Statistics from Altmetric.com
Recently, Karban and colleagues1 reported an association of a common NFKB1 gene polymorphism, −94ins/delATTG, with ulcerative colitis (UC) in a non-Hispanic, non-Jewish North American population. The deletion was significantly associated with disease in both family based and case control studies: in the combined case control cohort, the allele frequency of −94delATTG (D) was significantly increased in 350 non-Jewish UC cases (45.3%) compared with 802 non-Jewish controls (38.8%, p = 0.002). In a recessive model of inheritance, the homozygous (DD) genotype was significantly increased in UC cases (21.4%) compared with controls (14.8%) (p = 0.0043), giving an odds ratio of 1.57 for the DD genotype (95% confidence interval 1.14–2.16).
Nuclear factor κB (NFκB) is an important transcription factor implicated in the inflammatory response.2 The NFKB1gene, which encodes the p105/p50 subunit of the NFκB family of proteins, maps to chromosome 4q24, in a region showing linkage to inflammatory bowel disease3–5; a mouse locus for colitis, cdcs1, maps near the mouse homologue of human NFKB1. The −94ins/delATTG polymorphism in the promoter region of NFKB1 near transcription factor binding motifs may regulate expression of the gene. As NFKB1 is a plausible inflammatory bowel disease candidate gene, we sought to replicate the findings of Karban and colleagues.1
We genotyped the −94ins/delATTG polymorphism in 472 independent British UC cases (for ascertainment and diagnosis see Cuthbert and colleagues6) and 657 ethnically matched healthy controls. This compares with 350 cases and 802 controls in the Karban study. Case control studies have increased power to detect association compared with family based tests (for example, the transmission disequilibrium test).7 The χ2 test was used to analyse differences in allele and genotype frequencies between cases and controls, and to test for Hardy-Weinberg equilibrium. Our study was well powered to replicate this association, with 86% power to detect a significant difference in D allele frequency (significance level 5%) based on the allele frequencies of allele D observed by Karban et al, and 79% power to detect a significant difference in DD genotype frequency (significance level 5%) in a recessive model of inheritance.
The NFKB1 promoter region was amplified by polymerase chain reaction (PCR) using the primers promoter e forward (labelled with FAM fluorescent dye) and promoter f reverse described by Karban and colleagues,1 and PCR products sized by electrophoresis on an ABI 3100 Prism Genetic Analyser. The size of the product determined the presence or absence of the −94ATTG deletion: 286/286 bp = WW, 282/282 bp = DD, and 286/282 bp = WD.
Both case and control genotypes were in Hardy-Weinberg equilibrium (p>0.2). There was no significant difference in allele D frequency (40.1% v 39.7%, χ2 = 0.04, p>0.5, 1 df) or in the frequency of the DD genotype (16.3% v 14.6%, χ2 = 0.62, p>0.5, 1 df) (see table 1) between UC cases and controls. The odds ratio (OR) for the DD genotype in our sample was 1.14 (95% confidence interval 0.822–1.579) compared with an OR of 1.57 (95% confidence interval 1.14–2.16) in the Karban study. The confidence intervals for the two studies overlap, with the OR estimate of Karban et al lying at the upper end of the range for our study.
There are several possible reasons for non-replication of association studies.8 There could be phenotypic differences in the case population from the two studies, such as different proportions of patients with limited or extensive disease. Data on site of disease were available from 251 patients in our study; the frequency of allele D was very similar in patients with distal (n = 92, f = 40.8%) or extensive (n = 159, f = 39.9%) disease. There may also be population specific differences in the contribution of this variant to UC susceptibility although other loci such as CARD15 and IBD5 have been widely replicated in North American and British populations.9 Alternatively, the original report may be a false positive: it involved multiple testing against various phenotypes and Jewish versus non-Jewish populations that has not been corrected for. However, the UC association was detected in both family based and case control study designs. Lastly, the size of the effect may be much smaller than indicated by the Karban study, requiring a very large sample size to replicate.10
In summary, we found no evidence for association of the −94ins/delATTG NFKB1 polymorphism with ulcerative colitis in the British population. A more detailed survey of the NFκB activation pathway is in progress to assess its contribution to susceptibility to inflammatory bowel disease.
Conflict of interest: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.