GOLDN study design

The GOLDN study used a longitudinal design [13, 14] to study the TG level response to Fb. Participants were middle-aged, self-reported white individuals, likely to be genetically homogenous, who were recruited through a previous family study [12] from 2 centers in the United States. Individuals with extreme TG measures and/or a recent history of severe cardiovascular disease were excluded. Participants were asked to fast and abstain from alcohol, and not take lipid-lowering drugs for at least 4 weeks prior to the inception of the study. A high-fat meal challenge was given twice, 3 weeks apart, during which time Fb treatment was given to lower lipids.

Lipid levels, including TGs, were measured approximately 1 day prior to the drug intervention, with times 1 and 2 corresponding to measurements taken prior to treatment and times 3 and 4 corresponding to measurements taken after the treatment, within each pair (TG1 and TG2, TG3 and TG4) only 1 or 2 days apart. One can assume that TG1 and TG2 as well as TG3 and TG4 measure the same levels, except for random variation. The 8 papers in this group focused on treatment effects by using either T2 or T4, or the means of T1 and T2 or T3 and T4. These are denoted as preTG and postTG, respectively. Because TG levels have a skewed distribution, some authors used log-transformed triglyceride (lnTG).

Genetic markers were array-based SNPs. Epigenetic marker ML were beta-scores of CpG markers using a methylation array, based on CD4+ T cells and measured only at time points 2 and 4. The GAW20 simulations used the same data, except that posttreatment TG levels were simulated based on linear models.