Obesity and eating behavior patterns : associations and mediation considering genetic susceptibility

Background: Obesity prevalence has dramatically increased during the past decades and is currently a major global public health challenge. Both genes and environmental factors influence weight gain, but our understanding of how these factors impact body weight is still incomplete. Unhealthy diet is a key risk factor in the development of obesity and eating behaviors have been described as predictors of weight gain. Genes identified so far are likely to influence weight partly through appetite traits, representing one behavioral pathway of the genetic susceptibility to obesity. However, some genes are expressed in the adipose tissue, suggesting that there might be other pathways that could lead to weight gain and increased appetite in genetically susceptible individuals. Aims: This dissertation aims to 1) examine whether a diet quality score derived from a food frequency questionnaire is a valid brief instrument to estimate diet quality by comparing it with obesity measures, nutrient intakes and eating styles (eating behaviors and dietary patterns) (Study I); 2) examine whether eating behavior patterns (diet quality, eating behaviors and dietary patterns) are related to obesity measures in a cross-sectional setting of young adults (Study II) and a prospective setting of children (III); and 3) examine the relationship between genetic susceptibility to obesity, eating behaviors and obesity cross-sectionally in adulthood (Study II) and longitudinally in childhood (Study III). Materials and methods: This thesis was based on data from the FinnTwin16 study (FT16) at wave 5 (n = 4,407), a representative national longitudinal cohort of young adult Finnish twins (Studies I and II), and the IDEFICS/I.Family cohort (n = 21,293), a European multicenter study of children and adolescents (Study III). Both datasets include obesity measures [body mass index (BMI) and waist circumference], eating-related traits questionnaires (eating behaviors, dietary patterns, food frequency questionnaires, and food diaries), and family-level factors. For 1,055 twin individuals and 2,656 children with genome-wide data (Studies II and III, respectively), two polygenic risk scores for BMI were constructed using ~1 million (Study II) and 2.1 million (Study III) single nucleotide polymorphisms irrespective of genome-wide significance. Linear regression models and logistic regression models were calculated to test the associations between the diet quality score and obesity measures and eating styles in Study I (n = 3,592 twin individuals, n = 764 dizygotic twin pairs and n = 430 monozygotic twin pairs). Pearson’s correlations were calculated between the diet quality score and nutrient intakes in a subsample of 249 twin individuals and in n = 45 same-sex dizygotic twin pairs and n = 60 monozygotic twin pairs, who provided food diaries in Study I. Principal component analyses were used to derive eating behavior patterns in Studies II and III. To examine the relationship between eating behavior patterns and obesity measures, heritability estimates and Cholesky decomposition were estimated in 1,500 twin pairs (Study II) and cross-lagged path models were calculated in 2,355 children (Study III). Structural equation modeling (Study II) and causal mediation analyses (Study III) were used to identify the potential mediation models between the polygenic risk scores for BMI, eating behavior patterns and obesity measures. Results: A higher diet quality score was inversely associated with obesity measures, a lower risk of being overweight or abdominally obese and was associated with healthier eating styles (Study I). Further, the diet quality score was associated with key nutrient intakes, such as lower intakes of sucrose and total fat and higher intakes of magnesium; it can thus be used to rank individuals and twins according to diet quality (Study I). Analyses of twin pairs showed that the co-twin with a higher diet quality score tended to have healthier eating styles and nutrient intakes compared to their twin sibling (Study I). Eating behavior patterns were moderately heritable in adults (Study II). The cross-sectional associations between snacking and emotional and external eating behavior patterns with obesity measures were largely explained by genetic factors in young adults (Study II). The prospective associations between parental concern of overeating and obesity measures in children were bi-directional (Study III). The genetic susceptibility to obesity was partly mediated by the snacking eating behavior pattern, and to a lesser extent by the infrequent and unhealthy and the emotional and external eating behavior patterns during adulthood (Study II) and by parental concern of overeating during childhood (Study III). Futhermore, obesity was also tested as a mediator in the association between the polygenic risk score for BMI and parental concern of overeating, and obesity measures partly mediated the prospective association between genetic susceptibility and parental concern of overeating during childhood (Study III). Conclusions: This thesis provides new evidence that diet quality, eating behaviors and dietary patterns are important determinants of obesity during childhood and young adulthood. It provides a detailed picture of the complex associations between obesity, dietary risk factors and genetic susceptibility, by showing how eating behavior patterns and obesity measures are temporarily associated and share a common genetic liability. Moreover, it confirms that eating behavior patterns partly mediate the genetic susceptibility to obesity in both children and adults. Further, it suggests that there might be pathways other than eating behaviors by which the genetic susceptibility may lead to weight gain and the increased weight might subsequently increase appetite. A better understanding of the pathways that lead to weight gain and their impact and influences on early- and long-term health will be beneficial for future research and health professionals.