1 INTRODUCTION

The evidence suggests that coronary heart disease (CHD) is the most important cause of death among women aged 60 years old and older.1 Because of low level of oestrogen, which has vascular protective effects, the incidence of CHD rises in post-menopausal women.2 Scientists suggested that hormone replacement therapy might increase the risk of cancer, cardiovascular disease (CVD), and dementia; therefore, a lot of alternative factors have been assessed for their potential for reducing the morbidity and mortality of CVD among post-menopausal women.3

Soy food intake in Chinese women might be associated with a reduction in a number of CHD risk factors.4, 5 Daily intake of soy products has decreased the serum concentration of cholesterol in Japanese adults and has been associated with a reduction in mortality caused by CHD in Japanese women.6, 7 Scientists also suggested that substituting the soy food with non-soy protein in a therapeutic lifestyle changes (TLC) diet has reduced systolic and diastolic blood pressure in normotensive, prehypertensive and hypertensive women. They also proposed that mentioned beneficial effects may cause the inhibition of inflammation and oxidation with soy protein and soy isoflavones.8, 9

Tumour necrosis factor-alpha (TNF-α), C-reactive protein (CRP), and interleukin-6 (IL-6) are the most important inflammatory mediators. Lymphocytes and macrophages secrete IL-6 and TNF-α10, 11 and can stimulate the liver cells to secrete CRP into the blood circulation.12 The enhancement in these inflammatory mediators besides oestrogen reduction in post-menopausal women leads to chronic diseases such as CHD, dyslipidemia and blood pressure.13, 14

Soy isoflavones might reduce inflammation by inhibiting cytokine-induced signal transduction in immune cells and suppressing cell-mediated inflammatory responses.15, 16 Soy isoflavones might also conduct anti-inflammatory effects through antioxidant activities.17 Therefore, scientists have considered soy isoflavones as cardioprotective agents in post-menopausal women because of their anti-inflammatory effects. Besides isoflavones, soy peptides may also have anti-inflammatory effects. Some studies have indicated that the reduction in animal protein intake and the increase of soy protein intake might reduce circulating levels of the inflammatory biomarkers.18, 19

There are a lot of randomized clinical trials (RCTs) regarding soy consumption and inflammatory markers among post-menopausal women, but their results are far from conclusive. Some RCTs revealed the effectiveness of soy products on pro-inflammatory mediators,20-24 whereas others did not.25-31 In one meta-analysis in 2011, the effect of soy isoflavones on serum concentration of CRP among post-menopausal women was examined. Its findings showed a non-significant effect of soy isoflavones on CRP.32 What is more, the effect of soy products on the serum concentration of IL-6 and TNF-α among post-menopausal women was not reported in that study. Another meta-analysis, also, assessed the effect of soy products on inflammatory mediators, but they did not show the effect of these products among post-menopausal women.33 Therefore, a meta-analysis is necessary to summarize the overall effects of the combination of soy isoflavones and soy protein on IL-6 and TNF-α among this population.

2 METHODS AND MATERIALS

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA, Appendix 1) guidelines were used for conducting this systematic review and meta-analysis. To find the eligible English-language articles, we searched for all RCTs in the databases such as Cochrane Library, clinicaltrials.gov, Web of Science, PubMed and Scopus up to December 2020. We searched these databases with the following main keywords: “Interleukin-6”, “Tumour Necrosis Factor-alpha”, “Soy Foods”, “Soybean proteins”, “Isoflavones”, “Genistein”, “Equol” and “Phytoestrogens”. All keywords for searching these databases are presented in Table S1. Because a few articles might be missed when searching with only TNF-α and IL-6 in databases, we searched with keywords related to the most important inflammatory mediators. Additional eligible studies were found by manual searching the reference list of all published reviews and meta-analyses of soy products as well as searching the included original articles. The search strategy was designed by using Boolean operators, asterisks, quotation marks and parentheses. All found articles by a systematic search were imported into EndNote software (reference manager software, version X6), then two reviewers (M.H. and A.G.) performed a complete review of both titles and abstracts. Any disagreement was resolved after a discussion session. This systematic search was not limited to the time of publication. The protocol was registered with PROSPERO (No. CRD42020166053).

2.1 Inclusion criteria

The articles that met PICOS (population, intervention, comparison, outcome, study design) framework were included in this study: (a) population: healthy or unhealthy adults; (b) intervention: the combination of soy isoflavones and soy protein; (c) comparison: control group; (d) outcomes: assessing IL-6 and TNF-α; (e) study design: parallel or cross-over RCTs.

2.2 Exclusion criteria

Articles that met the following criteria were excluded from the study including: (a) those that did not report any information concerning the serum concentration of IL-6 and TNF-α at baseline or after intervention in comparison or intervention groups and any data for computing them; (b) those that presented the serum level of IL-6 and TNF-α in figures; (c) the articles with intervention <1 week; (d) those that did not have any comparison group; (e) the articles that took other food supplements besides soy products in the intervention group; (f) the articles in which the participants took only soy isoflavones or soy protein, but not the combination of soy isoflavones and soy protein; (g) those that did not report the dose of soy isoflavones or soy protein with no presented information for estimating them; (h) studies on participants with acute inflammatory disease; and (i) the articles were not written in English.

2.3 Quality assessment

Two reviewers (M.H. and A.G.) independently scored RCTs using Cochrane Collaboration’s tool,34 according to the following criteria: (a) randomizing in sequence generation; (b) concealing the allocation; (c) blinding of participants and personnel; (d) reporting selectively; (e) blinding of outcome assessors; and (f) presenting outcome data incompletely. For each item, a judgment (‘unclear risk of bias’, ‘high risk of bias’ or ‘low risk of bias’) was considered. Moreover, the included RCTs were scored as indicating ‘good’ if at least three items were categorized by low risk of bias, ‘fair’ if at least two items met the low risk of bias and ‘weak’ if less than two items met the low risk of bias.

2.4 Data extraction

Screening, selecting, and assessing the methodological quality of studies were performed independently by two reviewers (M.H. and A.G.) Selected articles were carefully read to extract the following data: author’s first name, publication year, study design, sample size in each group and total, participants’ age, health status and body mass index (BMI), study duration, geographical region, soy isoflavones and protein dose, as well as serum concentration of IL-6 and TNF-α before and after the intervention. Any disagreement was resolved after a discussion session. Regarding IL-6 and TNF-α, all units were converted into pg/mL. RCTs that were conducted on more than one comparison or intervention group were included in this systematic review and meta-analysis as separated articles. Clarification of any unclear information was done by sending an email to the corresponding author.

2.5 Data synthesis and statistical analysis

Mean differences (MDs) and their standard deviations (SDs) as a summary statistic were used in this meta-analysis and calculated by subtracting baseline values from post-intervention values of included articles.

The effect sizes of soy isoflavones plus soy protein on IL-6 and TNF-α were calculated using MDs and their SDs of both intervention and comparison groups. Hozo’s method was applied to estimate the mean using the values of range or median when the mean was not reported; moreover, some articles in which only the standard error (SE) was reported, the SD was calculated by multiplying SE in the square root of the sample size.35 DerSimonian and Laird random-effects model was used to estimate the summary of overall effect and its heterogeneity; furthermore, Cochran’s Q test and I2 statistics were used to determine the statistical heterogeneity of the intervention effects.36 A significant heterogeneity was revealed by P-value ≤0.10 and value ≥50% for Cochran’s Q test and I2 statistics, respectively.37

The sources of heterogeneity were found using subgroup analysis based on intervention duration, study design, soy isoflavones dose, soy protein dose, sample size, geographical region, publication year, quality assessment, participants’ age, BMI and health status. Eggar’s weighted regression tests were used to assess the publication bias in this meta-analysis.38, 39 Sensitivity analysis was used to evaluate the effect of each study on the overall effect size. Statistical analyses were performed using STATA 15 software (Stata Corp). The statistical significance was set at 0.05 and all calculated effect sizes were presented with 95% confidence intervals (CIs).

3 RESULTS

Totally, 4387 relevant articles were identified by systematic search in databases. After removing duplicate and irrelevant articles, which were examined by reading the titles and the abstracts, 38 articles remained for reading the full text. At full text level concerning inclusion and exclusion criteria, 29 articles were excluded for the following reasons: (a) the articles in which participants took other supplements besides soy intervention (n = 6); (b) the articles in which participants were not post-menopausal women (n = 11); (c) studies conducted on patients with acute inflammatory disease (n = 2); (d) those in which participants took only soy isoflavones or soy protein as intervention (n = 8); (e) the articles that were not written in English (n = 1); and (f) those that did not have a comparison group (n = 1). Therefore, nine RCTs remained eligible for our systematic review.9, 21, 27, 40-45 Because one trial did not report enough data for meta-analysis9 and one trial had a very large effect size in comparison with other trials,42 seven articles met the inclusion criteria21,27,40,41,43–45 (Figure 1). The effect of soy isoflavones plus soy protein on IL-6 and TNF-α was assessed in eight21, 27, 40-45 and seven21, 23, 27, 40-43 trials, respectively. There was more than one intervention group in two studies21, 40 and more than one comparison group in one article41; moreover, one trial was conducted on two different populations9; therefore, these trials were reported as separate studies in this systematic review and more than one effect size was calculated in this meta-analysis. In total, 10 data points were included in this meta-analysis (Table 1).

Flowchart of study selection process.

TABLE 1. Randomized controlled trial studies included in the systematic review and meta-analysis

Code

Author (year)

(country)

Subjects RCT Intervention Placebo Duration (week) Results

1.1

Azadbakht

2007

Iran21

Post-menopausal women with the metabolic

Syndrome

n = 42

Randomized cross-over clinical trial 11.25 g/day protein with 84 mg/day isoflavones Red meat 8

IL-6 did not change significantly

TNF-α decreased significantly in intervention group in compared with placebo group

1.2

Azadbakht

2007

Iran21

Post-menopausal women with the metabolic

Syndrome

n = 42

Randomized cross-over clinical trial 15 g/day protein with 102 mg/day isoflavones Red meat 8

IL-6 did not change significantly

TNF-α decreased significantly in intervention group in compared with placebo group

2

Beavers

2009

USA44

Healthy post-menopausal women

n = 31

Single-blind, randomized, controlled trial 18 mg/day soy protein with 90 mg/day isoflavones Dairy milk 4 IL-6 did not change significantly

3

Charles

2009

USA43

Healthy post-menopausal women

n = 75

Randomized, double-blind, placebo-controlled trial 20 g/day soy protein with 160 mg/day of total isoflavones Milk protein 12 IL-6 and TNF-α did not change significantly

4

Christie

2010

England27

Obese post-menopausal White and African American women

n = 33

Randomized, double-blind, controlled trial 20 g/day soy protein with 160 mg/day isoflavones Casein 12

IL-6 decreased significantly in intervention group in compared with placebo group

No significant changes were seen in TNF-α between the two groups

5.1

Jenkins

2002

Canada40

Post-menopausal women

n = 18

Randomized crossover trial 52 g/day soy protein with 10 mg/day isoflavones Regular diet 4 IL-6 and TNF-α did not change significantly

5.2

Jenkins

2002

Canada40

Post-menopausal women

n = 18

Randomized crossover trial 50 g/day soy protein with 73 mg/day isoflavones Regular diet 4

IL-6 increased significantly in soy food group in compared with control group

No significant changes were seen in TNF-α between the two groups

6

Mangano

2013

USA45

Healthy older women

n = 47

Randomized, double-blind, placebo-control, clinical trial 18 g/day soy protein with 105 mg/day isoflavone Maltodextrin and animal protein 48 IL-6 did not change significantly

7.1

Nadadur

2016

USA41

Healthy post-menopausal

women

n = 37

Single blind randomized clinical trial 15 g/day soy protein with 50 mg/day soy isoflavones Very low fat diet 8 IL-6 and TNF-α did not change significantly

7.2

Nadadur

2016

USA41

Healthy post-menopausal

women

n = 37

Single blind randomized clinical trial 15 g/day soy protein with 50 mg/day soy isoflavones Control diet 8 IL-6 and TNF-α did not change significantly

8.1

Nasca

2008

USA9

Healthy post-menopausal women

n = 48

Randomized, placebo-controlled, crossover 25 g/day soy protein with 101 mg/day Isoflavones Control diet 8 IL-6 did not change significantly

8.2

Nasca2008

USA9

Hypertension post-menopausal women

n = 12

Randomized, placebo-controlled, crossover 25 g/day soy protein with 101 mg/day isoflavones Control diet 8 IL-6 did not change significantly

9

Ryan-Borchers

2006

USA42

Healthy post-menopausal women

n = 52

Randomized, double-blind, placebo-controlled trial 18 g/day soy protein with 71.6 mg/day isoflavones Cow milk 16 No significant changes were seen in TNF-α between the two groups Abbreviations: IL-6, interlukin-6; N/M, not mention; RCT, randomized clinical trial; TNF-α, tumour necrosis factor-α; Φ, median (IQR).

The dose of soy isoflavones and soy protein ranged from 10 to 160 mg/day and from 11.25 to 52 g/day, respectively. It took 4 weeks for treatment in two studies,40, 44 8 weeks in three studies,9, 21, 41 12 weeks in two studies,27, 43 16 weeks in one study42 and 48 weeks in one study.45 Of nine trials, three trials9, 21, 40 and six trials27, 41-45 used the cross-over design and parallel design, respectively.

3.1 Risk of bias assessment

Of the nine RCTs included in this systematic review, four RCTs were scored as ‘good’,27, 42-44 two RCTs as ‘fair’,21, 45 and three RCTs as ‘weak’.9, 40, 41 Blinding of participants and outcome assessment were the source of bias in four9, 21, 40, 44 and three RCTs,9, 40, 41 respectively. Regarding allocation concealment, three RCTs were rated at high risk of bias.9, 40, 41 Bias associated with selective reporting and incomplete outcome data seemed in four9, 27, 40, 45 and three RCTs,41, 43, 45 respectively. More details of the quality assessment of included RCTs are presented in Table 2.

TABLE 2. Quality of bias assessment of the included studies according to the Cochrane guidelines Author name, year of publication, references Random sequence generation Allocation concealment Blinding of participants and personnel Blinding of outcome assessment Incomplete outcome data Selective reporting Overall quality Azadbakht21 U U H U L L F Beavers44 U U H L L L G Charles43 L L L L H U G Christie27 L L L U L H G Jenkins40 U H H H L H W Mangano45 L U L U H H F Nadadur41 U H U H H U W Nasca9 U H H H L H W Ryan-Borchers42 L U L U L L G Abbreviations: G, good; H, high risk of bias; L, low risk of bias; U, unclear risk of bias. F, fair; W, weak. 3.2 Findings of this meta-analysis

Seven RCTs involving 10 comparisons assessed the effect of soy isoflavones plus soy protein on serum IL-6 levels (Figure 2A). The intake of soy isoflavones plus soy protein was not associated with significant change of IL-6 levels compared to the comparison group (WMD = 0.07 pg/mL; 95% CI = −0.03, 0.17 pg/mL; P = 0.190). A significant heterogeneity was observed in included studies (I2 = 79%, P < 0.001). The results of subgroup analysis indicated that the serum concentration of IL-6 increased after taking soy isoflavones plus soy protein in studies with soy isoflavone dose ≤87 mg/day (WMD = 0.22 pg/mL; CI = 0.03, 0.4 pg/mL; P = 0.023), cross-over design (WMD = 0.16 pg/mL; 95% CI = 0.03, 0.29 pg/mL; P = 0.019), weak quality (WMD = 0.28 pg/mL; CI = 0.16, 0.4 pg/mL; P = <0.001), and studies on at-risk or sick participants (WMD = 0.14 pg/mL; CI = 0.02, 0.27 pg/mL; P = 0.028; Table 3).

A, Forest plot of the effect of soy isoflavones plus soy protein consumption on serum interleukin-6 (IL-6) concentrations. B, Forest plot of the effect of soy isoflavones plus soy protein consumption on serum tumour necrosis factor-α (TNF-α) concentrations

TABLE 3. Subgroup analyses for studies evaluating the effect of soy isoflavones plus soy protein on serum IL-6 Subgroup No. of trial Change in IL-6 (95% CI) P-value I2 (%) Pheterogeneity Total – 10 0.07 (−0.03, 0.17) 0.190 79 <0.001 Soy isoflavones dose ≤87 mg/d 5 0.22 (0.03, 0.4) 0.023 82.9 <0.001 >87 mg/d 5 −0.1 (−0.28, 0.08) 0.270