Family-based genome-wide association of inflammation biomarkers and fenofibrate treatment response in the GOLDN study

In this paper we analyzed whole-genome genetic information provided by GAW20 from the Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) study for family data. Lipid levels such as triglycerides (TGs) and high-density lipoprotein (HDL) are measured at different time points before and after administration of an anti-inflammatory drug fenofibrate. Apart from that, the data contain some covariates and whole-genome genotype information. We propose 2 novel approaches based on Henderson’s iterative mixed model to identify associated loci corresponding to (a) inflammatory biomarkers like TGs and HDLs together over time, and (b) the response to fenofibrate treatment. We developed a mixed-model approach using both TG and HDL phenotypes at all 4 time points for a genetic association study whereas we used TGs only to study genetic association with response to the drug. We expect that use of complete family data in a longitudinal framework under a single model involving the appropriate correlation structures would be able to exploit the maximum possible information contained in the sample. Our analysis of whole-genome single nucleotide polymorphisms (SNPs) and genomic regions corresponding to drug treatment finds no significant locus after multiple correction. Arguably, the moderately small sample size of the data set, as compared to the sample size usually used in genome-wide association studies (GWAS), could be a reason for such a result. Nevertheless, we report the top 20 SNPs associated with the phenotypes, and the top 20 SNPs and genomic regions associated with a response to fenofibrate treatment. Application of our methods to larger GWAS and further functional validation of the reported top SNPs and genomic regions might provide important biological insight into the genetic constitution of the trait.