Abstract
As part of GAW20, we analyzed the familiality and variability of methylation to identify cytosine-phosphate-guanine (CpG) sites responsive to treatment with fenofibrate. Methylation was measured at approximately 450,000 sites in pedigree members, prior to and after 3 weeks of treatment. Initially, we aimed to identify responsive sites by analyzing the pre- and posttreatment methylation changes within individuals, but these data exhibited a confounding treatment/batch effect. We applied an alternative indirect approach by searching for CpG sites whose methylation levels exhibit a genetic response to the drug. We reasoned that these sites would exhibit highly familial and variable methylation levels posttreatment, but not pretreatment. Using a 0.1% threshold, posttreatment sibling correlation (scor) and standard deviation (SD) distributions share 16 outliers, while the corresponding pretreatment distributions share none. Comparing the pre- and posttreatment CpG outliers, 36 (8%) of SD distributions, and 449/450 (nearly 100%) of scor distributions differ. Combined, these results identify methylation sites within the KIAA1804 and ANAPC2 genes. Each gene also has a highly significant methylation quantitative trait locus (meQTL) (KIAA1804: p < 1e-200; ANAPC2: p < 3e-248), indicating that methylation levels at these CpG sites are driven by meQTL and fenofibrate.