Genome-wide association study of uncontrolled eating

Uncontrolled Eating (UE) is a unifying heritable trait linked with obesity, overeating, personality and the brain. UE is a combination of eating-related traits with different names, food addiction, disinhibition, hedonic hunger, emotional eating, binge eating and the like. However, they all emphasise loss of control over intake and their questionnaire measures correlate r>0.5. Therefore, datasets from each of those measures can be combined to map the genomics of Uncontrolled Eating (Vainik et al., 2019, Eur J Neurosci). We used Estonian Biobank (EstBB) data of 108,733 participants who had responded to surveys on personality or mental health that included items on overeating. We built a single UE factor model from the following items from four different questionnaires: 1)NEO-PI-3 “Sometimes I am not able to control my appetite”; 2) RED/TFEQ: “When I start eating, I just cannot seem to stop; 3) AEBQ: “I enjoy eating”, “I get full up easily” and “I eat slowly”; and 4) SCOFF: “Do you worry you have lost control over how much you eat?”. The final model's fit statistics were CFI=0.986, RMSEA=0.028, SRMR = 0.023. We used the latent UE factor score for GWA analyses with Regenie. Genetic correlations were calculated with LDSC. UE had SNP heritability of 0.092, which is similar to other self-reported phenotypes, such as personality traits or educational attainment. Loss of control over eating was associated with 9 genome-wide significant loci, annotated by the nearest genes: PTEBP2, AMPD2, TMEM18, ROBO2, TFAP2B, FOXP2, BCDIN3D, FTO, MC4R. While mostly associations relate to BMI, the genetic correlation with BMI was rg=.57, so UE can be considered related to BMI but still independent. We used genetic correlations to map UE to phenotype space. Expectedly, we replicated several associations known from phenotypic literature, such as positive genetic correlations with Type 2 diabetes, smoking, preference for sweets and salty foods, and negative correlation with subjective well-being. More unexpected findings were positive genetic correlations with spicy food preference (capsicum), self-reported exercise intensity, overall health score, and intelligence. UE was associated with mostly BMI-related loci, but genetic correlation analysis suggested certain independence from BMI. Genetic correlation analysis also demonstrated known relationships with poor health and diet but also outlined several novel associations, widening the UE concept. Namely, people may report UE for other reasons, such as exercising and needing more calories, or eating too much spicy food. Furthermore, calorie surplus as opposed to calorie deficit may support positive overall health and intelligence. While unexpected, this result may be explained by the health and intelligence measured in UKB with an elderly population, where eating too little may indicate frailty. Current results need replication in other samples. Should they hold, current findings pave the way to map the causal inputs and outputs of UE.