A similar tendency (P = 0.06) inhibitor Sorafenib was observed in the genotypic and dominant model. In contrast, in the present study we failed to replicate the weak association between IRGM and UC (rs13361189 P = 0.0069, pooled OR = 1.16; rs4958847 P = 0.014, pooled OR = 1.13) that was reported recently in a Spanish meta-analysis[10]. Of note, the present study was underpowered to detect such small differences, however, there was even no trend for a difference between UC and controls. In addition, in an Italian study[27], the rs4958847 polymorphism was associated with fistulizing behavior (P = 0.037, OR = 1.54, CI = 1.02-2.31) and perianal fistulas (P = 0.045, OR = 1.55, CI = 1.01-2.38) in a logistic regression analysis. This was later partially confirmed by the Leuven group.

Henckaerts et al[28] have reported in a very elegant study a significant association between the IRGM rs4958847 variant, U7 gene desert rs12704036 T-allele, NOD2/CAR15 mutation, ileal involvement at diagnosis, male sex, and time to development of non-perianal fistulas in a Cox regression analysis. In a further study from New Zealand[29], rs13361189 variant increased the risk for ileal CD in 507 CD patients and 576 controls. The OR in ileal CD was 1.92 (95% CI = 1.27-2.96). Moreover, Peterson et al[30] have suggested an association between IRGM and pediatric disease onset (< 17 years) of CD in a North American cohort. However, only one of the two IRGM variants studied (rs13361189) was weakly associated with CD in their study (uncorrected P = 0.03).

In the present study, we could not confirm the association between age at onset and the presence of the IRGM variant, in accordance with Canadian, Italian and Scottish results[29,31]. In contrast, we found an association between IRGM carriage and disease location in CD; colonic location was significantly more common in carriers of the variant allele (OR = 1.62, 95% CI = 1.07-2.44). Of note, however, the rs4958847 variant was not investigated in the present study. The association between the rs10883365 variant of NKX2-3 gene and susceptibility of IBD was first reported in CD in Caucasian patients and in a Japanese study with an OR of 1.2-1.6[12-14]. In addition, using the UC panel and the expanded WTCCC control panel, Fisher et al[15] have reported a modest association (P = 3.3 �� 10-4 and P = 2.4 �� 10-6) between rs10883365 and UC.

In the present study, we confirmed these findings; the rs10883365 variant was associated GSK-3 with UC and CD susceptibility in the allelic and genotypic models with an OR of 1.53 and 1.24, respectively. Based on our data, the association between NKX2-3 and IBD is stronger in UC compared to CD, at least in patients from Eastern Europe. In a recent Dutch study, the association between the rs10883365 variant of NKX2-3 and CD was linked to smoking status, with the risk being more pronounced in active and passive smokers[32].