Dalin Li, Dermot P.B. McGovern, Esther A. Torres, Suna Onengut-Gumuscu, Stephan R. Targan, Kyuyoung Song, Suk-Kyun Yang, Kent D. Taylor, Yoichi Kakuta, Byong Duk Ye, Stephen S. Rich, Jerome I. Rotter, Talin Haritunians, and Marla Dubinsky
Background: Studies have confirmed associations with TNFSF15 and CD in Caucasian and Asian populations, but these studies have been limited by the number of markers tested. Decoy Receptor 3 (DcR3), previously identified as a susceptibility gene from a Caucasian pediatric CD GWAS, is a decoy receptor that can neutralize pro-inflammatory ligands including, TL1A (TNFSF15). Aim: To perform transethnic fine mapping across TNFSF15 and DcR3 in Caucasian, Puerto Rican and Korean CD. Methods: Genotyping was performed in Non-Jewish Caucasians (NJ; 779 CD, 4182 controls), Ashkenazi-Jews (AJ; 492 CD, 395 controls), Puerto Ricans (PR; 300 CD, 235 controls), and South Koreans (SK; 729 CD, 469 controls) using the 200K SNP ImmunoChip genotyping array. Analysis of TNFSF15 (416 SNPs) and DcR3 (13 SNPs) was conducted. Principal components analysis controlled for population structure, and logistic regression models tested for association of each SNP with CD. Results: TNFSF15 analyses: 4 rare variants (MAF,1%) and 9 common variants (MAF= 27.7-34.9%) were significantly associated with NJ CD after correction for multiple-testing. Analyses in NJ CD confirmed that rare and common variants associations could each be tagged by a single SNP (rare: rs1322057, MAF=0.83%, p=3.02E-6, OR=6.97; common: rs7869487, MAF=22.6%, p=7.26E-5, OR=0.78). Further, the associations with CD of these SNPs were independent of each other. All 13 NJ CD SNP associations were replicated in SK CD (all P ,3.15E-10) but with higher MAF (0.13% vs 34.6% for rs1322057 and 32.4% vs 40.9% for rs7869487 in controls). The independence of the 2 SNPs seen in NJ was replicated in SK. The rare variants were also associated with PR CD but with higher MAF than in NJ (6.6% for rs1322057, p=3.19E-2, OR=1.62), but no association was seen at 9 common variants in PR. No significant SNP associations were observed with AJ CD. DcR3 analyses: rs6062496, located in intron 6 of DcR3, was significantly associated with NJ CD (p=5.83E-4, OR=1.22). This association was confirmed in SK CD (p=2.01E-2, OR=1.32) but not in AJ or PR CD. The associated SNP was also significantly associated with protection against stricturing CD (p=2.95E-2, OR=0.75). There were no significant gene-gene interactions between rs6062496 in DCR3 and the 2 independent SNPs from TNFSF15 in both NJ (p.0.18) and SK CD (p.0.61). Conclusion: We have identified novel associations of rare variants in TNFSF15 with NJ CD that are independent of previously reported common variants, and are also found in PR and SK. These variants are more common in SK and also associated with SK CD, perhaps explaining the ‘dominant' role of this locus in Asian IBD susceptibility. DcR3 is significantly associated with NJ CD, a finding replicated in SK and variation at this locus modifies a stricturing phenotype.