June 5, 2001 Access past issues

Reviewed by: Marta L. Gwinn and Muin J. Khoury Office of Genomics and Disease Prevention, Centers for Disease Control and Prevention

The Health Outcome

Crohn’s disease (CD) is a chronic, inflammatory bowel disease that affects about 1 per 1000 people in western countries.  Causes of CD are un

known but likely to involve complex gene-environment interactions.

The Finding

Two groups of investigators recently reported finding variants of the NOD2 gene associated with increased susceptibility to CD (1, 2). Detailed abstractions of these two studies are available on-line (3, 4) and are part of the HuGE database of published literature  (in preparation) (5).  In a family study, Ogura et al. (1) found that a frameshift mutation resulting from a 1-bp insertion at nucleotide 3020 was associated with CD.  The investigators also conducted a case-control study comparing the frequency of this mutation in CD case from 416 independent families with a group of 287 controls.  Characteristics of cases and controls were not well described.  Re-analysis of the study using epidemiologic notations is summarized in the table (6). The data suggest a modest increased risk for heterozygotes (odds ratio OR of 1.5) and a markedly increased risk for homozygotes (OR 17.6).  Hugot et al. (2) found three NOD2 variants independently associated with CD: one frameshift and two missense mutations.  The frameshift mutation seems identical to that found by Ogura et al, although the notation differs (7). The investigators compared the frequency of these mutations in 468 CD cases and 103 controls.  Controls were “unaffected [and] unrelated,” but selection criteria and characteristics of subjects were otherwise not described.  The data (6) suggest a moderately increased risk for heterozygotes (OR 2.9) and a markedly increased risk for homozygotes (OR 40+). Both articles present arguments for the biological plausibility of these findings, suggesting that the interaction of NOD2 gene products with bacterial components provides a molecular model for the pathogenesis of CD.

Public Health Implications

The implications of these studies for the prevention and treatment of CD are presently unclear. The findings have to be replicated in well-designed population-based studies.  In the present studies, potential confounding and selection bias resulting from the lack of comparability of cases and controls can affect inference.  Moreover, even if the associations presented here are causal, both authors acknowledge that the NOD2 variants they identified account for only a fraction of CD cases.  The highest attributable fraction (AF) that could be calculated from the published data (6) is 33% for the combined group of simple and compound heterozygotes and homozygotes. In addition, even with high relative risks, the positive predictive values for CD (absolute risk, penetrance) for these variants appear to be low (< 3 per 1000 for heterozygotes and <5% for homozygotes), suggesting that other genetic and environmental factors are associated with CD risk.  Nevertheless, these studies could shed light on the etiology and pathogenesis of CD.

References

1. Hugot JP et al. Nature, 2001;411:599-603. 
2. Ogura Y et al. Nature 2001:411:603-606. 
3. Hugot JP et al. electronic journal club abstraction form 
4. Ogura et al. electronic journal club abstraction form 
5. HuGE published literature
6. Table with epidemiologic analysis 
7. NOD2 protein OMIM entry