Associations between Phthalate Metabolite Concentrations in Follicular Fluid and Reproductive Outcomes among Women Undergoing in Vitro Fertilization/Intracytoplasmic Sperm Injection Treatment

Background: Phthalates have been reported to impair fertility in various studies. However, evidence exploring the associations between phthalate metabolites in follicular fluid (FF) and reproductive outcomes is lacking. Objectives: To investigate the associations between phthalate metabolite concentrations in FF and in vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI) outcomes among women recruited from a fertility clinic. Methods: We included 641 women undergoing IVF/ICSI treatment from December 2018 to January 2020. The levels of eight phthalate metabolites, including monoethyl phthalate (MEP), mono-isobutyl phthalate (MiBP), mono-n-butyl phthalate (MBP), monobenzyl phthalate (MBzP), mono(2-ethylhexyl) phthalate (MEHP), mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono(2-ethyl-5-oxohexyl) phthalate (MEOHP), and mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), were quantified in FF collected on the oocyte retrieval day. Associations between quartiles of individual phthalate metabolite concentrations and nine IVF/ICSI outcomes, including oocyte yield, mature oocyte number, two distinct pronuclei (2PN) zygote number, fertilization rate, blastocyst formation rate, implantation, clinical pregnancy, miscarriage, and live birth, were estimated with generalized linear models. The effects of phthalate mixtures on IVF/ICSI outcomes were assessed using Bayesian kernel machine regression (BKMR) models. Results: After adjusting for relevant confounders, elevated quartiles of MBzP, MEHHP, and MEHP in FF were inversely associated with the numbers of retrieved oocytes, mature oocytes, and 2PN zygotes (all p for trends <0.10). In comparison with the lowest quartile, the highest quartile of molar sum of di(2-ethylhexyl) phthalate metabolites (ΣDEHP) was associated with a reduction of 9.1% [95% confidence interval (CI): –17.1%, –0.37%] and 10.3% (95% CI: –18.8%, –0.94%) in yielded oocyte and mature oocyte numbers, respectively. Furthermore, the BKMR models revealed inverse associations between phthalate mixtures and the numbers of retrieved oocytes and mature oocytes. We generally found null results for implantation, clinical pregnancy, miscarriage, and live birth. Discussion: Certain phthalate metabolites in FF are inversely associated with the numbers of retrieved oocytes, mature oocytes, and 2PN zygotes among women undergoing IVF/ICSI treatment. https://doi.org/10.1289/EHP11998


Introduction
2][3] This alarming prevalence of infertility places significant physical, psychological, and economic burdens on affected couples.Assisted reproductive technology (ART) has emerged as an effective therapy for patients with infertility, with the annual number of ART cycles surpassing 849,811 cycles in Europe and 1.15 million cycles in China. 4,5In vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI) is the most common treatment in ART. 5 Despite the rapid increase in the use of IVF/ICSI treatment, the success rates for pregnancy and delivery remain under 40%, 6 which emphasizes the importance of investigating the risk factors associated with IVF/ICSI failure.8][9] Previous studies have suggested that endocrine-disrupting chemicals (EDCs) can impair human fecundability in the general population. 10][13] Phthalates, as one class of EDCs, are widely used as plasticizers and additives in numerous consumer products, such as food packages, personal care products, and pharmaceutical tablet coatings. 14,15More than 25 types of phthalates are commonly used in industry, contributing to an annual production of 8 million metric tons worldwide. 16,17Because of their unconjugated binding to consumer products, phthalates can easily migrate into environmental media, such as soil, water, dust, and air, ultimately entering the human body via inhalation, ingestion, or dermal absorption. 15][20] Several in vivo and in vitro studies have found that phthalate exposures at environmentally relevant doses can impair germ cell quality and female reproductive function.For instance, Cuenca et al. reported that 100 lmol=L of di(2-ethylhexyl) phthalate (DEHP) can damage chromosomes in oocytes and disrupt early embryogenesis in Caenorhabditis elegans. 21Phthalate mixtures, even at 1 lg=mL, were found to inhibit ovulation and prostaglandin production in mouse antral follicles. 22In addition, gestational exposure to 20 lg=kg=d-500 mg=kg=d of phthalate mixtures can accelerate reproductive ageing in mouse offspring, manifesting as decreased antral follicle count and altered hormone levels. 23A previous human study found that urinary DEHP metabolites were inversely associated with ovarian reserve. 24Several epidemiological studies also reported inverse associations between urinary phthalate metabolites and the odds of implantation, clinical pregnancy, and live birth among women undergoing ART, whereas others reported null associations, as reviewed by a recent literature. 11However, human evidence investigating the associations between phthalate metabolites in FF and IVF/ICSI outcomes is still lacking.
FF is an essential microenvironment for oocyte development, because it directly contacts with oocytes. 25Thus, FF provides a unique matrix for assessing phthalate exposures within the target organ.Human studies have demonstrated that there is the widespread presence of phthalate metabolites in FF samples 18,19 and that their distribution patterns are distinct from those in urine samples. 20,26Therefore, we hypothesized that phthalates in FF can directly impair oocyte and embryo quality, thereby leading to adverse reproductive outcomes.To test this hypothesis, we quantified the concentrations of eight phthalate metabolites in FF samples and used generalized linear models (GLMs) and Bayesian kernel machine regression (BKMR) models to assess the associations between these individual and mixture of metabolites and nine reproductive outcomes among a cohort of 641 women undergoing IVF/ICSI treatment.

Study Population
Our participants were from the Tongji Reproductive and Environmental (TREE) study, which has been described previously. 27This prospective cohort recruited 2,057 women >20 y of age who planned to undergo ART treatment at the Reproductive Center, Tongji Hospital, Wuhan, China, from December 2018 to January 2020.Among them, 958 individuals had a medical indication for IVF/ICSI (e.g., infertility, bilateral salpingectomy, bilateral oviduct obstruction, and poor semen quality) 6 and underwent IVF/ICSI treatment with fresh embryo transfer.We excluded women who had no available embryo or underwent frozen embryo transfer.We further excluded 255 women due to missing or insufficient FF samples, 14 women due to the presence of blood contamination in FF samples, 43 women due to missing detailed data on clinical characteristics or reproductive outcomes, 2 women due to age >45 y, and 3 women for whom donated sperm were used, leaving 641 participants in the current analysis (Figure S1).Each participant signed an informed consent form at enrollment.This study was approved by the Ethical Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology.

Covariates
At recruitment, the research staff used a questionnaire to obtain basic information including age, ethnicity, gravidity, educational level, household income, smoking status, and drinking status.Except for the Han ethnicity, Chinese minorities were classified into the "other" group (i.e., Hui, Mongolian, Miao, Tujia, Uygur, Yao, Yi, and Zhuang ethnicity).Participants who reported no history of smoking or drinking were defined as "never," who currently smoking or drinking at least once a week for at least 1 y were categorized as "current," 28,29 and who had ceased smoking or drinking for 1 y were classified as "quitting." 30Women exposed to secondhand smoke for >15 minutes per day at home or in the workplace were classified as passive smokers. 31Data on height, weight, infertility duration, reasons for infertility, controlled ovarian hyperstimulation (COH) protocol, fertilization methods, number of embryos transferred, and stage of embryo transferred were extracted from the medical records.Weight and height were used to calculate the body mass index (BMI; kg=m 2 ).
The duration of infertility was self-reported during the treatment process.The reasons for infertility were categorized into four categories (female factors, male factors, mixed factors, or unexplained factors) according to the diagnoses by the clinicians at the hospital.Female factors encompassed ailments affecting the female reproductive system, including pelvic factors and ovulation disorders.Male factors referred to conditions affecting the male genital system, including dysspermia, sexual dysfunction, and other factors.Cases where both female and male factors coexisted within the diagnosis of one couple were classified as mixed factors. 32Infertility lacking a definitive diagnosis was classified as unexplained infertility.The detailed classification of infertility factors is listed in Table S1.

IVF/ICSI Protocol
The COH protocols included agonist, antagonist, and others (i.e., natural cycle, luteal phase ovarian stimulation, and minimal ovarian stimulation).For the agonist and antagonist protocols, a gonadotropin-releasing hormone (GnRH) agonist (Decapeptyl; Ferring Pharmaceuticals) or antagonist (Cetrotide; Merck Serono) was used to restrain follicular development, respectively.When the down-regulation was achieved, ovarian stimulation generally began with 150-300 international units (IU) of follicle-stimulating hormone (FSH) (Gonal-F, Merck-Serono). 33Then, the progression of follicle development, uterine endometrial thickness, and serum hormone levels was monitored to determine the optimal timing of ovulation.Other protocols did not use the GnRH agonists or antagonists.Once two or more follicles with a diameter of ≥18 mm or three or more follicles with a diameter of ≥17 mm were observed, 10,000 IU of human chorionic gonadotropin (HCG) (Ovidrel, Merck-Serono) was administered to induce ovulation.Oocyte retrieval was performed 34-36 h later. 33IVF/ICSI procedures were carried out according to the standard requirements of the Reproductive Medicine Center, Tongji Hospital, which was also consistent with the international guidelines. 34Cleavage-stage embryos and blastocysts were used for embryo transfer.Cleavagestage embryos refer to blastomeres on Day 3, whereas blastocysts denote blastomeres on Day 5 or Day 6.

IVF/ICSI Parameters and Pregnancy Outcomes
IVF/ICSI parameters and pregnancy outcomes were extracted from the electronic medical records.The IVF/ICSI parameters included retrieved oocyte number, mature oocyte number, two distinct pronuclei (2PN) zygote number, fertilization rate, and blastocyst formation rate.A mature oocyte was defined as an oocyte exhibiting the first polar body that could be fertilized.The 2PN referred to the presence of two pronuclei in a zygote after fertilization.The fertilization rate was calculated by dividing the 2PN zygote number by the retrieved oocyte number.The blastocyst formation rate was the ratio of blastocyst number divided by the number of embryos on Day 3. 35 Pregnancy outcomes included implantation, clinical pregnancy, miscarriage, and live birth.Serum HCG levels were assessed 14 or 16 d after embryo transfer, with an HCG concentration ≥10 mIU=mL defined as implantation.Visualization of one or more gestational sacs via transvaginal ultrasound between 6 and 7 gestational weeks was regarded as clinical pregnancy.Miscarriage referred to the spontaneous loss of a clinical Environmental Health Perspectives 127019-2 131(12) December 2023 pregnancy under transvaginal ultrasound before 28 completed gestational weeks, excluding those who had a positive hCG but not a positive ultrasound pregnancy.A live birth was defined as the delivery of a newborn after 28 completed gestational weeks, exhibiting life signs such as breathing, heartbeat, or observable voluntary muscle movements. 27The implantation rate was defined as the ratio of implantation cycles to aspirated cycles.The miscarriage rate was computed by dividing the number of miscarriage cycles by the number of clinical pregnancy cycles.The live birth rate was calculated by dividing the number of live birth cycles by the aspirated cycles. 35

Statistical Analyses
Baseline characteristics are presented as the mean ± standard deviation (SD) or number (%) as appropriate.The chi-square test or Student's t-test was used to compare the differences in basic information between women included in this study and the entire TREE cohort.Given the skewed distribution, phthalate concentrations were ln-transformed for subsequent analyses.Phthalate concentrations below the LODs were assigned to LOD= p 2. 38 Correlations between phthalate metabolites in FF were assessed using the Spearman test.GLMs and BKMR were included in the exposure-response analysis.The GLMs were used to estimate the relationships of eight individual phthalate metabolites and two summary measures of DEHP metabolite concentrations in FF with nine reproductive outcomes, with the "stats" package (version 4.2.1). 39Each exposure was categorized into quartiles to minimize the effects of skewed distribution and extreme values, with the lowest quartiles serving as the reference groups.Tests for trends were calculated by entering the quartiles of exposure as ordinal categorical variables (1-4) in the regression models. 40The oocyte number, mature oocyte number, and 2PN zygote number were modeled using a Poisson distribution and log link, whereas the fertilization rate and blastocyst formation rate were modeled using a binomial distribution and probit link.Implantation, clinical pregnancy, miscarriage, and live birth were modeled using a binomial distribution and logit link.The percent changes were calculated using the formula 100% × ½exp ðbÞ − 1, where b represents the regression coefficients.Women who had missing exposure information or outcome data were excluded before the analyses (Figure S1).One missing data for educational level was imputed using random forest regression with the R package "mice" (version 3.13.0), in which the method argument was set to rf. 41 Regarding blastocyst formation rate, women who did not undergo blastocyst culture were excluded, and for miscarriage, women who failed to have clinical pregnancy were excluded.
6][47] These covariates were then identified using a directed acyclic graph. 48Ethnicity and smoking status were excluded because there were few women with other ethnicities (n = 30) and few active smokers (n = 39), which can lead to sparse data bias.Instead, we adjusted for passive smoke exposure as an indicator of tobacco exposure.Finally, age, BMI, passive smoke exposure, educational level, household income, drinking status, COH protocols, and reasons for infertility were included in the regression models (Figure S2). 48Age and BMI were retained as continuous variables.Passive smoke exposure (never or ever) and educational level (high school and below or college and above) were treated as dichotomous variables.Drinking status (never, current, or quit), COH protocols (agonist, antagonist, or others), and reasons for infertility (female factors, male factors, mixed factors, or unexplained) were included as categorical variables.Household income (≤3,000, 3,001-5,000, 5,001-10,000, or ≥10,001 Yuan/month) was included as an ordinal categorical variable.No passive smoke exposure, high school educational level or below, never drinking, agonist COH protocol, female-factor infertility, and household income ≤3,000 Yuan/ month were treated as the reference groups.
Several sensitivity analyses were conducted to assess the robustness of our results.First, we evaluated the impact of confounding factors on associations between phthalate metabolites in FF and IVF/ICSI outcomes by comparing the regression coefficients from the crude models and the adjusted models.Second, given that both lower and higher BMI can adversely affect female fertility and IVF outcomes, 49 we performed a restricted analysis among participants within the normal BMI range (18:5-24:0 kg=m 2 , n = 404).Third, we restricted our analysis to those couples with both female factors and unexplained factors (n = 474) to minimize the effects of male reproductive factors on IVF/ICSI outcomes. 50inally, we used BKMR models to evaluate the associations between mixtures of phthalate metabolites and IVF/ICSI outcomes using the "bkmr" package (version 0.2.2) in R software. 51BKMR, an innovative statistical approach, enables the estimation of joint and nonlinear effects of multiple pollutants, as well as potential interactions within multipollutant mixtures in a flexible and parsimonious manner.Other statistical approaches, such as weighted quantile sum (WQS) provides only unidirectional assumptions for mixtures.Because certain phthalate metabolites were significantly correlated in the Spearman correlation analysis (Figure S3), we conducted a hierarchical variable selection with 50,000 iterations.To avoid the effect of the burn-in phase, we dropped the first 25,000 iterations.MEHP, MEHHP, MEOHP, and MECPP, as the metabolites of DEHP, 52  studies. 53,54The BKMR models adjusted for the same covariates as in the GLMs, and all phthalate metabolite concentrations were lntransformed.Continuous exposures, outcomes (i.e., oocyte yield, number of mature oocytes, 2PN zygote number, fertilization rate, and blastocyst formation rate), and covariates (i.e., age and BMI) were centered and scaled for standardization.We used a Gaussian distribution for continuous outcomes and a binomial distribution for binary outcomes (i.e., implantation, clinical pregnancy, miscarriage, and live birth).The posterior inclusion probability (PIP) is a parameter measuring the relative importance of each phthalate metabolite in the mixtures in defining the exposure-response associations between mixtures and each outcome, with higher values (closer to 1) indicating higher importance.The group posterior inclusion probability (groupPIP) represented the relative importance of each group in defining the associations between mixtures and IVF/ICSI outcomes.The conditional posterior inclusion probability (condPIP) represented the relative importance of a particular phthalate metabolite in defining the associations between phthalate metabolites and IVF/ICSI outcomes within the assigned group.Univariate exposure-response relationship was estimated by plotting a single metabolite when fixing other metabolites at their median values using the function "PredictorResponseUnivar."The bivariate exposure-response interactions were evaluated by plotting the exposure-response function of a main exposure where the second exposure was fixed at various quantiles using the functions "PredictorResponseBivar" and "PredictorResponseBivarLevels."Furthermore, the function "OverallRiskSummaries" was employed to visualize the overall risk effects of the mixtures.The details of the BKMR setting are presented in the Supplementary Materials.All analyses were performed using R software (version 4.1.1;R Development Core Team).In addition, p < 0:05 was regarded as statistically significant, and p < 0:10 was regarded as suggestively significant.

Population Characteristics
The characteristics of the study population are presented in Table 1.The average age and BMI of the study population were 30.6 y and 22:4 kg=m 2 , respectively.Among these women, 95.3% were of Han ethnicity, 92.5% were nonsmokers, and 76.6% were nondrinkers.Regarding clinical characteristics, 65.5% of participants were diagnosed with female-factor infertility, and 63.3% used an agonist protocol.In comparison with all women from the TREE cohort, we generally observed no differences in basic information except for gravidity and reasons for infertility (Table S2).In the treatment process, 73.6% of participants underwent IVF, 83.9% of participants accepted single embryo transfer, and 91.7% of participants chose to transfer cleavage-stage embryos (Table 1).The overall rates of implantation, clinical pregnancy, and live birth were 77.7%, 49.6%, and 42.6%, respectively.

Associations of Mixtures of Phthalate Metabolites in FF with IVF/ICSI Outcomes
The scaled concentrations of phthalate metabolites in FF were presented in Excel Table S1.In the BKMR models, we observed that mixtures of phthalate metabolites in FF were negatively associated with oocyte yield and mature oocyte number (Figure 1; numeric results in Tables S3-S4).The DEHP factors appeared to have a higher contribution in defining the exposure-response relationships (groupPIPs ≥0:64; Tables S5-S6).Among the DEHP factors, MEHP and MEHHP had larger conditional PIP (condPIP ≥0:37; Tables S5-S6).Consistent with the results of GLMs, we also found inverse associations of MiBP, MEHHP, and MEHP in FF with oocyte yield, mature oocyte number, or 2PN zygote number, and the univariate exposure-response relationships were linear when other compounds were fixed at median values (Figure S4, numeric results in Excel Table S2).However, we observed positive associations of MEP in FF with numbers of retrieved oocytes, mature oocytes, and 2PN zygotes.There were no significant associations between phthalate mixtures and pregnancy outcomes.No evidence of obvious interaction effects between the phthalate metabolites was detected (Figure S5, numeric results in Excel Table S3).

Sensitivity Analysis
As with the full covariates adjusted models in GLMs, quartiles of MEHHP, MBzP, and MEHP in FF, we observed, were inversely associated with numbers of retrieved oocytes, mature oocytes, and 2PN zygotes in crude models (Tables S7-S8), though some effects were attenuated (all p for trends ≤0:16), indicating that the covariates did not substantially change our results.Our results remained generally unchanged when restricting the analysis to participants within a normal BMI range (18:5-24:0 kg=m 2 ) (Tables S9-S10), and to participants without male factor infertility (Tables S11-S12).

Discussion
In this study among women undergoing IVF/ICSI treatments, we observed that the concentrations of MEHHP, MBzP, and MEHP in FF were inversely associated with the numbers of retrieved oocytes, mature oocytes, and 2PN zygotes in single-pollutant models.The BKMR models further revealed that mixtures of phthalate metabolites in FF were inversely associated with oocyte yield and mature oocytes.However, we generally found null associations between phthalate metabolites in FF and pregnancy outcomes including implantation, miscarriage, clinical pregnancy, and live birth.
In our research, phthalate metabolites were detected in >86% of the FF samples, indicating the ubiquitous presence of phthalate metabolites in the ovary.This presence is not unexpected, because phthalates in the bloodstream can be transferred to FF through the thecal capillaries in the ovary. 55To our knowledge, phthalate metabolites have been extensively detected in FF samples in studies from Israel, 56 Italy, 19 and China. 20However, the concentrations of phthalate metabolites in our current study were generally lower than those reported in previous studies.For instance, the median concentration of MEHP among our study participants was 0:17 lg=L, which was lower than the corresponding values reported in Israel (0:86 lg=L), 56 Italy (7:78 lg=L), 19 and China (2:80 lg=L). 20The varied metabolite concentrations may be attributed to the differences in study region, population characteristics, and lifestyle factors.We found null or inverse correlations between MECPP, MEOHP, and MEHHP in FF, which is different from the correlation patterns reported for urine, where these DEHP metabolites show strong correlations. 54,57The transfer efficiencies of environmental chemicals from blood to follicles can vary based on their molecular structures, physicochemical properties, and other biological factors (e.g., age), 58 which may contribute to the differential permeation of   DEHP metabolites at varying concentrations.The permeability of environmental chemicals can also be influenced by dynamic changes in the composition of the basement membranes during FF development. 59n the current analysis, we found that individual or mixtures of phthalate metabolites in FF were inversely associated with oocyte yield, mature oocyte number, or 2PN zygote number.Consistent with our findings, in vivo and in vitro studies have found that DEHP or phthalate mixtures can disrupt the quality and function of oocyte.For instance, Kalo et al. found that follicular development was retarded in DEHP-treated cows and that MEHP was detected in FF samples, and the embryo and blastocyst development were also disrupted in cumulus-oocyte complexes cultured with FF aspirated from the treatment group. 60Lu et al. found that oral exposure to DEHP significantly decreased the maturation and fertilization of oocytes in mice.DEHP was also found to trigger oxidative stress, impair mitochondrial function, promote early apoptosis, and induce DNA damage in oocytes. 61In addition, Meling et al. reported that phthalate mixtures (MEP, MEHP, MBP, MiBP, MiNP, and MBzP) could inhibit follicle growth and disrupt sex hormone production in mice. 62revious human studies have also investigated the association between phthalate exposure and IVF/ICSI outcomes.4][65][66][67][68][69] In line with our results, Machtinger et al. reported that urinary MEHHP was negatively associated with oocyte yield and mature oocyte numbers among 136 women undergoing IVF/ICSI treatment in Israel.They further found that urinary concentrations of RDEHP, MEOHP, MECPP, MEP, and MBP were negatively associated with the numbers of total oocytes, mature oocytes, and fertilized oocytes. 66Likewise, the EARTH study team measured urinary phthalate metabolites of 256 women from a reproductive center in the United States and observed that RDEHP, MEHP, MEHHP, MEOHP, MECPP, and monocarboxynonyl phthalate (MCNP) were inversely associated with oocyte yield and number of mature oocytes. 36Given the weak correlation between the concentrations of phthalate metabolites in urine and FF, 26 FF may provide a unique medium to assess phthalate exposure in target organs.To date, only two human studies have investigated the associations between phthalate metabolites in FF and oocyte and embryo quality among women undergoing IVF/ICSI treatment. 20,56In our previous research, Du et al. reported a generally null association between phthalate metabolite concentrations in FF and oocyte and embryo development among 112 women from the Reproductive Center of Tongji Hospital. 20Barnett-Itzhaki et al. found that MEP concentration in FF was inversely associated with the number of blastomeres on Day 3 among 105 women recruited from an infertility center in Israel. 56The limited sample size in these previous studies may lead to imprecise estimation effects. 70Inconsistencies could also be attributed to differences in population characteristics, FF sampling strategies (pooled vs. single follicle), stimulation protocols (agonist vs. antagonist protocol), and exposure levels.In addition, our multipollutant analyses revealed that phthalate mixtures were also inversely associated with the numbers of retrieved oocytes and mature oocytes in BKMR models, but the magnitude of association estimates was not notably higher than those in GLMs.MEP was positively associated with the numbers of retrieved oocytes, mature oocytes, and 2PN zygotes in the exposure-response analysis from the BKMR models, which may weaken the adverse effects of other phthalate metabolites on outcomes.Similarly, a previous publication also reported a positive association between urinary MEP concentration and the number of retrieved oocytes. 63However, more research is required to confirm our findings.
To our knowledge, no previous studies have investigated the associations between phthalate metabolites in FF and pregnancy  among women undergoing IVF treatment. 63,66However, Al-Saleh et al. found that urinary MEHP, MEP, and %MEHP were positively associated with risks of failed clinical pregnancy and live birth among 599 couples who underwent in vitro fertilization. 64The EARTH study team also found that urinary RDEHP was inversely associated with clinical pregnancy and live birth among women recruited from a fertility clinic in the United States. 36Based on BKMR models, the EARTH team further reported that DEHPrelated factors were positively associated with pregnancy failure before live birth. 65The discrepant results may be attributed to the differences in phthalate exposure levels and biological specimens for phthalate exposure assessment.Notably, concentrations of phthalate metabolites in FF are significantly lower, ranging from one to three orders of magnitude, when compared with the urine samples. 26In addition, phthalates can impair ART outcomes through pathways at a more global level, such as introducing systematic oxidative stress responses, 61 altering circulating hormone levels, 62 and modulating gene expressions. 71Further research can simultaneously determine phthalate metabolites in paired urine and FF samples, thereby elucidating their synergetic effects on oocyte quality and ART outcomes.The strengths of this study included the prospective study design, the relatively large sample size, and the direct quantification of phthalate metabolites in FF samples.We also collected FF from withinsubject pools of follicle samples instead of single follicle samples to reduce the potential measurement error.However, some limitations should be noted.First, we collected only one FF sample from each participant, which may lead to exposure misclassification because of the considerable within-individual variability observed in phthalate metabolite concentrations in biological samples. 72Second, our participants were from a fertility center and underwent IVF/ICSI treatment, which limits the generalizability of the results.The concentrations of phthalate metabolites in FF samples may also be affected by IVF/ICSI procedures such as medication administration and COH. 73hird, even though we discarded contaminated FF samples and centrifuged the biospecimens immediately after collection, trace amounts of blood contamination and systematic pollution caused by surgical operation, COH stimulation, or sample measurement may still exist. 74gure 1.Estimate effects (95% credible intervals) of the mixtures of phthalate metabolites on (A) oocyte number, (B) mature oocyte number, (C) 2PN zygotes number, (D) fertilization rate, (E) blastocyst formation rate, (F) implantation rate, (G) clinical pregnancy, (H) miscarriage, and (I) live birth by BKMR models when all ln-transformed phthalate metabolite concentrations at particular percentiles were compared with all the phthalate metabolites at their 50th percentile.The BKMR models were adjusted for age, BMI, passive smoke exposure, drinking status, household income, education level, reasons for infertility, and COH protocols.Numerical values are presented in Supplemental Materials, Table S3-S4.Note: 2PN, two distinct pronuclei; BKMR, Bayesian kernel machine regression; BMI, body mass index; COH, controlled ovarian hyperstimulation.

Environmental Health Perspectives
127019-9 131(12) December 2023 Fourth, our results may be biased by other unmeasured factors, such as psychological stress, 75 physical exercise, 76 and coexposure to other EDCs (e.g., bisphenol A, parabens, and organochlorine pesticides). 77n the hierarchical BKMR analysis, some metabolites (e.g., MEHP and MEOHP) had low within-group correlation, which is not best suited to the settings of these applied models and may bias the results of mixtures.In addition, some baseline and clinical characteristics, such as passive smoke exposure and infertility duration, were selfreported.Although the self-reported information was easy to obtain and showed moderate accuracy, the recall bias may still exist. 78

Conclusion
Based on individual and mixture analyses, we observed that certain phthalate metabolites in FF were inversely associated with number of retrieved oocytes, mature oocytes, and 2PN zygotes, suggesting that phthalate exposures in the ovary may impair oocyte and embryo quality.Our study provides new evidence for the toxic effects of phthalates on IVF/ICSI outcomes, whereas more epidemiological studies are warranted to verify our findings.

-
were classified into group 2, and the other metabolites were classified into group 1, namely DEHP factors and non-DEHP factors separately as reported in previous epidemiological

Table 2 .
Distribution of phthalate metabolite concentrations in follicular fluid among 641 women undergoing IVF/ICSI treatment from the Tongji Reproductive and Environmental (TREE) cohort, 2018-2020.

Table 3 .
Associations between phthalate metabolite concentrations in follicular fluid and IVF/ICSI parameters in the Tongji Reproductive and Environmental (TREE) cohort, 2018-2020.

Table 3 .
Generalized linear regression models were adjusted for age, BMI, passive smoke exposure, drinking status, household income, education level, reasons for infertility, and COH protocols.P Note: