Soy isoflavones are naturally occurring oestrogen-like compounds, and supplements are currently marketed as a way of reducing symptoms of the menopause and offer an alternative to hormone replacement therapy.
Conflicting reports however have clouded the picture about the beneficial effects of soy isoflavones, with some studies indicating that breast cancer cells in mice were stimulated by the isoflavones. Population studies have shown that women with a high-soy diet generally have lower rates of breast cancer.
"Because soy foods and soy supplements are widely used, we conducted this first true meta-analysis to understand what role soy foods might have on breast cancer risk," said Robert Clarke from Georgetown's Lombardi Comprehensive Cancer Center.
The meta-analysis, published in the April issue of the Journal of the National Cancer Institute (Vol. 98, pp. 459-471), pooled 18 epidemiological studies (12 case-control and six cohort or nested case-control) dating from 1978 to 2004.
Since the original studies used varying measures of soy intake, the scientists converted all the measures to an estimate of soy protein per day, an exercise that required multiple assumptions.
A pooled analysis of all women reported that high soy intake was linked to a "modest" 14 percent reduction in breast cancer risk. Subgroup analysis of Asian women only did not produce a significant association. However, high soy intake for pre-menopausal women was associated with a 30 per cent reduction in breast cancer risk.
When the researchers examined the relationship between risk and estimated soy protein exposure, a small benefit was seen only in the pre-menopausal group.
"While the data were too inconsistent to recommend soy as a breast cancer preventive, there is no evidence to suggest that consumption of soy foods in amounts consistent with an Asian diet is detrimental to breast health," wrote Bruce Trock from John Hopkins School of Medicine.
However, in a release from John Hopkins Medical Institutes, Brock said: "At this point, women should not be taking high-dose soy supplements, especially those who are breast cancer survivors and women at increased risk for the disease. We don't have long-term data on the effects of these supplements, and there is some evidence that they could be harmful."
In an accompanying editorial by Maria Martinez, Cynthia Thomson from the University of Arizona, and Stephanie Smith-Warner from Harvard School of Public Health, the meta-analysis was commended for tackling the "complex and controversial, yet highly important, issue."
The editorial points out that, due to the variability of the original studies, such a meta- analysis posed "remarkable challenges." The heterogeneity of the of the originals however "limits the ability to interpret the findings."
Martinez, Thomson and Smith-Warner called for future studies to report links between both soy foods and nutrients such as protein and isoflavones to help make the data comparable with other studies.
Dr Daniel Fabricant, vice president of scientific affairs for the National Nutritional Foods Association (NNFA), an industry association, told NutraIngredients-USA.com that the new meta-analysis was "irresponsible and misleading."
"It appears that the authors of this article do not have a clear understanding of what is actually in a soy isoflavone supplement or how the isoflavones are metabolized. The recommendation of the authors that "women should not be taking high-dose soy supplements" is not scientifically based on clinical data," said Fabricant.
Dr Fabricant said that the meta-analysis also cast a cloud over the 187m Americans who regularly take dietary supplements as a part of their healthy lifestyle.
"It is unfortunate that those who are ignorant regarding soy and the components of soy (isoflavones), dietary supplements, and botanicals, feel the need to comment on a meta-analysis they performed, in which even the editorial board of the journal states that "assumptions were made", without doing any basic or clinical research of their own on the subject," said Fabricant.