We found weak positive association between CETP and PLTP activities and mean particle diameter of HDL and other variables related to extra-large HDL particles at baseline. Overall, CETP can be seen as one among the factors affecting reverse cholesterol transport pathway and high CETP activity is reportedly associated with decreased HDL cholesterol levels. CETP inhibitors have recently gained great interest due to their ability to increase HDL cholesterol concentration, but it remains to be seen whether their use will result in functionally protective HDL populations and clinical benefit in large trials. Although the effects of intervention on clinical characteristics and measured 20S-Notoginsenoside-R2 metabolites were modest, we found a strong association between increased fish intake and variables related to large HDL particles. However, as a potential limitation, our study design does not allow for total separation of fish intake from increased intake of whole grain and low insulin response grain products and bilberries, but this was considered in statistical analyses by including the intervention group into the model. In fact, not all subjects in the third tertile with the greatest increase in fish intake were in the HealthyDiet group. The average consumption of fish was 74 grams per day in the third tertile, which is more than 500 g per week. This suggests that the instructed three fish meals per week were not enough for these changes. The mean consumption of fish in the third tertile was more than double of average fish consumption among the Finnish population. We have seen that people who are willing to participate in dietary interventions are more interested in healthy eating than people in average. This phenomenon is also observed when comparing the baseline nutrient intakes in the 4-(Benzyloxy)phenol present study with the respective average intakes in Finland. This is one obvious explanation why it might be difficult to see significant changes in the intervention groups with healthy diet. For the same reason, people in the Control group had to limit their rye bread, fish and berry intake, so that the control diet resembled more an average Finnish diet. For example, rye bread intake in men in the Northern Savo area is more than double compared to corresponding intake in Southern Finland. Besides lipoprotein subclass distribution and detailed information on serum lipids, the NMR platform enables detection of lowmolecular-weight metabolites including e.g. citrate, acetate, urea and amino acids. We did not see changes in these 22 measured metabolites during the intervention. The battery of 22 lowmolecular-weight metabolites was not selected particularly for this dietary intervention study and we did not have preliminary hypothesis related to these metabolites. Furthermore, the power calculation was based on fasting glucose concentration, and it is possible that there was not enough power to see all changes in these secondary outcome variables. Of these metabolites, branched chain amino acids were probably the most interesting ones, because they are related to type 2 diabetes risk and we have earlier found that serum isoleucine decreased after the 12 week consumption of diet high in rye bread and pasta. Despite similar carbohydrate modification of 
the diet in the HealthyDiet and WGED groups of the present study, we could not see any changes in isoleucine. This confirms our earlier conclusion that decreased isoleucine may rather be due to decreased intake of protein rich foods.