Skip to main content
European Commission logo print header

The role of social, genetic and environmental factors in healthy eating: a multicentre analysis of eating disorders and obesity

Deliverables

Anorexia nervosa and bulimia nervosa were linked to different neuropsychological impairment. However, the findings in the previous studies were not consistent, included non-specific deficits in different cognitive functions and were as such difficult to interpret. On the other hand, there was increasing evidence, that anorexia nervosa is closely related to some personality characteristic, such as perfectionism and compulsivity. Furthermore, obsessive-compulsive personality was found to be possible risk factors for their development (Anderluh et al, 2003). Our hypothesis was that neuropsychological deficits in eating disorders mirror the behavioural and personality characteristics of obsessive-compulsive type. A battery of neuropsychological tests that measure different aspects of cognitive inflexibility was selected, which included Trail Making B test, Brixton test, Verbal Fluency test, Haptic Illusion test, a Cognitive Set Shifting test and a Visual Set test. Four of these tests include the conceptual shift component, in one a perceptual inflexibility is tested and in additional test of dysdiadodokinesis was added that tracks a soft neurological dysfunction of the motor system. In the first part, three groups were testing using this battery of tests, a group of people with acute anorexia nervosa, with acute bulimia nervosa and a healthy control group. Measures from the tests clustered into four factors: simple alteration, mental flexibility, preservation and semantic shifting. The anorexia nervosa group showed global impairment on three of four factors, compared to healthy controls, while patients with bulimia nervosa experienced impairment in a more selected areas that present a cognitive set shifting dysfunction. This supports the previous studies that showed cognitive dysfunction as a common feature of anorexia nervosa (Strupp et al, 1986, Maxwell et al, 1984, Braun et al, 1992, Tchanturia et al, 2001, Tchanturia et al, 2002), as well those that reported finding of a soft neurological sign of dysdiadodokinesis in people with anorexia nervosa (Gillberg et al, 1994, Rastam, 1992). In the second part, patients with acute anorexia nervosa were compared to the group of patients with anorexia nervosa soon after the weight restoration and to the fully recovered patients that were without any symptom of the disease for at least a year. While recovered patients showed better performance than weight restored or acutely ill patients, their performance on tests that cluster into perceptual set shifting factor and preservation factor persisted to be significantly worse than in healthy control group. This results confirm the prediction that at least some aspects of cognitive inflexibility persists after weight restoration and full recovery and thus possible presents a vulnerability factor for eating disorders. In the third part, performance on the selected cognitive tests was correlated to the measured level of current obsessive-compulsive symptomatology, as well as obsessive-compulsive personality in childhood before the onset of an eating disorder. Interesting, cognitive inflexibility as measured by selected tests, was found to correlate with both measures of obsessive-compulsive personality. The temperamental features may thus together with deviation of cognitive flexibility presents an underlying and possible vulnerability factor for eating disorders. Our findings suggest that cognitive inflexibility may present an important manifestation of biological substrate of anorexia and bulimia nervosa and may add to the broader phenotypic definition for future genetic studies. Furthermore, our neuropsychological model stresses the importance of cognitive impairment in eating disorders and offers additional confirmation that eating disorders may present neurodevelopmental disorders, similar to other complex psychiatric illness, such as for example schizophrenia. It also offers new areas for therapeutic intervention, which can importantly contribute to the higher efficiency of the treatment strategies in anorexia nervosa, the disease which today still lacks successful treatment approaches. First treatment interventions based on the enhancement of cognitive flexibility are underway as part of our next research projects.
Previous studies have suggested genetic and a range of environmental risk factors predisposing eating disorders. However, few studies have addressed the risk factors of eating disorders in unselected population samples of adolescents at high-risk period for development of eating disorders. Information about developmental factors and heterogeneity of disease characteristics at different ages remains scarce. In this study, we present preliminary results on potential risk factors for self-reported eating disorders at age 17 based on follow-up questionnaire data at age 12 and 14 in a longitudinal setting. We also assess what risk factors at baseline (age 12) were associated with the following outcomes at age 14: DSM-IV eating disorders, dieting behaviour, and overweight/obesity. Predictors of dieting at age 12 and correlates at age 14: Based on the diagnostic SSAGA interviews in 1848 twins, 113 girls (12.1%) and 126 boys (13.7%) reported having been on a diet. Compared to non-dieters, dieters had a higher BMI (OR 1.24 per unit increase, 95% CI: 1.18-1.30), the median BMI being 20.6 among 14-year old dieters and 18.9 among non-dieters. Dieting was significantly associated with full and partial syndromes of eating disorders (p<0.0001), depression and depressive symptoms (p< 0.0001) and suicidal behaviour( p=0.0002). Suicidal thoughts and acts were reported by 12.1% (29 of 239) 14-y-o dieters, compared to 5.5% of non-dieters at the same age. Early pubertal stage (measured at age 12) significantly predicted dieting at age of 14 in girls (OR 2.09, CI 95% 1.15-3.76), after BMI was statistically controlled for. In boys, there was no association with dieting and pubertal stage. BMI at age 14 in boys as well as in girls was positively and significantly associated with dieting. Parental monitoring; family environment; satisfaction with leisure time activities, academic performance, and relationships to family and friends; frequency of watching TV and videos, and frequency of exercise at age 12 were not statistically significantly associated with dieting at age 14. Predictors and correlates of obesity at ages 12-14: Using questionnaire data from FinnTwin12, the BMI of 4593 14-year old twins was calculated. Adolescents with BMI> 95th percentile were defined as obese. The BMIs of girls at age 14 (n=2299) ranged from 12.5 to 49.4 and of boys (n=2247) from 11.6 to 35.8 (mean 19.34). The 229 obese adolescents had BMIs in excess of 24.5 (girls) and of 24.6 (boys). Statistically significant predictors of obesity at age 14 were the puberty development scale (PDS) scores at age12 (for girls OR 1.67, 95%CI: 1.16-2.40; for boys, OR 2.83, 95%CI: 1.31-6.11). Obesity in 14-y-o girls was positively associated with daily TV watching (OR 2.88;1.06-7.80), self-reported unpopularity among peers (OR 1.8; 1.22-2.69) and among peers of opposite sex (OR 2.3; 1.43-3.70). In boys, obesity was also associated with sleeping disturbances (OR 2.42; 1.46-4.03) and daily outdoor activities (OR 0.61; 0.40-0.93). The intrapair, tetrachoric correlations of obesity at age 14 were 0.87 for monozygotic (MZ) female, 0.52 for dizygotic (DZ) female, 0.92 for MZ male, 0.73 for DZ male and 0.44 for opposite-sex twin pairs. The intraclass correlations for BMI at age 14 were 0.81 for MZ female, 0.47 for DZ female, 0.84 for MZ male, 0.53 for DZ male and 0.36 for opposite sex twin pairs. Risk factors for self-reported eating disorders at age 17: Of lifetime diagnoses, depressive symptoms and depression at age 14 were over represented in individuals with self-reported ED at age 17. However, lifetime diagnosis of alcohol dependence or conduct disorder seemed not to predict ED at age 17. Overall, numbers of individuals with DSM-IV diagnoses at the age of 14 were very low. Of individuals with self-reported eating disorder at age 17,59% reported unpopularity among peers of opposite sex at age 14. As character trait, low self-esteem analysed by Rosenberg’s scale at age 14 was studied. Adolescents with self-reported eating disorder scored generally lower than adolescents without eating disorders, but the association was not statistically significant. PDS scores at age 12 and 14 were analysed as well as behavioural correlates; they did not predict self-reported eating disorder at 17, nor did daily TV watching, daily outdoor activities, smoking, dieting, self-reported unpopularity among peers, sleeping disturbances, or other psychosomatic symptoms. Conclusions: Our results suggest that larger samples are needed to analyse premorbid and prodromal eating disorders in adolescents. Dieting is associated with many unhealthful behaviours at the age 14, but at this early risk age mood problems are a better predictor of future eating disorder symptoms than dieting.
The aim of this work is to define the interaction between genetic factors and environmental/personality factors. This includes interaction between measures such as reward dependence and polymorphisms in the dopamine system, or childhood feeding difficulties and genes in the serotonin system. The first analysis, using novelty seeking harm avoidance and reward dependence is complete. Two associations are nominally significant before correction for multiple testing, DRD4 VNTR polymorphisms allele 7 and harm avoidance, and DRD4 VNTR polymorphisms allele 7 and reward dependence. Analysis is now continuing for a series of personality and risk factors. These include personal vulnerability traits and exposure to high parental expectations and sexual abuse as well as poor feeding in childhood. We are interested in further joint research and development of our scientific findings in eating disorders with both European and third country bodies. In the long term we expect that these findings will lead to improvements in treatments and preventative measures, for example by addressing the effects of environmental risk factors in genetically vulnerable individuals such that gene-environment interaction leading to disease is prevented. We will publish this data in scientific publications, conferences, meetings with patients, their carers, families, the general public and healthcare professionals. Web-based dissemination is also foreseen. We are interested in developing simple, personal CD-ROM, WEB or SMS based treatments for eating disorders based on our research findings. This includes early intervention in high risk groups, and use of genetics and genomics to identify those at high risk of eating disorders. We would expect to collaborate with an SME involved in e-health or new genetic technologies.
The major objective was to determine whether caloric restriction (dieting) and weight loss are triggers for the development of anorexia or bulimia nervosa (AN, BN) or binge eating disorder (BED) in patients with childhood onset of obesity. While there is no shortage on data pertaining to the relationship between self-conducted dieting and the onset of eating disorders, it is unclear whether the findings of these studies can readily be applied to young obese inpatients seeking weight-loss treatment. It has previously been discussed that being overweight might have a protective effect from developing eating disorders during caloric restriction, however, no work has been done so far that focuses on inpatient treatment for weight reduction in adolescents. In this study, we had the opportunity to examine patients who experienced a documented weight loss achieved by caloric restriction as well as by increased energy expenditure during their inpatient treatment. This is of paramount interest, because treatment in these facilities is very resource-consuming and potential adverse effects should be known with a reasonable amount of certainty. The completed statistical analyses confirmed the initial trend apparent in the preliminary evaluations, showing that inpatient treatment for weight reduction is not followed by a significant rise in the incidence of eating disorders. When asked "Were there times when you consumed exceptionally large amounts of food, i.e. had a 'binge'?", 81.3% answered negatively and 18.7% (n=66) reported they had experienced binges. Of these 66 patients 31 (47%) reported binges with a consumption of an amount of food that would qualify as a binge, and 97% (n=30) of these 31 affected individuals stated that these binges had been present before the treatment began, 1 patient (3%) was unable to date the commencement of binge eating prior to or after initiation of treatment. Additionally, a sample of 89 patients was assessed prospectively by application of the EDE (Eating Disorder Examination) at admission and three months after leaving the facility. We were again able to document that eating disorders do not ensue in the three-month period after inpatient treatment. Because we consider this period as the greatest risk period for the development of both progressive weight loss subsequently leading to anorexia nervosa and for the development of binge eating, we conclude that weight loss in the magnitude of on average 6.5kg during inpatient treatment of adolescents does not increase the risk for development of an eating disorder. In an attempt to identify factors that are associated with both positive and negative outcomes we additionally compared the retrospectively ascertained patients whose therapy was successful after one year to those who had no success. Successful weight losers were able to change of eating behaviour, exercise more, and had lower school stress; unsuccessful weight losers had higher stress at school, exercises less, has physical disease more often. The results indicate that depending on whether or not patients were classified as successful or unsuccessful weight losers different items were endorsed as contributing to their current body weight. Even though the answers to the questions are not indicative of the real reasons underlying the probands' respective weight loss, it would be highly interesting to further investigate methods to differentiate between these potential subgroups.
We have analysed a series of candidate genes for association with anorexia nervosa using family-based association analysis. These are listed below: BDNF Brain-derived neurotrophic factor (BDNF) has been implicated in the regulation of food intake and body weight in rodents. Two single nucleotide polymorphisms (SNPs), –270C>T and Val66Met, have been identified within the BDNF gene. A combined population and family-based study in seven European countries reveals that the Met66 variant is strongly associated to all eating disorders subtypes [anorexia nervosa (AN), restricting AN, binge-eating/purging AN and bulimia nervosa (BN)]. Moreover, the –270C BDNF variant has an effect in AN, BN and late age of onset of weight loss. This is the first consistent identification of a gene in the pathophysiology of ED in different populations, and indicates a role for BDNF as a susceptibility factor for aberrant eating behaviours. Serotonin receptor 5-HT2A. We found no evidence for a significant role of the 5-HT2A gene in anorexia nervosa. Previous results may have been exposed to stratification bias (which we controlled by the TDT method) and/or the risk of type 1 error (from which we were less exposed because of the sample size). Catechol-O-methyltransferase Overall we found no support for the hypothesis that the Val158 allele of COMT gene is associated with anorexia nervosa in our combined European sample. Agouti related protein. The orexigenic neuropeptide agouti-related protein (AGRP), a MC4-r antagonist, plays a crucial role in maintaining body weight, by inducing food intake. The sequence of the coding region of the human AGRP gene (AGRP) was determined and the AGRP of 100 patients with AN was screened for variations. Three single nucleotide polymorphisms (SNPs) were identified and screened in a further 45 patients and 244 controls. However in a larger European-wide sample we found no evidence for association. Noradrenaline transporter. Because of the functional relevance of NET to anorexia and the association observed by Urwin et al., (2002) we have analysed the NETpPR in 142 family trios from London, UK and Vienna, Austria. The 142 trios consisted of 67 with restricting anorexia nervosa (RAN), 48 with binge-purging anorexia nervosa (ANBP) and 27 AN patients who were not classified into either group. We see no evidence for transmission distortion in our samples of family trios with anorexia nervosa and we failed to replicate the association Urwin et al. (2002) saw between alleles of the noradrenaline transporter polymorphism, NETpPR. We also do not see a trend towards a difference between RAN and ANBP as the L4 allele is transmitted to ANBP probands more often than the S4 allele, i.e. the opposite direction to that observed by Urwin et al. (2002). Melanocortin 4 receptor. Obesity has been identified as a risk factor for the development of bulimia nervosa (BN). A single patient with both extreme obesity and BN had a haplo-insufficiency mutation in the MC4R. Comparison of current and maximal body mass index (BMI) of all patients with cross-sectionally obtained BMI in the general population revealed an age appropriate distribution for current BMI and a substantially increased frequency for overweight at time of maximal BMI. Our findings suggest that overweight is a risk factor for BN in clinically ascertained patients. For the first time a genotype predisposing to obesity has been detected in an extremely obese patient with BN. 5-HT2c. Objective: To examine a non-conservative Cys23Ser in the 5-HT2c gene for association with anorexia nervosa in females. Methods: We used case-controls and family based association analysis in a sample of 118 patients with DSMIV anorexia nervosa, 244 controls and 47 family trios. Results: We found a significant increase in the frequency of the Ser23 allele in the patients by allele (18.8% vs 12.8%; P = 0.026; OR 1.58 95%CI 1.0-2.4) and genotype (32.2% vs 19.9%; genotype-wise P = 0.027). There was significant correlation between genotype and minimum BMI (r2 = 0.056; P = 0.01), indicating that the Ser23 allele has an effect on severity of illness. We performed TDT analysis using a sub-sample of the cases (58) for whom both parents were available and we found an excess transmission of the Ser23 allele (P = 0.05). Conclusion: Our findings are consistent a role for the Ser23 allele of 5-HT2c in mediating susceptibility to and increasing severity of anorexia nervosa.
In order to identify neural correlates underlying the eating pathology and to define endophenotypes for further genetic research, a functional neuroimaging study has been performed in a large representative sample encompassing the spectrum of eating disorders. All functional experiments included visual presentation of images: - Food and drink images compared to matched non-food images. - Images of female bodies of varying proportions compared to matched drawings of houses. - Standard aversive emotional images from the International Affective Picture System (IAPS, Lang et al 1999). Structural images have also been collected to measure the brain volume and specific structures such as the hippocampus (Connan et al, in preparation). In total 54 participants have been scanned: - 10 female patients with bulimia nervosa. - 16 female patients with anorexia nervosa. - 9 women long-term recovered from anorexia nervosa. - 19 healthy women (comparison group). All data have been analysed and interpreted with the result of one paper in press (Uher et al, in press), second paper submitted (Uher et al, submitted) and a third paper in preparation. Summary of main findings: - Medial prefrontal response to symptom provoking stimuli has been identified as a common feature of anorexia and bulimia nervosa. - Abnormal prefrontal reaction is specifically associated with symptom-related material; it is independent of current weight status, medication or depressive symptoms. - Abnormal propensity to activate medial prefrontal circuits in response to inappropriate stimuli may be common to eating, obsessive-compulsive and addictive disorders. - Decreased response to food in the anterior and lateral prefrontal cortex is specific to bulimia nervosa and differentiates it from both anorexia nervosa and the healthy comparison group. - The medial prefrontal response to disease-specific stimuli is present even after long-term recovery from anorexia nervosa and may be related to trait vulnerability. - Lateral and apical prefrontal involvement differentiates recovered people from chronic patients and is associated with a good outcome.
We have created an aetiological model of anorexia nervosa (AN) and bulimia nervosa (BN), based on genes, environment and function of the brain. Patients who develop eating disorder have a set of identified environmental risk factors together with genetic vulnerability which together lead to the illness. In addition, there are characteristic traits such as childhood perfectionism, cognitive inflexibility, and dieting behaviour, which result from the actions of both genes and environment. We identified genetic risk factors, including BDNF, which appears common to both disorders, and MCR4, which is a risk factor for BN. Thus, AN and BN appear to differ for some factors (BN for example having genetic risk factors in common with obesity) and have common risks for others (such as cognitive inflexibility and many environmental risk factors). These risks manifest themselves though specific abnormalities in neural process, which can be viewed by neuroimaging. Specific results: Twin population analysis: Dieting is associated with many unhealthy behaviours at the age 14, but at this early risk age mood problems are a better predictor of future eating disorder symptoms than dieting. Predictors of obesity at age 14 were early puberty, daily TV watching, self-reported unpopularity among peers, sleeping disturbances (boys) and reduced daily outdoor activities. Both full and partial syndromes of anorexia and bulimia were very significantly associated with depression, depressive symptoms in girls and suicidal behaviour in both sexes (p<0.0001). Smoking/snuff use was positively and significantly associated with eating disorders (p=0.002). Dieting was a very significantly associated with eating disorders in both sexes (p<0.0001), and much more prevalent than full syndromes of eating disorders. We found mild to moderate heritability and some common environment effects in girls, little heritability in boys, and sex-specific differences in the transmission of dieting behaviour. Functional magnetic resonance imaging: We found that medial prefrontal cortex response to symptom provoking stimuli is a common feature of anorexia and bulimia nervosa and common with obsessive-compulsive and addictive disorders. It is present even after long-term recovery from anorexia nervosa and may a trait vulnerability. Prefrontal reaction is specifically associated with food. Decreased response to food in the anterior and lateral prefrontal cortex is specific to bulimia nervosa. Lateral and apical prefrontal involvement differentiates recovered people from chronic patients and is associated with a good outcome. Neuropsychology: A battery of neuropsychological tests of mental flexibility was used, and uncovered four factors: simple alteration, mental flexibility, preservation and semantic shifting. The AN group showed global impairment on three of four factors, compared to healthy controls, while BN had impairment in selected areas that present a cognitive set shifting dysfunction. Our findings suggest that cognitive inflexibility may present an important manifestation of biological substrate of AN and BN and may add to the broader phenotypic definition for future genetic studies. Risk factors: Nearly all risk factors previously found were replicated in this study using a within-family design corroborating that these risk factors are non-shared in nature. The occurrence of major depressive episodes was significantly more common in the pre-onset period of AN patients. Patients were high in harm avoidance, persistence, and self-transcendence, and low in novelty seeking, self-directedness and cooperativeness. They did not differ in reward dependence. Furthermore, being low in self-directedness predicted strongly the comorbid presence of a personality disorder in these patients with AN. Cross cultural variation and gender. There are cross-cultural differences in childhood eating patterns and parental attitudes to food in people with an ED. There are few gender differences in personality and psychological traits in eating disorders. However, higher premorbid obesity and overweight are more common antecedent factors in males. Does dieting trigger and eating disorder? Inpatient treatment for weight reduction is not followed by a significant rise in the incidence of eating disorders. Genetic analysis: We have analysed 13 different genes in the project. One of these genes, BDNF, was highly significant across all samples and a paper is in preparation. We have also finished collecting data and performed preliminary analysis for gene environment interaction. Two associations are significant DRD4 VNTR polymorphisms allele 7 and harm avoidance, and DRD4 VNTR polymorphism allele 7 and reward dependence.
In terms of personality traits, males with eating disorders demonstrated less harm avoidance (p<.025), compared to females. Males showed significantly lower scores in drive for thinness (p<.007), impulse regulation (p<.043) and body dissatisfaction (p<.01). There were no gender differences on other symptomatological and psychopathological measures. The male group gave a history of premorbid overweight or obesity more frequently (45% vs. 15%, p<038). The results of this study indicate that, there were few gender differences in personality and psychological traits. However, higher premorbid obesity and overweight are more common antecedent factors in males.
We have retrospectively investigated 45 sister pairs discordant for AN using the Oxford Risk Factor Interview developed by Fairburn and the Sibling Inventory for Differential Experiences (SIDE) developed by Daniels and Plomin. Nearly all risk factors for AN found by Fairburn using a case-control design with unrelated controls were also found by us in this study using a within-family design therefore, corroborating that these risk factors are non-shared in nature. In particular, the risk factors in the personal and environmental vulnerability domains were predictive of being the eating disordered sister (also after using stepwise regression procedures). The comparison of the discordant sister-pairs with the SIDE measure revealed no differences between the sisters in parental treatment and peer group characteristics, but much more jealousy in childhood towards the (later) healthy sister in the sister with AN. In addition, we have investigated the psychiatric disturbances of these sister-pairs having occurred before onset of AN. The occurrence of major depressive episodes was significantly more common in the pre-onset period of AN patients. The rates of anxiety disorders and substance related disorders did not premorbidly differ between the sisters. Temperament and character has also been investigated in these sister-pairs and differed as expected from results of other studies. Patients were high in harm avoidance, persistence, and self-transcendence, and low in novelty seeking, self-directedness and cooperativeness. They did not differ in reward dependence. Furthermore, being low in self-directedness predicted strongly the comorbid presence of a personality disorder in these patients with AN. A retrospective pilot study in sister-pairs discordant for eating disorders using the Child Behaviour Checklist (CBCL) regarding childhood before onset of the eating disorder has found significantly higher scores in the overall scale, the externalising, and the internalising behavioural problems scale in female AN/R, AN/BP, and BN patients compared with their healthy sisters. Aggressive behaviour, attention problems, social problems, anxious/depressive symptoms, somatic complaints, and social withdrawl were prominent in particular in the AN/R group (n=28). Within-family studies offer the ideal opportunity to control for cultural bias and problems with ascertainment of appropriate controls and give the additional value tom study genetic risk factors in the same samples. We have proposed a design, which should enable interesting results once the sample is larger and once, as mentioned above, some genetic associations have been replicated in AN patients. Protective factors: It is indeed important to elucidate factors contributing to resistance of youth to unhealthy dieting and eating disorders. However, despite some advances, which have been made between the two reviews by diagnosing complete non-existence of research on this topic and giving a short summary of research since 1990 up to 2000, there are only a few papers dealing with protective factors for eating disorders so far. Furthermore, there are the same methodological problems within numerous studies: e.g. some studies have cross-sectional designs and give correlates instead of risk-factors, not specifying and attempting to assess factors within the premorbid period (e.g. most recently. How protective factors act has been summarized: they can decrease dysfunction directly, interact with the risk factors interrupting its effects, can disrupt the mediational chain through which the risk factors act or prevent the occurrence of risk factors itself. A larger number of general and unspecific protective factors for mental problems and disorders have been identified. However, more specific protection factors should be identified, which would be informative for developing prevention strategies. Participation in non-elite sports seems to serve as protective factor for eating problems. High self-esteem and high self-efficacy are protective. Problem-solving skills are important in protecting someone to get an eating problem. Conclusions and further research: There are numerous methodological considerations to bear in mind in order to enhance the value of risk and protection factor study in the eating disorders. Risk factors have been much more intensely investigated than protective factors. Beside retrospective studies, prospective longitudinal studies would be needed also in this rare but very severe disorder, cause good prevention programs are based on well designed studies on risk and protection factors. This should be taken into consideration by funding institutions. The inclusion of genetic measures in order to elucidate further gene environment interaction and correlation would be sensible.

Searching for OpenAIRE data...

There was an error trying to search data from OpenAIRE

No results available