Meta-analysis of the effects of soy protein containing isoflavones on the lipid profile.

BACKGROUND: Convincing evidence shows that soy protein intake has beneficial effects on lipid changes, but it is unclear which components of soy protein are responsible.

OBJECTIVE: We conducted a meta-analysis to identify and quantify the effects of soy protein containing isoflavones on the lipid profile.

DESIGN: Twenty-three eligible randomized controlled trials published from 1995 to 2002 were identified from the PUBMED database (National Library of Medicine, Bethesda, MD). Weighted mean effect sizes were calculated for net changes in serum lipid concentrations by using fixed-effect or random-effect models. Pre-specified subgroup analyses were performed to explore the influence of covariates on net lipid change.

RESULTS: Soy protein with isoflavones intact was associated with significant decreases in serum total cholesterol (by 0.22 mmol/L, or 3.77%), LDL cholesterol (by 0.21 mmol/L, or 5.25%), and triacylglycerols (by 0.10 mmol/L, or 7.27%) and significant increases in serum HDL cholesterol (by 0.04 mmol/L, or 3.03%). The reductions in total and LDL cholesterol were larger in men than in women. Initial total cholesterol concentrations had a powerful effect on changes in total and HDL cholesterol, especially in subjects with hypercholesterolemia. Studies with intakes >80 mg showed better effects on the lipid profile. The strongest lowering effects of soy protein containing isoflavones on total cholesterol, LDL cholesterol, and triacylglycerol occurred within the short initial period of intervention, whereas improvements in HDL cholesterol were only observed in studies of >12 wk duration. Tablets containing extracted soy isoflavones did not have a significant effect on total cholesterol reduction.

CONCLUSIONS: Soy protein containing isoflavones significantly reduced serum total cholesterol, LDL cholesterol, and triacylglycerol and significantly increased HDL cholesterol, but the changes were related to the level and duration of intake and the sex and initial serum lipid concentrations of the subjects.