Current Evidence on Associations of Nutritional Factors with Ovarian Reserve and Timing of Menopause: A Systematic Review

Abstract

Ovarian aging is thought to be influenced by environmental factors, including nutrition. The aim of this study was to systematically review current evidence on the associations between nutritional factors, ovarian reserve, and age at menopause. PubMed and Scopus were structurally searched until May 2016. Original studies, with either observational or interventional designs, that examined the associations of nutritional factors (serum or dietary nutrients, food groups, and/or dietary patterns) with different ovarian reserve markers and/or timing of menopause were considered eligible. Twenty-six studies met the inclusion criteria: 17 studies on ovarian reserve markers and 9 studies on menopausal age. Significant diversity was observed in nutritional factors examined across studies. In the study of nutritional factors, associations of serum 25-hydroxyvitamin D [25(OH)D] concentration and intakes of soy or soy products with ovarian reserve have been the most investigated. For associations with menopausal age, intakes of total fat, fiber, and soy products have been mainly examined. Significant associations with ovarian reserve markers were found in 4 of 7 studies on serum 25(OH)D, 2 of 6 studies on soy or soy products, 1 of 2 studies on fiber intake, 1 study on serum zinc and copper concentrations, and 1 study on serum antioxidant concentrations. Studies on nutritional factors and menopausal age provided inconsistent findings, some of which suggested modest associations. Although there is some promising evidence on the influential role of nutrition in ovarian aging, a limited number of studies, heterogeneous in their design and study of nutritional factors, makes it difficult to draw definite conclusions. To better understand this issue, examination of associations of dietary intakes or dietary patterns with more precise markers of ovarian reserve, such as anti-mullerian hormone and antral follicle count, with age at menopause is needed. In addition, to explore whether nutritional factors alter the process of ovarian aging, an examination of changes in ovarian reserve markers should be considered.

Introduction

Ovarian reserve shows ovarian function by representing the quantity and quality of ovarian follicles (1). It can be used to predict the risk of female infertility, the success of infertility treatment (1, 2), reproductive life-span length, and menopausal age (3–5). Although it gradually decreases with increasing age (2), there is a high interindividual variability in the ovarian reserve among females of similar age, and the rate of decline in ovarian reserve varies widely among reproductive-aged women (2, 5, 6).

When the number of ovarian follicles is reduced to a critical threshold, menopause occurs (3, 4). Menopause affects both reproductive and nonreproductive health status, and therefore the age at which natural menopause occurs is important for women. Early menopause has been attributed to an increased risk of diabetes, cardiovascular diseases, neurologic diseases, and mortality (7–9). In contrast, the increased incidence of hormone-related cancers has been reported in women with late menopause (10, 11).

In addition to genetic and age influences on ovarian reserve, environmental factors are thought to modify the rate of decline in ovarian reserve (2, 12). More attention has focused on determining influential environmental factors to investigate how female follicular recruitment and its quality can be optimized. The impact of nutrition on processes that induce follicular atrophy, including oxidative stress and inflammation, supports the possibility that nutritional factors may influence ovarian reserve and menopausal age (13, 14). In addition, some nutritional factors have been shown to be associated with serum estrogen concentrations (15, 16) and the duration of the menstrual cycle (17). Therefore, nutrition is likely to affect the female reproductive life span by determining both the age at menarche (18) and the age at menopause (2).

Since the first cross-sectional study conducted in 1994, which showed that meat consumption was significantly related to odds of menopause (19), more studies have examined the associations of nutritional factors, including nutrients and food groups, with biological measures of ovarian reserve and timing of menopause. However, these studies provided inconsistent findings. Hence, the aim of this study was to systematically review the observational and interventional evidence available from human studies on the associations of nutritional factors with 1) ovarian reserve and 2) age at natural menopause. To our knowledge, this is the first study that comprehensively examines the current evidence on the potential effects of nutritional factors on ovarian aging in humans.

Methods

Search strategy

This study was conducted based on the standard protocol for systematic reviews and meta-analyses (20). Searches were performed in PubMed and Scopus databases combining specified reproductive and nutritional keywords from inception until May 2016. The reproductive-related keywords were as follows: ovarian reserve, premature ovarian failure, premature ovarian aging, poor ovarian, ovarian aging, menopause, menopaus*, follicle stimulating hormone, antimullerian hormone, FSH, AMH, and ovul*. Nutritional related-keywords were as follows: nutrition, diet*, carbohydrate, fat*, protein, nutrient, food, vitamin, and mineral. All keywords were defined by the collaboration of 2 investigators. The search was restricted to published studies written in English. The reference lists of published articles were also checked to find further relevant studies that had not been retrieved by using the search strategy.

Inclusion and exclusion criteria

Original studies that examined the associations of nutritional factors (serum and dietary intakes of nutrients, food groups, and/or dietary patterns) with 1) ovarian reserve markers [antral follicle count (AFC) , anti-mullerian hormone (AMH), follicle-stimulating hormone (FSH)] and/or 2) timing of menopause, with either an observational or interventional design, were considered eligible. Dietary intakes from both foods and supplemental sources were considered eligible for inclusion in this study. Duplicate articles, review articles, and in vitro and animal studies were excluded.

Study selection and data extraction

First, the title and abstract of the studies were screened for inclusion and exclusion criteria. Then, the full texts of eligible articles were retrieved and the information on study characteristics, participant characteristics, number of participants, study design, exposures or interventions, outcomes, covariates included in the full model (for observational studies), and results were extracted. The study selection and data extraction process were performed independently by 2 investigators.

Results

By using the search strategy, 1662 published articles were identified; 1467 articles were screened after excluding duplicates. Of 1467 articles, 11 were found to be eligible for this systematic review after assessment of their titles, abstracts, and full texts. By reference checking, 15 additional articles were identified (Figure 1).

FIGURE 1.

Flowchart of study selection.

Nutrition and ovarian reserve

Seventeen studies (10 observational and 7 interventional) investigated the association between nutritional factors and ovarian reserve. Nutritional factors examined in these studies included intakes of fat, fiber, and soy or soy products and serum concentrations of 25-hydroxyvitamin D [25(OH)D], zinc, copper, and antioxidants (α-tocopherol, ascorbic acid, lycopene, β-carotene, and γ-tocopherol) (Tables 1 and 2).

TABLE 1.

An overview of studies that examined the associations of nutritional factors with ovarian reserve in premenopausal women¹

	Observational studies		Interventional studies
Nutritional factors/measures of ovarian reserve	Number of studies	Association2	Number of studies	Association2
Dietary intakes
Soy products/FSH	1	NS	5	− (2)/NS (3)
Fat/FSH	2	NS	2	NS
Fat/AMH	—	—	1	NS
Fiber/FSH	2	− (1)/NS (1)	—	—
Serum measures
25(OH)D/FSH	3	− (2)/NS (1)	—	—
25(OH)D/AMH	3	+ (1)3/NS (2)	2	+ (1)/NS (1)4
25(OH)D/AFC	2	NS	—	—
Zinc/FSH	1	−	—	—
Copper/FSH	1	+	—	—
α-Tocopherol, ascorbic acid, lycopene, and β-carotene/FSH	1	−	—	—
γ-Tocopherol/FSH	1	+	—	—

¹AFC, antral follicle count; AMH, anti-mullerian hormone; FSH, follicle-stimulating hormone; 25(OH)D, 25-hydroxyvitamin D; +, positive association; −, negative association.

²Numbers of studies in each category are presented in parentheses.

³A significant positive correlation between serum 25(OH)D and AMH was observed only in women aged >40 y.

⁴Vitamin D supplementation could reduce AMH significantly in women with polycystic ovary syndrome.

TABLE 2.

Summary of studies that examined the associations of nutritional factors with ovarian reserve in premenopausal women¹

First author, year (ref)	Study design	Participants, n	Age,2 y	Intervention/exposure	Ovarian reserve markers	Duration	Adjustment (full model)	Findings
Soy and soy products
Cassidy et al., 1995 (21)	Randomized crossover	15	20–29	A: 60 g TVP/d; 45 mg conjugated isoflavones (n = 6)	Plasma FSH	1 menstrual cycle	—	Reduction in midcycle FSH significant only in group A (mean ± SD: 8.9 ± 4.4 vs. 15.5 ± 5.9 IU/mL; P < 0.01)
				B: 50 g fermented soybean paste (miso); 25 mg unconjugated isoflavones (n = 3)
				C: 60 g isoflavone-free soy product (n = 5)
				D: 28 g TVP; 23 mg conjugated isoflavones (n = 6)
Duncan et al., 1999 (22)	Randomized crossover	14	26.5	Usual diet + soy protein:	Plasma FSH	3 menstrual cycles + 9 d (13 wk)	—	FSH reduced significantly in midcycle in group B compared with A (least-squares mean ± SE: 3.92 ± 0.2 vs. 4.62 ± 0.2 IU/L; P = 0.04).
				A: 10 ± 1.1 mg isoflavones/d (control)
				B: 64 ± 9.2 mg isoflavones/d (low-isoflavones)
				C: 128 ± 16 mg isoflavones/d (high-isoflavones)				No significant changes were observed in group C compared with either A or B
Brown et al., 2002 (23)	Randomized crossover	14	27.6	A: high-saturated-fat Western diet (control diet)	Serum FSH	2 menstrual cycles	—	Nonsignificant reduction in FSH in group C compared with group A [geometric mean (95% CI): 0.2 IU/L (0.1, 0.3 IU/L) vs. 0.4 IU/L (0.2, 0.6 IU/L); P = 0.076]
				B: control + soy protein (soy diet)
				C: control + PUFAs (PUFA diet)				No significant differences between groups B and A [geometric means (95% CI): 0.3 IU/L (0.1, 0.4 IU/L) vs. 0.4 IU/L (0.2, 0.6 IU/L)]
				Control diet: PUFA:SFA ratio of 1:4, 15% SFAs and 7% PUFAs
				PUFA diet: PUFA:SFA ratio of 2:1, 7% SFAs and 15% PUFAs
				Soy diet: 31 g soy protein/d (40 mg isoflavones)
Maskarinec et al., 2002 (24)	Randomized, parallel, double-blind trial	34	42.4	100 mg isoflavones or placebo	Serum FSH	1 y	—	No significant changes were observed in the intervention group compared with placebo [mean difference in change (95% CI): −0.82 (−2.53, 0.897); P = 0.34]
Zittermann et al., 2004 (25)	Randomized crossover	14	24.0	A: 5 soy cookies (52 mg isoflavones)	Serum FSH	1 menstrual cycle 1 wk before menstruation (4 wk)	—	FSH concentrations were not significantly different between groups A and B after intervention (mean ± SD: 5.11 ± 1.87 vs. 5.59 ± 1.16 mIU/mL
				B: 5 wheat cookies
Tsuji et al., 2012 (16)	Cross-sectional	393	39.8	Usual intake of isoflavones	Plasma FSH	—	Age, BMI, phase of the menstrual cycle, number of births, age at first birth, breastfeeding, smoking	Geometric means of FSH were not significantly different across quartile intakes of soy isoflavones (P-trend = 0.54)
Serum 25(OH)D
Merhi et al., 2012 (26)	Cross-sectional	388 premenopausal women with regular menses	37.8	Serum 25(OH)D	AMH	—	HIV status, BMI, race, smoking, illicit drug use, glucose, insulin concentratons, GFR, geographic site	Vitamin D and AMH positively correlated only in women aged ≥40 y (β = +0.011, SE = 0.005; P = 0.028)
Dennis et al., 2012 (27)	Intervention	33 premenopausal women	19–39	1000 IU vitamin D2 (n = 7), 1000 IU vitamin D3 (n = 16), or placebo (n = 10) daily in autumn and winter	AMH	6 mo	Age, initial concentrations of 25(OH)D and AMH	Significant seasonal variations in vitamin D and AMH in both placebo and vitamin D2 groups
								No reduction in vitamin D and AMH during the winter in vitamin D3 group
								Positive correlation between vitamin D changes and AMH changes in all 33 women (r = 0.45, P = 0.014) after adjusting for initial AMH and vitamin D concentrations and age
Kebapcilar et al., 2013 (28)	Case-control	35 POI and 28 women with regular menses	37.2	Serum 25(OH)D	FSH	—	—	Significantly lower serum vitamin D in POI women vs. controls (9.5 ± 4.0 vs. 18.5 ± 7.5 ng/mL; P < 0.001)
								FSH was negatively correlated with serum vitamin D (r = −0.55, P < 0.001) in all 63 participants
Irani et al., 2014 (29)	Intervention	51 premenopausal women (16 PCOS and 35 non-PCOS)	29.9	50,000 IU vitamin D3/wk	AMH	8 wk	—	In PCOS women, AMH significantly reduced to normal concentrations after supplementation (P = 0.003)
								In non-PCOS women, no significant change was observed
Chang et al., 2014 (30)	Cross-sectional	73 healthy nonobese women	33.8	Serum 25(OH)D	FSH, AMH, AFC	—	—	25(OH)D did not correlate with FSH (r = −0.101, P = 0.399), AMH (r = 0.001, P = 0.991), or AFC (r = 0.066, P = 0.606)
Pearce et al., 2015 (31)	Cross-sectional	340 women of reproductive age (58 PCOS and 248 non-PCOS)	32.3	Serum 25(OH)D	AMH, AFC	—	Age, BMI, skin color, menstrual cycle length, occupation	No seasonal variations in AMH (36.9 ± 3.3 vs. 38.5 ± 2.7 pmol/L) and AFC (24.3 ± 2 vs. 22.2 ± 1.6) in summer vs. winter despite the seasonal variation in serum 25(OH)D (71.1 ± 2.7 vs. 49.3 ± 1.7 nmol/L; 30% variation; P < 0.001)
								25(OH)D did not correlate with AMH (r2 = 0.04, P = 0.4) and AFC (r2 = 0.03, P = 0.85)
Jukic et al., 2015 (32)	Cross-sectional	527 premenopausal women	42	Serum 25(OH)D	Urinary FSH	—	Age, race, BMI, smoking, menstrual cycle day of urine collection, season of blood draw	14% reduction in urinary FSH/10-ng/mL increase in serum vitamin D (95% CI: −23, −5; P = 0.003)
								The association was stronger in women aged <40 y vs. women ≥40 y [−19%; 95% CI: −33%, −6% (P = 0.06) vs.−12%; 95% CI: −24%, −0.1% (P = 0.05) per 10-ng/mL increase in 25(OH)D]
Dietary fat
Brown et al., 2002 (23)	Randomized crossover	14	27.6	A: high-saturated-fat Western diet (control diet)	Serum FSH	2 menstrual cycles	—	Nonsignificant reduction in FSH in group C compared with group A [geometric means (95% CI): 0.2 (0.1, 0.3) vs. 0.4 (0.2, 0.6); P = 0.076]
				B: control diet + soy protein (soy diet)
				C: control diet + PUFA (PUFA diet)
				Control diet: PUFA:SFA ratio of 1:4; 15% SFAs and 7% PUFAs
				PUFA diet: PUFA:SFA ratio of 2:1; 7% SFAs and 15% PUFAs
				Soy diet: 31 g soy protein/d (40 mg isoflavones)
Tsuji et al., 2012 (16)	Cross-sectional	393	39.8	Total fat, SFAs, PUFAs, MUFAs, long chain ω-3 PUFAs	Plasma FSH	—	Age, BMI, phase of the menstrual cycle, number of births, age at first birth, breastfeeding, smoking	No significant associations; geometric means of FSH were not significantly different across quartile intakes of total fat (P-trend = 0.13), SFAs (P-trend = 0.08), MUFAs (P-trend = 0.21), PUFAs (P-trend = 0.64), and long-chain ω-3 PUFAs (P-trend = 0.95)
Mumford et al., 2016 (33)	Prospective	259 women with regular menses aged 18–44 y	27.3	Average cycle-specific fat intakes including total fat, SFAs, PUFAs, MUFAs,trans-fat, ω-3, ω-6, EPA, DHA	FSH	Followed ≤2 menstrual cycles	Age, race, BMI, energy (energy substitution)	No significant association between fat intakes and FSH
							Age, race, BMI, energy, intakes of the remaining types of fat and protein (carbohydrate substitution)
Al-Safi et al., 2016 (34)	Intervention	15 obese and 12 normal-weight women aged 28–34 y with regular menses	31.6	Long-chain ω-3 PUFA supplementation (including 1860 mg EPA, 1500 mg DHA) for 2 menstrual cycles	FSH, AMH	1 mo	—	Mean AMH did not change significantly in either normal-weight (4.7 vs. 5.3 ng/mL; P = 0.85) or obese (3.9 vs. 4.2 ng/mL; P = 0.43) participants compared with baseline
								Mean FSH did not change significantly in either normal-weight (4.0 vs. 4.8 IU/L; P = 0.51) or overweight (4.2 vs. 4.4 IU/L) participants compared with baseline
								Decreases in FSH concentration and its peak by 17% and 20%, respectively, after administration of exogenous GnRH in normal-weight participants
Dietary fiber
Gaskins et al., 2009 (35)	Prospective	250 women aged 18–44 y	27.5	Fiber intake (total fiber, soluble fiber, insoluble fiber, vegetable fiber, grain fiber, fruit fiber)	FSH around ovulation	Followed ≤2 menstrual cycles	Energy intake, race, age, and vitamin E	FSH decreased by 0.034 nIU/mL per each 5-g increase in total fiber intake (P = 0.05)
Tsuji et al., 2012 (16)	Cross-sectional	393	39.8	Fiber	Plasma FSH	—	Age, BMI, phase of the menstrual cycle, number of births, age at first birth, breastfeeding, smoking	Geometric means of FSH were not significantly different according to quartile of dietary fiber intakes (P-trend = 0.88)
Serum nutritional markers
Kebapcilar et al., 2013 (28)	Case-control	35 POI and 28 women with regular menses	37.2	Serum zinc, copper, copper:zinc ratio	FSH	—	—	Significantly lower serum zinc in POI vs. control women (11.9 ± 1.4 vs. 13.8 ± 1.9 μg/dL; P < 0.001)
								Significantly higher serum copper in POI women (152.8 ± 22.3 vs. 112.6 ± 17.3 μg/dL; P < 0.001)
								FSH was negatively correlated with serum zinc (r = −0.38, P = 0.002) and positively correlated with serum copper (r = 0.66, P < 0.001) and copper:zinc ratio (r = 0.65, P < 0.001) in all 63 participants
Mumford et al., 2016 (36)	Prospective	259 women with regular menses aged 18–44 y	27.3	Serum antioxidants (α-tocopherol, γ-tocopherol, retinol, lutein, lycopene, β-carotene, ascorbic acid)	FSH around ovulation	Followed ≤2 menstrual cycles	Age, race, BMI, parity, time-varying sleep, pain medication use, total energy intake, serum cholesterol, F2-isoprostanes, other antioxidants, and concurrent hormones with the use of inverse probability weights	Inverse significant associations were observed between FSH and α-tocopherol, ascorbic acid, lycopene, and β-carotene
								Positive significant association was observed between FSH and γ-tocopherol

¹AFC, antral follicle count; AMH, anti-mullerian hormone; FSH, follicle-stimulating hormone; GFR, glomerular filtration rate; GnRH, gonadotropin-releasing hormone; mIU, milli-international units; nIU, nano-international units; PCOS, polycystic ovary syndrome; POI, primary ovarian insufficiency; ref, reference; TVP, textured vegetable protein; 25(OH)D, 25-hydroxyvitamin D.

²Age values are presented as means, medians, or ranges.

Soy products.

Five clinical trials (21–25), of which 4 were crossover trials (21–23, 25), examined the effects of soy products on serum and plasma FSH. Sample sizes ranged from 14 to 34 participants. In these studies, the effects of soy were examined as soy foods (21, 25), an isoflavone-free soy product (21), soy protein added to diets (22, 23), and soy isoflavones (24). Except for one study (24), the studies examined short-term effects of supplementation with durations <3 menstrual cycles. The consumption of 60 g textured vegetable protein/d, containing 45 mg isoflavones, for 1 cycle reduced midcycle FSH concentrations significantly, whereas no significant change was observed when 60 g of an isoflavone-free soy product was consumed (21). A low-isoflavone diet, containing 64 ± 9.2 mg isoflavones, reduced serum FSH concentration by 28% in women who consumed the diet for 3 menstrual cycles compared with women who consumed a control diet consisting of 10 ± 1.1 mg isoflavones/d (P = 0.009). However, FSH concentrations did not change significantly in women who consumed a high-isoflavone diet containing 128 ± 16 mg isoflavones/d compared with women who consumed either the control or the low-isoflavone diet (22). Other studies reported no significant effects on FSH concentrations after supplementation (23–25). In a cross-sectional study, associations of usual dietary intakes of soy isoflavones during 1 y, assessed by an FFQ, with serum FSH concentrations were examined in 393 Japanese premenopausal women. The findings of this study showed no significant association between the mean daily intake of 40.3 ± 28.6 mg soy isoflavones and FSH concentrations (16).

Serum 25(OH)D.

Five observational (26, 28, 30–32) and 2 interventional (27, 29) studies investigated the associations between serum 25(OH)D concentrations and ovarian reserve markers. Ovarian reserve markers studied were serum AMH (26, 27, 29), serum FSH (28), urinary FSH (32), a combination of AMH and AFC (31), and a combination of FSH, AMH, and AFC (30). The sample sizes in these studies were between 33 (27) and 527 (32) participants. In all of the studies, participants were reproductive-aged women with a mean or median age of <40 y, except for one study, in which the mean age of participants was 42 y (32). Women aged >40 y were included in 2 studies. Most of the above study participants were vitamin D deficient because their serum vitamin D concentration was <30 ng/mL. The mean BMI (in kg/m²) of participants was >25, with the exception of 2 studies that were conducted in nonobese women (27, 30). Female participants in 3 studies had regular menses (26, 28, 30), and in the other studies the menstrual cycle status of participants was not specified. In 3 studies, women with low AMH concentrations were excluded, although the threshold definition for AMH varied across studies (27, 29, 31). Two studies did not mention any threshold for ovarian reserve for participants to be eligible (28, 30); and in 1 study, 5% of participants had AMH concentrations below the threshold (26). Two studies assessed serum AMH and 25(OH)D concentrations in different seasons (27, 31); one of them reported a significant reduction in AMH concentrations during winter that could have been prevented by vitamin D₃ supplementation (27), whereas the other found no seasonal variation in AMH concentrations, despite a high decline in serum 25(OH)D during the winter (31).

No significant correlations between serum 25(OH)D and any of the ovarian reserve markers (FSH, AFC, and AMH) were observed in 2 cross-sectional studies (30, 31), whereas in another cross-sectional study a weak positive correlation between serum vitamin D and AMH concentrations was found only in women aged ≥40 y (26). An inverse correlation was observed between serum vitamin D and serum FSH concentrations in a case-control study of primary ovarian insufficiency (POI) (28). In addition, an inverse association between serum vitamin D and urinary FSH concentrations was reported in a cross-sectional study, with the suggestion that the association was stronger in women aged <40 y (32). Daily supplementation of 1000 IU vitamin D₃ for 6 mo during autumn and winter in 16 women with an age of 19–39 y prevented the decline in serum AMH concentrations during these seasons (27); however, weekly supplementation of 50,000 IU vitamin D₃ for 8 wk did not change AMH concentrations in 35 women without polycystic ovary syndrome (PCOS) but reduced it to normal concentrations in 16 women with PCOS (29).

Dietary fats.

Dietary intakes of total fat or fat subtypes, such as SFAs, MUFAs, and PUFAs, either as usual intakes during the past year (grams per day) (16) or as mean intakes per menstrual cycle (% of energy) (33) were not associated with concentrations of FSH in 2 observational studies conducted in Japanese and American women with normal weight. Two intervention studies with small sample sizes examined the effect of fat intake on FSH concentrations (23, 34). The replacement of 8% of energy from SFAs with PUFAs for 2 menstrual cycles reduced FSH concentrations in a crossover clinical trial in 14 premenopausal women, although the change did not reach significance (P = 0.076) (23). Omega-3 FA supplementation at a dose of 4 g/d, containing 465 mg EPA and 375 mg DHA, for 1 mo in 15 obese and 12 normal-weight women did not change basal AMH concentrations, whereas FSH concentrations showed a marginally significant decrease in normal-weight women (P = 0.06). In this study, the peak FSH was reduced after exogenous gonadotropin-releasing hormone stimulation in normal-weight women (34). All of the above studies were conducted in women with regular menses.

Dietary fiber.

Two observational studies examined total dietary fiber intake (16, 35). One reported a significant association between total fiber intake and FSH concentrations (35). Fiber subtypes, including soluble and insoluble, vegetable, fruit, and grain fiber, were also investigated in one of these studies, but no significant association was observed (35). Both of the studies were conducted in women with normal weight and regular menses. FSH concentrations decreased by 0.034 nano-international units per milliliter (nIU/mL) (95% CI: −0.068, 0.005 nIU/mL) per each 5-g increase in total fiber intake per cycle in young American women, as followed for ≤2 cycles in the BioCycle study. However, the association was marginally significant (P = 0.05) (35). Usual total fiber intake, assessed by a 169-item FFQ, was not associated with plasma FSH concentrations in 393 Japanese women (16). Mean concentrations of FSH and fiber intake in these Japanese women were 6.5 ± 7.7 milli-international units (mIU)/mL and 16.2 ± 6.7 g/d, respectively, which was comparable to a mean concentration of 7.8 ± 4.8 mIU FSH/mL and 13.6 ± 6 g mean total fiber intake/d of American female participants in the BioCycle study (35).

Serum nutritional markers and ovarian reserve.

Serum zinc concentrations were significantly lower in women with POI than in controls, whereas serum copper concentrations were higher in women with POI in a case-control study. In combined POI and control participants (n = 63), FSH concentrations were inversely correlated with the concentration of serum zinc (r = −0.38; P = 0.002) and positively correlated with serum copper concentration (r = 0.66, P < 0.01) (28). From the serum antioxidants examined in the BioCycle study, serum α-tocopherol and ascorbic acid were inversely associated with FSH concentrations, as measured at the same cycle in 259 healthy young women (36).

Nutrition and menopause

Eight observational studies, of which 2 were cross-sectional (19, 37) and 6 were prospective (13, 14, 38–41), and 1 interventional study (42) investigated the association between nutritional factors and age at menopause (Tables 3 and 4). The primary outcome of these studies was natural menopause; therefore, women with surgical menopause due to hysterectomy or ovarectomy were either excluded or censored. Menopause was defined as the cessation of menses for ≥6 consecutive months in 2 studies (38, 42) and for ≥12 mo in 4 studies (13, 14, 39, 40), 1 study used both definitions (19), and 2 studies did not specify how menopausal status was defined (37, 41). Participants who used oral contraceptive pills (OCPs) or hormone replacement therapy (HRT) were either excluded or censored, except for 3 studies that considered the use of OCPs (14, 41) or both OCPs and HRT as covariates (13). Four studies provided no information about the use of OCPs (38–40) or HRT (41). The use of OCPs and/or HRT did not change the results in 4 studies that performed sensitivity analyses with respect to these variables (37, 39, 41, 42).

TABLE 3.

Overview of studies that examined the associations of nutritional factors with occurrence of natural menopause¹

	Cross-sectional studies		Prospective studies		Interventional studies
Dietary intakes	Number of studies	Association2	Number of studies	Association2	Number of studies	Association
Nutrients
Total fat	1	−	4	Nonlinear − (1)/NS (3)	—	—
Animal fat	—	—	1	Nonlinear +	—	—
Dairy fat	—	—	1	−	—	—
MUFAs/PUFAs	—	—	1	+	—	—
Cholesterol	1	−	1	Nonlinear −	—	—
Carbohydrate	1	NS	3	− (1)/NS (2)	—	—
Low-fat, high-carbohydrate diet	—	—		—	1	NS
Protein	1	NS	3	− (1)/NS (2)	—	—
Vegetable protein	—	—	1	Nonlinear +	—	—
Dairy protein	—	—	1	−	—	—
Fiber	1	NS	4	Nonlinear + (1)/NS (3)	—	—
Calcium	1	+	2	NS	—	—
Vitamins A, C, and E	1	NS	1	NS	—	—
Vitamin D	1	NS	1	NS	—	—
Food groups
Meat	1	−	3	− (1)/NS (2)	—	—
Dairy products	—	—	2	− (1)/NS (1)	—	—
Vegetables	—	—	3	Nonlinear + (1)/NS (2)	—	—
Green/yellow vegetables	—	—	1	—	—	—
Fruit	—	—	2	− (1)/NS (1)	—	—
Soy products	1	+	4	NS	—	—
Beverages
Alcohol	2	− (1)/NS (1)	3	− (1)/NS (2)	—	—
Coffee	1	−	—	—	—	—
Tea	—	—	1	NS	—	—

¹+, positive association; −, negative association.

²Numbers of studies in each category are presented in parentheses.

TABLE 4.

Summary of studies that examined the associations of dietary intakes with occurrence of natural menopause¹

First author, year (ref)	Study design	Participants, n	Age,2 y	Intervention/exposure	Duration, y	Adjustment (full model)	Findings
Torgerson et al., 1994 (19)	Cross-sectional	2074 (1227 premenopause, 258 postmenopause)	45–49	Meat, alcohol	—	Stepwise analyses including age, smoking, age of maternal menopause, parity, social class, meat consumption, and alcohol	Meat intake >1 time/wk was associated with 73% lower risk of menopause compared with individuals who never consumed meat (P = 0.0098)
							Everyday alcohol consumption was associated with 50% lower odds of menopause compared with never
Torgerson et al., 1997 (38)	Prospective	1227	47–51	Meat, alcohol	2	Not specified	No significant association was observed between meat intake and onset of menopause
							Alcohol consumption reduced risk of menopause (OR: 0.91; 95% CI: 0.84, 0.98; P = 0.009)
Nagata et al., 1998 (37)	Cross-sectional	3704	45–55	Total energy; macronutrients; cholesterol; calcium; crude fiber; vitamins A, C, D, and E; carotene; soy product; retinol; coffee; alcohol	—	Age and total energy	Risk of menopause reduced from T1 to T3 of intakes of fat (P-trend = 0.01), cholesterol (P-trend = 0.02), and coffee (P-trend = 0.03)
							Risk of menopause increased from T1 to T3 for intakes of calcium (P-trend = 0.03) and soy products (P-trend = 0.002)
Nagata et al., 2000 (39)	Prospective	1130	42.7	Energy, macronutrients, animal protein/fat, vegetable protein/fat, fat from fish, cholesterol, calcium, crude fiber, vitamin A, retinol, vitamin C, vitamin E, green and yellow vegetables, other vegetables, soy products	6	Age, BMI, smoking, age at menarche	Risk of menopause was 51% and 41% higher in T2 of vegetable protein (HR: 1.51; 95% CI: 1.13, 2.02) and animal fat (HR: 1.41; 95% CI: 1.06, 1.87) vs. T1
							Risk of menopause was 26% lower in T2 of cholesterol (HR: 0.74; 95% CI: 0.54, 0.99) vs. T1
							Green and yellow vegetables inversely reduced the risk of menopause (P-trend = 0.02)
Nagel et al., 2005 (13)	Prospective	5568	35–65	Macronutrients, alcohol, meat, dairy products, fish, vegetables, fruit, cereal products, fiber, soy products, sweets, added animal fat, added vegetable fat	5.8	Age, education, OC use, HRT use, parity, BMI, time of breastfeeding, age at first full-term pregnancy, smoking habit, alcohol intake, leisure-time physical activity, total energy	Higher intake of carbohydrate increased risk of earlier age at menopause but the trend become nonsignificant (P-trend = 0.119)
							Q3 of vegetable intake vs. Q1 was associated with earlier menopause (HR: 1.32, 95% CI: 1.09, 1.60)
							High intakes of cereal, fiber and soy products were associated with earlier menopause, but the trend become nonsignificant (P-trend = 0.06, 0.16, and 0.06, respectively)
							Q3 of total fat intake vs. Q1 was associated with later menopause (HR: 0.78; 95% CI: 0.59, 0.79)
							Higher intake of meat was associated with later menopause (P-trend = 0.046)
Martin et al., 2006 (42)	Randomized clinical trial	2611	44.7	LFHC: CHO%:fat% was 65%:15% of energy with no changes in total energy or protein intakes	7	—	Dietary intervention did not influence timing of menopause
							An interaction between diet and baseline BMI was observed: the LFHC was associated with earlier menopause in women with low BMI but higher BMI was associated with later menopause in women in the LFCH group
Dorjgochoo et al., 2008 (14)	Prospective	33,054	60.2	Energy, macronutrients, vegetables, fruit, red meat, saturated fat, total soy, total fiber, tea, alcohol	Not specified	Age, education, occupation, age at menarche, number of live births, OC use, weight gain between age 20 and 50, smoking, physical activity pattern in adolescence and adulthood, energy intake	Total energy intake was significantly associated with later menopause and longer reproductive span (P-trend < 0.01)
							Higher fruit intakes were associated with later menopause (P-trend = 0.04) and longer reproductive years (P-trend = 0.03)
							Higher protein intake was associated with later menopause (P-trend = 0.02) and longer reproductive years (P-trend < 0.01)
							Higher carbohydrate intake was associated with later menopause (P-trend = 0.04) and longer reproductive years (P-trend = 0.06)
Nagata et al., 2012 (40)	Prospective	3115	43.0	Energy, fat, SFAs, PUFAs, MUFAs, long ω-3 FAs, fiber, isoflavones, alcohol	10	Age, BMI, smoking, parity, years of education, age at menarche, lifelong irregular menstrual cycle, and physical activity	Higher intake of PUFAs was associated with earlier onset of menopause (HR for Q4 vs. Q1: 1.15; 95% CI: 1.01, 1.31; P-trend = 0.002)
							Higher intake of MUFAs was associated with earlier onset of menopause (HR for Q4 vs. Q1: 1.12; 95% CI: 0.98, 1.28; P-trend = 0.05)
Carwile et al., 2013 (41)	Prospective	46,059	48.3	High-fat dairy, total low-fat dairy, skim milk, whole milk, dairy fat, dairy protein, calcium, vitamin D, lactose	20	Energy, age at menarche, age at first birth, parity, moderate to vigorous activity, OC use, BMI, smoking, marital status, red meat consumption, egg consumption	Only in women aged <51 y:
							Individuals who consumed >3 servings low-fat dairy/d were 14% less likely to report natural menopause in the next month relative to those with intakes of 0.1–1 servings/d (HR: 0.86; 95% CI: 0.77, 0.96; P-trend < 0.0001)
							Individuals with skimmed milk intake >3 servings/d were 7% less likely to report natural menopause in the next month relative to those with intakes of 0.1–1 servings/d (HR: 0.93; 95% CI: 0.89, 0.97; P-trend < 0.0001)
							Relative to those in the lowest quintile, those with the highest quintiles of dairy fat (HR: 0.94; 95% CI: 0.89, 0.99; P-trend = 0.02), dairy protein (HR: 0.91; 95% CI: 0.86, 0.95; P-trend < 0.0001), and lactose (HR: 0.92; 95% CI: 0.87, 0.97; P-trend = 0.0004) were less likely to reach natural menopause in the next month

¹CHO, carbohydrate; HRT, hormone replacement therapy; LFHC, low-fat, high-carbohydrate; OC, oral contraceptive; Q, quartile; T, tertile.

²Age values are presented as means, medians, or ranges.

Macronutrients.

Five observational studies investigated the association between intake of total fat and risk of menopause (13, 14, 37, 39, 40). One showed a gradual, significant reduction in the risk of menopause from study participants in the first tertile of total fat intake to the third tertile of intake (P-trend = 0.01) (37), another study reported a 22% (95% CI: 3%, 35%) lower risk of menopause for participants in quartile 3 of total fat intake than in quartile 1 after 5.8 y of follow-up (13), and other studies found no significant association (14, 39, 40).

One study suggested a higher risk of menopause for study participants in tertile 2 for animal fat intake relative to tertile 1 (41%; 95% CI: 6%, 87%; P-trend = 0.41) (39), but another study indicated that higher intakes of dairy fat contributed to a slightly older age at menopause in women <51 y (41). According to one study, higher intakes of MUFAs (P-trend = 0.05) and PUFAs (P-trend = 0.002) predicted a modestly earlier menopause after a 10-y follow-up (40).

Studies investigating intakes of carbohydrate and protein found no significant associations with the age at menopause (13, 37, 39), except for one study that reported that high intakes of carbohydrate and protein were weakly associated with later onset of menopause (14). The risk of menopause was 51% higher in study participants in the middle tertile of vegetable protein intake compared with the first tertile (P-trend = 0.81) according to the only study that examined the association (39). A 20-y prospective study in 46,059 premenopausal women showed that the probability of reaching menopause in the following month was 9% less (95% CI: 5%, 14%) in women aged <51 y who were in the highest quintile of dairy protein intake compared with those in the lowest quintile (P-trend < 0.0001) (41).

A randomized clinical trial examining the long-term effects of dietary macronutrients on the timing of menopause showed that adherence to a low-fat (15% of energy), high-carbohydrate diet (65% of energy) for 7 y with no changes in energy and protein intakes did not affect the time of menopause in 2611 women aged 30–65 y (42).

Cholesterol.

A higher intake of cholesterol was associated with later menopause in a cross-sectional study (37), even when the population was prospectively followed. A 26% (95% CI: 1%, 46%) lower risk of menopause was observed in study participants in the middle tertile for cholesterol intake relative to the first, although no significant trend was observed (P-trend = 0.32) (39).

Fiber.

The risk of reaching menopause was 30% (95% CI: 3%, 35%) higher in women in the highest quartile of fiber intake relative to those in the lowest quartile in a prospective study, with no significant trend (13). Findings from 4 other observational studies consistently showed that fiber intake was not related to the time of menopause (14, 37, 39, 40).

Calcium.

The risk of menopause was increased in study participants in tertile 1 of calcium intake compared with tertile 3 in a cross-sectional study (P-trend = 0.03) in Japanese women (37), but calcium did not predict the timing of menopause in this population that was followed prospectively for 6 y (39) or in the prospective study cohort of the Nurses’ Health Study (NHS) (41).

Vitamins A, C, E, and D.

There were no significant findings on the associations of intakes of vitamins A, C, and E and time of menopause on the basis of the limited studies available on intakes of these nutrients (37, 39). No significant association was observed between vitamin D intake and odds of menopause in a cross-sectional study (37). In the NHS, higher intakes of vitamin D predicted an increased age at menopause in women <51 y after adjusting for energy intake (P-trend = 0.007). However, this association was not significant after the addition of other covariates into the model (41).

Meat.

One cross-sectional (19) and 1 prospective (13) study suggested that higher intakes of meat were significantly related to an increased age at menopause, although 2 other studies found no significant association (14, 38). In only one study (14) was the type of meat specified as red meat.

Dairy products.

Dairy intake and menopausal age were investigated in 2 studies (13, 41), one of which examined dairy products in general with no clear definition for this group (13), while the other examined low-fat dairy, high-fat dairy, skimmed milk, and whole milk separately (41). Dairy products were not related to menopausal age in the first study (13), whereas the second showed that higher intakes of low-fat dairy (P < 0.0001) and skimmed milk (P < 0.0001) delayed menopause in women aged <51 y (41).

Vegetables and fruit.

Vegetable intake was not associated with age at menopause in 2 prospective studies (14, 39), one of which suggested that only green and yellow vegetables were associated with later menopause (P-trend = 0.02) (39). However, the age at menopause was decreased in the participants in quartile 3 of vegetable intakes compared with those in the first quartile after 5.8 y of follow-up in the Nagel et al. (13) study (P-trend = 0.43), which was the only other study on this topic. Fruit intake was not associated with age at menopause in a prospective study (13) and was only weakly associated with later menopause in another prospective study (14).

Soy products.

The odds of menopause were 39% higher (95% CI: 12%, 71%) in Japanese women with the highest intake of soy products (tertile 3) compared with those with the lowest intake (P-trend = 0.002) (37). However, when the population was followed prospectively, no significant association was observed (39). Soy (13, 14) and soy isoflavones (40) were not related to age at menopause in 3 other studies.

Alcohol, coffee, and tea.

Both cross-sectional and prospective studies in women in the United Kingdom suggest a later menopause with higher alcohol intake (19, 38). However, 3 other observational studies found no significant association for alcohol intake and time of menopause (13, 14, 37).

In a cross-sectional study, which was the only study reporting coffee intake, the odds of menopause were 32% lower (95% CI: 12%, 48%) in women in the highest tertile of coffee intake compared with those in the lowest tertile (P-trend = 0.03) (37). One study that examined tea intakes reported no significant associations (14).

Discussion

Current evidence on associations between nutritional factors and ovarian aging with the use of markers of ovarian reserve and age at menopause was systematically reviewed in this study. Some studies suggested significant associations between nutritional factors and ovarian reserve and timing of menopause. However, drawing definite conclusions at this time is difficult due to the limited number of studies, in consideration of the diverse single dietary exposures, and inconsistent findings.

In addition to the limited studies on the associations between nutritional factors and ovarian reserve, those available are hindered by the use of inappropriate markers of ovarian reserve and small sample sizes. Compared with other ovarian reserve markers, FSH was the most widely used in the reviewed studies. However, variability in FSH concentrations during the menstrual cycle and its changes due to nonrelated ovarian conditions make it an inaccurate and insensitive marker to measure ovarian reserve. Currently, AMH is suggested to be a better marker to predict ovarian aging and age at menopause (1). Except for the effect of ω-3 FA supplementation on concentrations of AMH examined in an interventional study (34), only the association of AMH with serum 25(OH)D has been investigated. Therefore, to understand the role of dietary intakes on ovarian reserve, further studies measuring AMH or AFC are needed. Associations between nutritional factors and ovarian reserve were mostly examined by cross-sectional studies, and interventional studies were conducted on a small scale. Of nutritional factors, intake of soy and serum 25(OH)D concentration has been predominantly evaluated.

Soy and soy products can potentially affect reproductive health and fertility because they are a rich source of phytoestrogens, which can interfere with the effects of estrogen by altering its concentrations and bioavailability and by acting as its agonists or antagonists (43). However, the potential effects of soy or isoflavone consumption on serum FSH concentrations seem rather unlikely on the basis of current data from interventional studies. Low statistical power due to small sample sizes and a high dropout rate are the main limitations of these studies. In addition, except for one study (24), other studies investigated the short-term effects of soy. A meta-analysis including 7 interventional studies suggested that soy or isoflavone consumption reduced FSH concentrations, but the modest reduction in FSH is unlikely to be clinically significant (44). In the only observational study of soy isoflavones, no significant association was found between usual intakes of soy isoflavones and serum FSH concentrations in a population of Japanese women (16). Most previous studies examined the effects of soy isoflavones, whereas soy foods or other soy substances may not have the same effects.

In vitro studies showed that vitamin D increased the expression of the enzymes involved in biosynthesis of steroids (45, 46). Vitamin D response element also has been identified in the promoter of the AMH gene in human prostate cancer cells (47) and granulosa cells of hens (48), suggesting that vitamin D acts as a regulatory factor of AMH production (26, 27). Concentrations of serum 25(OH)D and AMH were suggested to be positively correlated in a previous systematic review (45); however, the conclusion was made on the basis of the findings of 2 intervention studies with small sample sizes (29, 27), one of which observed the significant association only in women with PCOS (29) and a cross-sectional study that found a weak association only in late-reproductive-aged women (26). Since that systematic review, more studies on serum 25(OH)D reported no significant associations with concentrations of AMH or AFC (30, 31). In addition, limited studies that used serum and urinary FSH concentrations as ovarian reserve markers also reported inconsistent findings (28, 30, 32). Whether vitamin D is associated with female causes of infertility, including PCOS and POI, remains to be answered. Pharmacologic doses of 50,000 IU vitamin D₃/wk for 8 wk reduced serum AMH in vitamin D–deficient patients with PCOS, whereas no significant change was observed in women without PCOS, suggesting that the association of vitamin D with ovarian reserve may differ in women on the basis of their PCOS status (29). However, another study in this area could not find any significant association in women with or without PCOS, which may be because 60% of the population was vitamin D sufficient (31). Serum vitamin D concentrations were decreased in women with POI in one study (28), whereas in another study it was not significantly different in women with POI compared with those without POI (31).

If vitamin D regulates ovarian reserve, it can also influence the onset of menopause. Higher intakes of vitamin D predicted increased age at menopause in the NHS after adjusting for total energy intake, although the association became nonsignificant after adjusting for other covariates (41). Clarifying the role of vitamin D in the prediction of onset of menopause needs further investigation of the association between serum 25(OH)D and age at menopause, particularly in populations with high variations in serum vitamin D.

In cross-sectional studies, fat and meat intakes were associated with lower odds of menopause, whereas calcium and soy intakes were associated with higher odds of menopause (19, 37). When the associations were examined prospectively, no significant associations were observed (13, 14, 38–41) suggesting that menopause status may affect the diet in cross-sectional studies. Most studies on carbohydrate and protein intakes found no significant associations (13, 37, 39), with the exception of one prospective study that indicated later menopause with higher intakes of carbohydrate and protein (14). A randomized clinical trial provided strong evidence that long-term adherence to a low-fat, high-carbohydrate diet did not influence the timing of menopause (42). In most studies, fiber intakes were also not associated with menopause (14, 37, 39, 40), although one prospective study suggested a nonlinear inverse association between fiber intakes and age at menopause (13). Of vegetable intakes, green and yellow vegetables were suggested to be associated with later menopause (39), although the association has not been examined in other studies. Fruit intake was associated with later menopause in one study (13) but not in another (14). Low-fat dairy and skimmed milk could significantly predict later menopause only in women aged <51 y, which suggests that the effects of dietary intakes may vary according to age. In that same study, dairy fat, dairy protein, and lactose also predicted later menopause in women aged <51 y (41). The potential of altering estrogen concentration and its action as an agonist or antagonist and combating the toxic effects of oxidative stress on ovarian follicles by higher intakes of antioxidants from dietary sources are some of the mechanisms proposed by which nutritional factors may affect menopausal age (13, 14, 38, 40). However, on the basis of the current evidence, it seems that even if dietary intakes can affect menopause, the magnitude is modest, and the clinical significance of the effect is currently uncertain. Differences in the mean age of participants, the distribution of age at menopause, mean intakes of food components and their variations, combinations of food items in diet, and covariates entered to models as potential confounders can explain some of the inconsistencies in findings. In addition, it is possible that some associations became significant due to the large sample sizes.

The main strengths of this review are its systematic design, investigation of a broad range of nutritional factors, provision of evidence for both ovarian reserve markers and menopausal age to assess the effect of nutritional factors on ovarian aging, and the inclusion of both observational and interventional studies. However, the variability across studies in nutritional factors, study design, and ovarian reserve markers limited our ability to pool findings of these studies. Limiting our search results to published articles and articles in the English language also represent limitations of this systematic review. Most of the interventional studies included in this review had low power and a short duration. Recall bias for age at menopause and dietary intakes is a main limitation of observational studies. To date, studies that examined the role of nutritional factors on ovarian aging considered single foods or nutrients without controlling for the effect of other food components; evaluating the effect of dietary patterns that show a combination of food items is necessary in future studies. In addition, the associations between nutritional factors and longitudinal changes in ovarian reserve by multiple measurements of ovarian reserve markers have not been investigated. Therefore, it is unclear whether nutritional factors affect the rates of changes in ovarian reserve.

In conclusion, to date, an insufficient number of studies have examined the associations of nutritional factors and ovarian reserve, especially measuring AMH as a marker, and timing of menopause. The findings of some studies suggest modest associations of some single nutrients or food items with ovarian reserve and age at menopause. To better understand this issue, more studies examining the associations of dietary intakes and dietary patterns with concentrations of AMH and age at menopause are needed. Furthermore, to explore whether nutritional factors alter the process of ovarian aging, examining the rates of changes in ovarian reserve markers should be considered.