Statistics from Altmetric.com
Editor,—Lesage and colleagues reported failure to detect linkage to the IBD2 locus on chromosome 12 in a panel of 95 families with two or more relatives affected by Crohn's disease, and offered some possible explanations (
). Linkage of inflammatory bowel disease (IBD) to this region was first detected in a panel of 160 families containing multiple cases of Crohn's disease, ulcerative colitis, or both.1 Lesage et aljustify the study of Crohn's disease families alone on the grounds that “genetic heterogeneity in susceptibility cannot be ruled out”, and they imply that studying the Crohn's disease subgroup of IBD should thus maximise their chance of successful replication. We concur entirely that genetic heterogeneity is important, and we have recently reported strong evidence that it does indeed apply to chromosome 12.2 However, our study of 367 multiply affected families suggested a significantly stronger contribution of this locus to ulcerative colitis than Crohn's disease.2 The difference between the linkage results for ulcerative colitis (LOD=3.91) and Crohn's disease (LOD=1.66) reached statistical significance in two separate tests for heterogeneity. In the light of these results, the validity of the exclusion map drawn by Lesage et al is undermined. The exclusion map was based on an assumed locus specific λs of 2.0, but this value was derived from a panel containing Crohn's disease, ulcerative colitis, and mixed pairs.1 Given the evidence for a substantially stronger contribution to ulcerative colitis than Crohn's disease, it is likely that the true λs value for this locus with regard to Crohn's disease is much less than 2. Thus the contention that Lesageet al can exclude a contribution of IBD2 to Crohn's disease susceptibility is probably not valid. As pointed out in the accompanying editorial, simulation studies have demonstrated that lod scores can be expected to vary, particularly when the study population is relatively small.3 Furthermore, the implication that a panel of 157 affected relative pairs should provide sufficient power to detect linkage if this locus is contributing to disease susceptibility is at marked variance with the power calculations derived by Suarez et al, Mandalet al, and others.4 5
In many respects, the surprising feature is that IBD2 has been replicated in as many as five independent panels.3 The datasets that have failed to detect linkage at this locus have all contained predominantly or exclusively Crohn's disease pairs.6-8 Although IBD2 probably does contribute to Crohn's disease susceptibility, the effect is likely to be weak and thus would require very large panels of multiply affected families to have a realistic expectation of replicating (or excluding) the linkage result.
It is our view that attempts at fine mapping IBD2 probably have the greatest chance of success if they concentrate on panels of families and individuals with ulcerative colitis, which appears to be significantly more strongly linked to this locus than Crohn's disease.
Editor,—In 1996, Satsangi et al reported a positive linkage between inflammatory bowel disease (IBD) multiplex families (including Crohn's disease (CD), ulcerative colitis (UC) and mixed families) with a locus (called IBD2) located on chromosome 12.1-1 The attributable risk in siblings (λs) of this IBD2 locus was calculated to be 2.1-1 In a recently published study in the journal, we failed to demonstrate a positive linkage on chromosome 12 using an independent panel of 95 CD multiplex families (
). This result was different to the previous report and we proposed several explanations for the observed discrepancy.
The first explanation may be lack of statistical power in our replication study. We investigated a similar number of affected relative pairs (n=157, all CD pairs) compared with the first linkage analysis (n=186, 81 CD pairs, 64 UC pairs, and 41 mixed pairs). Because linkage tests may exhibit large fluctuations when applied to family sets of similar size for complex genetic disorders,1-2 we tested if a gene with a λs risk of 2 was compatible with our observation and we were able to reject this hypothesis. We thus concluded that genetic heterogeneity may occur in Caucasian family panels for IBD susceptibility.
Parkes et al have recently demonstrated that this genetic heterogeneity may be related to phenotypic heterogeneity.1-3 In their proposed susceptibility model, UC is more tightly linked to IBD2 than CD. This study confirms our conclusion that there is genetic heterogeneity in familial IBD. As expected, this heterogeneity may be in part reduced by an accurate phenotypic classification. Thus from a methodological point of view, Parkes' report demonstrates that working on homogeneous phenotypic groups may be preferable than pooling several phenotypes for linkage studies. Considering CD and UC families as separate subgroups, Parkeset al suggested that the IBD2 locus has only a marginal role in CD susceptibility. This conclusion is in complete accordance with our demonstration that the relative risk attributable to IBD2 in CD multiplex families is low.
In practice, it is difficult to know what is the weight of this IBD2 locus in both CD and UC. A line of evidence, including the above mentioned reports,1-3 and a large collaborative work performed on more that 600 multiplex IBD families1-4 clearly suggests that the role of the IBD2 locus is weak in CD families. In contrast, its role in UC is difficult to estimate to date. In their recent work, Parkes et al pooled previously investigated families from UK and US panels.1-1 1-5 Because these families were a priori known to be positively linked to IBD2, this study provides a biased estimate of the risk attributable toIBD2. Thus additional works using unselected family panels are required to answer this question.
Interestingly, the IBD1 locus1-6 has been postulated to play a major role in CD and to be less important in UC families. Thus it would be postulated that IBD1 is a CD susceptibility locus and IBD2 is a UC gene. Some truth may reside in this assertion. However, a line of evidence including analysis of mixed families1-7 suggests that CD and UC have common familial risk factors and does not allow a simple dichotomic classification of UC and CD genes. Many additional steps, including gene identification, are now required before we can understand the underlying genetic model for IBD which will certainly be confirmed as a complex genetic disorder.