (lowest among Asians based on U.S. biomonitoring data) (17) and prevalence of GDM (highest among Asians) (18).

We aimed to investigate the associations of urinary concentrations of BPA, two BPA substitutes (BPF and BPS), BP-3, and triclosan in early to midpregnancy with risk of GDM in a case-control study nested within a contemporary multiracial/ethnic cohort of pregnant individuals. We also a priori examined effect modification by Asian/Pacific Islander race/ethnicity resulting from the case-control matching (Asian/Pacific Islander vs. non-Asian/Pacific Islander) given the highest GDM prevalence among Asians/Pacific Islanders.

Study Design and Population

The Pregnancy Environment and Lifestyle Study (PETALS) is a population-based longitudinal multiracial/ethnic pregnancy cohort study, with a primary objective of investigating in a nested case-control study the associations of phenol concentrations in early to midpregnancy with risk of GDM. The study design has been previously described in detail (19). The study population was drawn from the membership of Kaiser Permanente Northern California (KPNC), an integrated health care delivery system serving 4.5 million members, who are highly representative of the population living in the served geographic area (20,21). Eligibility criteria included carrying singletons and no evidence of recognized pre-existing cancer, diabetes (given the GDM outcome), or hepatic diseases, such as hepatitis C or liver cirrhosis (which may reduce the ability of the liver in metabolizing phenols) based on review of medical records. Individuals were invited to participate in the study before 11 weeks’ gestation. Urine samples were collected at study visit 1 in the first trimester (gestational weeks mean ± SD 14.0 ± 2.3) and visit 2 in the second trimester (20.5 ± 2.4). The study was approved by the human subjects committee of the Kaiser Foundation Research Institute. All participants provided written informed consent. The analysis of deidentified specimens at the Centers for Disease Control and Prevention laboratory was determined not to constitute engagement in human subjects research.

In this clinical setting, pregnant individuals were universally screened for GDM with the 50-g, 1-h glucose challenge test at ∼24–28 weeks’ gestation. If the screening test was abnormal (>7.8 mmol/L), a 100-g, 3-h oral glucose tolerance test (OGTT) was performed after a 12-h fast. We identified GDM cases by any of the following criteria used at KPNC: 1) ≥2 plasma glucose values after the OGTT meeting or exceeding the Carpenter-Coustan thresholds recommended by the American College of Obstetricians and Gynecologists (22) or 2) fasting glucose ≥5.1 mmol/L performed alone or during the OGTT recommended by the International Association of Diabetes and Pregnancy Study Groups and American Diabetes Association (23).

Of 3,346 pregnant individuals who completed visit 1 (baseline) in the PETALS cohort, 162 (4.8%) did not have data on GDM screening; specifically, 32 had a pregnancy loss (1.0%), 46 (1.4%) were no longer KPNC members, and 84 (2.5%) were not screened. Among individuals screened for GDM, there were 310 with GDM. We conducted a nested case-control study by quantifying urinary concentrations of phenols among 111 individuals with GDM and 222 individuals without GDM who delivered between September 2015 and June 2017. GDM case subjects and control subjects without GDM were matched 2:1 according to age (±5 years), race/ethnicity (Asian/Pacific Islander vs. non-Asian/Pacific Islander), calendar time for enrollment (±3 months), gestational weeks at the first study visit (±3 weeks), and facility. Notably, we matched on Asian/Pacific Islander race/ethnicity because the highest prevalence of GDM was observed in this group compared with other racial/ethnic groups. Selected GDM case subjects and control subjects without GDM did not differ from their counterparts in the parent study with respect to major sociodemographic or medical characteristics. All 111 GDM case subjects and 222 control subjects without GDM contributed urine samples at study visit 1. At study visit 2, 6 GDM case subjects had missing urine samples and 12 of the pregnant individuals who developed GDM were diagnosed before urine collection. Therefore, 54 individuals (18 GDM case subjects and 36 matched control subjects without GDM) were excluded from analysis in the second trimester.

Urine Sample Collection

Urine specimens were collected during a typical day as defined by participants. Following Centers for Disease Control and Prevention practices, urine samples were collected in sterile polypropylene cups and stored at −80°C (24). A rigorous protocol was implemented to reduce external chemical contamination in the collection process. Participants were instructed not to use a wipe before urine collection, because wipes may contain triclosan, among other phenols; participants and study staff were asked not to use sunscreen or lotions on the day of the study visit, because they may contain BP-3. Every 3 months, we ran procedure field blanks, which entailed running and testing a solution of blank water (i.e., OmniSolv liquid chromatography–mass spectrometry water) through our collection process to ensure it did not contain detectable levels of BPA, BP-3, or triclosan. Throughout the study period, we did not find detectable levels of the target phenols in the procedure field blanks.

Quantification of Urinary Phenol Biomarkers

The total (free plus conjugated) concentrations of phenols at each study visit were determined using a method based on online solid-phase extraction coupled to high-performance liquid chromatography-atmospheric pressure chemical ionization-isotope dilution tandem mass spectrometry, as specified in detail previously (25). We imputed phenol concentrations below the limit of detection (LOD) (Supplementary Table 1), with the LOD divided by the square root of 2. Accuracy (percent of spike recovery) and precision (coefficients of variation from repeated measurements of quality control materials over 1 month) ranged from 91–107% and 2.5–8.5% for all analytes, respectively. Urinary creatinine, measured by a kinetic Jaffe reaction as previously described, was used to account for the degree of urine dilution (26). To estimate cumulative exposure to the environmental phenols across the two study visits, we calculated areas under the curve (AUCs) (formula in Table 2) to incorporate both the concentration and timing of exposure (27).

Covariates

Potential covariates were selected based on biologic plausibility and prior knowledge: maternal age at childbirth (continuous), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, Asian/Pacific Islander, or other/unknown), education (high school or less, some college, or college graduate or above), nulliparity (yes or no), prepregnancy BMI (<25.0, 25.0–29.9, or ≥30.0 kg/m2), and dietary quality and physical activity during pregnancy. Information on participants’ sociodemographic variables was collected by structured questionnaires administered at study visit 1. Prepregnancy BMI was calculated as prepregnancy weight (kg) measured by clinical staff on average 11 weeks before conception and abstracted from electronic health records (97.5%) or by self-report (2.5%), divided by squared height (m2) measured at study visit 1. Overall dietary quality was assessed by the Alternate Health Eating Index for Pregnancy based on habitual dietary intake during the previous 3 months via the Block food frequency questionnaire administered in the first trimester (Supplementary Table 2) (28). Moderate to vigorous physical activity (metabolic equivalent of tasks-hours per day) was assessed by the validated Pregnancy Physical Activity Questionnaire in the first trimester (29). A covariate was included in the final model if the coefficient of exposure of interest changed by ≥10%.

Statistical Methods

Differences in participant characteristics and concentrations and detection frequencies of urinary phenols between GDM case subjects and control subjects without GDM were assessed by Student t test for parametric continuous variables, Mann-Whitney U test for nonparametric continuous variables, and Pearson χ2 test for categorical variables.

Multivariable conditional logistic regression models were fitted to examine the association of individual urinary phenols (BPA, BPF, BPS, BP-3, and triclosan) quantified in samples collected in the first and second trimesters, respectively, and the AUCs across the two times with risk of GDM, adjusting for the aforementioned covariates and urinary creatinine. Covariate-adjusted standardization plus creatinine adjustment for urinary biomarkers have been recommended to account for urinary dilution (30). To ensure that exposures preceded the outcome, we excluded 18 case subjects and 36 matched control subjects from the second trimester and cumulative AUC analyses whose urine samples were collected in the second trimester after GDM diagnosis. We analyzed urinary concentrations of each phenol (except BPF) in tertiles based on the distribution among control subjects and continuously per interquartile range (IQR) increment of log-transformed values; BPF was parameterized as detected versus nondetected, and its AUC was not calculated because of its relatively low detection frequency. We further used nonparametric cubic spline regression to explore the potential nonlinearity in the associations between urinary phenols (except BPF) on a continuum and risk of GDM.

In addition, to evaluate whether the associations of urinary phenols with GDM risk varied by race/ethnicity (Asian/Pacific Islander vs. non-Asian/Pacific Islander) as a result of matching, we a priori included cross-product terms in the multivariable conditional logistic regression models to assess the potential effect modification. P for interaction was obtained by the likelihood ratio test. To rest the robustness of our findings, we performed several sensitivity analyses. Given that food is the primary source of bisphenols in the U.S. general population (31), we additionally adjusted for dietary quality during pregnancy. Similarly, given that sunscreen is the primary source of BP-3 and activity may be associated with sunscreen use and risk of GDM (32,33), we further adjusted for moderate to vigorous physical activity during pregnancy (Spearman ρ with first-trimester concentrations of BP-3: 0.19; P value <0.001) and season based on last menstrual period.

To assess the effects of phenols as mixture, we fit Bayesian kernel machine regression (BKMR) with probit regression models for risk of GDM (34). Following the same approach as our main analyses, we fit BKMR models among all, by race/ethnicity (Asian/Pacific Islander vs. non-Asian/Pacific Islander), and by first trimester, second trimester, and AUC. The BKMR phenol mixture analysis included four phenols: BPA, BPS, BP-3, and triclosan; BPF could not be included in the mixture analysis because of its low detection frequency. BKMR models were adjusted for urinary creatinine level, maternal age at childbirth, prepregnancy BMI, race/ethnicity, calendar time at enrollment, gestational weeks at urine collection, and facility. Of note, covariates included the matching factors, because probit models do not directly accommodate the matched case-control design. To obtain convergence in the BKMR models, we used a preadjustment method for covariates that first fit a frequentist probit regression model with all covariates and no phenols and then included the fitted values as a single adjustment covariate in the BKMR models. Each BKMR model fit used 50,000 Markov Chain Monte Carlo iterations, and the trace plots of model parameters were visually inspected to ensure model convergence. To summarize the results of the mixture analysis, we report the posterior inclusion probabilities (PIPs) and cross-section plots of the exposure response function. To examine the association of the phenol mixture with risk of GDM, we re-estimated the BKMR model with the phenol of interest forced into the mixture and used posterior samples to estimate the odds ratio (OR) and 95% credible interval for GDM by 10% increments of the specific phenol, with all other phenols set to the median.

All analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC) and R version 4.1.0 (R Core Team).

Data and Resource Availability

Extracted data are available within the publication and its online Supplementary Material. A deidentified analytic data set used in this study can be shared with qualified researchers subject to approval by the Kaiser Foundation Research Institute Human Subjects Committee and by the human subjects committees at the institutions requesting the data and a signed data sharing agreement.

Individuals with GDM were more likely to be overweight or obese before pregnancy (73.9 vs. 50.4%) and in the Hispanic ethnic group (43.2 vs. 28.4%) compared with control subjects without GDM (both P < 0.05), whereas the two groups did not differ by age, Asian/Pacific Islander racial/ethnic group, or gestational age at urine collection because of the case-control matching (Table 1). Compared with Asians/Pacific Islanders, non-Asians/Pacific Islanders were more likely to be younger at childbirth, physically active, and overweight or obese before pregnancy; have an education level lower than high school; and have daily use of food or beverage in cans and plastic or containers and less likely to be nulliparous (all P < 0.05) (Supplementary Table 3).

Detection Frequency and Concentrations of Urinary Phenols

Compared with control subjects without GDM, GDM case subjects had a higher detection frequency of BPA in the second trimester (79.6 vs. 66.7%) (Table 2), higher cumulative levels (i.e., AUC) of BPS across the two trimesters (145.7 vs. 103.9 ng/mL × day), and higher median concentrations of triclosan in the second trimester (4.3 vs. 2.3 ng/mL) and AUCs across the two trimesters (1,192.9 vs. 592.6 ng/mL × day; all P < 0.05). Because of the low detection frequency of BPF in both GDM case subjects (33.3–42.3%) and control subjects without GDM (29.7–31.7%) at either time point, no trimester-specific median or cumulative AUCs were calculated. Compared with Asians/Pacific Islanders, non-Asians/Pacific Islanders had significantly higher detection frequencies of BPA and BPS in both the first and second trimesters and consistently higher concentrations of BPA in the second trimester and cumulatively across the two trimesters, and higher concentrations of BPS regardless of the time point (all P < 0.05) (Supplementary Table 4). Despite similar detection frequencies, non-Asians/Pacific Islanders compared with Asians/Pacific Islanders had higher concentrations of BP-3 in the second trimester and cumulatively across the first two trimesters (all P < 0.05). Overall, urinary concentrations of BPF, BPS, and triclosan were similar, whereas concentrations of BPA were lower and BP-3 were higher in our study sample (2015–2017) as compared with a representative sample of noninstitutionalized U.S. women in the 2015–2016 National Health and Nutrition Examination Survey (NHANES) (Supplementary Table 5) (17).

Urinary Phenols in Early to Midpregnancy and Risk of GDM

Among all case and control subjects, after adjusting for covariates, only first-trimester urinary BPS was associated with risk of GDM (adjusted OR comparing the highest vs. lowest tertile [aORT3 vs. T1] 2.12 [95% CI 1.00–4.50]) (Table 3). Significant effect modification by race/ethnicity was observed, with significant associations between urinary phenols (BPA, BPS, BP-3, and triclosan) and GDM risk among non-Asians/Pacific Islanders only (P for interaction < 0.05). Among non-Asians/Pacific Islanders, aORT3 vs. T1 for first-trimester urinary BPA concentrations was 2.91 (95% CI 1.05–8.02), and the corresponding estimate comparing the second versus lowest tertile (aORT2 vs. T1) in the second trimester was 3.70 (95% CI 1.00–13.60). Urinary BPS concentrations were positively associated with GDM risk (first trimester: aORT3 vs. T1 4.60 [95% CI 1.55–13.70]; AUC across two trimesters: 8.21 [1.79–37.80]). Similarly, urinary triclosan concentrations in the first (aORT3 vs. T1 2.88 [95% CI 1.11–7.45) and second trimesters (4.12 [1.01–16.90]) and the AUC across the two trimesters (aORT2 vs. T1 5.51 [1.04–29.20]; aORT3 vs. T1 5.68 [1.40–23.00]) were positively associated with GDM risk. No association between BP-3 and GDM risk was found. Among non-Asians/Pacific Islanders, when urinary phenols were modeled continuously per IQR increase of log-transformed values, concentrations of triclosan, but not other phenols, during the first and second trimesters and across the two trimesters were positively associated with 1.02- (95% CI 1.00–1.04), 1.05- (1.00–1.11), and 1.14-fold (1.02–1.27) higher risk of GDM, respectively. Results remained robust after additional adjustment for dietary quality and physical activity (Supplementary Table 6). In a sensitivity analysis additionally adjusting for season, association between BP-3 and risk of GDM was slightly attenuated and remained null (data not shown).

We further explored among non-Asians/Pacific Islanders the nonlinear associations between urinary phenols and GDM risk using nonparametric cubic spline regression. When modeled on a continuum scale, urinary BPS in the first trimester and the AUC across the first two trimesters showed significant inverted U-shaped associations with GDM risk (Fig. 1). Similar to findings from the parametric models, urinary BPA and triclosan overall exhibited linear and positive associations with GDM risk, whereas BP-3 concentration was not associated with GDM risk (Supplementary Figs. 13).

Phenol BKMR Mixture Analysis

We further explored the association of urinary phenol mixture with GDM risk by conducting a mixture analysis using BKMR. Similar to the analysis of individual phenols, the strongest mixture effects were observed among non-Asians/Pacific Islanders, with weaker effects observed in the overall sample and Asians/Pacific Islanders. Among all, the PIPs for each phenol were low to moderate (ranging from 0.20 to 0.57) (Supplementary Table 7), indicating no strong evidence of contribution of each phenol in the mixture. The PIPs for each phenol were also low to moderate in the models among Asian/Pacific Islander individuals (ranging from 0.29 to 0.57) (Supplementary Table 7). Among non-Asians/Pacific Islanders, the PIPs for triclosan were high (0.82 for the first trimester, 0.74 for the second trimester, and 0.89 for AUC), demonstrating strong evidence for the contribution of triclosan to the mixture. The 80th vs. 20th percentile of triclosan exposure during the first and second trimesters and across the two trimesters was associated with 1.83- (95% CI 1.21–3.39), 1.74- (1.09–3.10), and 2.54-fold (1.40–4.89) increased risk of GDM, respectively, when all other urinary phenols were held at their median (Fig. 2). The cross-section plots of the exposure response function for each phenol showed a strong linear and positive association of triclosan with GDM risk, similar to our spline results (Supplementary Figs. 46). This finding is consistent with our analysis of individual phenols showing that triclosan was positively associated with GDM risk for triclosan exposures in the first trimester, second trimester, and AUC.

In this longitudinal nested case-control study within the prospective PETALS cohort, we provide, to our knowledge, the most extensive urinary concentration profiles of five phenols across the first and second trimesters of pregnancy in relation to risk of GDM. Among all, only urinary BPS was positively associated with GDM risk. However, we observed significant effect modification by race/ethnicity, with more pronounced associations of urinary BPA, BPS, and triclosan with GDM risk among non-Asians/Pacific Islanders compared with Asians/Pacific Islanders in both individual phenol and phenol mixture analyses. Our findings were robust after additional adjustment for lifestyle factors, including dietary quality and physical activity during pregnancy, that may affect exposure to BPA or BPS and triclosan, respectively.

Implications of Study Findings Within the Context of Previous Research

Previous data on bisphenols in relation to GDM risk were scarce and largely based on small-scale (including 18–89 GDM cases, with one exception of 167 cases) cross-sectional or retrospective studies with one point-in-time assessment of bisphenol concentrations at or after GDM diagnosis, generating inconsistent findings. Specifically, two cross-sectional studies that quantified urinary BPA at the time of OGTT in the U.S. (10) and China (35) reported null association with GDM, whereas one study in China (12) and one in Mexico (9) reported inverse associations of urinary BPA concentration during 1–3 days before delivery and on average at 23.3 weeks’ gestation (with unclear timing relative to OGTT) with GDM risk, respectively. Given that behaviors and metabolism may change after GDM treatment, it is difficult to establish causality based on cross-sectional or retrospective data. Therefore, we compared our findings with those from previous prospective studies, which illustrated potential racial/ethnic-specific BPA-GDM associations. Specifically, one Chinese study reported a null association of first-trimester urinary BPA with GDM risk (36), consistent with our observation of null associations of first- and second-trimester urinary BPA with GDM risk among Asians/Pacific Islanders. Furthermore, in line with our findings of positive associations of first- and second-trimester urinary BPA with GDM risk among non-Asians/Pacific Islanders, Chiu et al. (37) observed a positive association between second-trimester urinary BPA and blood glucose among primarily Caucasians in a U.S. fertility center. These racial/ethnic-specific BPA-GDM associations could be attributable to differences in participant characteristics and BPA concentrations, as observed in previous studies and our study showing significantly lower phenol concentrations among Asians/Pacific Islanders. The underlying reasons for lower concentrations of BPA and other phenols among Asians/Pacific Islanders could be partially explained by different behaviors, such as use of food and beverage in cans and plastic wraps or containers and level of physical activity, as shown in our study. In addition, emerging evidence among nonpregnant individuals suggests potentially different pathophysiology of hyperglycemia among Asians/Pacific Islanders versus other racial/ethnic groups, with impaired β-cell function playing a more pronounced role in the former as compared with insulin resistance among the latter (38,39). Importantly, given that race/ethnicity is a social, not biologic, construct (40), the potential roles of other lifestyle and environmental factors in the observed differences in phenol concentrations and risk of GDM by race/ethnicity remain unclear and warrant further investigation.

Furthermore, we report for the first time positive associations of urinary first-trimester BPS among all case and control subjects and first-trimester and cumulative levels across the first two trimesters of BPS with GDM risk among non-Asians/Pacific Islanders. The only previous study assessing BPS suggested a null association between first-trimester urinary BPS and GDM risk among pregnant Chinese individuals (36). Notably, the median first-trimester urinary BPS concentration among Asians/Pacific Islanders in our study was similar to that in the Chinese study (0.40 vs. 0.30 ng/mL) (36), which were both lower than concentration among non-Asians/Pacific Islanders in our study (0.60 ng/mL). A similar pattern was observed among U.S. individuals age ≥3 years in the 2015–2016 NHANES (17). These data may suggest a potential threshold effect of BPS. Interestingly, we did observe a U-shaped association between first-trimester BPS and GDM risk in our exploratory analysis.

Partially in line with our findings of a dose-response prospective relationship between urinary triclosan in the first or second trimester or across the two trimesters and GDM risk, Ouyang et al. (41) reported a positive but statistically nonsignificant association between urinary triclosan in late pregnancy and GDM risk. In contrast, studies among primarily Caucasians reported an inverse association of first-trimester triclosan with GDM in the U.K. (42) and a null association in Canada (43). Our findings extend the literature by illustrating the potential adverse metabolic effects of triclosan on increasing GDM risk, stimulating further investigation of its effects on other perinatal outcomes.

Previous prospective data on urinary BP-3 and GDM risk are lacking, with only one U.S. study reporting inverse associations of first- and second-trimester urinary BP-3 with blood glucose in late pregnancy (44); however, we did not observe a significant association between BP-3 in early to midpregnancy and GDM risk.

Of note, as shown by previous studies, BPA and substitutes, BP-3, and triclosan can transfer across the placenta (4548), imposing potential endocrine-disrupting risks to the developing fetus. Future studies investigating the associations of prenatal phenol exposure and fetal growth and development and the potential mediating role of GDM are warranted to better understand the potential adverse health effects of phenols.

Biologic Plausibility

Mechanisms whereby phenols may increase the risk of GDM remain to be fully elucidated. Phenols may act through several pathways, including dysregulation of glucose and insulin metabolism (49,50), oxidative stress and inflammation (51), and modification of epigenetic regulations (52). In vivo data have demonstrated that BPA may disrupt pancreatic β-cell function and induce insulin resistance (53), and in vitro data have suggested BPS may induce lipogenesis and hyperglycemia (54). Nascent data have also suggested the epigenetic toxicity of BPA via alterations of miRNA in placenta cell lines and serum samples among pregnant individuals (9,55). In animal models, triclosan has been shown to induce oxidative stress (56) and thyroid homeostasis impairment (57), which in turn have been linked to hyperglycemia. Given the ubiquitous human exposure to phenols, further research examining the comprehensive metabolic profile (including but not limited to glucose and insulin metabolism and the other potential pathways mentioned above) in early pregnancy is needed to better elucidate the underlying mechanisms for phenol-GDM associations.

Strengths and Limitations

Notable strengths of our study are the longitudinal design and the multiple urine sample collections across early to midpregnancy before GDM diagnosis, which allowed for more precise exposure assessment by using the cumulative levels across time points. Moreover, we had comprehensive data on lifestyle factors, including diet and physical activity during pregnancy. Despite the possibility of overadjustment because these lifestyle factors could be exposure routes for phenols, our sensitivity analyses showed that the findings were robust to potential confounding resulting from diet quality and physical activity.

Some limitations of our study merit discussion. Phenol concentrations were assessed using a single spot urine sample collected at each study visit. However, previous data have illustrated that a single spot urinary phenol assessment could predict the tertile categorization of a 24-h collection with an accuracy rate of 87% (58,59). Furthermore, given the precedence of phenol exposure assessment to GDM diagnosis, in the event of any exposure misclassification, it would be nondifferential to GDM status and thus could have drawn the effect estimates toward the null. Additionally, we calculated the AUC as a proxy of overall exposure to phenols in early to midpregnancy, which showed overall consistent results with time-specific analysis. Given the relatively low detection frequency of BPF, which is consistent with the 2015–2016 NHANES data (17), caution is warranted in interpreting findings of BPF-GDM association. To achieve a cost-efficient design, we matched GDM case subjects and control subjects without GDM by Asian/Pacific Islander race/ethnicity, given the highest prevalence of GDM in this population. Despite our larger sample size relative to most previous studies, we had limited statistical power to stratify by other racial/ethnic groups (White, Hispanic, and Black) and racial/ethnic subpopulations (e.g., Chinese, Filipino, Indian, and so on within Asians and Cuban, Mexican, Puerto Rican, and so on within Hispanics) that are potentially heterogeneous in exposure to phenols and risk of GDM (60). Future larger-scale studies of multiracial/ethnic pregnant individuals are needed.

Conclusions

In conclusion, in this prospective study, we observed positive associations of first-trimester BPS concentrations with GDM risk among all individuals and of BPA and triclosan concentrations in early to midpregnancy with GDM risk among non-Asians/Pacific Islanders, who had higher urinary phenol concentrations compared with their Asian/Pacific Islander counterparts. Our novel finding of a positive association of urinary BPS with risk of GDM may have timely public health ramifications. Given the emerging use of BPS as a BPA substitute, continued biomonitoring and increased awareness of its potential adverse health effects are warranted. Our finding of a positive association between triclosan in early to midpregnancy and GDM risk calls for further investigation of its effects on other perinatal outcomes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.21171337.

Acknowledgments. The authors thank all staff and participants in the Pregnancy Environment and Lifestyle Study and Gestational Weight Gain and Optimal Wellness randomized controlled trial for their valuable contributions.

Funding. This research was supported by grants from the National Institute of Environmental Health Sciences (R01ES019196) to A.F., National Institute of Child Health and Human Development (R01HD073572) to A.F., National Institutes of Health Office of Directors (UG3OD023289 and UH3OD023289) to A.F., National Institute of Diabetes and Digestive and Kidney Diseases (K01DK120807) to Y.Z., and National Heart, Lung, and Blood Institute (R01HL157666) to Y.Z.

The funding source had no role in the design of the study, analysis or interpretation of the findings, writing of the manuscript, or decision to submit the paper for publication. The findings and conclusions of this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC). Use of trade names is for identification only and does not imply endorsement by the CDC, Public Health Service, or U.S. Department of Health and Human Services.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.Z. drafted the manuscript. Y.Z., M.M.H., A.M.C., and A.F. contributed to data acquisition. Y.Z., M.M.H., and A.F. contributed to study concept and design. Y.Z., S.E.A., J.F., C.P.Q., and A.F. contributed to analysis and interpretation of data. All authors contributed to critical revision of the manuscript for important intellectual content. Y.Z. and A.F. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Ferrara
A
,
Kahn
HS
,
Quesenberry
CP
,
Riley
C
,
Hedderson
MM
.
An increase in the incidence of gestational diabetes mellitus: northern California, 1991-2000
.
Obstet Gynecol
2004
;
103
:
526
533
2.
Zhu
Y
,
Zhang
C
.
Prevalence of gestational diabetes and risk of progression to type 2 diabetes: a global perspective
.
Curr Diab Rep
2016
;
16
:
7
3.
McIntyre
HD
,
Catalano
P
,
Zhang
C
,
Desoye
G
,
Mathiesen
ER
,
Damm
P
.
Gestational diabetes mellitus
.
Nat Rev Dis Primers
2019
;
5
:
47
4.
Thayer
KA
,
Heindel
JJ
,
Bucher
JR
,
Gallo
MA
.
Role of environmental chemicals in diabetes and obesity: a National Toxicology Program workshop review
.
Environ Health Perspect
2012
;
120
:
779
789
5.
Song
Y
,
Chou
EL
,
Baecker
A
, et al
.
Endocrine-disrupting chemicals, risk of type 2 diabetes, and diabetes-related metabolic traits: A systematic review and meta-analysis
.
J Diabetes
2016
;
8
:
516
532
6.
Hwang
S
,
Lim
JE
,
Choi
Y
,
Jee
SH
.
Bisphenol A exposure and type 2 diabetes mellitus risk: a meta-analysis
.
BMC Endocr Disord
2018
;
18
:
81
7.
Lakind
JS
,
Goodman
M
,
Mattison
DR
.
Bisphenol A and indicators of obesity, glucose metabolism/type 2 diabetes and cardiovascular disease: a systematic review of epidemiologic research
.
Crit Rev Toxicol
2014
;
44
:
121
150
8.
Fisher
M
,
Arbuckle
TE
,
Mallick
R
, et al
.
Bisphenol A and phthalate metabolite urinary concentrations: daily and across pregnancy variability
.
J Expo Sci Environ Epidemiol
2015
;
25
:
231
239
9.
Martínez-Ibarra
A
,
Martínez-Razo
LD
,
Vázquez-Martínez
ER
, et al
.
Unhealthy levels of phthalates and bisphenol A in Mexican pregnant women with gestational diabetes and its association to altered expression of miRNAs involved with metabolic disease
.
Int J Mol Sci
2019
;
20
:
3343
10.
Robledo
C
,
Peck
JD
,
Stoner
JA
, et al
.
Is bisphenol-A exposure during pregnancy associated with blood glucose levels or diagnosis of gestational diabetes?
J Toxicol Environ Health A
2013
;
76
:
865
873
11.
Shapiro
GD
,
Dodds
L
,
Arbuckle
TE
, et al
.
Exposure to phthalates, bisphenol A and metals in pregnancy and the association with impaired glucose tolerance and gestational diabetes mellitus: the MIREC study
.
Environ Int
2015
;
83
:
63
71
12.
Wang
X
,
Wang
X
,
Chen
Q
, et al
.
Urinary bisphenol A concentration and gestational diabetes mellitus in Chinese women
.
Epidemiology
2017
;
28
(
Suppl. 1
):
S41
S47
13.
Di Renzo
GC
,
Conry
JA
,
Blake
J
, et al
.
International Federation of Gynecology and Obstetrics opinion on reproductive health impacts of exposure to toxic environmental chemicals
.
Int J Gynaecol Obstet
2015
;
131
:
219
225
14.
ACOG Committee Opinion No 575
.
Exposure to toxic environmental agents
.
Fertil Steril
2013
;
100
:
931
934
15.
Pergialiotis
V
,
Kotrogianni
P
,
Christopoulos-Timogiannakis
E
,
Koutaki
D
,
Daskalakis
G
,
Papantoniou
N
.
Bisphenol A and adverse pregnancy outcomes: a systematic review of the literature
.
J Matern Fetal Neonatal Med
2018
;
31
:
3320
3327
16.
Rochester
JR
,
Bolden
AL
.
Bisphenol S and F: a systematic review and comparison of the hormonal activity of bisphenol A substitutes
.
Environ Health Perspect
2015
;
123
:
643
650
17.
Centers for Disease Control and Prevention
.
Fourth National Report on Human Exposure to Environmental Chemicals Updated Tables, 2021
.
Accessed 23 October 2021. Available from https://www.cdc.gov/exposurereport/data_tables.html
18.
Hedderson
M
,
Ehrlich
S
,
Sridhar
S
,
Darbinian
J
,
Moore
S
,
Ferrara
A
.
Racial/ethnic disparities in the prevalence of gestational diabetes mellitus by BMI
.
Diabetes Care
2012
;
35
:
1492
1498
19.
Zhu
Y
,
Hedderson
MM
,
Feng
J
,
Mevi
AA
,
Ferrara
A
.
The Pregnancy Environment and Lifestyle Study (PETALS): a population-based longitudinal multi-racial birth cohort
.
BMC Pregnancy Childbirth
2017
;
17
:
122
20.
Gordon
N
,
Lin
T
.
The Kaiser Permanente Northern California Adult Member Health Survey
.
Perm J
2016
;
20
:
15
225
21.
Go
AS
,
Lee
WY
,
Yang
J
,
Lo
JC
,
Gurwitz
JH
.
Statin therapy and risks for death and hospitalization in chronic heart failure
.
JAMA
2006
;
296
:
2105
2111
22.
American College of Obstetricians and Gynecologists
.
Practice bulletin No. 180: gestational diabetes mellitus
.
Obstet Gynecol
2017
;
130
:
e17
e37
23.
Metzger
BE
,
Gabbe
SG
,
Persson
B
, et al.;
International Association of Diabetes and Pregnancy Study Groups Consensus Panel
.
International Association of Diabetes and Pregnancy Study Groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy
.
Diabetes Care
2010
;
33
:
676
682
24.
Centers for Disease Control and Prevention
.
Blood and urine collection
.
25.
Centers for Disease Control and Prevention
.
Laboratory procedure manual: benzophenone-3, bisphenol A, bisphenol F, bisphenol S, 2,4-dichlorophenol, 2,5-dichlorophenol, methyl-, ethyl-, propyl-, and butyl parabens, triclosan, and triclocarban
.
26.
Larsen
K
.
Creatinine assay in the presence of protein with LKB 8600 Reaction Rate Analyser
.
Clin Chim Acta
1972
;
38
:
475
476
27.
Smith
TJ
,
Kriebel
D
.
A Biologic Approach to Environmental Assessment and Epidemiology
.
Oxford, U.K.
,
Oxford University Press
,
2010
28.
Block
G
,
Hartman
AM
,
Dresser
CM
,
Carroll
MD
,
Gannon
J
,
Gardner
L
.
A data-based approach to diet questionnaire design and testing
.
Am J Epidemiol
1986
;
124
:
453
469
29.
Chasan-Taber
L
,
Schmidt
MD
,
Roberts
DE
,
Hosmer
D
,
Markenson
G
,
Freedson
PS
.
Development and validation of a pregnancy physical activity questionnaire
.
Med Sci Sports Exerc
2004
;
36
:
1750
1760
30.
O’Brien
KM
,
Upson
K
,
Buckley
JP
.
Lipid and creatinine adjustment to evaluate health effects of environmental exposures
.
Curr Environ Health Rep
2017
;
4
:
44
50
31.
van Woerden
I
,
Payne-Sturges
DC
,
Whisner
CM
,
Bruening
M
.
Dietary quality and bisphenols: trends in bisphenol A, F, and S exposure in relation to the Healthy Eating Index using representative data from the NHANES 2007-2016
.
Am J Clin Nutr
2021
;
114
:
669
682
32.
Ehrlich
SF
,
Ferrara
A
,
Hedderson
MM
,
Feng
J
,
Neugebauer
R
.
Exercise during the first trimester of pregnancy and the risks of abnormal screening and gestational diabetes mellitus
.
Diabetes Care
2021
;
44
:
425
432
33.
Calafat
AM
,
Wong
L-Y
,
Ye
X
,
Reidy
JA
,
Needham
LL
.
Concentrations of the sunscreen agent benzophenone-3 in residents of the United States: National Health and Nutrition Examination Survey 2003--2004
.
Environ Health Perspect
2008
;
116
:
893
897
34.
Bobb
JF
,
Claus Henn
B
,
Valeri
L
,
Coull
BA
.
Statistical software for analyzing the health effects of multiple concurrent exposures via Bayesian kernel machine regression
.
Environ Health
2018
;
17
:
67
35.
Hou
Y
,
Li
S
,
Xia
L
, et al
.
Associations of urinary phenolic environmental estrogens exposure with blood glucose levels and gestational diabetes mellitus in Chinese pregnant women
.
Sci Total Environ
2021
;
754
:
142085
36.
Zhang
W
,
Xia
W
,
Liu
W
, et al
.
Exposure to bisphenol A substitutes and gestational diabetes mellitus: a prospective cohort study in China
.
Front Endocrinol (Lausanne)
2019
;
10
:
262
37.
Chiu
Y-H
,
Mínguez-Alarcón
L
,
Ford
JB
, et al.;
EARTH Study Team
.
Trimester-specific urinary bisphenol A concentrations and blood glucose levels among pregnant women from a fertility clinic
.
J Clin Endocrinol Metab
2017
;
102
:
1350
1357
38.
Li
W
,
Zhang
S
,
Liu
H
, et al
.
Different associations of diabetes with β-cell dysfunction and insulin resistance among obese and nonobese Chinese women with prior gestational diabetes mellitus
.
Diabetes Care
2014
;
37
:
2533
2539
39.
Gujral
UP
,
Mohan
V
,
Pradeepa
R
, et al
.
Ethnic variations in diabetes and prediabetes prevalence and the roles of insulin resistance and β-cell function: the CARRS and NHANES studies
.
J Clin Transl Endocrinol
2016
;
4
:
19
27
40.
Flanagin
A
,
Frey
T
,
Christiansen
SL
;
AMA Manual of Style Committee
.
Updated guidance on the reporting of race and ethnicity in medical and science journals
.
JAMA
2021
;
326
:
621
627
41.
Ouyang
F
,
Tang
N
,
Zhang
H-J
, et al
.
Maternal urinary triclosan level, gestational diabetes mellitus and birth weight in Chinese women
.
Sci Total Environ
2018
;
626
:
451
457
42.
Fisher
BG
,
Frederiksen
H
,
Andersson
A-M
, et al
.
Serum phthalate and triclosan levels have opposing associations with risk factors for gestational diabetes mellitus
.
Front Endocrinol (Lausanne)
2018
;
9
:
99
43.
Shapiro
GD
,
Arbuckle
TE
,
Ashley-Martin
J
, et al
.
Associations between maternal triclosan concentrations in early pregnancy and gestational diabetes mellitus, impaired glucose tolerance, gestational weight gain and fetal markers of metabolic function
.
Environ Res
2018
;
161
:
554
561
44.
Wang
Z
,
Mínguez-Alarcón
L
,
Williams
PL
, et al
.
Perinatal urinary benzophenone-3 concentrations and glucose levels among women from a fertility clinic
.
Environ Health
2020
;
19
:
45
45.
Bai
X
,
Zhang
B
,
He
Y
, et al
.
Triclosan and triclocarbon in maternal-fetal serum, urine, and amniotic fluid samples and their implication for prenatal exposure
.
Environ Pollut
2020
;
266
:
115117
46.
Balakrishnan
B
,
Henare
K
,
Thorstensen
EB
,
Ponnampalam
AP
,
Mitchell
MD
.
Transfer of bisphenol A across the human placenta
.
Am J Obstet Gynecol
2010
;
202
:
393.e1
393.e7
47.
Liu
J
,
Li
J
,
Wu
Y
, et al
.
Bisphenol A metabolites and bisphenol S in paired maternal and cord serum
.
Environ Sci Technol
2017
;
51
:
2456
2463
48.
Song
S
,
He
Y
,
Huang
Y
, et al
.
Occurrence and transfer of benzophenone-type ultraviolet filters from the pregnant women to fetuses
.
Sci Total Environ
2020
;
726
:
138503
49.
Lind
PM
,
Lind
L
.
Endocrine-disrupting chemicals and risk of diabetes: an evidence-based review
.
Diabetologia
2018
;
61
:
1495
1502
50.
Ruiz
D
,
Becerra
M
,
Jagai
JS
,
Ard
K
,
Sargis
RM
.
Disparities in environmental exposures to endocrine-disrupting chemicals and diabetes risk in vulnerable populations
.
Diabetes Care
2018
;
41
:
193
205
51.
Watkins
DJ
,
Ferguson
KK
,
Anzalota Del Toro
LV
,
Alshawabkeh
AN
,
Cordero
JF
,
Meeker
JD
.
Associations between urinary phenol and paraben concentrations and markers of oxidative stress and inflammation among pregnant women in Puerto Rico
.
Int J Hyg Environ Health
2015
;
218
:
212
219
52.
Singh
S
,
Li
SS-L
.
Epigenetic effects of environmental chemicals bisphenol A and phthalates
.
Int J Mol Sci
2012
;
13
:
10143
10153
53.
Li
H
,
Park
H-M
,
Ji
H-S
, et al
.
Phenolic-enriched blueberry-leaf extract attenuates glucose homeostasis, pancreatic β-cell function, and insulin sensitivity in high-fat diet-induced diabetic mice
.
Nutr Res
2020
;
73
:
83
96
54.
Héliès-Toussaint
C
,
Peyre
L
,
Costanzo
C
,
Chagnon
MC
,
Rahmani
R
.
Is bisphenol S a safe substitute for bisphenol A in terms of metabolic function? An in vitro study
.
Toxicol Appl Pharmacol
2014
;
280
:
224
235
55.
Avissar-Whiting
M
,
Veiga
KR
,
Uhl
KM
, et al
.
Bisphenol A exposure leads to specific microRNA alterations in placental cells
.
Reprod Toxicol
2010
;
29
:
401
406
56.
Han
J
,
Won
EJ
,
Hwang
UK
,
Kim
IC
,
Yim
JH
,
Lee
JS
.
Triclosan (TCS) and triclocarban (TCC) cause lifespan reduction and reproductive impairment through oxidative stress-mediated expression of the defensome in the monogonont rotifer (Brachionus koreanus)
.
Comp Biochem Physiol C Toxicol Pharmacol
2016
;
185-186
:
131
137
57.
Rodríguez
PE
,
Sanchez
MS
.
Maternal exposure to triclosan impairs thyroid homeostasis and female pubertal development in Wistar rat offspring
.
J Toxicol Environ Health A
2010
;
73
:
1678
1688
58.
Mahalingaiah
S
,
Meeker
JD
,
Pearson
KR
, et al
.
Temporal variability and predictors of urinary bisphenol A concentrations in men and women
.
Environ Health Perspect
2008
;
116
:
173
178
59.
Weiss
L
,
Arbuckle
TE
,
Fisher
M
, et al
.
Temporal variability and sources of triclosan exposure in pregnancy
.
Int J Hyg Environ Health
2015
;
218
:
507
513
60.
Hedderson
MM
,
Darbinian
JA
,
Ferrara
A
.
Disparities in the risk of gestational diabetes by race-ethnicity and country of birth
.
Paediatr Perinat Epidemiol
2010
;
24
:
441
448
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.