Correspondence

Soy Protein and Serum Lipids

N Engl J Med 1995; 333:1715-1716December 21, 1995

Article

To the Editor:

In reporting the results of a meta-analysis of the effects of soy protein on serum lipid levels (Aug. 3 issue), Anderson et al.1 commented on the position of the American Heart Association (AHA) on this issue. The conclusion of the 1993 AHA Rationale of the Diet–Heart Statement was that although there is good evidence that soy protein can lower cholesterol levels in animal models, the effects in humans, with a few exceptions, are not nearly so convincing.2 Indeed, most studies included in the analysis by Anderson et al. did not show a significant reduction in levels of low-density lipoprotein cholesterol, as was the case with a well-controlled study that was excluded from the meta-analysis but cited in the AHA report.3 Furthermore, the meta-analysis fails to show a significant effect in subjects with plasma cholesterol levels below 200 mg per deciliter, the principal population group to whom the AHA report was directed. Regardless of the validity of the meta-analysis, it is not prudent to use such “averaged” information to make specific dietary recommendations for the public, rather than for people with high serum cholesterol concentrations.

On the other hand, the moderate use of soy protein is consistent with the dietary guidelines of the AHA if soy-based food products are included in a diet that is balanced overall. On the basis of stronger and more extensive evidence than that included in the meta-analysis, these recommendations call for limiting the intake of animal fat, which is associated with animal protein, and consuming more vegetables and fruits. The article by Anderson et al. may appropriately serve to call readers' attention to considering soy protein as a food option. As with other studies of specific foods, however, the AHA is concerned about the potential for medical reports to focus on a single nutrient and in so doing to deflect attention from established, broader-based principles of nutrition.

In addition, more information is needed on the mechanisms of the presumed effects of soy protein on serum lipids. For example, if plant estrogen components of soy products are important, as suggested,1 it will be necessary to understand their overall biologic effect, including the conceivable risks associated with high intake, as well as the variability of soy preparations with respect to the content of such estrogens. The AHA's Nutrition Committee can be expected to develop more specific guidelines for the intake of soy products as information accumulates from research.

Ronald M. Krauss, M.D.
Lawrence Berkeley National Laboratory, Berkeley, CA 94720

Alan Chait, M.D.
University of Washington, Seattle, WA 98195

Neil J. Stone, M.D.
Northwestern University Medical School, Chicago, IL 60611

3 References

1. 1

   Anderson JW, Johnstone BM, Cook-Newell ME. Meta-analysis of the effects of soy protein intake on serum lipids. N Engl J Med 1995;333:276-282
   Full Text | Web of Science | Medline

2. 2

   Chait A, Brunzell JD, Denke MA, et al. Rationale of the diet-heart statement of the American Heart Association: report of the Nutrition Committee. Circulation 1993;88:3008-3029
   Web of Science | Medline

3. 3

   Grundy SM, Abrams JJ. Comparison of actions of soy protein and casein on metabolism of plasma lipoproteins and cholesterol in humans. Am J Clin Nutr 1983;38:245-252
   Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: Krauss et al. note that most studies included in our report1 did not show significant decreases in serum concentrations of low-density lipoprotein cholesterol. Meta-analyses are specifically designed to address clinical questions when there is a disparity between studies.2 Traditional reviews rely heavily on personal judgment and are subject to the biases of the authors.3 Meta-analyses usually focus on single clinical questions.2 A basic objective of the approach is to increase the power of tests of significance by combining the results of different studies and thus to strengthen conclusions about the efficacy of treatment.

Often, single studies are too limited in size or scope to detect significant effects. The pooling process used in meta-analysis weights studies to maximize the precision of estimates of the average effect. Such estimates are an important index of the general efficacy of treatment and can assist in resolving inconsistencies between the results of individual studies.4,5

Although our conclusion that soy protein intake significantly decreases serum cholesterol concentrations is supported by the meta-analysis, trained clinical scientists would probably draw similar conclusions from these data even without statistics. In 34 of 38 controlled clinical trials (89 percent), serum cholesterol concentrations decreased more with soy protein than with the control treatment; in the 4 studies that did not report reductions, the initial serum cholesterol values averaged 185 mg per deciliter. The unweighted net reduction in serum cholesterol — the “average” reduction obtained without weighting for sample size or variance — was 10.0 percent, a figure quite similar to the weighted net reduction of 9.3 percent.

Our observations1 that reductions in serum cholesterol concentrations in patients with soy protein intake are highly correlated with base-line serum cholesterol concentrations have important clinical implications; persons with moderate-to-severe hypercholesterolemia are likely to have substantial benefits. However, even those with serum cholesterol concentrations below 200 mg per deciliter had an estimated decrease in the serum concentration of low-density lipoprotein cholesterol of 7.7 percent.

James W. Anderson, M.D.
Bryan M. Johnstone, Ph.D.
Margaret E. Cook-Newell, M.S., R.D.
University of Kentucky, Lexington, KY 40536-0086

5 References

1. 1

   Anderson JW, Johnstone BM, Cook-Newell ME. Meta-analysis of the effects of soy protein intake on serum lipids. N Engl J Med 1995;333:276-282
   Full Text | Web of Science | Medline

2. 2

   Goodman SN. Have you ever meta-analysis you didn't like? Ann Intern Med 1991;114:244-246
   Web of Science | Medline

3. 3

   Abramson JH. Meta-analysis: a review of pros and cons. Public Health Rev 1990;18:1-47
   Medline

4. 4

   DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177-188
   CrossRef | Medline

5. 5

   National Research Council. Combining information: statistical issues and opportunities for research. Washington, D.C.: National Academy Press, 1992.
   

Citing Articles (2)

Citing Articles

1. 1

   Hyang-Im Back, Soo-Ran Kim, Ji-Ae Yang, Min-Gul Kim, Soo-Wan Chae, Youn-Soo Cha. (2011) Effects of Chungkookjang Supplementation on Obesity and Atherosclerotic Indices in Overweight/Obese Subjects: A 12-Week, Randomized, Double-Blind, Placebo-Controlled Clinical Trial. Journal of Medicinal Food 14:5, 532-537
   CrossRef

2. 2

   Naoyoshi Takatsuka, Chisato Nagata, Yoko Kurisu, Shizuyo Inaba, Norito Kawakami, Hiroyuki Shimizu. (2000) Hypocholesterolemic Effect of Soymilk Supplementation with Usual Diet in Premenopausal Normolipidemic Japanese Women. Preventive Medicine 31:4, 308-314
   CrossRef