Plasma isoflavones and fibrocystic breast conditions and breast cancer among women in Shanghai, China

Abstract

Background: Proliferative benign breast conditions are associated with elevated risk of breast cancer, whereas nonproliferative conditions are not strongly associated with risk. Factors acting before onset of hyperplasia might be associated with both benign conditions and breast cancer, whereas those on the proliferative disease-to-cancer pathway would be associated only with cancer. Soy isoflavone exposure may influence breast cancer risk, but little is known of its association with benign conditions. Materials and Methods: We examined possible relationships between plasma genistein and daidzein concentrations and risk of breast disease in women, in a breast self-examination trial in Shanghai, China, diagnosed with breast cancer (n = 196) or a benign breast condition (n = 304), and 1,002 age-matched controls with no known breast disease. Benign conditions were classified as nonproliferative (n = 131) or proliferative with or without atypia (n = 173). Results: Isoflavone concentrations were inversely associated with risk of nonproliferative and proliferative benign fibrocystic conditions, as well as with breast cancer, both with and without concomitant proliferative changes in ipsilateral noncancerous mammary epithelium (Ptrend < 0.01 for all comparisons with controls). Women in the highest quartile of plasma genistein (>76.95 ng/mL) were less likely to have breast cancer (odds ratio, 0.26; 95% confidence interval, 0.13-0.50) or benign conditions (odds ratio, 0.40; 95% confidence interval, 0.23-0.70) compared with women in the lowest quartile (<9.42 ng/mL). Observed risks for breast cancer with and without surrounding proliferative changes were not different, respectively, from observed risks for benign proliferative and nonproliferative conditions alone. Conclusion: Isoflavone exposure was inversely associated with fibrocystic breast conditions and breast cancer, and the results suggest that effects on cancer risk occur early in carcinogenesis. (Cancer Epidemiol Biomarkers Prev 2007;16(12):2579–86)