Discussion:
To the best of our knowledge this is the first meta-analysis which reported the effect of soy products based on soy isoflavones and the combination of soy isoflavones plus soy protein on serum CRP levels among postmenopausal women.
Our systematic review contains 23 articles studying the effects of soy isoflavones and soy isoflavones plus soy protein on serum CRP concentration. Meta-analysis contains 19 articles with 33 datasets and its results indicated that soy isoflavones and soy isoflavones plus soy protein could not change CRP concentration among postmenopausal women. Our meta-analysis results confirm the result of a meta-analysis in year 2011(37) which reported the effect of overall soy products, furthermore; this results revealed that new published papers with longer treatment duration, using higher dose of soy isoflavones or soy protein, and more sample size could not change non-significant effect of soy products on serum CRP levels in previous article.
There was no significant change in serum levels of CRP in subgroup analysis based on dose, age, intervention duration, baseline CRP level, sample size, region, quality assessment, publication year, and health status. Dose response analysis revealed no association of higher dose of soy isoflavones with isoflavones effect on CRP levels.
Review articles revealed that some nutrients might change inflammatory mediators (56, 57). Evidence from in vitro (58) and animal studies (59) have indicated that soy products can reduce the activity of nuclear factor-kB as cytokine-induced signal transduction thereby decreases the concentration of pro-inflammatory cytokine. Furthermore, new evidence suggested that IL-6 by activating nuclear factor-kB is able to induce hs-CRP expression (60). Therefore, in participants with health condition which enhances IL-6 levels, soy products by inhibition of IL-6 effect on nuclear factor-kB and CRP expression can reduce CRP levels, but this results were not confirm in our meta-analysis.
Regardless of subgroup analysis, one of the main reasons for non-significant effect of soy isoflavones or the combination of soy isoflavones and soy protein on serum CRP levels might be due to different ability of participants in the metabolization of isoflavones to equol (61). Equol is an isoflavandiol estrogen produced from daidzein by bacterial flora in the intestines (61). Although equal has the superior antioxidant activity in compared with other isoflavones, most people do not have ability to produce equol. In western countries 20-25% of papulation have the capacity to produce equol and in Asian countries about 60% of populations produce equol (62). The result of univariate meta-regression analysis showed there is a significant association between region and soy isoflavones effect on serum CRP levels.
In a study by Acharjee et al., (25) revealed that taking soy products causes no significant changes in blood CRP levels among equol non-producers women, but equol-producers indicated significant reduction. Most articles in our meta-analysis did not report any information regarding equol production in participants and we could not perform subgroup analysis based on equol non-producers and equol-producers.
According to scientific proposal, soy products might be more effective among Asian papulation, but our analysis revealed non-significant changes in CRP levels in subgroup analysis based on region. One possible reason for this result might be participants age in this meta-analysis. Subjects in our article were postmenopausal women and intestinal microflora changes during aging might be a reason for low amount of equal production in intestine, therefore; soy isoflavones fail to exert anti-inflammatory effects. (63)
There was a considerable heterogeneity between included studies even in most subgroup analysis. The probable reasons might be related to discrepancies in the ability of study papulation in metabolization and absorption of isoflavones, participants’ intestinal flora, dietary habit and genetic background of the participants.
Our systematic review and meta-analysis has several limitations. First of all, most articles did not have any information related to equal production among participants, therefore; we did not assess whether equal production could change non-significant effect of soy products on CRP to significant effect among elderly subjects. Secondly, participants’ dietary habit was not reported in any article. Since dietary habit might change intestinal flora, neglecting confounding effect of dietary habit is an important issue. Thirdly, most articles did not have any information regarding drugs used by participants. Fourthly, trials included in this meta-analysis had small or moderate sample size. Small sample size does not have enough statistical power to detect a significant effect in trials and in the meta-analysis. Fifthly, even though changes in fat mass are very important factor in inflammatory mediators levels, most RCTs did not report any information related to body composition changes. Sixthly, we did not include non-English articles.
Our meta-analysis had a number of strengths in comparison with previous meta-analysis. First of all, the large number of studies were included in meta-analysis, therefore; we could conduct subgroup analysis based on dose, age, intervention duration, baseline CRP level, sample size, region, BMI, quality assessment, and publication year. Secondly, we included RCTs which used only soy products in form of supplements or natural soy products and we excluded studies with other food supplements beside soy products. Thirdly, our meta-analysis included 19 trials from different countries; therefore, we might limit the differences in habits and lifestyles in this analysis. Fourthly, we reported the effect of soy isoflavones or the combination of soy isoflavones and soy protein separately with sufficient effect sizes. Fifthly, we did not any limitation on publication time. Finally, Egger’s test revealed no evidence of significant publication bias.
In conclusion according to our results, published RCTs did not provide strong evidence regarding decreasing effect of soy isoflavones or the combination of soy isoflavones plus soy protein on CRP concentration among postmenopausal women. Future studies should report soy effect among equal producers and non-producers and assess gut microbial composition; furthermore, confounding effect of diet should be adjusted.