ANCESTRY-SPECIFIC GENETIC EFFECTS IDENTIFIED IN A COHORT OF HISPANICS WITH INFLAMMATORY BOWEL DISEASE

Background: Although incidence of Inflammatory Bowel Disease (IBD) is rising in underrepresented populations, including Hispanics, genetic studies of IBD have primarily focused on Northern European ancestral populations. The genetic admixture of Hispanics provides a unique opportunity to examine the ancestral origins of risk and may further elucidate the underlying pathophysiology of IBD. We sought to determine whether heterogeneity of effect exists by ancestral origin for previously identified IBD susceptibility alleles and to identify novel alleles which may be ancestry-specific.
Methods: Whole Genome Sequence (WGS) data were available for 7273 Hispanic subjects including 1659 cases [938 Crohn’s Disease (CD), 684 Ulcerative Colitis (UC), 37 other IBD] and 5614 controls. Following stringent quality control, WGS data were phased with BEAGLE and local ancestry was computed with RFMixV2. Genome-wide association was performed with Tractor to obtain ancestry-specific effect size estimates for alleles from an African (AFR), European (EUR), and Amerindian (AI/AN) origin.
Results: For CD, the strongest associations (p < 1.0E-07) were identified for alleles residing on a EUR haplotype within the previously identified NOD2 (6 variants), IL23R (55), CYLD (1), NKD1 (2), and HLA-DRA (1) loci. All identified variants had low frequency (<1%) among alleles from an AFR or AI/AN origin, except for 48 of the IL23R variants which had frequency >10% amongst AFR or AI/AN alleles yet showed no evidence for association within those ancestral subsets. For alleles residing on an AFR or AI/AN haplotype, novel associations (p < 1.0E-06) were seen both within (i.e. KIAA119 for AFR, R2 < 0.2 with the previously identified variant) and outside of previously identified loci (i.e. GAD2 for AI/AN, >500 KB from a previously identified variant). Similarly for UC, the strongest associations (p < 1.0E-07) were identified for alleles residing on a EUR haplotype within the previously identified HLA locus (22 variants on chr6 from 31-33 MB), where all but one (SKIV2L) had low frequency among alleles from an AFR or AI/AN origin. For alleles residing on an AFR or AI/AN haplotype, novel associations were again seen both within (i.e. DOCK8 for AFR) and outside of previously identified loci (i.e. CSMD1 for AFR).
Conclusions: We have demonstrated the power of ancestry-informed regression to identify novel AFR and AI/AN specific risk alleles both within and outside of previously identified IBD loci. We have additionally shown that some previously identified IBD alleles are observed only on a EUR haplotype while others are observed across ancestral backgrounds but only exert an effect on IBD amongst the EUR allele subset. These findings highlight the important role genetic ancestry may play in elucidating the biological underpinnings of IBD and facilitating treatment for all patient populations.