Dietary patterns, biomarkers of atherosclerosis, cardiovascular and all-cause mortality

Summary belonging to the thesis entitled ‘Dietary patterns, biomarkers of atherosclerosis, cardiovascular and all-cause mortality’ The long history of epidemiologic studies on diet and cardiovascular disease (CVD) has traditionally relied on analysis of specific nutrients or foods. Dietary patterns are multiple dietary components operationalized as a single exposure; they reflect the entire diet. In general, two methods are used to define dietary patterns: 1) theoretically, or a priori, defined dietary scores and 2) empirically, or a posteriori, derived dietary patterns. A priori dietary scores were developed to assess diet quality based on adherence to dietary patterns or recommendations. An example of an ‘a posteriori’ approach is factor analysis (e.g. principal components analysis (PCA)). Factor analysis reduces data into patterns based upon intercorrelations between nutrients or foods. The aim of this thesis was to create, examine and compare several dietary patterns and indices and assess these in relation to both early stage markers of CVD (markers of endothelial function and oxidative stress) and to mortality from CVD and all-causes. In chapter 2 we described the creation of the A Priori Diet Quality Score, representing overall diet quality in the Coronary Artery Risk Development in Young Adults (CARDIA) study. The CARDIA study included 5115 black and white men and women, aged 18-30 at baseline (1985-86). Diet was assessed diet at baseline, year 7(1992-93) and 20 (2005-06) examinations. The A Priori Diet Quality Score summed 46 food groups rated by investigators as positive or negative on the basis of hypothesized health effects. In 2652 participants with 3 diet assessments, the mean (±SD) A Priori Diet Quality Score increased from 64.1± 13.0 at year 0 to 71.1 ± 12.6 at year 20, which was primarily attributable to increased age. However, the secular trend, which was estimated from differences of dietary quality scores across time at a fixed age (age matched time trend), decreased. The diet score was higher in whites than in blacks and in women than in men and increased with education, but demographic gaps in the score narrowed over 20 y. Consumption of positively rated food groups tended to increase and negatively rated food groups tended to decrease, and were similar in direction across demographic groups. In chapter 3 we used the ‘A Priori Diet Quality Score’ and two dietary patterns derived using principal components analysis (PCA) the ‘Fruit and Vegetables’ dietary pattern and the ‘Meat’ dietary pattern in the CARDIA study. We studied prospective associations of the ‘A Priori Diet Quality Score’, the ‘Fruit and Vegetables’ dietary pattern and the ‘Meat’ dietary pattern with cellular adhesion molecules (CAMs). The ‘Fruit and Vegetables’ dietary pattern was characterized by high intakes of fruit, vegetables, and whole grains and the ‘Meat’ dietary pattern by high intakes of red meat, refined grain, and butter. The ‘A Priori Diet Quality Score’ was related to all CAMs. The ‘Fruit and Vegetables’ dietary pattern was related to E-selectin and sICAM-1 but not to P-selectin and VCAM. The ‘Meat’ dietary pattern was related to all CAMs except VCAM. Strongest associations were for the ‘Meat’ dietary pattern with E-selectin (effect size 28% of an SD (+3.9/13.7 ng/mL)) and P-selectin (effect size 37% of an SD (+4.1/11.2 ng/mL)) and the ‘A Priori Diet Quality Score’ with sICAM-1 (effect size 34% of an SD (-15.1/44.7 ng/mL)) and VCAM (effect size of 26% of an SD (-45.1/170.3 ng/mL)). Chapter 4 described prospective associations of the A Priori Diet Quality Score, ‘Fruit and Vegetables’ dietary pattern and ‘Meat’ dietary pattern and a plasma biomarker of lipid peroxidation, F2-isoprostanes also in the CARDIA study. We estimated associations between each dietary pattern and plasma F2-isoprostanes cross-sectionally (at year 20, n=2736) and prospectively (year 0/7 average diet and year 15/20 average F2-isoprostanes, n=2718). In the cross-sectional analysis, the A Priori Diet Quality Score and the ‘Fruit and Vegetables’ dietary pattern were inversely, and the ‘Meat’ dietary pattern was positively, associated with F2-isoprostanes (all p values In chapter 5 we described a food classification system derived from the Food-based Dietary Guidelines in the Netherlands that can be used to systematically and objectively classify foods in relation to their effects on health. Classification criteria for each food group were developed based on presumed positive, neutral or negative effects on chronic diseases of five nutrients: four that likely increase (saturated fatty acids, mono-trans unsaturated fatty acids, sodium, and added sugar) and one that likely decreases (dietary fiber) the risk of chronic diseases. This classification system also provided a framework to create food-based dietary scores for epidemiologic research on diet and chronic disease relationships. Chapter 6 describes the creation of two dietary scores the ‘Dutch Healthy Nutrient and Food Score’ and the ‘Dutch Undesirable Nutrient and Food Score’ based on the food classification system described in chapter 5 in the Alpha Omega Trial. The Alpha Omega Trial is a randomized controlled trial; however the current analyses were done from an observational prospective cohort perspective (with adjustment for intervention groups). We included 4307 cardiac patients aged 60-80 years and monitored mortality for 10 years. Patients in the highest quintile of the ‘Dutch Healthy Nutrient and Food Score’ had 30% (HR 0.70; 95% CI 0.55-0.91) lower CVD and 32% (HR 0.68; 95%CI 0.47-0.99) lower all-cause mortality risk compared to patients in the first quintile. The ‘Dutch Undesirable Nutrient and Food Score’ was unrelated to both CVD and all-cause mortality. In Chapter 7 we also created a ‘Dutch Healthy Nutrient and Food Score’ and a ‘Dutch Undesirable Nutrient and Food Score’ in the Zutphen Elderly Study. We assessed the association of these scores with 25 year CVD and all-cause mortality and life-years gained. We divided the men (age 65-84 years) into those with (n=210) and without (n=616) cardiovascular-metabolic diseases at baseline in 1985. During a median follow-up of 10.6 years (IQR 5.8-15.9) 806 participants died, of whom 359 from CVD. Diet scores did not predict death in all men. Among men with cardiovascular-metabolic diseases, ‘Dutch Healthy Nutrient and Food Score’ was associated with lower CVD (HR: 0.57; 95%CI: 0.35-0.93) and all-cause mortality risk (HR: 0.64; 95% CI: 0.44-0.94) comparing highest vs. lowest tertiles of the score. Men with cardiovascular-metabolic diseases in the highest vs. lowest tertile of the ‘Dutch Healthy Nutrient and Food Score’ lived 2.5 year longer. The ‘Dutch Healthy Nutrient and Food Score’ was not associated with CVD and all-cause mortality in men without cardiovascular-metabolic diseases. The ‘Dutch Undesirable Nutrient and Food Score’ was not associated with any of the outcomes. In Chapter 8 we summarized the main findings of this thesis and reflected on some methodological considerations. First, we discussed the different approaches to derive dietary scores and patterns and the advantages and disadvantages of these methods. Second, we reflected on important aspects for creating a priori dietary scores and on further research. Finally, the general conclusions and implications were presented. From the results presented in this thesis we conclude that adherence to a healthy diet is inversely associated with early stage markers of CVD (markers of endothelial function and oxidative stress), CVD and all-cause mortality. In summary, a healthy diet consists of plenty of vegetables and fruit, legumes, whole grains, nuts and seeds, moderate intake of fish/poultry/lean meats and low fat dairy, and limited intake of processed meats, refined grains, sugar sweetened beverages, ready meals and snacks. However, this thesis also showed that a high quality dietary pattern can be achieved in several different ways, and may differ among populations.