Soy isoflavones consumption and risk of breast cancer incidence or recurrence: a meta-analysis of prospective studies
Numbers of epidemiologic studies assessing soy consumption and risk of breast cancer have yielded inconsistent results. We aimed to examine the association between soy isoflavones consumption and risk of breast cancer incidence or recurrence, by conducting a meta-analysis of prospective studies. We searched for all relevant studies with a prospective design indexed in PUBMED through September 1st, 2010. Summary relative risks (RR) were calculated using fixed- or random-effects models. Pre-specified stratified analyses and dose–response analysis were also performed. We identified 4 studies of breast cancer recurrence and 14 studies of breast cancer incidence. Soy isoflavones consumption was inversely associated with risk of breast cancer incidence (RR = 0.89, 95% CI: 0.79–0.99). However, the protective effect of soy was only observed among studies conducted in Asian populations (RR = 0.76, 95% CI: 0.65–0.86) but not in Western populations (RR = 0.97, 95% CI: 0.87–1.06). Soy isoflavones intake was also inversely associated with risk of breast cancer recurrence (RR = 0.84, 95% CI: 0.70–0.99). Stratified analyses suggested that menopausal status may be an important effect modifier in these associations. We failed to identify a dose–response relationship between total isoflavones intake and risk of breast cancer incidence. Our study suggests soy isoflavones intake is associated with a significant reduced risk of breast cancer incidence in Asian populations, but not in Western populations. Further studies are warranted to confirm the finding of an inverse association of soy consumption with risk of breast cancer recurrence.
KeywordsSoybeans Isoflavones Breast cancer Prospective studies Meta-analysis
Breast cancer is the leading cause of death with respect to cancer among women in Europe and North America. The American Cancer Society’s estimates indicate that approximately 1.3 million new cases of invasive breast cancer were diagnosed globally in 2007 and nearly 500,000 died from this disease . Meanwhile, the population of breast cancer survivors has become noticeably large. For example, there are more than 2.5 million breast cancer survivors in the United States currently, and this population is expected to grow to 3.4 million by 2015 .
It has been suggested that soy foods, which are highly consumed in Asia where the prevalence and incidence of breast cancer are substantially low, may contribute to the prevention of breast cancer. In fact, isoflavones, a class of phytoestrogens abundant in soy foods, have been demonstrated to exert weak estrogenic effect and have anti-carcinogenic properties. A number of epidemiologic studies examining the association between dietary soy foods or isoflavones intake and risk of breast cancer have yielded inconsistent results. Although previous meta-analyses [3, 4, 5] have showed a modest protective effect of soy intake on breast cancer, they were largely based on case–control studies. For example, of total 19 individual studies included in the latest meta-analysis , 14 studies used a case–control design. Case–control studies inevitably suffer selection and recall biases, therefore the quality and strength of evidence from previous meta-analyses are relatively low, and the results should be treated with caution.
On the other hand, genistein, a major form of isoflavone, has been shown to enhance the proliferation of breast cancer cells in vitro and to promote estrogen-dependent mammary tumor growth in ovariectomized rats [6, 7]. The safety of soy therefore has drawn great attention, especially among breast cancer survivors. Results from several epidemiologic studies [8, 9, 10, 11] that investigated the association between soy or isoflavones intake and breast cancer survival were conflicting. A benefit effect of soy on breast cancer survival was observed in one study , but not in others [8, 9, 10].
Thus, our objective is to examine the association of soy isoflavones intake with risk of breast cancer incidence or recurrence, by conducting a meta-analysis of prospective studies. Pre-specific stratified analyses were performed to assess the impacts of various study characteristics on outcomes. We also attempted to explore a dose–response relationship between soy isoflavones intake and risk of breast cancer incidence.
We conducted a systematic literature search of the PUBMED databases (National Library of Medicine, Bethesda, MD) through September 1st 2010 by using the following search terms: “soybeans,” “soy foods,” “isoflavones,” “phytoestrogens,” and “breast cancer.” We also carried out a manual search using reference lists of original articles and recent reviews. Only full-length original journal articles were considered, no attempt was made to include abstracts or unpublished studies.
Studies were eligible for our analysis if: (1) the study design is prospective, i.e., cohort or nested case–control study; (2) data related to dietary consumption of soy isoflavones were available; (3) the endpoint was breast cancer incidence or recurrence; (4) the association of isoflavones with breast cancer risk was specifically evaluated; and (5) relatively complete assessment of total isoflavones intake was performed.
We recorded study characteristics as follows: (1) name of first author, publication year, and country of origin; (2) study design and endpoint; (3) length of follow-up; (4) number of cases; (5) confounders adjusted for in multivariate analysis; (6) relative risk (RR), hazard ratio, or odds ratio from the most fully adjusted model for the highest versus the lowest soy exposure and their corresponding 95% confidence interval (95% CI). RR is used to represent ratio measures of effect, including hazard ratio and odds ratios. Soy consumption was expressed uniformly as milligrams of soy isoflavones intake per day.
We investigated the associations between the overall soy isoflavones intake and the risk of breast cancer incidence and recurrence separately. Homogeneity of effect size across studies was tested by Q statistics (P < 0.10). We also computed the I2, a quantitative measure of inconsistency across studies. If substantial heterogeneity exists, the random-effects model is appropriate; otherwise, the fixed-effects model is preferred . Pre-specified stratified analyses were performed to assess the impacts of various study characteristics, including menopausal status, regions, case numbers, length of follow-up, and study design, on outcomes. A sensitivity analysis was conducted using both fixed- and random-effects models to evaluate the robustness of results.
We conducted a dose–response analysis based on data for categories of isoflavones intake levels on median dose, number of cases and participants, and adjusted logarithm of the RR with its SE [13, 14]. This analysis was restricted to studies reporting three or more exposure levels. Studies were not eligible if necessary data were not reported or could not be estimated. One study  was subset of another main study  or had overlapping data and was excluded from the meta-analysis, but eligible for the dose–response analysis because the main study  did not report numbers of cases by different levels.
The potential publication bias was examined by the funnel plot and Egger’s test  (P < 0.10). All analyses were performed using STATA version 10.0 (Stata Corp, College Station, Texas). A P value < 0.05 was considered statistically significant, except where otherwise specified.
Characteristics of the included studies
Characteristics of the included studies
First author and year
Country or region
Mean follow-up (years)
RR (95% CI)
den Tonkelaar  2001
Urinary genistein (3rd vs. 1st tertile)
Age at diagnosis, stage of disease, radiotherapy, ER status, and total energy intake
Horn-Ross  2002
Genistein (1.1 vs. 0.29 mg/day)
Age at diagnosis and dietary energy intake
Yamamoto  2003
Isoflavones (25.3 vs. 6.9 mg/day)
Age, age at menarche, age at first pregnancy, parity, menopausal status, smoking, alcohol, physical activity, education, energy, consumption of meat, fish, vegetables, and fruit
Grace  2004
Urinary genistein (3rd vs. 1st tertile)
Age, soy supplement use, BMI, menopausal status, tobacco, tumor stage, ER status, race, and kilocalories
Keinan-Boker  2004
Isoflavones (0.77 vs. 0.19 mg/day)
Age, race, energy intake, family history of breast cancer, age at menarche, nulliparity/age at first pregnancy, physical activity, and an interaction term for BMI and menopausal status
Boyapati  2005
Total isoflavones (3rd vs. 1st tertile)
Age, age at menarche, age at first pregnancy, parity, menopausal status, smoking, alcohol, physical activity, energy, meat, fish, vegetables, and fruit
Touillaud  2006
Total isoflavones (36–112 vs. 1–22 µg/day)
Age at enrollment, age at first full-term delivery, height, weight, parity, physical activity, oral contraceptives or HRT, marital status, and energy intake
Fink  2007
Total isoflavones (≥0.6 vs. ≤0.15 mg/day)
Age, BMI, oral contraceptives, age at first pregnancy, age at menarche, parity, cancer in sisters or mothers, and energy intake, alcohol, and saturated fat
Verheus  2007
Plasma genistein (3rd vs. 1st tertile)
Age, BMI, age at menarche, family history of breast cancer, oral contraceptive, age at first pregnancy, parity, alcohol, and dietary energy intake
Hedelin  2008
Total isoflavonoids (4th vs. 1st quartile)
Height, BMI, age at menarche, age at first birth and parity, alcohol consumption, and daily energy intake, menopausal status and, and current HRT use
Iwasaki  2008
Dietary genistein (27.3 vs. 15.7 mg/day)
Age, years of interview, dialect, education, family history of breast cancer, parity, age when period became regular, menopausal status, BMI, and n-3 fatty acid
Travis  2008
Isoflavones (>20 vs. <10 mg/day)
Age, education, physical activity, age at first live birth, BMI, season of recruitment, family history of breast cancer, and total energy intake
Ward  2008
Serum total isoflavones (higher vs. lower)
Crude OR. Adjustment for age, height, weight, parity, age at menopause, benign breast disease, family history, HRT, smoking did not change OR (<10% change)
Wu  2008
Soy isoflavones (≥10.6 vs. <10.6 mg/1,000 kcal)
BMI, menopausal status, parity, HRT, smoking, family history of breast cancer, and saturated fat intake
Goodman  2009
Urinary total isoflavones (75th vs. 25th percentile)
Crude OR. Adjustment for various possible confounders did not change OR (<10% change)
Guha  2009
Genistein (≥13.03 vs. 0 mg/day)
Number of births and age at first birth
Lee  2009
Isoflavones (≥44.24 vs. ≤15.93 mg/day)
Age, location, race-ethnicity, date and time of urine collection, HRT use, and fasting hours BMI, alcohol use, parity, and family history of breast cancer
Shu  2009
Isoflavones (62.68 vs. ≤20.00 mg/day)
Weight, oral contraceptive use, HRT, menopausal status, parity, menarche, breast feeding, family history of breast cancer, daily intake of fat and energy, and batch
The studies eligible for analysis were published between 2001 and 2009. Six studies from Asia, 12 from Europe or North America. Of the 18, 4 were cohort studies of recurrence in breast cancer patients [8, 9, 10, 11], and the rest were cohort or nested case–control studies of breast cancer incidence. Results were presented by menopausal status in 10 of these studies, 5 in overall women, 1 in pre-menopausal women, and 2 in post-menopausal women. The median length of follow up ranged from 1.57 to 13 years. The average intake of soy isoflavones was much lower in Western populations than that in Asian populations, except for one British study  that enrolled numbers of vegetarian women. Most individual studies adjusted for a wide range of potential confounders, including age, BMI, physical activity, smoking, alcohol use, and hormone replacement therapy.
Study of breast cancer incidence
Study of breast cancer recurrence
Stratified analyses and sensitivity analysis
Results of stratified analyses of breast cancer incidence
No. of Studies
Summary relative risk (95% CI)
P for heterogeneity
Length of follow up (years)
No. of cases
We failed to identify a significant dose–response relationship between total isoflavones intake and risk of breast cancer incidence using data from 6 studies [15, 20, 22, 23, 26, 27]. The risk of breast cancer incidence decreased, on average, by 4% for every 10 mg/day increase of soy isoflavones intake (RR = 0.96, 95% CI: 0.90–1.02, P = 0.176, P for heterogeneity = 0.230).
The present meta-analysis of prospective studies supports a significant inverse association between soy isoflavones intake and risk of breast cancer incidence, as well as risk of breast cancer recurrence. Soy isoflavones intake was associated with a significant reduced risk of breast cancer incidence in Asian populations, but not in Western populations. Moreover, the inverse association between soy exposure and risk of breast cancer incidence or recurrence was stronger in post-menopausal women than that in pre-menopausal women. Menopausal status as an effect modifier may play an important role in the soy and breast cancer association.
Our finding that soy intake was inversely associated with breast cancer risk in Asian but not Western populations is consistent with a previous meta-analysis , despite the different study design. This finding may largely be attributed to the different amount of soy consumption between Asian and Western populations as higher level of consumption among Western populations is obviously far below even the lower levels of intake for Asian populations. Another explanation is that the protective effect observed in Asian studies results from lifelong or early life exposure to soy. To date, five epidemiologic studies [16, 33, 34, 35, 36] examining the effect of early soy intake on adult breast cancer risk provide consistent evidence that early isoflavone exposure is protective against breast cancer and early life influences play an important role in the cause of breast cancer. Additionally, equol, a metabolite of daidzein, may also affect the soy and breast cancer association because it is superior to all other isoflavones in its antioxidant activity . Asian populations have higher ability to produce equol than Western populations; however, evidence is insufficient [28, 30] to determine whether equol-producer status contributed to the inconsistent literature.
The current study suggests that menopausal status may be an important modifier of the effect of isoflavones on the risk for breast cancer, because mechanisms that might mediate the effect involve the ovarian synthesis of sex hormones or the alteration of other menstrual cycle characteristics [38, 39]. The lack of effect in pre-menopausal women implies that the mechanism by which isoflavones act may be effective only at low sex hormone concentrations as shown in post-menopausal women. Further studies are strongly recommended to assess soy and breast cancer risk separately in pre- and post-menopausal women.
A dose–response relationship, if present, can strongly support causality of association. The lack of dose–response relationship in the present study may be due to limited number of studies eligible for this analysis; for example, Wu et al.  reported only two levels of soy exposure, and Lee et al.  did not provide number of cases in different levels. Alternatively, it might imply the presence of a threshold level of effect as a wide range of average soy isoflavones intake from none to 33.9 mg/day was covered in our analysis.
As for breast cancer recurrence, four studies [8, 9, 10, 11] to date have investigated the effects of soy intake in women with breast cancer. Although combining data from individual studies showed a significant risk reduction and little evidence of heterogeneity, caution should be taken since existing data are insufficient and few modifiable lifestyle factors for breast cancer survival have been identified. Menopause status might modify the association of soy isoflavones with risk of breast cancer recurrence as a benefit effect was observed in post-menopausal women but not in pre-menopausal women. The interpretation of the differences by menopausal status is challenging and the play of chance could not be excluded. With regard to tamoxifen, Guha et al.  reported that consumption of soy foods may be associated with a reduced risk of recurrence among patients with tamoxifen treatment, while Shu et al.  found that intake of soy foods was associated with improved survival regardless of tamoxifen use, indicating soy does not appear to negative tamoxifen efficacy. Additionally, in Boyapati et al. and Shu et al.’ studies [8, 11], the association between soy food intake and prognosis of breast cancer did not vary by estrogen receptor (ER) status, while Guha et al.  showed the benefit of soy food intake on survival was more pronounced among women with ER-positive breast cancer. Further confirmation is required in large prospective studies, especially randomized controlled trials (RCT), before definitive recommendations can be made.
Well-conducted RCTs provide much stronger support for a causal association than observational studies. A recent meta-analysis  of eight RCTs assessing the effects of isoflavone-rich foods or supplements on women’s mammographic density, a biomarker of breast cancer risk, concludes that isoflavone intake does not alter breast density in post-menopausal women, but may cause a small increase in breast density in pre-menopausal women. Short duration and timing of intervention of individual RCTs included in that study tended to be the major limitations because existing evidence, as previously mentioned, strongly suggests that the protective effects of soy may relate to lifetime or early life exposure.
Our study has strengths. The original studies included in this meta-analysis were all prospective, which greatly reduces the likelihood of recall bias and selection bias. Furthermore, they were all population-based except for one study , thereby minimizing the selection bias. Most studies employed validate and reliable food frequency questionnaire targeting soy isoflavones especially in Western populations among which soy consumption is substantially low, and assessment of soy intake therefore was relatively complete.
Limitations of this study include the high degree of heterogeneity, which made the results complicated to interpret. However, our stratified analysis indicated that the geographical variations reflecting differences in the amount of soy intake, timing of soy exposure, and equol-producer status, as discussed previously, largely contributed to the substantial heterogeneity. A small number of studies [18, 23, 25, 28, 29] assessing the role of the ER status in relation to isoflavones and breast cancer risk reported non-significant differences, and further investigations of the soy and breast cancer association by ER status are needed. Moreover, the apparent protective effect of isoflavones on risk of breast cancer may be due to other healthy lifestyles related to soy intake, such as high vegetable and fruit intake, more physical activity, and reduced alcohol use. However, as shown in Table 1, many individual studies have adjusted for a wide range of these potential confounders, thereby reducing the likelihood that residual confounding can explain the findings. Additionally, soy consumption was assessed only once except for one cohort study , and changes in soy consumption may have weakened the observed associations during the long follow-up period. However, exclusion of the first 1, 3, or 5 years [16, 20, 27] of follow-up did not substantially change the results.
In conclusion, the present study suggests soy isoflavones intake is associated with a significant reduced risk of breast cancer incidence in Asian populations, but not in Western populations. Further studies are warranted to confirm the finding of an inverse association of soy consumption with risk of breast cancer recurrence.
J-Y D was responsible for the literature search, data analyses, and writing of the manuscript. L-Q Q was responsible for designing the research, interpreting the data and results, and revising the manuscript.
Conflicts of interest
None of the authors had declared a conflict of interest.
- 1.American Cancer Society (2007) Global cancer facts and figures 2007. American Cancer Society, AtlantaGoogle Scholar
- 2.American Cancer Society (2009) Breast cancer facts and figures 2009–2010. American Cancer Society, AtlantaGoogle Scholar
- 18.den Tonkelaar I, Keinan-Boker L, Veer PV, Arts CJ, Adlercreutz H, Thijssen JH, Peeters PH (2001) Urinary phytoestrogens and postmenopausal breast cancer risk. Cancer Epidemiol Biomarkers Prev 10(3):223–228Google Scholar
- 21.Grace PB, Taylor JI, Low YL et al (2004) Phytoestrogen concentrations in serum and spot urine as biomarkers for dietary phytoestrogen intake and their relation to breast cancer risk in European prospective investigation of cancer and nutrition-norfolk. Cancer Epidemiol Biomarkers Prev 13(5):698–708PubMedGoogle Scholar
- 25.Hedelin M, Lof M, Olsson M, Adlercreutz H, Sandin S, Weiderpass E (2008) Dietary phytoestrogens are not associated with risk of overall breast cancer but diets rich in coumestrol are inversely associated with risk of estrogen receptor and progesterone receptor negative breast tumors in Swedish women. J Nutr 138(5):938–945PubMedGoogle Scholar
- 26.Iwasaki M, Inoue M, Otani T, Sasazuki S, Kurahashi N, Miura T, Yamamoto S, Tsugane S (2008) Plasma isoflavone level and subsequent risk of breast cancer among Japanese women: a nested case–control study from the Japan Public Health Center-based prospective study group. J Clin Oncol 26(10):1677–1683CrossRefPubMedGoogle Scholar