Functional foods for coronary heart disease risk reduction: a meta-analysis using a multivariate approach

Abstract

It has been difficult to identify the appropriate bioactive substance for the development of new functional foods associated with coronary heart disease, because the results of many clinical studies are contradictory. The objective of this study was to use the multivariate statistical approach known as principal component analysis (PCA) followed by a mixed model to process data obtained from a meta-analysis aimed at evaluating simultaneously the effect of ingestion of 1 of 3 types of bioactive substances (n-3 fatty acids, soluble fibers, and phytosterols) on 1 or more of 4 biomarkers (plasma total cholesterol, triacylglycerol, LDL cholesterol, and HDL cholesterol). Five independent variables (number of patients per study, dose, age, body mass index, and treatment length) and 4 dependent variables (percentage change in blood total cholesterol, LDL, HDL, and triacylglycerol) from 159 studies and substudies were organized into a matrix. The original values were converted to linear correlation units, which resulted in a new matrix. Two principal components were enough to explain 63.73% and 84.27% of the variance in the independent and dependent variables, respectively. Phytosterols and soluble fibers had a hypocholesterolemic effect, whereas n-3 fatty acids lowered triacylglycerol and increased total, LDL, and HDL cholesterol. The PCA and mixed model showed that this behavior was independent of dose, number of patients per study, age, and body mass index but was associated with treatment length. PCA is useful for summarizing available scientific information in examinations of health claims for foods and supplements.