Meta-analysis of 46 FDA Identified Studies on Lipid Markers

Background: The FDA in their re-evaluation of the scientific evidence for a relationship between soy protein and reduced risk of coronary heart disease (CHD), has revised the data included in the 1999 soy protein and CHD health claim, as well as studies published after this authorized health claim and concluded that a heart health claim is no longer justified based on more recent studies. The 46 identified studies have not been meta-analyzed.

Need for proposed research: To meta-analyse the 46 identified studies to provide the highest level of evidence to aid in the final decision of the soy protein and reduced risk of CHD health claim.

Objective: To provide evidence-based guidance for publich health policy, health claims, and nutrition guidelines relating to soy, we will conduct a meta-analysis of the 46 FDA identified studies of the effect of soy protein on surrogate markers of cardiovascular disease.

Design: The conduct of the meta-analysis will follow the Cochrane handbook for Systematic Reviews of interventions. The reporting will follow the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines.

Data sources and study selection: We will include the 46 studies pre-selected by the FDA in the Proposed Rule, Soy Protein and Coronary Heart Disease (Docket No. FDA-2017-N-0763).

Data extraction: Two independent investigators will extract study information (study design, sample size, subject characteristics, soy form, dose, comparator, follow-up, background diet, outcome data).

Outcomes: Effects of soy protein intake on LDL-C and TC.

Data synthesis: Mean differences will be pooled using the generic inverse variance method with random-effects model even in the absence of statistically significance between-study heterogeneity, as they yield more conservative summary effect estimates in the presence of residual heterogeneity. Paired analysis will be applied fro crossover trials. Heterogeneity will e assessed by the Cochran Q statistic and quantified by the I2 statistic. To explore sources of heterogeneity, we will conduct sensitivity analysis, in which each study is systematically removed. We will also explore sources of heterogeneity by a priori subgroup analyses by baseline LDL-C levels (<135 mg/dl vs >=135mg/dl), soy dose (<25 g/d vs >= 25 g/d), soy protein food (Isolated soy protein, soy food, soy milk, etc), comparator (animal protein, dairy protein, meat protein), follow-up (median), and categories given by the FDA (baseline cholesterol level [desirable or borderline, high, or normal, borderline, and high], how soy was provided in the studies [feeding or substitution, added to diet], by a mixture of type of soy, how it was provided and baseline cholesterol level [1)Desirable or borderline cholesterol level, added ISP; 2) Desirable or borderline cholesterol level, feeding or substitution studies with ISP; 3) Desirable or borderline cholesterol level, added soy foods; 4)Desirable or borderline cholesterol level, feeding or substitution studies with soy foods; 5) Normal, borderline, and high cholesterol levels, feeding or substitution studies with ISP; 6) High cholesterol level, added ISP; 7) High cholesterol level, feeding or substitution studies with ISP; 8) High cholesterol level, added soy foods; and 9) High cholesterol level, feeding soy foods]), and risk of bias. Meta-regression analysis will assess the significance. Publication bias will be assess by visual inspection of funnel plots and by the use of Egger and Begg's tests. If publication bias is suspected, then we will attempt to adjust for funnel plot asymmetry by imputing the missing study data using the Duval and Tweedie trim and fill method. Dose-reponse analyses will be conducted using the metareg and mkspline commands.

Knowledge translation plan: The results will be disseminated in two stages, first, by the Soy Nutrition Institute in a response to the Proposed Rule, Soy Protein and Coronary Heart Disease, and by a publication in a high impact factor journal.

Significance: The meta-analysis will aid in knowledge translation related to the role of soy protein intake and surrogate markers of cardiovascular disease (LDL-C and TC), strengthening the evidence-based for guidelines and health claims, and improving health outcomes by educating healthcare providers and patients.