Abstract
Objective: Although reduction of vasomotor symptoms (VMS; hot flashes and night sweats) has been reported in postmenopausal women who used isoflavones, a clear dose response has not been shown, has largely not been reported for perimenopausal women, and has largely only been reported for reducing prevalent VMS, not preventing newly developing VMS. We analyzed longitudinal data from the Study of Women’s Health Across the Nation for the relation of dietary phytoestrogen and fiber intake to incident VMS in this multiracial/ethnic cohort.
Methods: The Study of Women’s Health Across the Nation included 3,302 premenopausal and early perimenopausal women, 1,651 of whom reported no VMS at baseline and were followed with annual visits for 10 years. Dietary intakes of isoflavones, coumestrol, lignans, and fiber were assessed by a food frequency questionnaire at baseline and in annual visits 5 and 9 and interpolated for intervening years. The number of days experiencing VMS in the past 2 weeks was self-reported annually. Using multinomial logistic regression with generalized estimating equations, we modeled incident VMS in relation to isoflavones, lignans, fiber, coumestrol, or total phytoestrogen intake and covariates.
Results: No consistent monotonic relations were observed between any dietary phytoestrogen or fiber and incident VMS, although adjusted odds ratios for some individual quartiles were statistically significant.
Conclusions: For certainty of any effect of dietary phytoestrogens or fiber on the prevention of incident VMS, a randomized, placebo-controlled, double-masked trial with sufficient numbers of women in different racial/ethnic, menopausal status, and metabolic groups over years of follow-up is required, but our results suggest that a clinically significant or large effect is improbable.
Introduction
The North American Menopause Society recently issued a position statement on isoflavones for menopausal health in which they summarized the results of 14 randomized placebo-controlled trials of soy isoflavone supplements for vasomotor symptoms (VMS). Lower rates of VMS were reported in most studies (including randomized trials) of postmenopausal women who used soy isoflavone supplements. However, the effect has not shown a clear dose response, has largely not been reported for perimenopausal women (as noted in the position statement) even though perimenopausal women have rates of VMS at least as high as those of postmenopausal women, has only involved supplements and not dietary isoflavones, and has largely only been related to reducing VMS in women who have them, rather than preventing newly developing incident VMS.
Phytoestrogens (“plant estrogens”) are heterocyclic phenols found in many plant foods. Three major categories have been established within the general class of phytoestrogens: isoflavones (eg, daidzein and genistein), coumestans (eg, coumestrol), and lignans (eg, secoisolariciresinol and matairesinol). Isoflavones, lignans, and coumestrol bind to α and β estrogen receptors and are predominantly agonists. However, the ultimate effects of phytoestrogens on the estrogen pathway are likely to be determined by a variety of factors, including the following: their relative affinity for α and β estrogen receptors and effects on the conformation of these receptors; their agonist or antagonist properties related to their concentrations and to circulating levels of estradiol; and the different phytoestrogens (isoflavones, lignans, and coumestrol) that each modify the metabolism and bioavailability of endogenous estrogen by mechanisms that either increase or decrease such bioavailability.
In one early trial, postmenopausal women who received soy supplements (containing isoflavones) or wheat flour (containing enterolactones) had a statistically significant reduction in VMS (40% in the soy group and 25% in the wheat group), although the rates of decline did not differ significantly (partly because of the small sample size). Most, but not all, subsequent randomized, double-blind, controlled clinical trials of soy or soy isoflavones for VMS have observed similar reductions in VMS. In contrast, trials that included perimenopausal women have not observed lower rates of VMS in relation to greater phytoestrogen administration.
The apparent menopause stage–specific results suggest that the effects of dietary phytoestrogens vary according to the menopausal transition (MT) stage, perhaps reflecting differences in circulating endogenous estrogen levels. However, many other aspects of the design and methods of the above-mentioned trials may also have produced inconsistent results. Several types, doses, and formulations of phytoestrogens have been used. In addition, sample sizes in trials have been small to moderate, ranging from 39 to 177; thus, limited statistical power to detect effects may also have led to inconsistent outcomes. Most importantly, however, these trials have focused on women who already have VMS, and relatively few studies have examined the relation of phytoestrogen intake to developing incident VMS. Furthermore, studies have not examined whether the effects of phytoestrogens on VMS differ by race/ethnicity, perhaps reflecting underlying differences in metabolism, which is another possible source of differing results.
In addition, urinary excretion of phytoestrogens and the concentration of plasma sex hormone–binding globulin have been positively associated with dietary intake of fiber, which has been inversely related to circulating levels of estradiol. Few studies have investigated the relation of dietary fiber to VMS reporting, but one cross-sectional study of women with prior early-stage breast cancer showed a protective effect. Furthermore, fiber may be a proxy for lignan intake. If the measurement of fiber is more accurate than the measurement of lignans, the protective association between fiber and VMS reporting may be stronger than that for lignans.
Evidence from these small- to medium-sized studies suggests that soy or soy isoflavones suppress VMS in postmenopausal (but not in perimenopausal) women. The effects of other classes of phytoestrogens remain uncertain, and their effects on the development of new VMS—rather than on the reduction of existing VMS—have not been well studied. We undertook several analyses to build upon prior work by testing the relation between dietary phytoestrogens and development of incident VMS longitudinally, considering the relations in each stage of the MT. As outlined above, the effects of phytoestrogens on VMS would be expected to differ according to the specific phytoestrogen and MT stage—a theory that we tested in the multi-racial/ethnic longitudinal cohort of the Study of Women’s Health Across the Nation (SWAN). We improved the phytonutrient database used to determine dietary intake in SWAN to investigate whether other classes of phytoestrogens—in addition to isoflavones, particularly lignans and dietary fiber—were related to the incident reporting of VMS in SWAN. These analyses examined the longitudinal relations between each of the three main classes of phytoestrogens and dietary fiber (exposures) and incident VMS (outcome).
Specifically, we undertook analyses of SWAN’s annual data and dietary assessment to test the following hypotheses: (1) higher dietary intake of total isoflavones, total lignans, coumestrol, and/or dietary fiber would be associated with lower rates of reporting newly occurring VMS longitudinally; and (2) the relation of each class of phytoestrogens and dietary fiber to incident VMS would vary by MT stage, with stronger benefit observed in postmenopause.