Dietary patterns and ovarian reserve among women attending a fertility clinic.

AFC was ascertained by reproductive endocrinology and infertility physicians from the MGH Fertility Center using transvaginal ultrasound performed on the 3rd day of an unstimulated menstrual cycle or on the 3rd day of a progesterone withdrawal bleed. No fertility medications were used in the cycle prior to the ultrasonography assessment. Of the 363 women, 14 (4%) had an AFC >30. Because women in the current study had a median AFC (interquartile range) of 13 (8 to 19), we truncated AFC at 30 to reduce the influence of extreme values.

At study entry, the 363 women included in our analysis had a median (IQR) age of 35.0 (32.0-38.0) years, with a BMI of 23.2 kg/m 2 (21.3-26.2 kg/m 2) , and were predominantly Caucasian (84%), never smokers (73%), with a college or higher degree (92%). The DP adherence scores were weakly correlated to each other, with the highest correlation observed between the FD and the PFD ( r = 0.38, p < 0.001); and the lowest between the MedDiet and the FD ( r = 0.31, p < 0.001). Women with the highest adherence to the MedDiet and the PFD were more physically active, and those in the highest quartile of adherence to the FD were more likely to have graduate degrees and to the PFD were also slightly older than those with the lowest adherence. Higher adherence to all three scores were related to higher frequency of multivitamin use. Diet characteristics were related to each DP as expected based on their nutrient contribution ( Table 1 , Table 2 ). None of the DP adherence scores were associated with AFC in age and energy adjusted models, before or after adjustment for other covariates ( Table 3 ). Women with the highest adherence to the MedDiet had an adjusted mean (95% CI) for AFC of 13.9 (13.0-14.9) vs. women with the lowest adherence (AFC [95%CI]:13.5 [95%CI: 12.6-14.4]), in the multivariable model (P-trend=0.34). A similar trend was observed with higher adherence to the FD and the PFD (AFC [95%CI]: 14.0 [95%CI: 13.2-14.9] and 12.5 [95%CI: 11.6-13.5], respectively) when compared to AFC of women with the lowest adherence (AFC [95%CI]: 13.5 [95%CI: 12.7-14.3] and 13.3 [95%CI: 12.5-14.2], respectively) ( Table 3 ). Further exclusion of alcohol intake from the MedDiet score did not change the findings and this pattern remained unrelated to AFC (data not shown). We additionally investigated if the associations between DP and AFC differed by demographic characteristics. FD was negatively related to AFC in lean women (<25 kg/m 2 ) (P-interaction and P-trend=0.03). PFD was inversely associated with AFC among women aged <35 years (P-interaction=0.002; P-trend <0.001) and in never smokers (P-interaction < 0.001; P-trend=0.02), and positively related to AFC among women who had ever smoked (P-trend=0.01). However, these findings changed when using different cutoffs (37 and 40 years, BMI of 23 and 30 kg/m 2 ) or different smoking groups (current vs. not current smokers) (Data not shown).

Starting in 2004, couples seeking infertility treatment at the Massachusetts General Hospital (MGH) Fertility Center were invited to participate in the EARTH (Environment and Reproductive Health) Study; an ongoing prospective cohort investigating the effects of environmental and nutritional determinants of fertility ( 23 ). Participants included in the present study were females between the ages of 18 and 45 years. All participants reported information on demographic and lifestyle characteristics, medical and reproductive history in a nurse-administered general health questionnaire. Clinical assessments included anthropometry and AFC measurements. Starting in 2007, a validated food frequency questionnaire (FFQ) was introduced to the study protocol to assess diet. Of the 709 AFC scans from women who completed diet questionnaires, we excluded scans done on Lupron (n = 25), women with a history of oophorectomy (n = 5), diagnosis of polycystic ovary syndrome (n = 43), difficult to visualize the ovary in the scan (n=6) and incomplete AFC data (n = 8), as well as women who completed the FFQ after the transvaginal ultrasound (n = 230), women with implausible total energy intake (3,500 kcal/day) (n=9) and repeated scans from women (n = 20), final analytical sample of 363 women. Ethical approval was obtained from the Human Studies Institutional Review Boards of the MGH, and the Harvard T. H. Chan School of Public Health and all participants provided written informed consent. Participants filled out an extensively validated ( 24 , 25 ) semi-quantitative FFQ. Women reported their usual intake of foods, beverage and supplements over the past year in 9 response categories per food item, ranging from never or less than once a month to > 6 times per day. Nutrient content of each item was obtained from the nutrient database of the US Department of Agriculture ( 26 ) with supplemental information from food manufacturers. We computed three adherence scores consistent with DP previously related to better fertility or infertility treatment outcome: the Mediterranean diet (MedDiet) ( 27 ), the Fertility diet (FD) ( 18 ) and the Profertility diet (PFD) ( 21 ) ( Table 1 ). The DP scores were calculated based on previously specified ( 18 , 21 , 27 ) combinations of food and nutrient intakes. Points were assigned for each food/nutrient intake criterion based on how closely intake matched these criteria, with higher scores indicating closer alignment between an individual’s diet and the DP. The overall adherence score for each DP was calculated as the sum of points across all individual intake criteria.

Among women presenting at a fertility clinic, we found that predefined dietary patterns that were related to ART success and infertility disorders were not associated with AFC, a marker of ovarian reserve. Due to the limited and heterogeneous current evidence it is important to evaluate this association in further studies, and in particular among women from the general population.

In this prospective study of women from a fertility clinic we did not observe a relation between three DP with previous relations to infertility outcomes, with AFC, as a marker of ovarian reserve. Although some interactions were found according to baseline characteristics, these findings were not consistent across categorizations which decreased our confidence that these identified interactions represented true biological effects. To our knowledge, this is the first study to examine the role of DP on ovarian reserve. The study of DPs can be advantageous since it can capture interactive or synergistic effects between nutrients that analyses focused on single nutrients may miss ( 30 ). In addition, this strategy can result in dietary recommendations that are easier to understand and adopt ( 31 ). In this work, DPs were predefined according to previously documented associations with risk of infertility and outcomes of infertility treatment. In addition, this a priori method does not necessarily represent overall dietary intake since scores usually focus on specific aspects of the diet ( 32 ). While we are not able to determine the underlying mechanisms of our results, the absence of associations of these DPs with AFC may suggest that the dietary factors related to risk of infertility and outcomes of infertility treatment may differ from those related to ovarian reserve. On the other hand, there may still be effects of a specific nutrients and foods on AFC that were diluted ( 30 , 33 ) by examining the relationship between DPs and AFC. Therefore, pattern analysis should be considered as a complementary strategy to evaluate the role of diet along with the traditional food or nutrient-based approaches ( 32 , 33 ). Previous studies have assessed the relation of specific nutrients, supplements, or different foods with AFC or anti-Müllerian hormone (AMH) levels, but the literature on these relations as a whole remains scarce and inconsistent. For example, in women from the EARTH Study, we previously reported an inverse association between intake of protein mainly from dairy sources and AFC ( 34 ) while another study that looked at carbohydrate intake among subjects in their late reproductive years ( 35 ), found a positive relation to AMH levels. However, these findings were inconsistent with a prospective study among women with a broader age range ( 36 ). Both serum levels ( 37 , 38 ) and supplementation with 25-hydroxy vitamin D ( 39 , 40 ) has been related to higher ovarian reserve in some studies, but other studies have failed to replicate these relations ( 10 , 41 – 44 ). Also, supplementation with omega-3 fatty acids did not appear to influence AMH levels ( 45 ), although a couple of observational studies reported inverse associations between omega-6 and total polyunsaturated fatty acids with AMH levels ( 35 , 36 ). Last, one study suggested a significant inverse association between coffee intake and AMH in healthy young women ( 46 ). While taken together, these findings suggest that dietary factors may be linked to ovarian reserve, the scarcity of the literature, the contradictory results, the lack of attempted replication for many dietary factors, and, importantly, the vast methodological heterogeneity across studies, including wide variation in study design, population settings and characteristics, the short duration of the interventions in clinical trials, and lack of covariate adjustment in many observational studies, limits the interpretation of these findings. Clearly, the scientific literature evaluating the potential role of diet on ovarian reserve is still in its infancy and additional research is much needed. It is important to consider the null findings in the context of the study limitations. Habitual dietary intake was assessed using a FFQ which, like all diet assessment methods relying on self-report, is not immune to measurement error despite extensive validation ( 24 , 25 ) and exclusion of subjects with implausible energy intakes ( 33 ). Since women are for all practical purposes blinded to their AFC at the time they completed the diet evaluations, the most likely effect of measurement error in this study is to bias the results towards the null. We therefore cannot exclude small effects of these diet patterns on AFC. Second, although the prospective design constitutes a strength, women reported intakes only during the past year; therefore, the impact of diet preceding our measurement may not have properly captured. Third, it is important to keep in mind that, being an observational study, our measures of DP reflects how closely each participant’s diet aligned with a pre-specified DP. While this methodology is useful and widely implemented ( 30 , 33 ), it is not the same as assigning individuals to follow a particular diet. Fourth, because we lacked data on AMH levels for all women in our study (of the 363 women, 82 (23%) had AMH determination), we were unable to examine the consistency of associations of diet with AFC and with AMH. Nevertheless, both biomarkers have shown good predictive values and are superior to FSH ( 16 ). Finally, because the study was conducted among women presenting to a fertility clinic, it limits the generalization of our results to women from the general population. However, the homogeneous sample may enhance the internal validity of this analysis.

Differences in participant characteristics across quartiles of DP adherence scores were assessed by Kruskal-Wallis test for continuous variables and χ 2 tests for categorical variables. Correlation coefficients between the DP scores were determined by the Spearman’s rho (r) test. We used multivariable generalized linear models with Poisson distribution and log link function to estimate AFC predicted marginal means and 95% confidence intervals (95% CIs) by quartiles of adherence scores. After analysis, the estimates were back-exponentiated to represent their original scale. We estimated tests for linear trend across DPs quartiles as the median value corresponding to each category as continuous variables ( 28 ). Confounding was assessed using substantive knowledge assisted by directed acyclic graphs ( 29 ) and descriptive statistics. Using these criteria, the final multivariable model included age, smoking status, education, body mass index (BMI), physical activity, and total energy intake. We further assessed effect modification by demographic characteristics (age, BMI, smoking status) through cross-product terms in the multivariable adjusted models. Analyses were performed using the statistical software SAS v. 9.4 (SAS Institute Inc, Cary, NC, USA).

Global fertility rates have declined by 50% since the mid-twentieth century ( 1 ). This downward trend is accompanied with an increasing maternal age at first child ( 2 – 4 ). Several socioeconomical determinants have influenced on this demographic shift, leading women and couples to delay childbearing till their 30s or more ( 5 – 7 ). The decision to postpone motherhood confronts with the natural age-related decline of the quantity and quality of ovarian follicles ( 7 , 8 ) resulting in a higher risk of reproductive issues and also yielding poor outcomes among women undergoing assisted reproductive technologies (ART) ( 6 , 7 , 9 ). Therefore, it is important to identify modifiable factors that could impact fertility by maintaining ovarian reserve and prolong the woman’s reproductive lifespan ( 10 – 13 ). Several works have addressed the effect of nutritional factors, specifically single foods and nutrients, on ovarian reserve biomarkers, most of them using follicle-stimulating hormone (FSH) levels ( 12 ). However, a recent systematic review concluded that the current evidence is insufficient and future research is needed, including elucidating the role of dietary patterns (DP) and the use of more precise ovarian reserve biomarkers ( 12 ). The study of DP in nutritional epidemiology has gained attention due to its ability to capture complex interactions between nutrients and they would more closely parallel real-world conditions ( 14 ). Moreover, ovarian reserve biomarkers may provide an indirect estimate of ovarian aging ( 15 ) and ovarian stimulation response in ART ( 16 , 17 ), and multiple studies have linked healthy DP to prevent female infertility as well as higher success among women undergoing ART ( 18 – 22 ). Increased adherence to the Fertility diet score showed lower risk of infertility due to problems with ovulation in healthy women ( 18 ). The Mediterranean diet has also been related to lower risk of infertility among healthy women ( 22 ). Moreover, stronger adherence to the Mediterranean diet ( 19 , 20 ) and the Profertility diet score ( 21 ) have been positively associated with ART outcomes such as clinical pregnancy ( 19 – 21 ) and live birth ( 20 , 21 ). One plausible biological mechanism underlying these associations is a beneficial effect of healthy DPs on ovarian reserve. Therefore, our aim was to investigate the associations between DP previously related to infertility disorders or ART outcomes with antral follicle count (AFC), a marker of ovarian reserve, among women seeking infertility treatment.