Genome-wide gene-diet interaction analysis in the UK Biobank identifies novel effects on Hemoglobin A1c

Diet is a significant modifiable risk factor for type 2 diabetes (T2D), and its effect on disease risk is under partial genetic control. Identification of specific gene-diet interactions (GDIs) influencing risk biomarkers such as glycated hemoglobin (HbA1c) is a critical step towards developing precision nutrition for T2D prevention, but progress has been slow due to limitations in sample size and accuracy of dietary exposure measurement. We leveraged the large sample size of the UK Biobank (UKB) cohort and a diverse group of dietary exposures, including 30 individual dietary traits and 8 empirical dietary patterns, to conduct genome-wide interaction studies in [~]340,000 European-ancestry participants to identify novel GDIs influencing HbA1c. We identified five variant-dietary trait pairs reaching genome-wide significance (p < 5x10-8): two involved dietary patterns (meat pattern with rs147678157 and a fruit &vegetable-based pattern with rs3010439) and three involved individual dietary traits (bread consumption with rs62218803, dried fruit consumption with rs140270534, and milk type [dairy vs. other] with 4:131148078_TAGAA_T). All of these were affected minimally by adjustment for geographical and lifestyle-related confounders, and four of the five variants lacked any genetic main effect that would have allowed their detection in a traditional genome-wide association study for HbA1c. Notably, multiple loci near transient receptor potential subfamily M genes (TRPM2 and TRPM3) were identified as interacting with carbohydrate-containing food groups. Some of these interactions showed nominal replication in non-European ancestry UKB subsets, as well as association using alternative measures of glycemia (fasting glucose and follow-up HbA1c measurements). Our results highlight relevant GDIs influencing HbA1c for future investigation, while reinforcing known challenges in detecting and replicating GDIs.