Vol. 5. 63-70, JWZI4(U%’ 1996 Cancer Epidemiology, Biomarkers & Prevention 63
Effects of Soya Consumption for One Month on Steroid Hormones in Premenopausal Women: Implications for Breast Cancer Risk Reduction’
Lee-Jane W. Lu,2 Karl E. Anderson, James J. Grady, Introduction and Manubai Nagamani Ovarian hormones and other reproductive factors (e.g. , ages of Departments of Preventive Medicine and Community Health [L-J. W. L.. menarche, menopause, and parity) have important influences K. E. A.. J. J. G.j and Obstetrics and Gynecology [M. NI. University of on breast cancer development ( I ). Increased levels of estrogens Texas Medical Branch. Galveston. Texas 77555 in blood and urine are markers for high risk for breast cancer (2-5). For example, in a large prospective, case-control study, Toniolo et a!. (5) reported a significant relationship between Abstract serum estrogen levels and the risk for breast cancer in women Soybean consumption is associated with reduced rates of in New York. Goldin et a!. (6) reported 44% lower blood levels breast, prostate, and colon cancer, which is possibly of estrogens and androgens in oriental women who emigrated to related to the presence of isoflavones that are weakly the United States from areas of low breast cancer risk, when estrogenic and anticarcinogenic. We examined the effects compared to Caucasian Americans, who have a higher risk for of soya consumption on circulating steroid hormones in breast cancer. Women in rural China have a low risk for breast six healthy females 22-29 years of age. Starting within cancer and 36% lower plasma estrogen levels when compared 6 days after the onset of menses, the subjects ingested a to women in Britain where breast cancer is more common (7). 12-oz portion of soymilk with each of three meals daily Similarly, postmenopausal women in Japan have lower blood for 1 month on a metabolic unit. Daily isoflavone intakes estrogen levels and a lower risk for breast cancer than do white were 100 mg of daidzein (mostly as daidzin) and -100 women in the United States (8). Earlier studies (9) are in mg of genistein (mostly as genistin). Serum 17fi-estradiol agreement with these recent observations. levels on cycle days 5-7, 12-14, and 20-22 decreased by Dietary factors, such as legumes, may account at least in 31% (P = 0.09), 81% (P = 0.03), and 49% (P 0.02), part for these differences in hormone levels and breast cancer respectively, during soya feeding. Decreases persisted for risk. Consumption of legumes is greater in countries such as two to three menstrual cycles after withdrawal from soya China, Japan, and Thailand, where breast and prostate cancers feeding. The luteal phase progesterone levels decreased are less common than in Western countries (10-13). Legumes by 35% during soya feeding (P 0.002). are the major source of protein for vegetarians, another group Dehydroepiandrosterone sulfate levels decreased that is at low risk for many cancers (14, 15). A cross-sectional progressively during soya feeding by 14-30% (P 0.03). study in Australia found that legume consumption was associ- Menstrual cycle length was 28.3 ± 1.9 days before ated with lower risk for colon cancer ( I 5). Consumption of rice soymilk feeding, increased to 31.8 ± 5.1 days during the and tofu (a soy product) was inversely related to prostate cancer month of soymilk feeding (P = 0.06), remained increased risk among men of Japanese ancestry in Hawaii ( 10). Among at 32.7 ± 8.4 days (P 0.1 1) at one cycle after premenopausal women in Singapore, breast cancer risk was termination of soymilk feeding, and returned to pre-soya inversely related to soy protein intake (16. 17). These epide- diet levels five to six cycles later. These results suggest miological observations are supported by results of animal that consumption of soya diets containing phytoestrogens studies in which soya feeding was protective against experi- may reduce circulating ovarian steroids and adrenal mentally induced mammary and other organ cancers ( 13, 18). androgens and increase menstrual cycle length. Such Possible mechanisms for the cancer protective effects of effects may account at least in part for the decreased risk soya in humans are not established. However. soya contains of breast cancer that has been associated with legume significant amounts of the isoflavones daidzein and genistein consumption. ( 19). These isoflavones are weak estrogens with uterotropic potencies about I X l0 that ofdiethylstilbestrol and binding affinities for mammalian estrogen receptors, including those from MCF-7 cells (20), about 0.1-2 X lO_2 that of 17f3- Received 5/23/95: revised 9/25/95: accepted 9/27/95. estradiol (2 1 ). These isoflavones may act as anti-estrogens by The costs of publication of this article were defrayed in part by the payment of competing with endogenous estrogens for receptor binding, and page charges. This article must therefore be hereby marked ad,’ertise,nent in this may reduce estrogen-induced stimulation of breast cell accordance with 18 U.S.C. Section 1734 solely to indicate this fact. proliferation (20) and breast tumor formation. Alternatively. I This study was supported by USPHS Grants CA56273 and CA45181, and NIH National Center for Research Resources General Clinical Research Center Grant they may reduce breast cancer risk by decreasing endogenous MOI RROO()73. ovarian steroid levels. For example, genistein stimulates pro- 2 To whom requests for reprints should be addressed. at Department of Preventive gesterone synthesis by isolated bovine granulosa cells at a Medicine and Community Health, University of Texas Medical Branch. 2. I 02 Ewing Hall. Rt. I I 10. Galveston. TX 77555-I I 10. Phone: (409) 772-1730: Fax: concentration <0.2 ,.LM but inhibits progesterone synthesis at a (409 ) 772-9 108; E-mail: [email protected]. concentration >2 .LM (22) and antagonizes TGF-a-induced
Downloaded from cebp.aacrjournals.org on September 24, 2021. © 1996 American Association for Cancer Research. 554 Soya Diet and Ovarian Steroids
estrogen biosynthesis in granulosa and theca cells (23). 33.49 ± 14.48 mg (mean ± SD; n = 1 1) of daidzin plus Genistein inhibits the enzyme activity ofestrogen-specific 17f3- daidzein, and 38.39 ± 14.62 mg of genistin plus genistein/ hydroxysteroid oxidoreductase type I, an enzyme that converts l2-oz lot, mostly as daidzin and genistin, but 15-19% of the estrone to l7 -estradiol (24). Unlike some other flavonoids, total isoflavones were the aglycones daidzein and genistein. isoflavones are in general weak inhibitors for aromatase (25) Thus, the subjects ingested - 100 mg daidzin plus daidzein and
and inefficient inducers of monooxygenase and transferase - 100 mg genistin plus genistein/day. Soymilk ingestion began activities (26). Soya isoflavones may also decrease cancer in- 4.7 ± 1 .0 days (range, 3-6 days) after onset of menses. At cidence by mechanisms that do not involve alteration of steroid biweekly intervals and for 1-2 days, the subjects ingested all levels. For example, genistein is a specific inhibitor of tyrosine three l2-oz portions of soymilk within 30 mm for determination kinase (Ref. 27: an important enzyme in transmembrane signal of isoflavone absorption and disposition.4 Blood samples for transduction) induces cell differentiation (28), inhibits the ac- hormone determinations were obtained before and 1 day after tivity of topoisomerase II (29), and inhibits angiogenesis (30). starting soymilk and, thereafter, at weekly intervals during soya l7f3-Estradiol stimulates breast cell proliferation and may feeding. promote breast tumor growth (3 1 ). Progesterone antagonizes Hormone levels in normal women during a 28-30-day the proliferative effect of l7 3-estradiol on the endometrium but menstrual cycle are well described (38). 1 7/3-Estradiol levels its role in l7 -estradiol-induced breast cell proliferation is not are the lowest (<50 pg/ml) during the early follicular phase clear. Human breast epithelial cells proliferate two to three (days 1-7), peak during days 12-14 (> 200 pg/ml), decrease times more rapidly during the luteal phase when progesterone precipitously after the LH surge (to < 100 pg/ml), peak again levels are also high (32). This suggests that proliferation of during days 20-24 (100-200 pg/ml), and then decrease to breast cells induced by I 7 -estradiol may be further enhanced earlier follicular levels before the onset of menses. Progester- by high levels of progesterone (3 1). DHEA3 is an adrenal one levels are all <0. 1 ng/ml during the follicular phase, began androgen, an intermediate in the biosynthesis of l7 -estradiol to rise after the LH surge (near day 14), and peak during the (33) and has a biphasic dose effect on breast cell proliferation midluteal phase (5-12 ng/ml; days 20-24; Ref. 38). Therefore, in vitro (34). The role of DHEA as an influence on human blood samples were obtained from our subjects on days 5-7, breast cancer development is still controversial (35). However, 12-14, and 20-22 of each menstrual cycle. This was accom- a DHEA sulfate-supplemented diet reduced the incidence of plished for one cycle before soya feeding in three subjects (only mammary tumors in mice (36). Therefore, l7 3-estradiol, pro- on day 5 for one subject), and for the month of soya feeding and gesterone, and DHEA sulfate may be useful biomarkers for for at least 3 cycles within the first 6 months after termination breast cancer risk. The purpose of the present study was to of soya feeding in all subjects. We were able to follow four of determine if consumption of soya containing isoflavones can the six subjects as outpatients for up to 10 months by frequent alter these three circulating steroid levels in premenopausal phone contacts; the lengths of their menstrual cycles were women and, thereby, protect them from developing breast recorded, and blood samples were obtained for hormone cancer. analysis. Hormone Analysis. 17 -Estradiol and progesterone concen- Materials and Methods trations in the serum were measured by specific RIA after fractionation by microcelite column chromatography as de- Study Design. Six healthy, nonvegetarian women with regular scribed previously (39, 40). Briefly, 1 ml of serum was ex- menstrual cycles (26-3 1 days) were admitted for 33 days to the tracted twice with 3 volumes of ether after adding 3H tracers of General Clinical Research Center at The University of Texas 17j3-estradiol and progesterone for assessment of recovery. The Medical Branch. Four subjects were Caucasian, 1 African- American, and 1 Hispanic. Their age range was 22-29 years, extracts were dried under nitrogen and chromatographed on microcelite columns with ethylene and propylene glycol as the height 1.6-1.7 m, weight 53.8-70.5 kg, and body mass index stationary phase. This chromatographic separation is effective 18.9-26.2 kg/m2. Five subjects had never taken birth control pills and one had taken an oral contraceptive until 2 months in removing steroids that can interfere with the assay because of cross-reactivity. The fractions containing progesterone and before the study. History, physical examination, blood cell l7/3-estradiol were collected and dried, and the residue was counts, blood chemistry profiles, and serum ferritin were nor- mal. The study was approved by the Institutional Review Board dissolved in the assay buffer. Two aliquots of different volumes of this solution were used in the assay, and 0. 1 ml was used for of the University of Texas Medical Branch and written in- recovery calculations. Blank and control sera were run with formed consent was obtained from each subject. Subjects consumed a standard hospital diet and three each assay. The antibody used in I 7 -estradiol assay was 12-oz portions of soymilk (one with each meal) daily for 1 highly specific with <0. 1% cross-reactivity with estrone and month under the supervision of the dietary staff. Soymilk used estriol. Levels of DHEA sulfate were measured by direct RIA (41). The intra-assay coefficient ofvariation was 4-8% and the for this study was an homogenized, pasteurized preparation interassay variation was 5-9%. Steroid assays began only after containing no preservatives (Banyan Foods Co., Houston, TX). Isoflavone content in soymilk was analyzed as described (37). the collection of samples from all six subjects was completed. Briefly, an 80% methanolic extract was prepared from soymilk, All of the samples from each subject were analyzed in a single assay. dried and redissolved in water. Portions of the aqueous solution with or without f3-glucosidase digestion were purified by Statistical Analysis. To determine if the duration of soymilk ChemElut chromatography. The dried eluates after silylation ingestion had an effect on steroid hormone levels, Friedman’s were analyzed by gas chromatography equipped with a flame two-way ANOVA of the ranked data was used. If Friedman’s ionization detector. Selected lots used in the study contained test showed that the duration of treatment significantly affected
:‘ The abbreviations used are: DHEA. dehydroepiandrosterone: LH. luteinizing 4 L-J. W., Lu, S. N. Lin, J. J. Grady. M. Nagamani, and K. E. Anderson, hormone. manuscript in preparation.
Downloaded from cebp.aacrjournals.org on September 24, 2021. © 1996 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention 65
60
(I) >1
. 4 a) C C) . .2 4O 0 Fig. I. Individual (symbols) and mean menstrual cy- >i dc length during soya injestion and I and 2-3 months C.) (Mon) after soya ingestion in six women who ingested CC three I 2-oz portions of soymilk daily for I month. I- Cl) 3O C) =OO6.
-mean 20 - - -- Baseline During I Mon 2-3 Mon Soya Post Post Soya Soya
hormone levels, multiple comparisons of the various treatments soymilk feeding on individual and mean 17f3-estradiol levels, were made using paired t tests on the unranked data. Subjects when these were averaged over a cycle (Fig. 14 ) and analyzed with missing data at a particular time point were not included separately on specific cycle days (Fig. 2, B-D). One month of in the post hoc analysis by paired t tests for only that specific soya feeding reduced 17/3-estradiol levels either when all mea- time point, but their data from other time points were included surements were averaged during a cycle (Fig. 2A: P = 0.00 15, in the ANOVA and the calculation of the means shown in the Friedman’s test) or when they were analyzed on cycle days 5-7 figures. (Fig. 2B; P = 0.16), 12-14 (Fig. 2C; P = 0.0001), and 20-22 (Fig. 2D; P = 0.08). Fig. 14 shows that compared to the Results baseline, the mean cycle levels of 17j3-estradiol decreased by 62% during soymilk feeding (P = 0.03, paired t test), by 58% Fig. 1 shows the effect of I-month soymilk ingestion on mdi- during the first post-soymilk cycle (P = 0.05), and by 46% vidual and mean menstrual cycle lengths. The mean (± SD) during the second to the third cycles after soya (P = 0. 1 1). Fig. cycle length for this group of six women was 28.3 (± 1.9) days 2B shows that 17 -estradiol levels in the early follicular phase before soya feeding; it increased slightly to 31.8 (± 5.1) days (days 5-7) decreased by 31% during chronic soymilk feeding during the month of soymilk feeding (P = 0.06, paired t test) compared to control periods (P = 0.09). Fig. 2C shows that and to 32.7 (± 8.4) days one cycle after termination of soymilk compared to the baseline, l7f3-estradiol levels during the late feeding (P = 0. 1 1 ). Five women experienced increase in cycle length of 1-12 days during soymilk feeding, whereas one follicular phases (days 12-14) decreased by 81% during soya woman had a 2-day decrease. The subject who was taking an feeding (n = 5; P = 0.03), by 72% during the first cycle oral contraceptive 2 months before the study had a 3-day post-soya (n = 5, P = 0.056), and by 68% during the second increase in cycle length, which was similar to other subjects. to third cycles after soya feeding (n = 5; P = 0.045). The late All women returned to their pre-soymilk cycle lengths five follicular phase l7 -estradiol levels during soya feeding were cycles after termination of soymilk feeding (results not shown); also 41% lower than those during the second to third cycles this included a subject who had a cycle length of42 days during after soymilk feeding (P = 0.003). Fig. 2D shows that com- soymilk feeding and 49 days during the first cycle after soymilk pared to the baseline, 17f3-estradiol levels during the luteal feeding. This subject consumed soymilk for 4 extra days and phases (days 20-22) decreased by 49% during soya feeding also provided one blood sample after the last soymilk dose for (n 5; P = 0.02), by 68% during the first cycle post-soya hormone measurements. (n 4; P = 0.04), and by 29% during the second and third With only one exception, our subjects had baseline hor- cycles post-soya (n = 4; P = 0.24). Thus, chronic soya inges- mone levels within expected normal ranges (see “Materials and tion decreased l7 -estradiol levels that persisted for two cycles Methods;” Ref. 38). Hormone levels from one cycle before and after soya ingestion. During the soymilk feeding, only one for five to ten cycles after termination of soymilk feeding from woman (who was previously on an oral contraceptive) had a our study subjects were similar and, therefore, were averaged small l7 -estradiol peak (57 pg/ml) on day 14. During the and regarded as baseline values for each subject. The results of month of soymilk feeding, higher levels of l7f3-estradiol in all the second and third cycle after soymilk feeding were combined six subjects were detected on cycle days 22-37 (range, 60-135 because there were no detectable differences in hormone levels pg/ml) compared with those on cycle days 12-14 (range, 22-57 between these two cycles. pg/ml), and this was in contrast to the profiles observed during Fig. 2 shows the effect and persistence of 1 month of baseline periods. If the highest levels (e.g., independent of
Downloaded from cebp.aacrjournals.org on September 24, 2021. © 1996 American Association for Cancer Research. 66 Soya Diet and Ovarian Steroids
250 60 A: Cycle Average B: Cycle Days 5-7
200 E E 40 a) 0) 0. 150 0. . \ a \, p0.ll 0 0 V V (C CC U) 100 U) LU LU 20 F’.. I.’.
50 p=0.09