Abstract 051: Genome-wide Association Study Identifies 12 Loci for Habitual Consumption of Bitter and Sweet Beverages

Abstract

Introduction: Widely consumed beverages (e.g., soft drinks, coffee, tea) are critical sources of energy, added sugar and phytochemicals and are associated with obesity and chronic disease. Taste perception and preferences are highly heritable and strong determinants of food and beverage choice. We aimed to identify novel loci underlying habitual bitter and sweet beverage intake. Methods: We performed a genome-wide association study (GWAS) of self-reported bitter and sweet beverage intake in participants of European ancestry in the UK Biobank. Diet was assessed via multiple 24-h diet recalls (n=84703, subset) or questionnaire (n=335909, all). Bitter beverage intake was the sum of coffee, tea and grapefruit juice. Sweet beverage intake was the sum of artificially and sugar sweetened beverages and other fruit juice. Multivariable linear regression under an additive genetic model was applied. GW-significant (P < 5х10 -8 ) SNPs were followed-up for replication in independent studies of European ancestry. Results: Multiple SNPs spanning 11 loci were associated with bitter beverage intake (P <5х10 -8 , Table 1), and at least 5 of them reflected the caffeine content of coffee and tea. Multiple SNPs in the obesity candidate gene FTO were associated with sweet beverage intake (P <5х10 -8 ). The effect size per allele ranged from 0.02 to 0.2 cup per day. Loci in/near AHR, CYP1A2, and FTO were associated with both bitter and sweet beverage intake but in opposite directions. Replication efforts are ongoing. So far, associations at all loci, except 1q25.2 and 2q36.2, were replicated (P range: 0.04 to 1.8x10 -8 ) in independent studies (n=17322) which provided 80% power for replicating 8 of these 12 loci at P=0.05. Conclusions: Loci linked to caffeine metabolism and obesity predisposition rather than taste are major determinants of beverage intake. These and other identified loci have been linked to chronic disease and risk factors, suggesting causal or pleiotropic effects. Our findings have potential public health and methodological implications.