Brasher, M. S.; Sutton, K. J.; Patterson, W. B.; Cole, J. B.

Although dietary intake is a leading risk factor for many common diseases, adherence to dietary recommendations remains low. This may partly reflect limited consideration of individual differences in eating behavior that arise from both environmental and genetic factors. While genome-wide association studies (GWAS) of dietary intake have identified hundreds of associated loci, the X chromosome has largely been ignored. To address this gap, we applied multiple X-chromosome-wide association study (X-WAS) models on dietary intake phenotypes to identify novel associations. We performed X-WAS of 46 dietary intake traits from food frequency questionnaires in up to 424,758 European participants from the UK Biobank. Phenotypes included quantitative measures (e.g., fruit intake), binary traits (e.g., decaffeinated vs caffeinated coffee), and principal component-derived food groups. We tested for genetic associations using several models: a traditional sex-combined additive GWAS, additive models stratified by sex, and two joint models accounting for sex-interaction effects and non-additivity. We also conducted X-WAS in five additional genetic ancestry groups and performed a sex-combined multi-ancestry additive GWAS meta-analysis with up to 445,773 individuals. We identified 18 loci associated with 20 dietary intake traits (P < 5x10-8), including 17 variants without prior associations in the GWAS Catalog. Among these loci, 10 were significant across multiple X-WAS models, and 5 were strongest in a model other than the traditional sex-combined additive GWAS, highlighting the value of approaches that address known complexities of the X chromosome. These results demonstrate that incorporating the X chromosome in GWAS can reveal novel loci, even for complex behavioral traits such as dietary intake. Applying multiple association models further improves discovery by accounting for unique features of the X chromosome.