Abstract Background Dementia is a common disease influenced by both genetic and environmental factors. APOE ε4 is well-known to increase the risk of dementia, and it has been shown to attenuate the protective association of fish oil supplementation and the incidence of dementia. To identify more genetic factors with similar modifying effects, we performed a genome-wide scan. Methods We first performed time-to-event genome-wide association study (GWAS) of all-cause dementia and two of its subtypes, Alzheimer’s disease (AD) and vascular dementia, in the UK Biobank. GWAS were performed in all participants (N = 357,631) and in two subgroups with or without fish oil supplementation (N = 113,267 and 244,364, respectively). Single nucleotide polymorphisms (SNPs) suggestively associated with dementia were then evaluated for their interactions with fish oil status in Cox-regression models. Furthermore, we conducted gene set enrichment analysis to identify the relevant cell types for these interaction signals. Results Time-to-event GWAS identified 6, 5, and 2 genome-wide significant loci (p < 5e-8) for the incidence of all-cause dementia, AD, and vascular dementia, respectively. Most of them overlapped with previously known GWAS loci for AD and related dementia. A total of 178 suggestive GWAS loci (p < 1e-5) were passed onto interaction analysis, and 43 of them were found to significantly modify the association between fish oil supplementation and dementia incidence (p < 2.8e-4 with Bonferroni correction). One locus overlapped with a known AD GWAS locus ( EED / PICALM ) and two overlapped with GWAS loci for circulating omega-3 fatty acids ( SRSF4 , PSMG1 ). Gene set enrichment analysis found that candidate genes of interaction signals demonstrated tissue or cell-type specificity in the brain. Conclusion We identified 43 genetic loci that modify the association between fish oil supplementation and dementia. These findings indicate a need for genome-informed personalized nutrition of fish oil supplementation for the purpose of dementia prevention.
Fish oil supplements (FOS) are known to alter circulating levels of polyunsaturated fatty acids (PUFAs) among individuals but in a heterogeneous manner. These varied responses may result from unidentified gene-FOS interactions. To identify genetic factors that interact with FOS to alter the circulating levels of PUFAs, we performed a multi-level genome-wide interaction study (GWIS) of FOS on 14 plasma measurements in 200,060 unrelated European-ancestry individuals from the UK Biobank. From our single-variant tests, we identified genome-wide significant interacting SNPs (P < 5e-8) in the FADS1-FADS2 gene cluster for total omega-3, omega-3%, docosapentaenoic acid (DHA), DHA% and the omega-6 to omega-3 ratio. Among the interaction signals for omega-3%, the lead SNP, rs35473591 (C>CT, CT allele frequency = 0.34), had a lower association effect size in the FOS-taking group (beta = 0.35 for allele C) than that in the group without FOS (beta = 0.42). Likewise, the effect sizes of associations between FOS and omega-3% varied across the three genotype groups (beta = 0.45, 0.50, and 0.59, respectively, in C/C, C/CT, and CT/CT). Our gene-level aggregate and transcriptome-wide interaction analyses identified significant signals at two loci, around FADS1-FADS2 and GRP12. The contribution of genome-wide gene-FOS interactions to phenotypic variance was statistically significant in omega-3-related traits. This systemic gene-FOS GWIS contributes to our understanding of the genetic architecture of circulating PUFAs underlying FOS response and informs personalized dietary recommendations.