Associations between Urinary Concentrations of Disinfection Byproducts and in Vitro Fertilization Outcomes: A Prospective Cohort Study in China

Background: Experimental studies show that disinfection byproducts (DBPs) can inhibit oocyte maturation, decrease fertilization capacity, and impair embryo development, but human evidence is lacking. Objectives: We aimed to evaluate the associations between exposure to drinking water DBPs and in vitro fertilization (IVF) outcomes. Methods: The study included 1,048 women undergoing assisted reproductive technology (ART) treatment between December 2018 and January 2020 from a prospective cohort study, the Tongji Reproductive and Environmental study in Wuhan, China. Exposure to DBPs was assessed by dichloroacetic acid (DCAA) and trichloroacetic acid (TCAA) in up to four urine samples, which were collected on the day of both enrollment and oocyte retrieval. Multivariable generalized linear mixed models, accounting for multiple IVF cycles per woman, were applied to evaluate the associations between urinary biomarkers of DBP exposures and IVF outcomes. Stratified analyses were used to explore the potential effect modifiers. Results: The included 1,048 women underwent 1,136 IVF cycles, with 960 (91.6%), 84 (8.0%), and 4 (0.4%) women contributing one cycle, two cycles, and three cycles, respectively. We found that elevated quartiles of urinary DCAA and TCAA concentrations were associated with reduced numbers of total oocytes and metaphase II oocytes and that urinary DCAA concentrations with a lower proportion of best-quality embryos (all p for trends<0.05). Moreover, elevated quartiles of urinary DCAA concentrations were associated with decreased proportions of successful implantation, clinical pregnancy, and live birth (14%, 15%, and 15% decreases in adjusted means comparing the extreme quartiles, respectively; all p for trends<0.05). Stratification analyses showed that the inverse associations of urinary TCAA concentrations with multiple IVF outcomes were stronger among women ≥30 y of age (p for interactions<0.05). Discussion: Exposure to drinking water DBPs was inversely associated with some IVF outcomes among women undergoing ART treatment. Further study is necessary to confirm our findings. https://doi.org/10.1289/EHP12447


Introduction
Infertility, which is characterized by the inability to achieve a clinical pregnancy after 1 y of unprotected and regular sexual intercourse, affects 8%-12% of reproductive-aged couples worldwide. 1In China, it is estimated that 15%-20% of reproductive-aged couples are suffering from infertility. 2 Since the first baby was born in 1978 by use of in vitro fertilization (IVF), 3 assisted reproductive technology (ART) has become the most effective treatment method for infertility. 4lthough there are ∼ 2:8 million ART cycles performed worldwide annually, the success rate of live birth per ART cycle remains around 30%. 5 This raises an increasing concern that potential risk factors adversely affect ART outcomes. 6[9][10] Disinfection byproducts (DBPs) are a class of widespread drinking water chemical contaminants formed when disinfectants react with natural and/or synthetic organic matter in source waters. 11uman exposure to DBPs occurs via ingestion, dermal absorption, or inhalation during daily water usage activities depending on the type of DBPs considered. 12Among more than 700 identified DBPs, haloacetic acids (HAAs) are one of the prevalent species in drinking water. 13Dichloroacetic acid (DCAA) and trichloroacetic acid (TCAA) are the two most abundant compounds of HAAs and have half-lives of elimination in the body ranging from minutes to days. 14,157][38] However, to the best of our knowledge, no human studies to date have investigated the effects of exposure to DBPs on early reproductive outcomes, such as oocyte fertilization and embryo implantation, which are not available among women conceiving without medical assistance. 39herefore, the aim of the present study was to investigate the associations between DBP exposures and IVF outcomes among women undergoing ART treatment.We measured DCAA and TCAA concentrations in up to four urine samples as biomarkers of DBP exposures to reduce measurement error.As age, body mass index (BMI), treatment and fertilization protocols, and embryo transferred types are predictors of female fertility or IVF outcomes, [40][41][42][43][44] we also examined these factors as potential effect modifiers.

Study Population
Study participants were drawn from the Tongji Reproductive and Environmental (TREE) study, an ongoing prospective cohort study established in December 2018, which has been detailed in our prior study. 37In brief, couples were eligible to participate in this study if they were at least 20 y old and planned to undergo ART treatment, including IVF, intrauterine insemination (IUI), and intracytoplasmic sperm injection (ICSI) at the Reproductive Medicine Center of Tongji hospital in Wuhan, China.The TREE study was approved by the ethics committee of Tongji Medical College.All participants provided written informed consent and enrolled in the study at their preoperative health examinations, followed by controlled ovarian stimulation (COS), oocyte retrieval, embryo transfer, and pregnancy test.
The flowchart of study participants included in the present study is shown in Figure 1.A total of 2,195 women were invited and 2,057 women (participation rate 93.7%) agreed to participate in the TREE study between December 2018 and January 2020.Of those, women were excluded if they gave up ART treatment (n = 401), underwent IUI treatment (n = 210), had karyotype abnormality (n = 112), or cancelled oocyte retrieval (n = 20).Also, women were excluded if they did not provide a urine volume for exposure assessment (n = 228).We further excluded those women who had a history of occupational exposure to certain synthetic materials (i.e., dichloroethylene, trichloroethylene, tetrachloroethylene, 1,1,1-trichloroethane) (n = 38), as these chemicals may be metabolized to DCAA and TCAA.Finally, there were 1,048 women available for the current analysis.There were no significant differences between the included population in this study and total population in demographic characteristics except that BMI was lower in the subgroup population (Table S1).

Covariates
Demographic information including lifestyle factors (e.g., alcohol use and smoking), water-use activities (e.g., tap water consumption), socioeconomic status (e.g., education and household income), occupational exposure, as well as medical and reproductive history was collected from each subject by a questionnaire at enrollment.Age, height, weight, gravidity, parity, infertility causes, and duration of infertility were obtained from the electronic medical record.Passive smoking was defined as self-reported exposure to environmental tobacco smoke for more than 15 min per day at workplace or/and home. 45Alcohol use refers to drinking alcohol at least once a week for a period of 6 months or longer.BMI was calculated as weight (in kilograms) divided by the square of height (in meters) and divided into four categories: underweight (<18:5 kg=m 2 ), normal weight (18:5-24:9 kg=m 2 ), overweight (25-29:9 kg=m 2 ), and obesity (≥30 kg=m 2 ) according to World Health Organization classifications. 46Infertility causes were assessed by clinical physicians based on the results of preoperative health examinations and categorized into female factor, male factor, mixed factors, and unexplained factor.

Exposure Assessment
Urine samples were collected from each participant on the day of enrollment and oocyte retrieval by a 50-mL polypropylene container.The women receiving multiple IVF cycles provided an additional urine sample on the day of oocyte retrieval at each IVF cycle.The 1,048 women provided a total of 2,197 urine samples, with 42 (4.0%),883 (84.3%), 103 (9.8%), and 20 (1.9%) contributing one urine sample, two urine samples, three urine samples, and four urine samples, respectively (Table S2).The mean ½ ± standard deviationðSDÞ time interval between urine sample collections was 79.6 ( ± 42:2) days.All urine samples were aliquoted to 15-mL polypropylene tubes and frozen at −20 C until further analysis.
The analytical approach for urinary DCAA and TCAA concentrations has been described in detail previously. 37,47In short, a 5-mL urine sample was adjusted to pH <0:5 with 1 mL concentrated sulfate acid.After adding 1:5 g of anhydrous copper sulfate and 5 g of anhydrous sodium sulfate, the sample was extracted with 5 mL methyl-tert-butyl-ether containing 1,2-dipropyl bromide as the internal standard.After centrifugation, 3 mL of the upper organic layer was derivatized by adding 1 mL acidic methanol at 50°C for 2 h, followed by the addition of saturated sodium bicarbonate to neutralize the derivatization reaction.Finally, the targets were measured by the gas chromatography (Agilent 6890N) equipped with an electron capture detector.Quality control samples including one blank and two spiked targets (Sigma-Aldrich Inc.) were analyzed along with each analysis batch.The limits of detection (LODs) were 1:00 lg=L for DCAA and 0:50 lg=L for TCAA, and the targets below LODs were replaced by LOD= p 2. The spiked recoveries for the targets ranged from 92.00% to 118.19%.The interday and intraday variation was lower than 10.00%.We measured specific gravity (SG) from each urine sample with a hand-held digital refractometer (Atago PAL-10S) to correct for the variation in urine diluteness. 36

Outcome Assessment
The IVF outcomes, including early IVF outcomes [total oocyte numbers, metaphase II (MII) oocyte numbers, the proportions of fertilization, and best-quality embryos] and clinical outcomes (implantation, clinical pregnancy, and live birth), were assessed in this hospital and extracted from its electronic medical record.According to age, ovarian reserve function, and infertility diagnosis, women adopted the appropriate COS protocols, including long gonadotropin-releasing hormone (GnRH) agonist and nonlong GnRH agonist (e.g., GnRH antagonist, mild stimulation, lutealphase stimulation, and natural cycle). 48When more than 2 dominant follicles reached 18 mm in diameter, recombinant human chorionic gonadotropin (hCG) (250 mg; Ovidrel; Serono) was administered to induce ovulation.Oocyte retrieval was conducted under transvaginal ultrasonographic guidance 34-36 h after hCG injection.After 4-6 h, fertilization was obtained by conventional IVF or ICSI based on clinical indication and sperm quality (e.g., infertility causes and severity of sperm abnormalities). 49The ICSI is a specialized form of IVF that involves the injection of a single sperm directly into an egg for fertilization.On day 2 or day 5 after oocyte retrieval, one to two embryos were selected for fresh embryo transfer, and the remaining embryos would be frozen.These frozen embryos were thawed and subsequently transferred into the individuals' uterus during a cycle with a prepared uterine lining.The frozen embryo transfer was selected for the participants with high likelihood of ovarian hyperstimulation syndrome, uterine effusion, endometrial factors, and so on.Pregnancy test was performed 14 d after embryos were transferred.
The number of total oocytes was counted on the day of the egg retrieval.Oocytes were classified by embryologists as degenerated, MII (first polar body extruded), metaphase I (germinal vesicle breakdown without extrusion of the first polar body), or germinal vesicle. 50Normal fertilization was confirmed 16 to 18 h after insemination by the appearance of a fertilized egg with two pronuclei (2PN).The proportion of fertilization was defined as the number of 2PN divided by the number of MII oocytes.Embryo quality was assessed at 3 d after fertilization according to the cell number and degree of cytoplasmic fragmentation.Best-quality embryos were classified as those with four cells on day 2 or seven to nine cells on day 3, no multinucleation, and <20% fragmentation. 51The proportion of best-quality embryos was defined as the number of best-quality embryos divided by the number of 2PN.Implantation was denoted as a serum b-hCG level >10 mIU=mL at ∼ 14 d after embryo transfer.The confirmation of an intrauterine pregnancy by ultrasound at ∼ 28 d after embryo transfer was considered as clinical pregnancy.The delivery of at least one liveborn infant after 28 wk of gestation was considered as live birth.

Statistical Analysis
The demographic and clinical characteristics of the study population were described as mean ± standard deviation (SD) or as number (percentage) where appropriate.The following formula was used to adjust for urinary dilution 52 : Ps = P½ð1:019 − 1Þ=ðSGc −1Þ, where Ps is the SG-adjusted urinary HAA concentration, P is the measured urinary HAA concentration, SGc is the measured urinary SG concentration, and 1.019 is the mean SG concentration in all of the study participants.The HAA concentrations were SG adjusted to correct for urine dilution and then ln transformed to satisfy the assumption of normality.Our previous study showed that urinary HAA concentrations had high within-person variability, and collection of multiple urine samples improved the exposure classification during 3 months. 16herefore, for those women who provided two or more urine samples (96%), the SG-adjusted urinary HAA concentrations were averaged before ln transformation to reduce exposure measurement error.
Multivariate generalized linear mixed models with random intercepts, accounting for within-person correlations in outcomes, were fitted to estimate the associations between urinary HAA concentrations and IVF outcomes.A binomial distribution and logit link function was applied to analyze the associations with the proportions of fertilization and best-quality embryos.A Poisson distribution and log link function were applied to examine the associations with the number of total and MII oocytes.A binary distribution and logit link function were applied to assess the associations with implantation, clinical pregnancy, and live birth.Urinary HAA concentrations were included as categorical variables in these regression models.Tests for linear trend were conducted across increasing exposure quartiles by assigning the quartiles of urinary HAA concentrations (Q1-Q4) as ordinal variables (1-4), and the lowest quartiles were treated as the reference.The regression coefficients for the number of total and MII oocytes were back-transformed f100 × ½exp ðbÞ − 1g to obtain the percent changes.To enhance the interpretation of the results, the proportions of fertilization and best-quality embryos as well as clinical outcomes were displayed as population marginal means adjusted for covariates.The percentage decreases were computed by dividing the differences in marginal means between Q1 and Q4 by the marginal mean of Q1.We also used penalized regression splines with smooth curve based on generalized additive mixed models to assess the nonlinear dose-response associations between Environmental Health Perspectives 097003-3 131(9) September 2023 continuous urinary HAA concentrations (ln transformed) and IVF outcomes.
We reported the results from the crude model (model 1) in which no potential confounders were included.4][55][56] These covariates have been reported to be associated with DBP exposures and/or IVF outcomes and, thus, were included in the models (model 2).Moreover, infertility causes were also included in the models (model 3) based on the "change-in-estimate" method with >10% change in the estimated effects of urinary HAA concentrations on IVF outcomes (Table S3-S4).Since there were very few women who were former or current smokers in the present analysis (n = 62), we did not include active smoking in the regression models.
Several sensitivity analyses were carried out to assess the robustness of our findings.First, we restricted to those women who provided repeated urine samples to evaluate the potential exposure misclassification introduced by a single urine sample.Second, we excluded those women with the highest 5% of urinary HAA concentrations to reduce the effects of extreme values.Finally, because repeated IVF cycles and infertility causes may affect IVF outcomes, we restricted to women within the first IVF cycle and female factors as infertility causes.
We tested whether the associations between urinary HAA concentrations and IVF outcomes were modified by age (<30 y old vs. ≥30 y old), BMI (<25 kg=m 2 vs. ≥25 kg=m 2 ), and treatment protocols (long GnRH agonist vs. Nonlong GnRH agonist), since these factors are predictors of female fertility or IVF outcomes. 40,41,43Moreover, we stratified analyses for clinical outcomes by insemination types (IVF vs. ICSI) and embryo transfer types (fresh embryo transferred vs. frozen-thawed embryos transferred), as previous studies have suggested that ICSI fertilization causes damage to the oocyte during the process of sperm injection  Environmental Health Perspectives 097003-4 131( 9) September 2023 into the oocyte cytoplasm 57,58 and that frozen embryo transfer has higher probabilities of implantation, clinical pregnancy, and live birth rate than fresh embryo transfer. 44p-Values for interaction between exposures and stratified variables were evaluated by adding a cross-product term in the regression models.R software (version 3.6.2;R Development Core Team) was used to perform all of the statistical analyses.p-Value <0:05 (two-tailed) was defined as statistically significant.
The crude model and the adjusted models showed that elevated urinary quartiles of DCAA and TCAA concentrations were associated with lower total oocyte numbers and MII oocyte numbers (all p for trends <0:05) (Figure 2; numeric data in Table S5).After adjusting for age, BMI, alcohol use, education level, passive smoking status, household income, and infertility causes, compared with those in the first quartiles, women in the fourth quartiles of urinary DCAA and TCAA concentrations had 12.37% [95% confidence interval (CI): −21:01%, −2:80%] and 13.91% (95% CI: −22:37%, −4:53%) lower total oocyte numbers, as well as 13.41% (95% CI: −22:15%, −3:69%) and 11.87% (95% CI: −20:76%, −1:99%) lower MII oocyte numbers, respectively.These negative associations persisted and were linear for urinary HAA concentrations modeled as continuous variables (Figure S1).Moreover, women had lower adjusted means (5% decrease, p for trends = 0:041) in the proportion of best-quality embryos comparing the extreme quartiles of urinary DCAA concentrations.There was no evidence of associations between urinary DCAA and TCAA concentrations and the proportion of fertilization or between urinary TCAA concentrations and the proportion of best-quality embryos.
As shown in Figure 3, elevated quartiles of urinary DCAA concentrations were associated with lower proportions of successful implantation, clinical pregnancy, and live birth (all p for trends <0:05) in the crude model and the adjusted models (numeric data in Table S6).After adjusting for age, BMI, alcohol use, education level, passive smoking status, household income, and infertility causes, compared with women in the lowest quartiles, women in the highest quartile of urinary DCAA concentrations had declines in adjusted means for the proportions of successful implantation (14% decrease, p for trends = 0:013), clinical pregnancy (15% decrease, p for trends = 0:009), and live birth (15% decrease, p for trends = 0:032) compared to women in the lowest quartile.These negative associations also persisted and were linear when urinary DCAA concentrations were included as continuous variables (Figure S2).There were null associations of urinary TCAA concentrations with proportions of successful implantation, clinical pregnancy, and live birth in the adjusted models.
The above-mentioned negative associations were not materially changed when restricting to those women with repeated urine samples (Table S7 and S8), excluding the highest 5% urinary HAA concentrations (Table S9 and S10), the first IVF cycle (Table S11 and S12), and the female factor as infertility causes (Table S13 and S14), except that the observed association between urinary DCAA concentrations and the proportion of best-quality embryos became not statistically significant among women with the female factor as infertility causes.
In the age-stratified analyses (Table S15 and S16), the negative associations of urinary TCAA concentrations with the numbers of total oocytes and MII oocytes, as well as the proportion of successful implantation were stronger among older women (>30 y of age) than younger women (≤30 y of age) (all p for interactions <0:05).There were no significant differences in associations between urinary HAA concentrations and IVF outcomes stratified by BMI,  The denominator for the proportions of all pregnancy outcomes is the oocyte retrieval cycles, with a total of 1,136 cycles.Note: The 1,048 women provided a total of 2,197 urine samples.Of these, 43 urine samples were undetectable for DCAA and 7 urine samples were undetectable for TCAA.DCAA, dichloroacetic acid; HAAs, haloacetic acids; SG, specific gravity; TCAA, trichloroacetic acid.

Discussion
In this prospective cohort study among women undergoing IVF/ ICSI treatment, we observed that urinary DCAA and TCAA concentrations were associated with reduced numbers of total oocytes and MII oocytes and that urinary DCAA concentrations were associated with a lower proportion of best-quality embryos.We also found that urinary DCAA concentrations were related to lower proportions of successful implantation, clinical pregnancy, and live birth.Furthermore, the stratified analyses indicated that urinary TCAA concentrations in negative associations with the numbers of total oocytes and MII oocytes were stronger among women ≥30 y of age.
Our findings suggested that exposure to DBPs reflected by urinary HAAs was negatively associated with multiple indicators of IVF outcomes.These results were in agreement with previous experimental studies.In in vitro studies, exposure to HAAs, including chloroacetic acid, bromoacetic acid, and iodoacetic acid, has been observed to impair ovarian steroidogenesis, 22 suppress antral follicular growth, 23 and inhibit mouse oocyte maturation. 25Similarly, in vivo studies have shown that dibromoacetic acid or chlorinated drinking water organic extracts (including identified and unidentified DBPs) diminish primordial follicle populations in rabbits, 26 inhibit oocyte maturation, and reduce the IVF capacity of mouse oocytes. 27Moreover, in zebrafish embryo models, regulated DBPs or nitrogen DBPs can cause embryo developmental disorders and mortality, 59 decrease embryo hatching rate, and increase embryo morphological abnormalities. 28o the best of our knowledge, no epidemiological studies to date have used IVF as a model to examine the associations between DBP exposures and early pregnancy outcomes.However, a few human studies have shown evidence of the association between DBP exposures and decreased female fecundity.In a prospective study among 403 women, Windham et al. 60 found that exposure to high levels of total THMs, another abundant species of DBPs, was associated with reduced mean cycle length and follicular phase length.In our previous studies from the same cohort, we observed that urinary DCAA and TCAA were associated with menstrual cycle disturbances, as reflected by irregular cycles and prolonged cycles and bleeding duration. 36Moreover, we found that urinary DCAA and TCAA were associated with decreased antral follicle count and that urinary TCAA was related to reduced anti-Mullerian hormone levels, suggesting declined ovarian reserve of exposure to DBPs. 37][63] The median urinary DCAA and/or TCAA concentrations (5:42 lg=L and 4:85 lg=L, respectively) in our study were comparable to those of healthy young Chinese adults (4:83 lg=L and 6:75 lg=L, respectively), 64 pregnant women in England (TCAA: 6:1 lg=L) 65 and in Xiaogan, China (DCAA: 7:1 lg=L), 66  adults in the United States (TCAA: 3:3 lg=L), 67 whereas they were significantly lower than those of pregnant women in South Africa (TCAA: 201 lg=L). 68It seems that stronger effects were observed for urinary DCAA compared to TCAA, particularly for clinical outcomes.The possible reasons may be due to more toxic and higher levels in DCAA.An in vitro fertilization trial showed that exposure of oocytes to DCAA at low doses (100 ppm) significantly lowered fertilization rates, while such effect was only observed at high doses of TCAA (1,000 ppm). 69The median concentrations of SG-adjusted urinary DCAA (5:63 lg=L) in this study population were slightly higher than TCAA (5:20 lg=L).
In the stratified analysis, we observed that the negative associations between urinary TCAA concentrations and multiple indicators of IVF outcomes were more pronounced among older women (age ≥30 y) than younger women (age < 30 y).Our results suggested that older women may be more susceptible to the adverse reproductive effects of DBP exposures.The age-related differences are biologically plausible since increasing age is associated with reduced female fecundability 70 and higher internal levels of DBP exposures. 56In support of our finding, among women undergoing IVF treatment from the China National Birth Cohort in Shanghai, China, Hu et al. 8 showed that organophosphate pesticide exposures in negative associations with clinical outcomes were stronger among older women.Several human studies have also shown that older women have more pronounced adverse effects of exposure to environmental chemicals on fecundability. 71,72owever, our findings should be interpreted cautiously because the multiple stratified analyses may increase the rate of type 1 error.
Although the exact mechanisms underlying the adverse effects of DBP exposures on early pregnancy outcomes are not clear, hormone disruption has been proposed as one of mechanisms of action.Both animal and human studies have shown that DBPs can induce endocrine disruption by exhibiting estrogenic properties. 38,73,74][77] Another potential biological mechanism is oxidative stress.9][80] Our two previous human studies have also shown that urinary TCAA concentrations were positively associated with oxidative stress biomarkers including 8-hydroxy-2-deoxyguanosine and 8-iso-prostaglandin F2a. 81,82Excessive oxidative stress has been documented to impair oocyte maturation and embryo development. 25,83he strengths of our study included the prospective study design, large-scale population, and comprehensive consideration of potential effect modifiers.Moreover, the use of DCAA and TCAA concentrations as biomarkers of DBP exposures, especially the repeated urinary measurements, can improve the accuracy of exposure assessment. 16The use of IVF as a model also enabled us to focus on those early pregnancy outcomes, including oocyte retrieval, oocyte fertilization, embryo quality, and embryo implantation, which are not available among women conceiving without medical assistance. 39owever, our study had several limitations.First, our study population was a subset of the overall population, and the differences in BMI and potential unmeasured demographic characteristics  S6.Model 1 was a crude model.Model 2 was adjusted for age, BMI, alcohol use, education level, passive smoking status, and household income.Model 3 was adjusted for age, BMI, alcohol use, education level, passive smoking status, household income, and infertility causes.p For trend was performed using the quartiles of urinary HAA concentrations as ordinal variables in the models.Note: BMI, body mass index; CI, confidence interval; DCAA, dichloroacetic acid; HAA, haloacetic acids; IVF, in vitro fertilization; TCAA, trichloroacetic acid.
Environmental Health Perspectives 097003-7 131(9) September 2023 may introduce selection bias.In addition, our study participants were recruited from an infertility clinic, which may limit the generalizability of our findings to the general population.However, the prevalence of infertility in reproductive-age couples is 15-20% in China. 2 Thus, our findings still had relevant health implications for the significant proportion of infertile couples undergoing ART treatment.Second, although urinary DCAA and TCAA can effectively reflect ingestion of drinking water THMs and HAAs, 14,18,84 it remains unclear whether such biomarkers can reflect other exposure routes (e.g., via inhalation and dermal absorption) and other species of DBPs (e.g., brominated and iodinated DBPs).Moreover, there is significant temporal and spatial variability in drinking water DBP concentrations, and the individual's DBP exposures are influenced by water-use activities.Therefore, the findings extrapolated to DBP exposures in general should be interpreted with caution.Third, the current study did not consider paternal DBP exposures, which have been reported to be associated with decreased semen quality. 85eteriorated semen quality may contribute to the observed results of certain early IVF outcomes (e.g., the proportions of fertilization and best-quality embryos) and clinical outcomes.Finally, we cannot rule out the possibility that some unmeasured confounding factors (e.g., the specific time of urine sample collection, psychological factors, and different reproductive health conditions) or coexposure to other environmental pollutants (e.g., organophosphate pesticides and phthalates), 8,86,87 which may have biased the results of our findings.

Conclusions
In this prospective cohort study, we first examined the effects of exposure to DBPs on IVF outcomes among women undergoing IVF/ICSI treatment.We observed that urinary DCAA and TCAA were inversely associated with multiple indicators of IVF outcomes, including numbers of total oocytes and MII oocytes, the proportion of best-quality embryos, as well as proportions of successful implantation, clinical pregnancy, and live birth.Our findings provided novel evidence that preconception exposure to DBPs might have adverse effects on early pregnancy outcomes.However, further investigation is warranted to confirm our findings and to explore the involved mechanisms of action.

ap-
Values were calculated using chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables.b A total of 2 women had missing information on total tap-water consumption. b

Figure 3 .
Figure 3. Associations between urinary HAA concentrations and clinical outcomes among 1,048 women undergoing 1,136 IVF cycles.All analyses were conducted using generalized linear mixed models with random intercepts, binary distribution, and logit link function.Data are presented as adjusted mean proportions (95% CI).Dots: the adjusted mean proportions; Lines: 95% CI; The corresponding numeric data are shown in TableS6.Model 1 was a crude model.Model 2 was adjusted for age, BMI, alcohol use, education level, passive smoking status, and household income.Model 3 was adjusted for age, BMI, alcohol use, education level, passive smoking status, household income, and infertility causes.p For trend was performed using the quartiles of urinary HAA concentrations as ordinal variables in the models.Note: BMI, body mass index; CI, confidence interval; DCAA, dichloroacetic acid; HAA, haloacetic acids; IVF, in vitro fertilization; TCAA, trichloroacetic acid.

Table 2 .
Cycle-specifc clinical characteristics from 1,136 IVF cycles among 1,048 women in the TREE study enrolled between 2018 and 2020.

Table 3 .
Distribution of urinary HAA concentrations among 1,048 women.
and Associations between urinary HAA concentrations and early IVF outcomes among 1,048 women undergoing 1,136 IVF cycles.(A)Totaloocyte count (1,048 women; 1,136 cycles) and (B) MII oocyte count (1,048 women; 1,136 cycles) were conducted using generalized linear mixed models with random intercepts, Poisson distribution, and log link function, and data are presented as percent changes (95% CI).(C) Fertilization proportion (1,042 women; 1,129 cycles) and (D) best-quality embryo proportion (1,003 women; 1,084 cycles) were conducted using generalized linear mixed models with random intercepts, binomial distribution, and logit link function, and data are presented as adjusted mean proportions (95% CI).Dots: the percent changes for total oocytes and MII oocytes or adjusted mean proportions for fertilization and best-quality embryos; Lines: 95% CI; Dash lines: Reference line.The corresponding numeric data are shown in TableS5.Model 1 was a crude model.Model 2 was adjusted for age, BMI, alcohol use, education level, passive smoking status, and household income.Model 3 was adjusted for age, BMI, alcohol use, education level, passive smoking status, household income, and infertility causes.p For trend was performed using the quartiles of urinary HAA concentrations as ordinal variables in the models.Note: BMI, body mass index; CI, confidence interval; DCAA, dichloroacetic acid; HAA, haloacetic acid; IVF, in vitro fertilization; MII, metaphase II; TCAA, trichloroacetic acid.