DNA methylation loci associated with atopy and high serum IgE: a genome-wide application of recursive Random Forest feature selection

Todd M. Everson, Genevieve Lyons, Hongmei Zhang, Nelís Soto-Ramírez, Gabrielle A. Lockett, Veeresh K. Patil, Simon K. Merid, Cilla Sӧderhӓll, Erik Melén, John W. Holloway, S. Hasan Arshad, Wilfried Karmaus

The prevalence of allergic diseases are increasing worldwide, emphasizing the need to elucidate their pathogeneses. The aims of this study were to use a two-stage design to identify DNA methylation levels at cytosine-phosphate-guanine (CpG) sites across the genome associated with atopy and high serum immunoglobulin E (IgE), then to replicate our findings in an independent cohort. Atopy was assessed via skin prick tests and high serum IgE. Methylation levels were measured from whole blood using the Illumina Infinium HumanMethylation450 BeadChip from 18-year-old women (n = 245) and men (n = 122) in the Isle of Wight birth cohort. After data cleaning and processing, and removing probes with possible single nucleotide polymorphisms, DNA methylation levels from 254,460 CpG sites from the 245 women were subjected to recursive Random Forest feature selection for stage 1. The sites selected from stage 1 were tested in stage 2 for associations with atopy and high IgE levels (>200 kU/L) via logistic regression adjusted for predicted cell-type proportions and sex. Sites significantly associated with atopy in stage 2 underwent replication tests in the independent Swedish birth cohort BAMSE (n = 464). In stage 1, 62 sites were selected, of which 22 were associated with atopy in stage 2 (P-value range 6.5E-9 to 1.4E-5) and 12 associated with high IgE levels (P-value range 1.1E-5 to 7.1E-4) at the Bonferroni adjusted alpha (0.05/62 = 0.0008). Of the 19 available sites, 13 were replicated. We identified 13 novel epigenetic loci associated with atopy and high IgE that could serve as candidate loci for future studies; four were within genes with known roles in the immune response (cg04983687 in the body of ZFPM1, cg18219873 in the 5'UTR of PRG2, cg27469152 in the 3'UTR of EPX, and cg09332506 in the body of COPA).