March 1, 2015

High Isoflavone Intake May Increase Breast Cancer Growth

As another study obfuscates the connection, what do we tell patients?
Add another recent study on the effect of soy isoflavones on breast-cancer patients to the debate over whether soy intake is dangerous for this population. Here's an analysis of the study and guidance on how to talk about soy with patients as the debate rages on.


Shike M, Doane AS, Russo L, et al. The effects of soy supplementation on gene expression in breast cancer: a randomized placebo-controlled study. J Natl Cancer Instit. 2014;106(9). pii: dju189.


Randomized, placebo-controlled trial of 7 to 30 days duration


Of the 140 women with biopsy-proven breast cancer enrolled in the study, a total of 132 were evaluable. Participants included premenopausal (39.4%) and postmenopausal (60.6%) women with estrogen-receptor (ER) positive and ER negative status. Mean age was 56.2 years (±11.9 y). 


Participants consumed either a soy protein supplement (25.8 g/pkt) or placebo (milk protein, 25.8 g/pkt) from time of positive breast biopsy until the day before each participant’s surgical resection (7-30 d). Each packet also contained unknown quantities of maltodextrin, sucrose, fructose, artificial flavor, calcium phosphate, magnesium phosphate, riboflavin, vitamin A palmitate, folic acid, vitamin D3, and vitamin B12. Concentrations of isoflavones in the soy protein were as follows: genistein, 1.8 mg/g soy protein and diadzein, 0.8 mg/g soy protein. Daily doses of isoflavones for those in the intervention group totaled 98.44 mg/day and 16.51 mg/day each genistein and diadzein. 

Outcome Measures

Primary outcome measures were changes of proliferation and apoptosis measured by immunohistochemistry for Ki67 (proliferative index) and Cas3 (apoptotic index) between the 2 study groups (N=104; soy, n=54; placebo, n=50). Secondary outcome measures included changes in gene expression as evaluated by NanoString analysis (NanoString Technologies, Seattle, Washington) between the groups (N=14; soy, n=8; placebo, n=6). The only other secondary outcome measures were overall expression of genes through DNA microarray (N=51; soy, n=28; placebo, n=23) and quantitative PCR (qPCR) (N=46; soy, n=27; placebo, n=19) analyzed from surgical samples (not biopsied tissue). Plasma isoflavones were measured at the time of surgical resection for most participants (soy, n=63; placebo, n=62). Ad hoc analysis included dividing participants by plasma genistein levels referred to as “high” (n=11) and “low” (n=23) genestein subsets. The “high” genistein plasma level was set at 16 ng/mL.    

Key Findings

The primary outcome measures Ki67 and Cas3 were not different between the intervention group vs placebo group nor in the ad hoc analysis of those with high-circulating genistein. NanoString analysis of gene expression (n=14) showed a trend for variance between groups with several gene groupings having opposing directions (overexpressed/underexpressed). There was a large variance in circulating levels of genistein in the intervention group, with a median level of 6.3 ng/mL and 25% of the group having circulating levels less than 0.5 ng/mL. For this reason, the authors chose to create a “high” genistein group by including only those participants with circulating levels of genistein in excess of the 95th percentile of the placebo group. Comparing this high-genistein subset (n=11) to the low-genistein subset (n=23), there was a statistically significant difference (P<.01) between the gene signatures on microarray analysis. Further, the gene signature of the high-genistein group had overrepresentation of pathways that regulate cell growth and proliferation (P<.001). While groups were well matched without any variation in patient or tumor characteristics, there was a trend for luminal A‒type breast cancer in the low-genistein group and luminal B in the high-genistein group (P=.06). DNA microarray suggested overexpression of fibroblast growth factor receptor (FGFR2) and qPCR confirmed a 2.3-fold overexpression of FGFR2 in the soy protein group (n=27) vs the placebo (n=19).

Practice Implications

There is ongoing debate and contradictory information regarding soy and breast cancer in the literature, and even more so in the media. This study, while interesting, will no doubt add to the confusion. A thorough parsing of the data from this study in the context of the outcome data on soy consumption and breast cancer may clarify some of the confusion.
I sometimes emphasize to my patients that when we talk about soy, we are talking about 'just a bean.'
First, the present study used very large amounts of soy protein isolate (51.6 g/d) that contained very high levels of genistein (98.44 mg/d) and diadzein (16.51 mg/d). The consumption of nearly 100 mg of genistein from dietary sources would be extreme. What are considered “high soy‒intake diets” in previous studies did not reach such levels. For example, in the Shanghai Breast Cancer Survival Study, the highest quintile of intake was less than 15.41 g soy protein and over 62.68 mg total isoflavones daily.1 The Women’s Healthy Eating and Lifestyle Study considered high consumption to be 16.33 mg to 86.9 mg per day of total isoflavones as the highest intake group.2 The current study may be relevant to isolates of isoflavones (ie, supplements and fortified foods) but dietary amounts, even in high soy consumers, are unlikely to reach the levels studied. 
Another caveat to using this study to inform our general dietary advice is that the study failed to show any difference in primary outcome measures of proliferation (Ki67) or apoptosis (Cas3) between groups. It was only when the intervention group was divided into a subset of women (11 in the soy group vs 23 in placebo) with high circulating genistein (>16 ng/mL) that there were significant differences found in proliferation signatures via microarray. In the original cohort (N=104), there was no difference. 
Of interest, the variability of plasma isoflavones was tremendous between participants in the intervention group, with levels ranging from 0 ng/mL to 400 ng/mL at the time of surgery. This may reflect protocol adherence, but there is likely a variation in metabolism that is also contributing. So, what this study shows is that when 51 women consume nearly 100 mg of genistein per day, 11 of them (20%) have high-circulating levels of genistein. These same 11 women also have tumors with DNA expression patterns reflective of increased cellular proliferation.
When molecular subtypes were assessed, there was a trend for luminal B in the high-circulating genistein subset vs luminal A in the low-genistein subset (P=.06). It is possible that the randomization of samples resulted in this selection bias by chance. It is also possible that the genistein influenced the expression of genes associated with luminal B. If the former is correct, then any implication of genistein influencing expression patterns was skewed by the molecular subtype. If the latter is true, however, it provides a reason to caution against high doses of genistein.
While molecular profiling can inform our understanding of mechanisms, clinical outcome information should trump suppositions based on any mechanistic data. There is little contradiction in the ongoing publications of soy intake and breast cancer recurrence.3 This population is at particularly high risk of breast cancer because of their personal history of treatment for the disease. Even so, trials in this population have shown no increase in breast cancer recurrence in either Asian cohorts or more heterogeneous cohorts in the United States.
For now, the best advice for our patients about soy and breast cancer can be summarized in these 2 statements: 
  1. Evidence suggests that isoflavones in excessive amounts (100 mg/d) are contraindicated and may carry some risk. 
  2. Soy foods—in moderation—have been shown to be neutral or beneficial in all of the studies that tracked soy intake and breast cancer recurrence. 
I sometimes emphasize to my patients that when we talk about soy, we are talking about “just a bean.” There is a lot of opinion and even emotion wrapped in the soy debate. As far as the evidence regarding breast cancer, moderation of the quantity (1-2 servings/d) of preferably organically-grown whole-food forms (edamame) or fermented sources (miso) are the only stipulations I advise. 

Categorized Under


  1. Shu XO, Zheng Y, Cai H, et al. Soy food intake and breast cancer survival. JAMA. 2009;302(22):2437-2443.
  2. Caan BJ, Natarajan L, Parker B, et al. Soy food consumption and breast cancer prognosis. Cancer Epidemiol Biomarkers Prev. 2011;20(5):854-858.
  3. Kaczor T. The effects of soy consumption on breast cancer prognosis. Nat Med J. 2012;4(11). Accessed February 28, 2015.