Recalled maternal lifestyle behaviors associated with anti-müllerian hormone of adult female offspring.

Among 1228 women enrolled in the trial, 1202 completed the questionnaire regarding in utero exposures and had measured AMH concentrations available. Participants were primarily white (94.7%) with a mean body mass index (BMI) of 26.3 (standard deviation [SD] 6.5) kg/m 2 and mean age of 28.7 (SD 4.8) years ( Table 1 ). Participant’s mother’s age at time of questionnaire was 46 to 83 and their ages when they gave birth to participants in EAGeR ranged from 15 to 44. Women in the low AMH group tended to be older, had lower testosterone, and were more likely to be parous. There were no differences in BMI, race, treatment assignment, and number of prior pregnancy losses by AMH category. AMH values of the participants were then compared between those with and without the specific exposures ( Table 2 ). Participants who reported that their mothers were taking vitamin supplements while they were pregnant with them also had higher AMH than those who did not report exposure to vitamins (3.0 ng/mL versus 2.3 ng/mL, p=0.002). Self-reported in utero caffeine exposure was associated with lower AMH (2.7 ng/mL versus 3.0 ng/mL, p-value=0.04). Exposure to DES was very limited with only seven participants reporting this exposure and differences in AMH were similar (2.4 ng/mL versus 2.8 ng/mL, p-value=0.59). Geometric mean AMH concentrations were similar between those reported to be exposed and not exposed to tobacco, alcohol, and both maternal full and part time work. When evaluating the relationship of reported in utero exposures to low and high versus normal AMH categories, exposure to caffeine and DES were associated with an increased risk of low AMH in the offspring in unadjusted models ( Table 3 ), though after adjustment only caffeine was significantly associated with risk of low AMH. Conversely, report of mothers’ vitamin use during pregnancy was associated with high AMH in adjusted models. Further, while maternal employment was associated with a lower risk of high AMH before adjustment, results were attenuated after adjusting for potential confounders. No associations were noted for both in utero exposure to cigarette and alcohol use. Similar conclusions were observed across all exposures in sensitivity analyses to address missing exposure information ( Figures 1 and 2 for vitamin and caffeine exposure respectively; figures for other exposures not shown). Specifically, results indicated that under most scenarios the association between vitamin use and AMH would remain statistically significant, indicated by results from all multiple imputation scenarios lying in the white region of Figure 1 (as represented by the blue triangles). Further, under reasonable scenarios the association between caffeine and AMH would remain statistically significant, indicated by the majority of the blue triangles from multiple imputation scenarios lying in the white region of Figure 2 , though for some imputations the association could be attenuated.

This report is a secondary analysis of women participating in the Effects of Aspirin in Gestation and Reproduction (EAGeR) trial 21 , 22 .This trial was a multi-center, double-blind, block-randomized, placebo-controlled trial evaluating the effects of low dose aspirin (LDA) on live birth. The detailed study design and methods have been previously published 22 . In brief, 1228 women trying to conceive were enrolled from 2007–2011 at four U.S. clinical centers. Inclusion criteria were a history of one or two documented pregnancy losses and age 18–40 years. Exclusion criteria included a history of major medical disorders, infertility, pelvic inflammatory disease, tubal occlusion, endometriosis, anovulation, uterine abnormalities, or polycystic ovarian syndrome. Institutional Review Board authorization was obtained for the data coordinating center and at all clinical centers and all participants provided written informed consent. Patient safety was monitored by the Data Safety and Monitoring Board (DSMB) and the trial was registered with ClinicalTrials.gov , number NCT00467363 . Exposure information was collected via questionnaire at baseline. Women participating in the EAGeR trial were asked to answer based on their knowledge of their mother’s exposures during the participant’s in utero development. For example, women were asked: “To the best of your knowledge, did your mother take vitamins while pregnant with you?”, with possible responses of yes, no, or don’t know. In utero exposures that were investigated included DES exposure, vitamin use, smoking, alcohol intake, caffeine use, and working either full or part time during pregnancy. AMH concentrations in women in the EAGeR trial (the daughters of the women in whom exposures were assessed) were measured in serum samples collected during preconception at baseline, which occurred between days 2–4 of the menstrual cycle. Serum samples were stored at −80°C until analyses. Concentrations of AMH were determined in serum samples using the GEN II ELISA assay protocol with correction for complement interference (Beckman Coulter) 23 . AMH data were available for 1202 of the 1228 women in the cohort. All machine observed concentrations were used without substitution of concentration below the limits of detection (0.006 ng/mL) to avoid bias 24 . The inter-assay laboratory coefficients of variation (CVs) were 6.2% and 6.6% at mean concentrations of 8.9 and 3.1 ng/mL, respectively, for lyophilized manufacturer’s controls and 6.3% for an in-house pooled serum control. To maximize the measurement precision of AMH, we evaluated for any improper calibration curves or out-of-range values for manufacturer-provided control samples using a pooled standard curve and confirmed that sample recalibration was not required 25 . AMH concentrations for this analysis were categorized using clinical thresholds based on data from infertile women with low AMH as under 1.00 ng/mL, normal AMH as 1.00–3.5 ng/mL, and high AMH as over 3.5 ng/mL 1 , 3 , 23 . In prior studies, AMH is not a valuable tool for fecund women and therefore, the thresholds are importantly based on infertile women 3 . Categorizing, allows for generalization of high, normal, and low, in a variable that specific values are less meaningful than the category women may fall in after a diagnosis of infertility. Liquid chromatography and tandem mass spectrometry using a Shimadzu Prominence Liquid Chromatogram (Shimadzu Scientific Instruments, Inc., Columbia, MD) with an ABSceix 5500 tandem mass spectrometer (AB SCIEX, Framingham, MA) was utilized to measure the patient’s total testosterone for evaluations. Sex hormone binding globulin (SHBG) concentration was determined by SHBG reagent/sandwich immunoassay method/electrochemiluminescence (Roche Diagnostics, Indianapolis, IN) utilizing a Roche COBAS 6000 chemistry analyzer (Roche Diagnostics). Interassay CVs were 3.0% at 55.64 nmol/L and 3.8% at 19.74 nmol/L and limit of detection was 0.800 nmol/L. From this value, along with the patient’s total testosterone, the free testosterone was able to be calculated as 24.00314 × T/log10S – 0.0499 × T 2 and free androgen index (FAI) as 100*(T/S), where T = total testosterone in nmol/L and S = SHBG in nmol/L 26 . Demographic characteristics of the daughters (i.e. EAGeR participants) were compared by AMH category. ANOVA or Fisher’s exact test were used, as appropriate, to determine differences in characteristics across low, normal, and high AMH categories. Geometric mean AMH concentrations and 95% confidence intervals (CIs) were also compared between women with versus without the specific in utero exposures of interest. Multinomial regression was used to evaluate the associations between reported in utero exposures and the risk of low or high AMH compared to normal levels. Models were adjusted for relevant confounding factors including, EAGeR participant age, mother’s age at birth of daughter, mother’s history of fertility treatment, and mother’s history of vitamin use (except where history of vitamin use was the main exposure of interest), as these maternal factors may be related to offspring reproductive health as well as the given exposures of interest. As there was missing information on maternal exposures (ranging from 8% for smoking status to 40% for DES exposure; most factors were missing approximately 12%) we completed a sensitivity analysis for missing data to evaluate the degree to which this missing data may have influenced the findings. We imputed, from the digital data set of completed surveys, exposure information under every scenario in the low, normal, and high AMH groups, as well as using multiple imputation with 1000 imputed datasets. P-values for each scenario were plotted, along with the level of missing data imputed for each multiple imputation for the association between each exposure and AMH. SAS version 9.4 (SAS Institute, Cary, NC) and R version 3.5.2 (R Foundation, Vienna, Austria) were used for the statistical analyses.

As ovarian follicles develop in utero , there is concern that prenatal exposures during this follicular formation can have an impact on the future fertility of the offspring. Though our results are based on daughter’s report of the maternal exposures, our results are suggestive of the potential for in utero exposure to vitamins and caffeine to influence AMH concentrations in female offspring. Given the importance of prenatal health to optimize the health of the offspring, additional studies are needed to confirm these findings.

Overall, this study demonstrates potential intergenerational effects between reported in utero exposures during pregnancy and AMH in female offspring. Specifically, women who reported that their mothers were taking vitamin supplements while they were pregnant with them may have higher AMH in adulthood, whereas women whose mothers used caffeine during pregnancy may have lower AMH. This is one of two studies to evaluate the relationship between multiple in utero exposures and AMH in female offspring, and the results highlight the potential importance of lifestyle factors during pregnancy on reproductive health of the offspring 1 . Given that our findings were limited by reliance on daughters report of the maternal exposures, further studies with more detailed exposure information are needed to corroborate these results. Our findings are largely consistent with available evidence and proposed biological rationale, though direct comparison is difficult given the paucity of data for specific exposures. We observed that report of maternal caffeine use during pregnancy was associated with an increased risk of low AMH in the daughters. While caffeine use during pregnancy has been widely studied with respect to pregnancy outcomes, often with conflicting results, evaluation in relation to fertility of offspring is rare 27 , 28 . A study of rats exposed to caffeine showed no changes in fertility of offspring 27 , though a study in humans found that daughters of tea drinkers had delayed puberty, whereas daughters of caffeine drinkers did not 28 . To our knowledge, no other studies have evaluated intergenerational effects of caffeine use during pregnancy on fecundability or AMH in female offspring. The American College of Obstetricians and Gynecologists recommends limiting the amount of caffeine exposure in pregnancy, as it is known to limit blood flow to the uterus and placenta. This inhibition in blood flow may divert blood supply away from developing the reproductive organs. It is important to note that our data on caffeine exposure were limited by self-report of the daughters of any versus no exposure. Future studies should evaluate the dose of caffeine consumed and the timing during pregnancy. Though our findings with respect to DES exposure were not statistically significant and based on few events, our data indicate a potential increased risk of low AMH. This is consistent with the large body of evidence regarding the deleterious effects of DES on offspring fertility. Women exposed to DES in utero are also known to have menstrual irregularities and early menopause which may be due to the lack of complete development of the menstrual tract, and further support the negative, though insignificant, association we found in this study 29 . Maternal vitamin use during pregnancy was positively associated with an elevated AMH. This positive association is consistent with data showing that adequate maternal nutrition status in pregnancy is protective for healthy endocrine pathways that influence fertility 30 , though AMH in the daughters has not been previously evaluated. Our findings are in line with studies that have shown that maternal vitamin use is associated with beneficial health outcomes in the offspring, including healthy body weight 31 , reduced risk of asthma 32 , and leukemia 33 , among others. It has been suggested that maternal nutrition plays a critical role on these later offspring health outcomes through changes in DNA methylation 34 , and due to nutrient status, endocrine health of the mother, and oxidative stress on the developing fetus 35 . Though our exposure information was limited in that we do not know the formulation, dose, or frequency of vitamin use during pregnancy, these results add to our knowledge regarding the use of vitamins on offspring health and suggest that further research on the reproductive health outcomes of the offspring is needed. Caution should be noted as elevated AMH can also be a marker of other reproductive issues in a young female such as polycystic ovarian syndrome (PCOS). While there are women in the EAGeR trial who may have a subclinical phenotype with mild elevations in both AMH and testosterone ( Table 1 ), none of the women in this trial had known PCOS, so no comment can be made on associations with high AMH in that setting 36 , 37 . Further, we found that employment status of the mothers was not related to the AMH of the female offspring. There are no studies that directly examine the impact of maternal employment on the AMH or fertility of the offspring. Given the lack of detailed data regarding work during pregnancy, it is difficult to make strong conclusions and more research is needed in this area. Our finding that cigarette use during pregnancy was not associated with the AMH of the offspring is somewhat inconsistent with prior data that found decreased fertility and fecundability in the offspring of women who smoked while pregnant 7 , 8 , 12 , 17 , 19 , as well as both earlier and later menarche 4 – 6 , 14 – 16 , 18 , 20 , 38 , and earlier menopause 9 . Though it is important to note that not all prior studies have shown associations with these outcomes 16 , an association may be expected based on this collective prior evidence and biological rationale. However, AMH in the offspring has only been evaluated in girls 14–16 years old 1 . Animal studies show maternal tobacco exposure inhibits granulosa cell proliferation and promotes apoptosis 39 , 40 . The authors of the referenced study hypothesized that this effect on the ovaries may not affect AMH because the effects are not long term, and therefore adult AMH values are normal 1 . Interestingly, there was an association between paternal smoking during pregnancy and lower AMH of the offspring 1 . Cigarette smoking is known to be associated with prolonged time to pregnancy, adverse pregnancy outcomes, spontaneous abortion, preterm delivery, and low birth weight 12 , 17 , 41 . However, there are also several other studies that do not demonstrate these effects 1 , 42 , 43 . One possibility is that external smoke exposure damages the DNA of cord blood of infants 44 , but the authors of the referenced study believe that the change in AMH was also potentially from the effect of post-natal smoking and second-hand smoking during infancy and childhood with paternal smoking 1 . Our study found no relationship between smoking or alcohol use in pregnancy on the AMH of the adult daughters. While these data are self-reported by the adult daughter, this finding is surprising based on the well-known effects of smoking on personal fertility. However, this observation is consistent with the other two published studies on the intergenerational effects of smoking 1 , 38 . Alcohol abuse and fetal alcohol syndrome can have devastating effects on the child including severe developmental delays. Different mechanisms have been offered to explain the teratogenic effects of alcohol on the developing embryo including oxidative stress, disturbed metabolism of vital nutrients and proteins, impaired neurogenesis, increased cellular apoptosis, and effects on gene expression 45 . While there is evidence that spermatogenesis, and therefore male infertility in offspring, is adversely affected by prenatal alcohol exposure, there are a handful of reports of female infertility with alcohol exposure, however, they does not reference prenatal exposure, but rather exposure while trying to concieve 45 , 46 . This is consistent with the lack of association elucidated from this study. The strengths of this study include measurement of AMH in a large cohort of 1202 healthy women with an assurance of fecundity. Our study design allowed the evaluation to be adjusted for known confounders, BMI, and age, as well as maternal age at conception and maternal use of infertility treatments to conceive. This is consistent with the adjustments made by the other large study evaluating the effects of AMH 1 . Importantly, this study is limited in that it relies on the participants’ recall and knowledge of their mother’s health during their in utero development, which raises concerns regarding potential recall bias. Evaluations on maternal recall of medication use in pregnancy have varied in prior reports, but generally are universally present throughout similar studies 47 , 48 . Finally, as previously discussed, AMH values were selected based on studies of infertile women and therefore, the cut-offs used in this investigation are not applicable to fertile women 3 . In light of this limitation, our findings should be considered as hypothesis generating and as a proof of concept for potential in utero effects on ovarian reserve of the offspring and additional studies are needed to further refine the exposure assessment. Moreover, each question was listed only as a binary response of “yes” or “no”. This limits the ability to evaluate whether a dose-response or threshold effects are present. We also chose to include women with a history of one to two prior spontaneous abortions which limits this utility of this study to all healthy women and increases the chance for a selection bias. Finally, the utility of AMH as a marker for fertility is still under debate as it does not correlate directly with ability to conceive or quality of eggs, but at this time remains an important marker for ovarian reserve 2 .

All ovarian follicles are developed during fetal development and therefore, the maternal environment plays a critical role in the health of these developing follicles. A woman’s reproductive lifespan is related to the number of ovarian follicles, each containing a single oocyte, that are established in utero and how long it takes this number to undergo atresia 1 . The majority of follicles will never complete maturation. Given that the primordial follicle pool is established during fetal development, understanding the role of the maternal environment during pregnancy is critical for understanding factors associated with downstream fertility and reproductive health outcomes of daughters. AMH is a widely used marker related to ovarian reserve, and more specifically the current pool of antral follicles, that is secreted by the granulosa cells of the ovarian follicles 2 , 3 . It may be a particularly relevant endpoint for examining impacts of in utero exposures on reproductive health since it is a simple circulating marker. AMH does not measure egg quality or failure to conceive, but is more specific for ovarian reserve, which can be evaluated independently, but also with other hormonal markers like testosterone 1 , 2 . It is hypothesized that as ovarian follicular reserve is affected by certain lifestyle risk factors, AMH may aid in understanding mechanisms for infertility. Several studies have evaluated the impact of maternal tobacco and caffeine use during pregnancy on age of menarche, fecundability, and live birth rates of the offspring with conflicting results 1 , 4 – 20 . In utero exposure to diethylstilbestrol (DES), which was historically used to reduce the risk of pregnancy complications and losses, has also been extensively studied on the outcomes of exposed offspring noting increased risk for gynecological cancers, malformation of the genitourinary tract, and infertility related to these poor developmental outcomes. However, in women with normal anatomy and a DES exposure, specific relation to fertility hormones have not been evaluated. Despite that ovarian follicles are established during the fetal development, a potential role of maternal lifestyle risk factors during pregnancy for daughter’s reproductive lifespan is less understood. To date, only a single study evaluated maternal risk factors that were not external exposures in relation to AMH, but among daughters in 14–16 years of age 1 . As such it is unknown whether these lifestyle factors may influence ovarian reserve among reproductive age women, particularly among those without known infertility where effects may be more subtle 1 . Therefore, the objective of this study is to evaluate the associations of reported in utero lifestyle exposures, including prenatal smoking, alcohol use, vitamin intake, caffeine use, DES use, and maternal employment status as a proxy for stress and socioeconomic status, on the serum AMH, a widely used marker of ovarian reserve, of adult daughters.