Review: Soy protein intake decreases total and LDL cholesterol and triglyceride levels
ACP J Club. 1996 Mar-April;124:41. doi:10.7326/ACPJC-1996-124-2-041
Anderson JW, Johnstone BM, Cook-Newell ME. Meta-analysis of the effects of soy protein intake on serum lipids. N Engl J Med. 1995 Aug 3; 333:276-82. [PubMed ID: 7596371]
To determine the relation between soy protein consumption and serum lipid levels.
Studies were identified by searching the literature for articles that evaluated the effects of soy protein on serum cholesterol level.
Studies were selected if they had either a crossover or a parallel design, were controlled, used isolated soy protein or textured soy protein, and provided baseline values so that changes for each study group could be calculated. Studies were excluded if there was no control group, if several sources of vegetable protein were used, or if whole soybeans rather than soy protein were used.
Data were extracted on participant age, serum lipid levels, types and amount of soy protein used, type of diet, and the similarity of the control diet and the soy-containing diet.
A random-effects model was used to quantify the average effects of soy protein intake on serum lipids. Hierarchical mixed-effects regression models were used to predict variation as a function of the characteristic of the studies. 29 articles that included the findings of 38 trials met the selection criteria. In most of the studies, the intake of energy, fat, saturated fat, and cholesterol was similar between the control and soy-containing diets. Diets containing soy protein resulted in the following net changes (change during the soy diet minus change during the control diet): a decrease of 0.60 mmol/L (95% CI 0.35 to 0.85 mmol/L) for total cholesterol, a decrease of 0.56 mmol/L (CI 0.30 to 0.82 mmol/L) for low-density lipoprotein (LDL) cholesterol, and a decrease of 0.15 mmol/L (CI 0.003 to 0.29 mmol/L) for triglycerides. A trend existed toward increased high-density lipoprotein (HDL) cholesterol levels in patients who received soy protein. The initial serum cholesterol level was the only predictor of the change in the serum cholesterol and LDL cholesterol levels (P < 0.001).
The consumption of soy protein reduces serum cholesterol, low-density lipoprotein cholesterol, and triglyceride levels. The decreases in serum cholesterol and low-density lipoprotein cholesterol levels are related to initial serum cholesterol levels.
Source of funding: In part, Protein Technologies International.
For article reprint: Dr. J.W. Anderson, Medical Service, 111C, Veterans Affairs Medical Center, Leestown Road, Lexington, KY 40511, USA. FAX 859-233-3832.
Should we eat 16 ounces of tofu or drink a quart of soy milk per day? The meta-analysis by Anderson and colleagues suggests so. A significant reduction in serum LDL cholesterol levels was found in 6 of the 31 included studies, all of which enrolled patients with hypercholesterolemia; the pooled result, however, also shows modest benefit.
The included studies were mostly fixed rather than random crossover designs (including the 2 studies by Verillo and colleagues , which were incorrectly designated as randomized crossover designs). These studies are subject to a "regression to the mean" effect (2); that is, participants were selected for their apparently high cholesterol levels, which then "fell" during the soy protein phase.
Another problem with this review is a possible publication bias. The search strategy was not specified. The trials included were small and the intervention was simple; therefore, it is possible that other small "negative" trials exist that were not submitted for publication. Both of these biases suggest that this meta-analysis is likely to overstate any benefit of soy protein.
Ideally, we would have an adequately powered and properly designed randomized trial to answer the health care question regarding the effect of soy protein on lipid profile and clinically important outcomes such as cardiovascular and cerebrovascular events. Until then, I will continue to enjoy a small amount of tofu and will remain somewhat convinced that it helps my lipid profile. For my patients, however, I will continue to recommend the dietary changes shown to decrease all-cause mortality: fruits, vegetables, grains, and legumes (3, 4).
Paul Glasziou, MD
Harvard School of Public HealthBoston, Massachusetts, USA