Urinary phthalates, serum omega-3 fatty acids and ovarian reserve among women seeking fertility treatment.

Couples seeking fertility treatment at the Massachusetts General Hospital (MGH) Fertility Center were invited to enroll in the Environment and Reproductive Health (EARTH) Study, a prospective cohort designed to evaluate environmental and dietary factors that affect fertility ( Messerlian et al., 2018 ). Women between 18 and 45 years old were eligible to participate between 2004 and 2017. Approximately 60% of women contacted by staff enrolled. After excluding women taking medication for polycystic ovary syndrome (PCOS) (medication information extracted from electronic medical records by trained nurses), 271 women enrolled in EARTH had either provided one or several urine samples for phthalate biomarkers assessment before (N = 241) or one sample at the same time (N = 30) as the AFC assessment. Circulating fatty acids were quantified in randomly selected women with available specimens who underwent in vitro fertilization (IVF) as part of an awarded pilot study. The first fresh cycle/serum sample was selected for each woman. Finally, we included 139 women who had data on serum fatty acid measurements in this study. The study was approved by the Human Studies Institutional Review Boards of Massachusetts General Hospital (MGH), Harvard School of Public Health (HSPH), and the Centers for Disease Control and Prevention (CDC). Informed consent was signed by participants after study procedure explained and all questions answered at study enrollment. During each IVF cycle, participants provided up to two urine samples approximately one week apart. All urine samples collected before ovarian reserve determination were included in this study. Women provided a median (IQR) of 2 (1, 6) urine samples before ovarian assessment and 32% (N = 45) provided 5 or more samples. A total of 482 urine samples were included in this study. Urine collection, specific gravity (SG) measurement and phthalate metabolite quantification at CDC (Atlanta, GA, USA) were done as previously described ( Hauser et al., 2016 ). Along with study samples, each analytical batch included a set of standards, quality control materials (QCs) and reagent blanks. Precision over 3 months, calculated from the coefficient of variation of repeated analysis of QC samples, was 2.7%−14%, depending on the phthalate metabolite. The phthalate metabolites quantified between 2004 and 2017 were: mono-n-butyl phthalate (MBP), mono-isobutyl phthalate (MiBP), monoethyl phthalate (MEP), monobenzyl phthalate (MBzP), and four di(2-ethylhexyl) phthalate (DEHP) metabolites, namely 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). The limits of detection (LODs) were: 0.5–1.2 μg/L (MEHP), 0.2–0.7 μg/L (MEHHP, MEOHP), 0.2–0.6 μg/L (MECPP), 0.2–0.3 μg/L (MBzP, MiBP), 0.4–0.8 μg/L (MEP), and 0.4–0.6 μg/L (MBP). Women provided blood serum at study entry and each IVF cycle. One serum sample per woman was included in this study where concentrations of serum omega-3 fatty acids were measured. Serum samples were aliquoted, frozen, and stored at −80 °C until transfer to the Biomarker Research Laboratory at the Harvard T.H. Chan School of Public Health (Boston, MA) for analysis. Serum fatty acid concentrations were determined as previously described by Baylin et al ( Baylin et al., 2002 ). Briefly, fatty acids were extracted from serum samples and transmethylated with methanol and sulfuric acid as described by Zock et al ( Zock et al., 1997 , 1996 ). After esterification, the fatty acid methyl esters were re-dissolved in iso-octane and quantified by gas chromatography. Peak retention times were identified by injecting known standards of purity above 99 percent (NuCheck Prep, Elysium, MN). Agilent Technologies ChemStation A.08.03 software was used for analysis. CVs for all the fatty acids studied were monitored continuously by analysis of a pooled control sample, that were indistinguishable from other study samples, in each extraction and analysis batch. Quality control was done by external validation through programs offered by both the American Oil Chemists Society and the National Institute of Standards and Technology. The mean CVs for EPA and DHA in the 16 pairs of controls were 7.3% and 7.2% respectively. The sum of EPA and DHA (EPA+DHA, also referred to as the omega-3 index), was used based on the fact that EPA and DHA have similar dietary sources ( U.S. Department of Agriculture, Agricultural Research Service, 2012 ). All women included in this study underwent ovarian reserve evaluation as part of infertility diagnostic procedures at the MGH Fertility Center, including transvaginal ultrasound determination of AFC. We included the first AFC measurement for each participant. AFC describes the sum of antral follicles measuring 2–10 millimeters in diameter in both ovaries that are observed during an early follicular phase by transvaginal scan ( Practice Committee of the American Society for Reproductive Medicine, 2015 ). Transvaginal ultrasound was performed by reproductive endocrinology and infertility specialist physicians on the third day of an unstimulated menstrual cycle following a cycle without any fertility medications. Among women with menstrual dysfunction, AFC was determined when there was evidence(s) of early follicular phase (e.g., thin endometrial lining, no evidence of dominant follicle, and estradiol and progesterone levels reflecting early follicular phase). Women included in this study had a median (IQR) AFC of 11 (8, 16), and 7 (5%) of them had an AFC above 30. As previously done in several studies, AFC above 30 were truncated to 30 in the main analyses of this study to reduce the influence of very high AFC ( Jiménez-Cardozo et al., 2023 ; Maldonado-Cárceles et al., 2022 ; Mínguez-Alarcón et al., 2021 ). At study entry, each participant’s date of birth was collected, and weight and height were measured by trained staff. At enrollment, sociodemographic, lifestyle, and medical history questionnaires were administered to participants. Clinical information was extracted from electronic medical records by trained nurses. Infertility diagnoses were made by physicians and coded according to definitions by the Society for Assisted Reproductive Technology (SART), as previously described ( Hauser et al., 2016 ), and primary infertility diagnoses were reported in this study. Demographic and reproductive characteristics were reported as median ± inter-quartile ranges (IQRs) for continuous variables and count (percentage) for discrete variables. There were 2 missing values (1%) for body mass index (BMI). Given the low frequency of missingness, we imputed missing BMI as the median value of BMI of the cohort. We calculated the geometric mean of all urinary phthalate metabolite concentrations in urine samples provided prior to ovarian reserve assessment at a participant level. The distributions of urinary phthalate metabolite concentrations were reported as statistics of the geometric means at a participant level ( Supplemental Table 2 ). Each phthalate metabolite was detected in at least 93% of samples, and urinary phthalate metabolite concentrations below the LOD were replaced with the corresponding LOD divided by the square root of 2 ( Hornung and Reed, 1990 ). We used geometric means for molar sums of the four DEHP metabolites as each participant’s summary estimate of DEHP exposure, denoted by ∑DEHP (μmol/L), in all analyses. ∑DEHP was calculated by dividing each DEHP metabolite’s geometric mean by its molecular weight and then summing. Similarly, we calculated the molar sums of MiBP and MBP to create a summary measure of dibutyl phthalate (DBP) exposure, denoted by ∑DBP (μmol/L). The geometric mean concentrations of phthalate biomarkers were log-transformed (natural logarithm) to more closely approximate a normal distribution (tested by Shapiro-Wilk normality test) and were categorized into quartiles when assessed as a mixture. Serum concentrations of fatty acids were categorized into tertiles to minimize the influence of outliers when analyzed as effect modifier. The associations between phthalate biomarkers and ovarian reserve marker were evaluated both individually and as a mixture. To evaluate single chemical associations, we applied adjusted Poisson regression for AFC and population marginal means were calculated at the median within each quartile of phthalate biomarker. Quantile g-computation was used to assess the joint association of the phthalate biomarkers mixture with ovarian reserve marker ( Keil et al., 2020 ). Mixture effects were evaluated with bootstrapping method by simultaneously increasing all mixture components by one quartile. Effect modification was tested by adding a multiplicative term between phthalate biomarkers and tertiles of serum fatty acid and p-values of the interaction terms were determined by Wald tests. Stratification by tertiles of serum fatty acid levels was also conducted. P-values of the interaction terms and the estimated AFCs with 95% confidence intervals (CIs) at each quartile of the phthalate biomarkers mixture and tertile of serum fatty acid were reported. The set of confounding variables was selected based on previous knowledge regarding their impact on phthalate exposure and reproductive health of women and technical concerns and the use of directed acyclic graphs (DAGs) (Shrier and Platt, 2008) ( Supplemental Figure 1 ). Final models were adjusted for age (continuous), BMI (continuous), ever smoked (binary), number of urine samples (numeric, accounting for measurement error and length of treatment) and specific gravity (as a geometric mean of the specific gravities of urine samples included to account for urine dilution). Sensitivity analyses on untruncated AFC were conducted. To account for possible changes of lifestyle as well as ultrasound technique over time, sensitivity analyses further adjusting for year of AFC scan (binary, by 2010) were also conducted. The threshold of P value for statistical significance was 0.05. All analyses were performed in R (version 4.0.5). Single chemical models were built using package stats (version 4.0.5). Population marginal means were estimated by package emmeans (version 1.7.0). Quantile g-computation analyses were conducted by the R package qgcomp (version 2.15.2) and an extension package for effect modification called qgcompint (version 0.7.0).

The women included in our analysis were predominantly white/Caucasian (82%) and had never smoked (73%), with an average age of 36 years and an average BMI of 23.2 kg/m 2 at enrollment ( Table 1 ). The included and excluded women were generally comparable in age, BMI, AFC and other baseline and reproductive characteristics, although the excluded women tended to be less active on average, have fewer available urine samples and had a higher proportion of women with AFC greater than 30 ( Supplemental Table 1 ). Women with higher serum DHA and EPA more often had a graduate degree (48%, 61%, 70% across tertiles), consumed more dark meat fish (0.3 [0.0, 0.7], 0.7 [0.6, 1.1], 0.7 [0.3, 1.1] servings per week across tertiles), regularly took multi-vitamins (65%, 67%, 81% across tertiles) and had higher intake of EPA+DHA from supplements (75.3 [1.4, 182.8], 158.2 [1.0, 277.6], 176.2 [1.0, 384.7] mg/d across tertiles) compared to women with lower DHA+EPA. Overall, the women included have a median (IQR) AFC of 11 (8, 16). The median (IQR) urinary concentrations were 0.1 (0.1, 0.3) μmol/L (∑DEHP), 0.08 (0.04, 0.15) μmol/L (∑DBP),, 30.7 (14.1, 96.9) μg/L (MEP), 11.1 (4.9, 21.9) μg/L (MBP), 5.6 (2.9, 9.9) μg/L (MiBP), and 3.0 (1.3, 6.1) μg/L (MBzP) ( Supplemental Table 2 ). The distributions of phthalate biomarkers among women in this study were similar to those observed among females in the U.S. general population participating in the National Health and Nutrition Examination Survey (NHANES) ( National Center for Environmental Health, 2022 ), while the serum EPA and DHA concentrations [mean (SE) = 1.0 (0.07) and 2.9 (0.27) % of total serum fatty acids] were two-fold higher than reported levels among 20–55 years-old adults [EPA: 0.51 (0.02) %; DHA: 1.29 (0.06) %] ( Murphy et al., 2021 ). Serum EPA and DHA were moderately correlated with each other (Spearman r = 0.63) but showed weak correlation with ALA (Spearman r = 0.10 and 0.14, respectively). Serum EPA+DHA levels were weak to moderately reverse-correlated with the phthalate biomarkers (Spearman r from −0.36 to −0.11, Supplemental Figure 2 ). We observed negative associations of urinary concentrations of ∑DEHP, ∑DBP, MiBP and MEP with AFC when the phthalate biomarkers were modeled as continuous variables [percent change (95% CI) per log-unit increase = −7 % (−11, −2), −7% (−13, −1), −5% (−11, 1) and −4 % (−8, 0), P = 0.003, 0.03, 0.08 and 0.03, respectively] and stratified into quartiles [ P trend across quartiles = 0.004, 0.002, 0.005 and 0.03, respectively] ( Table 2 ). Similar non-significant decreasing trends were identified with urinary concentrations of MBP, while no association was found between urinary concentrations of MBzP and AFC. When evaluated as a mixture, we also observed a decreasing trend in mean AFC with a −12% (−23, 1) change per quartile increase of the phthalate biomarkers mixture. We found a suggestive effect modification by serum EPA+DHA on the associations between the phthalate biomarkers mixture and AFC ( P for interaction = 0.23) but not by ALA ( P for interaction = 0.94) ( Figure 1 and Supplement Table 3 ). Specifically, women in the lowest tertile of serum EPA+DHA had a decreasing estimated mean AFC across quartiles of the phthalate biomarkers mixture [18.4 (12.1, 27.9), 14.7 (11.6, 18.6), 11.8 (10.2, 13.6) and 9.4 (7.2, 12.3), respectively] ( P trend = 0.03). While similar decreasing trends were found among women in the middle tertile of serum EPA+DHA ( P trend = 0.07), no association was observed among women in the highest serum DHA and EPA group ( P trend = 0.56). Decreasing trends by phthalate mixture were observed across all tertiles of serum ALA but associations were not significant ( P trend = 0.16, 0.08 and 0.16, respectively), and no interaction was found ( P for interaction = 0.94). We then aimed to find the individual phthalate biomarker with the strongest effect modification. Urinary ∑DEHP appeared to drive the interaction with serum EPA+DHA in relation to AFC ( P for interaction = 0.01) ( Figure 2 and Supplemental Table 4 ). ∑DEHP was negatively associated with AFC among women in the low and middle tertile of EPA+DHA ( P trend < 0.001 and 0.002), but not among women in the highest tertile ( P trend = 0.93) ( Figure 2 and Supplemental Table 5 ). No significant interaction was found between ∑DEHP and serum ALA in relation to AFC. Potential interactions between MiBP and EPA+DHA and MBzP and ALA ( P for interaction = 0.08 and 0.07) were also worth noting. No significant interaction was found between other urinary phthalate biomarkers and serum fatty acid concentrations ( Supplemental Table 4 ). Sensitivity analyses with untruncated AFC yielded similar results on single chemical and mixture association and effect modification by serum EPA+DHA ( Supplemental Tables 6 and 7 ). Further adjusting for year of antral follicle scan showed similar effect modification ( Supplemental Table 8 ).

In this prospective study among women enrolled in the EARTH study, we investigated the relationships between urinary phthalate biomarkers, serum omega-3 fatty acids and marker of ovarian reserve. We found that the urinary phthalate biomarkers mixture and ∑DEHP were inversely related to AFC among only women in the low and middle tertiles of serum EPA+DHA, but not among women in the high tertile. No effect modification by serum ALA levels was found for AFC. These findings suggest that increased intake of certain serum n3PUFAs may attenuate the detrimental effect of phthalate exposure on ovarian reserve. Randomized trials are warranted to confirm the necessary dose, source and timing of such possible intervention. Phthalate exposure can disrupt normal ovarian function by multiple mechanisms ( Hannon and Flaws, 2015 ). It has been reported that exposure to phthalate mixtures increased oxidative stress and caused follicle death in cell culture studies ( Panagiotou et al., 2021 ) and altered gene expression on antral follicle growth and sex steroid synthesis in mice ( Meling et al., 2020 ). Also, DEHP was reported to disrupt critical timings of follicle growth and differentiation by activating PPARs in granulosa cells ( Froment et al., 2006 ; Lovekamp-Swan and Davis, 2003 ; Meling et al., 2022 ). Conversely, experimental studies have demonstrated the benefits of nPUFAs on ovarian health. In animal models, dietary n3PUFA prolonged the female reproductive lifespan and improved egg quality ( Nehra et al., 2012 ). It was also shown that n3PUFA supplementation can lower serum FSH among women with normal weight ( Al-Safi et al., 2016 ). An experimental study also revealed the protective effect of omega-3 fatty acids against cyclophosphamide’s ovarian toxicity as an antioxidant ( Nair et al., 2020 ). There was evidence that nPUFAs act through PPAR-γ to modulate inflammatory cytokine production and antioxidant signaling pathway in the reproductive system ( Frew et al., 2013 ; Ishihara et al., 2019 ). Thus, the potential mechanism of the interaction between nPUFAs and phthalate biomarkers might involve ligand competition for PPARs and modulation of oxidative stress in the ovary. However, further studies are warranted to confirm these hypotheses regarding the mechanisms of potential interaction. It is worth noting that the serum EPA and DHA concentrations in our population were two-fold higher than the concentrations reported in NHANES (2011–2012) among 20–55 year-old adults ( Murphy et al., 2021 ). The EARTH cohort was recruited at a fertility center in Boston, MA (USA), and the participants reported high consumption of dark meat fish and fish oil supplements (top sources of nPUFAs in the EARTH study) ( Y.-H. Chiu et al., 2018 ). It is possible that the generally high serum nPUFAs levels of EARTH participants made it possible to detect effects of nPUFAs that might need relatively high concentrations to interact with phthalates. Further studies are warranted to test if our findings are generalizable in the general population and women with lower intake of omega-3 fatty acids. To the best of our knowledge, this is the first prospective study to test the hypothesis that omega-3 fatty acids may attenuate the negative effects of phthalate biomarkers on ovarian reserve. Given the novelty of our study, we could not directly compare our results to any observations in previous studies. One recent study demonstrated high omega-3 PUFA measured in seminal plasma, notably elevated docosapentaenoic acid (DPA, C22:5n-3), moderated the association between benzophenone-3, mono(3-carboxypropyl) phthalate (MCPP) and sperm motion parameters among 155 male participants ( Gao et al., 2024 ), which corresponded to our findings among female. Another previous study by Stevens et al. explored the role of omega-3 fatty acids intake in the relationship between phthalate metabolites mixture and fetal growth ( Stevens et al., 2022 ). Contrary to their hypothesis, the negative association of phthalate metabolite mixtures with fetal growth was only observed among women with adequate omega-3 intake (dichotomized the population by reported number of meals with seafood per week and nPUFAs supplement use). This study had several strengths. First, the EARTH study is a well-established prospective cohort. EARTH was designed to evaluate nutritional and environmental determinants of reproductive health with detailed clinical information collected from medical records, allowing us to investigate complex interplay between objective biomarkers of dietary intake such as serum fatty acids and multiple phthalate biomarkers with important ovarian reserve measured (typically obtained when women attend the clinic and undergo the ultrasound). Second, the quantification of phthalate metabolites in multiple urine samples collected before ovarian reserve assessment better represented participants’ long-time exposure to phthalates and reduced exposure measurement error. Third, phthalate biomarkers were modeled both individually and as a mixture using modern statistical approaches to reflect real-world scenarios on chemical exposures. Fourth, in our study, we use serum fatty acid concentrations, which are relatively objective measurements and good biomarkers of habitual intake over the span of 15 months ( Murphy et al., 2021 ). However, we also recognize several limitations. First, these findings might not be generalizable to the general population because this study included women attending a fertility clinic. However, the population provided ovarian assessment data that were otherwise unavailable in the general population. Second, as in any observational study, residual confounding by other nutritional, environmental and genetic factors was possible. Third, although we restricted our analyses to women with urinary phthalate metabolites before ovarian reserve assessment to avoid reverse causation, we cannot interpret the associations as causal. Animal studies and intervention studies are warranted to further explore potential interactions as those we observed and their causal relationship with ovarian reserve.

In summary, among women seeking fertility treatment, we found concentrations of some urinary phthalate biomarkers were negatively associated with ovarian reserve marker, but only among women with low and middle serum levels of certain n3PUFAs. These findings suggest that certain serum n3PUFAs may attenuate the adverse effects of phthalate exposure on women’s fertility potential. Such interaction points toward select n3PUFAs as key modifiers of phthalate ovarian toxicity with potential implications for other women’s reproductive health endpoints. These results, if replicated by further studies, add to our understanding of the benefits of n3PUFA intake in ovarian health.

Phthalates are a class of endocrine disruptive chemicals (EDCs) ( Gore et al., 2015 ) widely used as plasticizers and solvents, and found in many consumer products and medical devices, such as flooring and wall coverings, food packaging and personal care products ( Hauser and Calafat, 2005 ), leading to ubiquitous exposure around the world ( National Center for Environmental Health, 2022 ; Silva et al., 2004 ; Zota et al., 2014 ). Exposure to phthalates has been widely associated with adverse reproductive health effects, including reduced ovarian reserve among women ( Basso et al., 2022 ; Génard-Walton et al., 2023 ; Hauser et al., 2016 ; Jiang et al., 2024 ; Jukic et al., 2016 ; Li et al., 2022 ; Messerlian et al., 2016 ; Sacha et al., 2021 ; Thomsen et al., 2017 ; Vélez et al., 2015 ; Zhan et al., 2022 ), though studies of phthalate mixtures effect on antral follicle count (AFC) yield discrepant results ( Génard-Walton et al., 2023 ; Jiang et al., 2024 ). Although efforts have been made to regulate or replace their use, phthalate exposure remains universal and difficult to avoid and eliminate in industrialized countries. Given this, studies exploring whether modifiable dietary components can reduce certain EDC exposure’s damage to reproductive health have been emerging ( Gao et al., 2024 ; Mínguez-Alarcón et al., 2016 ). Ovarian reserve can also be influenced by a number of dietary and behavioral factors ( Jiménez-Cardozo et al., 2023 ; Maldonado-Cárceles et al., 2022 ; Mínguez-Alarcón et al., 2018 ; Mitsunami et al., 2023 ). Long-chain n-3 polyunsaturated fatty acids (n3PUFAs) are essential fatty acids required for humans that must be obtained from diet. N3PUFAs are involved in important biological processes in human reproductive system and evidence has suggested that n3PUFAs are beneficial to human fertility ( Yu-Han Chiu et al., 2018 ; Gaskins and Chavarro, 2018 ; Sturmey et al., 2009 ; Wathes et al., 2007 ). Studies have shown that higher intake of n3PUFAs and n3PUFA-rich food (e.g. seafood and nuts) were associated with increased total E2 and lower risk of anovulation ( Mumford et al., 2016 ), lower follicle-stimulating hormone (FSH) ( Al-Safi et al., 2016 ) and higher fecundability ( Gaskins et al., n.d. ; Wise et al., 2018 ) in epidemiologic studies, and improved symptoms of polycystic ovary syndrome ( Albardan et al., 2024 ; Ma et al., 2023 ; Zhang et al., 2023 ) and delayed ovarian aging ( Nehra et al., 2012 ) in animal models. Interactions between phthalate metabolites and long-chain n3PUFAs are biologically plausible because both n3PUFAs ( Nakamura et al., 2014 ) and phthalate metabolites ( Casals-Casas et al., 2008 ; Corton and Lapinskas, 2005 ) can bind to PPARs, which are expressed in ovarian tissues and involved in key processes of ovarian function ( Abbott, 2009 ). Interestingly, it has been reported that high seminal plasma omega-3 fatty acids moderated the association between MCPP and sperm motion parameters ( Gao et al., 2024 ). Moreover, based on results from animal studies, it has been hypothesized that omega-3 supplementation during pregnancy may counteract the adverse effects of phthalate exposure on abnormal fetal development ( Latini et al., 2006 ). To our knowledge, whether intake of n3PUFAs ameliorates the negative effects of phthalates on ovarian reserve among women is unknown. To address this knowledge gap, we aimed to evaluate the effect modification of serum n3PUFA levels on the associations of urinary concentrations of phthalate biomarkers, individually and as a mixture, with marker of ovarian reserve among women seeking fertility treatment.