Background: Gene-gene interactions (G × G) potentially play a role in the etiology of complex human diseases, including inflammatory bowel disease (IBD), and may partially explain their 'missing heritability'.
Methods: Using the largest genotype dataset available for IBD (16,636 Crohn's disease (CD) and 12,888 ulcerative colitis (UC) cases) we analyzed G × G with the powerful case-only (CO) design. We studied 169 single nucleotide polymorphisms (SNPs) for CD (156 for UC), previously shown to be associated with the respective diseases. To ensure the validity of the CO design, we confined our analysis to pairs of unlinked SNPs. We used principal component analysis at the center level to adjust for possible causes of genotypic association other than G × G, such as population stratification and genotyping batch effects. Results from center-wise logistic regression analyses were combined by a random effects meta-analysis.
Results: A number of nominally significant (p < .05) G × G interactions were observed, but none of these withstood the Bonferroni multiple testing correction. However, one SNP pair, comprising rs26528 in the IL27 gene and rs9297145 in the KPNA7 gene region was characterized by an interaction odds ratio of 1.18 (95% CI: 1.10-1.27) for CD and a p-value of 7.75 × 10-6. Owing to the concurrent role of the IL27 and KPNA7 genes in NF-κB signaling, a master regulator of pro- and anti-inflammatory processes in IBD, the observed interaction also has biological plausibility.
Conclusions: We were able to exemplify the utility of the CO design for analyzing G × G, but had to recognize that such interactions are probably scarce for IBD.
Keywords: Crohn’s disease; Epistasis; effect modification; gene–gene interaction; ulcerative colitis.