We evaluated the association of soyfood intake and breast cancer risk in a
population-based case-control study among Chinese women in Shanghai. Includ
ed in the study were 1459 cases and 1556 age-matched controls, with respect
ive response rates of 91.1% and 90.3%. Usual soyfood intake was assessed us
ing a food frequency questionnaire (FFQ). Separate analyses were performed
for all subjects and for the subset who reported no recent change in soyfoo
d intake. The intake levels of soyfoods among women in Shanghai are high, w
ith 96.6% women reporting soyfood consumption at least once a week. A stati
stically non-significant reduced risk (odds ratio (OR) = 0.78 95% CI = 0.52
-1.16) of breast cancer was observed among those who reported eating soyfoo
d at least once a week. Compared to those in the lowest decile intake group
, women in the highest decile intake group had a 30% reduced risk of breast
cancer (OR = 0.66, 95% CI = 0.46-0.95), but no monotonic dose-response rel
ation was observed (P for trend, 0.28). Stratified analyses showed that the
inverse association was restricted primarily among women who had a high bo
dy mass index (BMI), with an adjusted OR of 0.30 (95% Cl = 0.10-0.94) obser
ved for the highest intake group. The reduction in risk was stronger for br
east cancer positive for both oestrogen receptor (ER) and progesterone, rec
eptor (PR) (OR = 0.44, 95% Cl = 0.25-0.78) than those with other ER/PR stat
us. More pronounced inverse associations were observed in analyses among th
ose who reported no recent change in soyfood intake than those conducted in
all subjects. A dose-response relation between soyfood intake and breast c
ancer risk was observed in this subset of women (P for trend, 0.02), with a
n OR of 0.46 (95% CI = 0.28-0.75) for those in the highest decile intake gr
oup. No clear monotonic dose-response relation was found between soyfood in
take and breast cancer risk among regular soy eaters, but nevertheless the
results suggest that regular soyfood consumption may reduce the risk of bre
ast cancer, particularly for those positive for ER and PR; the effect may b
e modified by body mass index. (C) 2001 Cancer Research Campaign.