The time between menarche and menopause, during which reproduction is possible, spans a period of almost 40 years in women. Although the presence of menstrual bleeding is often equated with the ability to conceive, establishing a pregnancy actually depends on the presence of a normal ovarian cycle (i.e., a normal ovulatory menstrual cycle). If vegetarianism affected women's reproduction, it would necessarily affect the characteristics of the ovarian cycle. To provide background information for an examination of this issue, the normal ovarian cycle will be described, as will subclinical and clinical disturbances of the cycle and their potential impact on reproduction. This will be followed by a discussion of the effects of various dietary and non-dietary factors on cycle characteristics, and finally, by a review of the available literature assessing whether differences exist between vegetarians and non-vegetarians.
The ovulatory menstrual cycle reflects a complex interplay of hormones.21 During menstrual flow, the onset of which defines the first day of the cycle, circulating levels of both estradiol and progesterone are low. Absence of feedback inhibition allows GnRH from the hypothalamus to stimulate the release of low levels of FSH and LH from the anterior pituitary. As a result of FSH stimulation, a number of ovarian follicles begin to mature and secrete estrogen. Then, estrogen levels gradually rise during the follicular phase of the cycle. The increasing estrogen levels inhibit FSH secretion and, consequently, all but the most mature of the follicles (the "dominant" follicle) undergo atresia. The dominant follicle releases large amounts of estrogen, which in turn appear to stimulate a major surge of LH at about mid-cycle (typically, days 12-14). Consequences of the LH surge include inhibition of estrogen production by follicular cells, initiation of changes leading to ovulation (rupture of the dominant follicle and release of the ovum), and transformation of the ruptured follicle to the corpus luteum, which synthesizes both estrogen and progesterone during the luteal phase of the cycle. If the ovum is not fertilized, estrogen and progesterone levels begin to fall and the corpus luteum undergoes luteolysis 10-16 days after its formation. The decline in ovarian hormones results in the shedding of the thickened uterine lining as menstrual flow, beginning the next cycle. The cycle averages 28 days in length, with a normal range from 21 to 35 days.
The normal ovulatory cycle described above does not invariably occur between menarche and menopause. Disturbances of cycle length and characteristics are most common in the years following menarche and preceding menopause,22 but can occur at any time in response to physiological or psychosocial triggers. These disturbances represent a continuum, ranging from a subtle decrease in progesterone secretion during the luteal phase (luteal phase defects) to frank amenorrhea (absence of menstrual flow for at least 6 months in a nonpregnant woman).
The least severe disturbances in the cycle are clinically silent; that is, they can occur within cycles of normal length and are not evident to women unless cycles are being monitored by techniques such as serial hormone measurements in blood or saliva. These disturbances include luteal phase defects (low progesterone levels during the luteal phase), a short luteal phase (< 10 days in duration) and anovulation (a cycle in which an ovum is not released).
Subclinical disturbances of the cycle are not uncommon. Data collected by Vollman,22 from Swiss women in the 1950s and 1960s indicated that 3% of cycles in mature women were anovulatory, and that an additional 14.6% of cycles had short luteal phases. More recently, in a study in which women were prescreened to have two consecutive normal ovulatory cycles, only 13 of 66 women consistently had normal ovulatory cycles over the subsequent year.23 The remaining women had one or more cycles that were either anovulatory (n = 13) or had a short luteal phase (n = 40). Examining the data another way, 29% of all cycles analyzed over the year had a short luteal phase or were anovulatory. However, only 3% of cycles analyzed fell outside the normal range for cycle length. Other studies have also documented a moderately high prevalence of subclinical disturbances in apparently healthy women.24,25
Anovulatory cycles are obviously associated with the inability to conceive, and cycles with luteal phase defects or a short luteal phase may also lead to reduced fertility. In the latter case, progesterone levels may not remain elevated long enough to allow the zygote to implant before the uterine lining is shed.
Oligomenorrhea (36-180 days between menses) and amenorrhea (absence of menses for at least 180 days in nonpregnant women) represent cycle disturbances that are apparent to women because of the absence or irregularity of menstrual flow. These disturbances occur less frequently than subclinical disturbances, with estimates of amenorrhea ranging from a 1-year incidence of 0.7% in Swedish women aged 18-45 years, to a 5%
incidence in selected groups such as young college students.26,27 Accurate data on the occurrence of oligomenorrhea are more difficult to obtain, especially since the term is usually used to refer to a pattern of long, irregular cycles rather than to the occurrence of one long cycle. Nevertheless, Vollman reported that approximately 15% of the cycles of Swiss women in their early 20s were longer than 36 days, and that this decreased to fewer than 5% of cycles in women aged about 30-45.22
Long cycles are often anovulatory, although this is not always the case. Clearly, however, fertility would be considerably reduced in women with oligomenorrhea or amenorrhea.
C. Effects of Dietary Components on the Ovarian Cycle
In this section, studies that have assessed the effects of various dietary components on the ovarian cycle will be described. It should be noted that there is relatively abundant literature on the effects of many of these components on serum hormone levels, some of which is presented elsewhere in this volume, but this review will focus on studies that have assessed or inferred effects on the ovarian cycle, per se. Because of the overwhelming influence of energy availability on the menstrual cycle, a brief overview of this topic will also be provided.
Either inadequate or excessive energy availability can lead to menstrual disturbances. Because the relative weights of vegetarians and non-vegetarians may differ, it is important that studies assessing the prevalence of menstrual cycle disorders in these groups ensure that any differences in relative weight are considered.
At one extreme, starvation and emaciation are almost always associated with amenorrhea, but dieting can induce missed cycles or irregularities even before substantial weight loss occurs.28,29 Even very short-term, acute energy shortages can interfere with LH pulsatility and may thereby affect ovarian function.30,31
At the other extreme, cycle disturbances also occur in association with obesity. Almost 50 years ago, before anorexia nervosa became common, a higher prevalence of obesity was observed among amenorrheic women.32 More-recent literature suggests that anovulatory cycles are more common among obese women,33,34 and that weight loss in these individuals results in improved ovulation and ability to become pregnant.35
The effect of diets containing 40% and 20% energy as fat on the menstrual cycle was assessed in 30 healthy premenopausal women.36,37 Diets were similar in energy, protein, the amount of meat provided and the ratio of polyunsaturated to saturated fatty acids (P/S), but the low-fat diet contained about 37% more crude fiber. Each woman consumed both the high- and low-fat diets for four menstrual cycles. Compared with the 40% diet, the 20%-fat diet significantly increased cycle length and duration of menstrual flow.36 The increased cycle length was due to an increase in the follicular phase length.37
The effect of diets high and low in fat was also assessed in another study of six lacto-ovo-vegetarian women.38 Diets containing 46% and 25% fat, and similar in protein, energy, P/S and fiber were provided to women for 1 month each in a crossover design. No changes in menstrual cycle length or in the lengths of the follicular or luteal phases were observed. Although this study was well controlled, the small number of subjects studied and the short duration of the intervention may have limited the potential to detect differences.
A low-fat diet intervention was also assessed in women with cystic breast disease.39 Sixteen women whose habitual diets contained about 35% fat were counseled on a low-fat diet, and reduced dietary fat to about 21% of energy. Serum hormones and gonadotropins were assessed before the intervention, as well as 2 and 3 months later. Specific data on the menstrual cycle were not reported, although the authors stated that the intervention had no discernable effects on the women's menstrual cycles.39
The effects of doubling dietary fiber from approximately 15 to 30 g per day was assessed in 62 premenopausal women whose baseline diets contained at least 25% energy as fat and less than 25 g fiber per day.40 Women were randomly assigned to receive a corn, oat, or wheat bran supplement (provided as muffins) for 2 months. Fat intake as a percentage of energy did not change during the intervention, and body weight was maintained. Data on menstrual cycle characteristics were not provided, but in the discussion, the authors suggest that changes did not occur.40
The combined effects of diets low in fat and high in fiber have also been studied, although data on menstrual cycle characteristics are not always available. In the study of Woods et al.,41 17 premenopausal women were first fed a "typical Western diet" with 40% energy as fat and 12 g dietary fiber for one menstrual cycle. They were then fed a diet containing 25% energy as fat and 40 g dietary fiber for two menstrual cycles. No specific results pertaining to menstrual cycle length were provided.
In the study of Goldin et al.,42 a larger number of women (n = 48) was studied using a similar protocol, and cycle length, follicular phase length, and luteal phase length were assessed. No significant changes were observed, although it is reported that increasing dietary fiber from 12 g to 40 g per day was associated with an increase in follicular phase length of 0.69 days (P = 0.27), and that decreasing fat from 40% to 20-25% was associated with an increased follicular phase length of 0.51 days (P = 0.47).
The effect of diets' containing or not containing meat was assessed in studies conducted by Hill et al.43,44 In the first study, 16 Caucasian women were switched from their usual meat-containing diets (with an average of 38% energy from fat) to a diet containing no meat or meat products (with an average of 33% energy from fat and no change in P/S, but presumably more fiber). Weight was maintained during the 2-month intervention. The vegetarian diet significantly decreased cycle length from 30.4 ± 0.6 days to 26.6 ± 0.4 days, and this was associated with a significant decrease in follicular phase length.43 In the same study, nine black South African women who customarily ate a vegetarian diet were provided with 150 g of cooked lean meat on a daily basis for 2 months. Whether fat or fiber intakes changed was not stated, but body weight was maintained. The meat-containing diet significantly increased cycle length and follicular phase length.43
These authors also compared the effects of a meat supplement and an isocaloric supplement of soybeans in 16 black South African women who habitually consumed a meat-free diet.44 Body weight was maintained during the 2-month intervention. Menstrual cycle length increased from 26.5 ± 0.9 to 29.7 ± 0.8 days (P < 0.01) when women received meat, and this was due to an increase in follicular phase length. No changes occurred in women who consumed the soybean supplement.
Isoflavones and lignans are two classes of phytogestrogens that are found in unusually high concentrations in soy and flaxseed, respectively. Phipps et al.45 assessed the effects on the menstrual cycle of adding 10 g flaxseed powder to the diets of 18 healthy premenopausal women. Each woman consumed her usual diet for 3 months and the flaxseed-enriched diet for 3 months in a randomized, cross-over design. During flaxseed supplementation, mean luteal phase length was significantly longer (12.6 ± 0.4 vs. 11.4 ± 0.4 days, P = 0.002), with no difference in follicular phase length.
Results virtually opposite to those of Phipps were found by Cassidy et al.,46,47 who assessed the effects of adding soy protein containing about 45 mg isoflavones per day to the diets of six healthy women. During soy supplementation, follicular phase length increased (17.5 ± 0.9 vs. 15.0 ± 0.4 days, P < 0.01), with no change in luteal phase length, and follicular phase serum estradiol levels also increased significantly. No changes were observed in women who consumed a soybean product from which the isoflavones had been chemically extracted.47
In another intervention study, six women were studied during a month on a metabolic ward, during which time they consumed soy milk that provided about 200 mg isoflavones, daily.48 Cycle length tended to increase during the month on soy milk (P = 0.06); however, this was largely due to a 12-day increase in one woman, compared with modest changes in the other five (including a 2 day decrease in 1 woman). Although cycle phase lengths were not reported, serum estradiol levels were significantly lower throughout the cycle, in contrast to Cassidy's finding of increased follicular phase serum estradiol.46 Serum progesterone levels were also lower with soy supplementation, and, in two of the six subjects, were consistent with anovulatory cycles.48
In contrast to studies assessing effects of acute interventions, associations between habitual soy product use and the menstrual cycle were assessed in a cross-sectional study of Japanese women.49 Fifty healthy premenopausal women reported consuming a mean of 37.9 ± 26.1 g per day from nine soy products during the past year. Although soy product use was inversely correlated with estradiol levels, there was no association with menstrual cycle length (r = -0.13, P = 0.45). Length of the follicular and luteal phases was not assessed.
In summary, the available data on the effects of various dietary components on the characteristics of the menstrual cycle are inconsistent. For example, in the study of Reichman et al.,37 a low-fat diet increased follicular phase length, but in Hill's study,43 in which dietary fat decreased in association with the removal of meat from the diet, follicular phase length decreased. Adding soy to the diets of black South African vegetarian women had no effect on cycle characteristics,44 while adding meat increased follicular phase length.44 In contrast, adding soy to the diets of Caucasian women increased follicular phase length.46 Finally, phytoestro-gens from soy increased follicular phase length,46 and increased46 or decreased48 follicular phase estradiol levels, whereas phytoestrogens from flaxseed increased luteal phase length.45
In addition to the discrepancies among studies, it is important to remember that there may not be consistent differences in the fat or fiber contents of vegetarian and omnivorous diets,50 and that changes may occur in combination (for example, diets might have increased amounts of both soy products and flaxseed). Although additional research is warranted to assess the effects of individual dietary components on the menstrual cycle, whether differences exist between vegetarians and omnivores can only be assessed by studying free-living populations consuming their habitual diets.
D. Effects of Nutrition-Related Cognitive Variables on the Ovarian Cycle
Because the human female ovarian cycle is controlled at the level of the hypothalamus, it can be affected by psychological, as well as by physiological variables.29,51 Stressors such as grief, travel, and moving are known to affect the menstrual cycle, but no data could be located to indicate whether these occur differentially between vegetarians and non-vegetarians. Stressors related to nutrition and food intake, however, could conceivably differ between these groups with different dietary patterns.
One such variable is cognitive dietary restraint. This is the perception that food intake is constantly being limited in an effort to control body weight, and it can be assessed using the restraint scale of the Three-Factor Eating Questionnaire.52 That dietary restraint is "cognitive" is reflected, at least in part, by findings that women with high and low restraint scores have similar values for BMI and report similar energy intakes.24,53-55
In at least three studies, women with high restraint scores were found to have a higher prevalence of subclinical menstrual disturbances than women with low restraint scores,24,54,55 despite values for BMI that were generally similar. It is possible that stress associated with monitoring food intake in women with high restraint scores leads to increased secretion of corticotropin-releasing hormone (CRH) and cortisol. CRH has been shown to inhibit gonadotropin secretion,56 and could thereby disturb menstrual function.
Whether levels of cognitive dietary restraint differ between vegetarians and omnivores may depend on the motivation for vegetarianism. Some women with high levels of cognitive dietary restraint may adopt a vegetarian diet as a means of limiting food intake. Supportive evidence is provided by studies of women with anorexia nervosa, a condition characterized by very high levels of restraint. Among consecutive cases in two patient series, 45% and 54% were vegetarian.57,58 And in a survey of 158
healthy women, those who had high restraint scores and were high in feminist values were more likely to be vegetarian.59 The authors of that study speculate that women high in feminist values may view dieting for weight loss as socially unacceptable, and for these individuals, becoming vegetarian may represent an attempt to conceal dieting behavior from others.59
In contrast, in a study of carefully selected vegetarians and omnivores (the inclusion criteria included being weight-stable and of normal body weight, maintaining chosen dietary pattern for at least 2 years, and no history of an eating disorder), vegetarian women had significantly lower scores for dietary restraint than did omnivores.50 Thus, levels of cognitive dietary restraint, and possibly the prevalence of eating disorders, may differ between vegetarians and omnivores, making it important to assess these variables in studies comparing menstrual function between these groups.
E. Studies of the Ovarian Cycle in Vegetarians and Omnivores
In the 1980s, several published reports suggested that menstrual disturbances were more common among vegetarian women than among omni-vores.60-62 In 1984, Brooks et al.60 reported on the diets of 11 amenorrheic and 15 regularly menstruating runners. Nine of the amenorrheic runners were "vegetarian" (defined as eating less than 200 g meat per week), compared with only two runners with regular cycles. In another study reported the same year, Slavin et al.61 found that the prevalence of amenorrhea was 31% among athletes who avoided meat, compared with only 4% among those who reported consuming all food categories. Other evidence was provided by a weight-loss intervention study in which normal-weight women were assigned to lose about 1 kg per week on either a vegetarian or an omnivorous diet.62 Although the amount of weight lost was similar between groups, only two of nine women assigned to the vegetarian diet had normal ovulatory cycles, compared with 7 of 9 women on the omnivorous diet. None of these studies, however, was specifically designed to assess whether the prevalence of menstrual disturbances differed between vegetarians and omnivores, and their results could have been confounded by other variables (e.g., prevalence of eating disorders).
One of the first studies designed to address this question was that of Pedersen et al.63 Vegetarian and non-vegetarian women, recruited through newsletters and newspaper advertisements, provided self-report data on whether they had regular cycles (11-13 menses per year), irregular cycles (three—10 menses per year), or were amenorrheic (< two menses per year). Compared with omnivores, a higher proportion of vegetarian women reported irregular cycles or amenorrhea (26.5% vs. 5.9%, P < 0.01).
Inferences from this study are limited, however, as a recruitment bias may have existed: If the study had been described as trying to determine whether menstrual disturbances were more common among vegetarians, vegetarian women with menstrual disturbances might have been more likely to volunteer. Furthermore, the current use of oral contraceptive agents (OCA) was not listed as an exclusion criterion, and OCA had been used for a significantly longer period of time by omnivores than vegetarians. Because OCA use results in "regular" cycles, this could have contributed to the higher prevalence of regular cycles among omnivores. Finally, it was not stated whether women with an eating disorder, or history of an eating disorder, were excluded. In a subsequent study by this group,64 in which OCA use during the previous 3 months was an exclusion criterion, irregular menses or amenorrhea was observed in four of 27 vegetarians and none of the non-vegetarians (P <0.05). Although the groups of women were well matched in many characteristics, the possibility of a recruitment bias remained, and history of an eating disorder did not appear to have been assessed.
The study the author and co-workers conducted was designed to attempt to avoid possible confounding variables.24 It was reasoned that if clinical menstrual disturbances were more common among vegetarians, subclinical differences would also be more common. Accordingly, the study was controlled for a recruitment bias by including only women who reported "regular" menstrual cycles. Other recruitment criteria included:
Women had been following their current dietary patterns for at least 2 years. Records of daily basal body temperature were kept for 6 months, and used to classify cycles as normally ovulatory, with a short luteal phase, or anovulatory.65 In this group of highly selected women, vegetarians had longer luteal phase lengths (11.2 ± 2.6 vs. 9.1 ± 3 days, P < 0.05) and fewer anovulatory cycles (4.6% vs. 15.1% of cycles, P < 0.01). Vegetarians in this study also had significantly lower levels of cognitive dietary restraint, which might have contributed to the findings.
Although the results of the preceding study suggest that vegetarianism, per se, is not associated with an increased prevalence of menstrual disturbances, this finding cannot be generalized to the population level. As mentioned earlier, some women adopt a vegetarian diet because of concerns about body weight, and such women would likely have been excluded from participating in the study. Population studies are needed to address this issue at the broadest level.
Was this article helpful?