Quantitative dependent variables were transformed to base-10 logarithmic scale to account for non-normal distributions. Selected confounding phenotypes was used to examine group related changes during the intervention. Control group was used as a reference group when comparing group differences. Benjamini�CHochberg false discovery rate was used to adjust the metabolomics measures and dietary intake results for multiple comparisons. Adjustment was done by one treatment effect at a time and in clinically relevant sets. FDRadjusted p-value # 0.05 was considered as statistically significant. All these analyses are secondary though aimed to be done in the original study protocol. The primary aim of the study was to examine effects on glucose metabolism. Associations between the changes in fish intake and changes in variables related to large HDL particles were tested using analyses of covariance. Intervention group, age, gender, BMI and fasting glucose were included into models. Analyses were performed using standardized variables. Differences between the fish intake tertiles were tested using one-way ANOVA and Bonferroni post hoc comparisons. Baseline associations were studied applying linear regression model including age, gender, BMI and fasting glucose into models. In the present study the effects of diets containing fatty fish, wholegrain and low insulin response grain products, and bilberries on clinical characteristics, serum metabolites and lipid transfer Gambogic-acid protein activities in persons with impaired glucose metabolism were investigated. We found significant changes in lipid metabolites in the HealthyDiet group reflecting increased polyunsaturation of plasma fatty acids, particularly increase in n-3 PUFAs. Atropine sulfate Furthermore, in the within-the-group comparisons we found an increasing trend in variables related to large HDL particles in the HealthyDiet group, and when analysing this in more detail we found that the increase of fish intake correlated strongly 
with the increased concentration of large HDL particles, larger average diameter of HDL particles and increased concentrations of large HDL lipid components. Significant changes in lipid metabolites in the HealthyDiet group were observed. All changes reflected increased polyunsaturation in plasma fatty acids, especially increase in n-3 PUFAs, while n-6 and n-7 fatty acids decreased. This is consistent with our earlier findings of increased plasma content of long chain polyunsaturated n-3 fatty acids and increase of plasma Tg incorporating the long chain PUFA in the HealthyDiet group. These findings also confirm that different analytical methods, give consistent results for comparable lipid measures. The observed changes confirm the good compliance to the instructions of fish intake, since the subjects in the HealthyDiet group were advised to eat fatty fish at least three times per week, whereas subjects in the Control group where instructed to restrict their fish intake for maximum of 1 fish meal per week. In the WGED group, dietary habits, besides the cereal products, were advised to keep unchanged, which did not cause changes in the measured metabolites. Serum total, LDL and HDL cholesterol levels, serum Tg or serum apoA-1 and apoB100 concentrations did not change during the intervention. We observed that a greater increase of fish intake was associated with increased concentration of large HDL particles, larger average diameter of HDL particles and increased concentrations of lipid components in large HDL particles.