BackgroundBreastfeeding is hypothesised to benefit child health and cognitive functioning by providing long-chain polyunsaturated fatty acids (LC-PUFA), which are essential for brain development. In 2007, Caspiet al.found evidence in two cohorts for an interaction between genetic variation in theFADS2gene (a gene involved in fatty acid metabolism) and breast feeding on IQ. However, subsequent studies have provided mixed evidence for the existence of an interaction.ObjectiveTo investigate the relationship between genetic variation in theFADS2region, breastfeeding, and their interaction on traits putatively affected by their interplay in a large, population-based cohort with appropriate control for confounders in genetic associations.MethodsWe tested for the interaction in up to 335,650 individuals from the UK Biobank, over a range of cognitive functioning tests, as well as educational attainment and other traits thought to be influenced by breastfeeding, including cardiometabolic traits, reproductive success, and atopic allergy.ResultsFADS2alleles associated with an increase in docosahexaenoic acid (DHA) in blood serum (the C allele of rs174575) were associated with decreased verbal-numerical reasoning (p=2.28× 10−5) and triglycerides (p=1. 40× 10−41), increased reproductive success (p=3. 40× 10−5), total cholesterol (p=5. 28× 10−36), HDL (p=1. 42× 10− 51), and LDL cholesterol (p=1. 46× 10− 21). We observed no evidence of an interaction in any of the traits, regardless of the modelling strategy.ConclusionsWe failed to replicate any breastfeeding by genotype interactions on any cognitive or non-cognitive traits. We postulate that the previous positive findings are likely to be spurious, perhaps due to lack of appropriate control for latent population structure.