Skip directly to local search Skip directly to A to Z list Skip directly to navigation Skip directly to site content Skip directly to page options
CDC Home

HuGENet™ Case Study

NOD2 and Crohn Disease

In May and June 2001, three groups of investigators reported that variants of NOD2 were associated with Crohn disease.

Ogura Y, Bonen DK, Inohara N, et al. A frameshift mutation in NOD2 associated with susceptibility to Crohn disease. Nature 2001;411:603-606.

Hugot J-P, Chamaillard M, Zouali H, et al. Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn disease. Nature 2001;411:599-603.

Hampe J, Cuthbert A, Croucher PJP, et al. Association between insertion mutation in NOD2 gene and Crohn disease in German and British populations. Lancet 2001;357:1925-1928.

Ogura et al. first described the NOD2 gene; they selected it as a candidate for Crohn disease because:

  • it maps to the IBD1 region on chromosome 16, previously identified as a susceptibility locus for Crohn disease.
  • it belongs to a class of genes whose products have a role in recognizing bacterial components, and an abnormal inflammatory response to enteric bacteria has been proposed as a pathogenetic mechanism in Crohn disease.

These investigators studied 416 Crohn disease families, of which 365 had more than one affected member. They sequenced key regions of NOD2 from 12 affected people and found that three had an insertion mutation (3020insC), which was absent in four controls. They then typed parents and one affected child from each family for the 3020insC variant. In 56 families, at least one parent had the variant, which was preferentially transmitted to offspring with Crohn disease (transmitted 39, not transmitted 17; p=0.0046 by transmission disequilibrium test).

 

Question 1: How does this finding implicate NOD2 in Crohn disease?
Could NOD2 be associated with Crohn disease
-- in the 17 people who did not inherit the 3020insC variant from their parents?
-- in people from families without other affected members?

[view answer]

Question 1 - Answer

The NOD2 3020insC variant was highly associated with Crohn disease (CD) in families in which at least one parent had the variant (13% of all families in the study).  

Other NOD2 variants—or other genes---could be involved in other cases of CD.   The authors noted that because 365 of the 416 families had more than one affected member, the association of NOD2 3020insC with “sporadic” CD requires further study.

No conclusions can be drawn from these data about the relative risk for CD in people with the variant, or the fraction of CD cases attributable to the variant.

[Close Window]

 

line

To further investigate the association of NOD2 3020insC with Crohn disease, the investigators performed a case-control study. Cases were not related (only one affected member was selected from each of the 416 families); controls were drawn from 4 available groups of DNA samples obtained from people without Crohn disease. The authors reported that all cases and controls were Caucasian, and that allele frequencies were similar in Jewish and non-Jewish cases. No further information was provided about the cases or controls.

 

Question 2: At minimum, what other information about cases and controls would be important to know?

[view answer]

Question 2 - Answer

CD can be difficult to diagnose, and no agreement exists about diagnostic criteria.  Therefore, knowing how CD was diagnosed  in cases (ie, the case definition) and how it was excluded in controls is particularly important.  More information is needed about the control populations: were they population-based or hospital-based?  Were participants asked about family history of IBD? 

Cases were from clinical populations studied at three major medical referral centers in Chicago, Baltimore, and Pittsburgh.  Allele frequencies in controls from San Francisco were similar to those in Chicago and Baltimore.  However, the German control group had a somewhat (though not statistically) higher frequency of the 3020insC variant.  Could any of these people have had undiagnosed CD?
[Close Window]

 

line

The distributions of NOD2 3020insC in case and control groups were presented as allele frequencies, which were significantly different by the large-sample approximation to a two-sample binomial test (p = 0.0018).

Allele frequency of 3020insC in unrelated Crohn disease patients and controls*

  Crohn disease   Controls
Source n % Source n %
Chicago
212
7.3
Chicago
65
3.8
Baltimore
88
6.8
Baltimore
46
3.2
Pittsburgh
116
10.8
San Francisco
81
3.1
      Germany
94
5.3
Total
416
8.2
Total
287
4.0

*Adapted from Ogura et al.

 

Question 3: How different were the distributions of the NOD2 3020insC variant in cases and controls?

[view answer]

Question 3 - Answer

The NOD2 3020insC allele was found among controls, but only about half as often as among cases.  The allele frequency among Baltimore cases (6.8%) was about twice as high as in Baltimore controls (3.2%), but not very different from German controls (5.3%).  These data demonstrate a clear association of the NOD2 3020insC variant with CD, but suggest it is neither a necessary nor sufficient cause. 

[Close Window]

 

line

Genotype frequencies among cases and controls were presented only in the text; they are summarized in the following table.

NOD2 genotypes in unrelated Crohn disease patients and controls*

Genotype Crohn disease Controls OR 95% CI
+/+
359
264
ref
ins/+
46
23
1.5
(0.8-2.5)
ins/ins
11
0†
17.6
(1.3-very high)
Total
416
287

*Adapted from Table 2, Ogura et al: + = wild type, ins = 3020insC, OR=odds ratio
†Expected number of controls=(0.04)2x287=0.4592, assuming Hardy-Weinberg equilibrium and prevalence of Crohn disease 1 per 1000 people.

Because no NOD2 3020insC homozygotes were found among controls, the expected number (assuming Hardy-Weinberg equilibrium) was used to estimate the corresponding odds ratio (OR).

 

Question 4: How strong was the association between NOD2 3020insC and Crohn disease? Are heterozygotes at increased risk?

[view answer]

Question 4 - Answer

NOD2 3020insC homozygotes were 18 times more likely than people without this variant to have CD, but accounted for only 11 of 416 cases.  Heterozygotes appeared to be at slightly increased risk, but this result was not statistically significant (95% CI 0.8-2.5).  Approximately 8% of controls were heterozygotes. 

[Close Window]

 

line

One way to quantify the contribution of the NOD2 3020insC variant to the occurrence of Crohn disease is to calculate the associated attributable fraction. One way to do this is with the formula of Miettinen:
attributable fraction = f c (R - 1) / R
where f c is the fraction of cases with the risk factor and R is the measure of relative risk (or odds ratio for rare diseases). If we perform this calculation using the data in the chart,

AF(homozygotes) = (11/416)(17.6 - 1)/17.6 = 2.5%
AF(homozygotes+heterozygotes) = (57/416)(1.8-1)/1.8 = 6.0%

 

Question 5: What assumptions are required to estimate an attributable fraction from these data? If the result is correct, what does it suggest about the role of NOD2 in Crohn disease at the population level?

[view answer]

Question 5 - Answer

Using case-control study results to estimate attributable fraction is reasonable only if the cases and controls are comparable and population-based; we can't really assume this from the data provided.  Although NOD2 3020insC is clearly a strong risk factor for CD (OR ~ 18 for homozygotes), it accounts for only a small proportion of cases, even in this selected group of families.  Although heterozygotes did not have a significantly elevated OR, they were much more prevalent than homozygotes among cases, contributing to a more than double estimate of attributable fraction. 

[Close Window]

 

line

Hugot et al., who first described the IBD1 locus on chromosome 16, had already examined and rejected two other candidate genes for Crohn disease (CD19 and CD43) in this region. In this study, they used a positional cloning strategy to identify three different single nucleotide polymorphisms (SNPs) associated with Crohn disease. Their approach involved typing 26 microsatellite markers in the IBD1 region in 77 multiplex families, followed by linkage analysis and linkage disequilibrium mapping to hone in on the most promising region.

This region did not include any obvious candidate genes but contained a number of transcribed regions that could be identified by expressed sequence tags. Eventually, 13 SNPs were examined for linkage disequilibrium, and three SNPs (8, 12 and 13) were independently associated with Crohn disease in 235 families. These three SNPs never occurred on the same haplotype. The strongest association was found for SNP 13, which codes for a one base-pair insertion identical to that found by Ogura et al. All three SNPs proved to be variants of the NOD2 gene.

The investigators presented allele and genotype frequencies for 468 unrelated people with Crohn disease, 159 people with ulcerative colitis, and 103 unaffected people (78 unaffected spouses of people with Crohn disease and 25 relatives of the French investigators). No additional information was provided about the study subjects.

Allele frequencies of three NOD2 variants associated with Crohn disease*

Condition # chromosomes SNP8 SNP12 SNP13† Total
Unaffected
206
0.04
0.01
0.02
0.07
Ulcerative colitis
318
0.03
0.00
0.01
0.05
Crohn disease
936
0.11
0.06
0.12
0.29

* Adapted from Hugot et al.
† Identical to the NOD2 3020insC allele described by Ogura et al.

 

Question 6: How did frequencies of these variants compare in the Crohn disease and unaffected control groups? How did frequencies of SNP 13 compare with the results of Ogura et al.?

[view answer]

Question 6 - Answer

All three variants were relatively rare but much more common in people with CD—among 936 chromosomes, 29% had at least one variant.  SNP13 (=3020insC) was somewhat more common in this population of patients with CD than in Ogura et al. (12% vs. 8.2%) and less common among controls (2% vs. 4%), but overall, the findings were similar.

[Close Window]

 

line

Question 7: Were any of the three variants associated with ulcerative colitis? Why is this finding important?

[view answer]

Question 7 - Answer

None of these variants appear to be associated with ulcerative colitis (UC), because their frequencies were similar to those in controls.  Ogura et al. reported a similar finding, which could help identify potential differences in pathogenetic mechanisms for CD and UC.  These disorders have overlapping phenotypes, and drug therapy is mostly undifferentiated. 

[Close Window]

 

line

Hugot et al. also presented genotype frequencies, summarizing data for the 16 possible genotypes to calculate odds ratios.

NOD2 genotypes in Crohn disease patients and controls*
Genotype Crohn disease Controls OR 95% CI
Wild type
267
88
ref
 
Simple heterozygous
133
15
2.9
(1.6-5.5)
Homozygous
28
0† a
42‡
(1.7-very high)
Compound heterozygous
40
0† b
45‡
(2.4-very high)
Total
468
103
   

*Adapted from Hugot et al.; OR=odds ratio.
†Assuming Hardy-Weinberg equilibrium and prevalence of Crohn disease 1 per 1000, expected values are a=0.22, b=0.29 .
‡Differ slightly from those reported (which may contain rounding error).

 

Question 8: How were these NOD2 variants associated with Crohn disease?

[view answer]

Question 8 - Answer

People with one of these variants were 3 times more likely to have CD. In people with two variants (whether the same or different), risk was increased 40-fold. The authors suggested that this pattern could indicate a “recessive” trait, at least in this sample of cases, of whom half came from multiplex families.

[Close Window]

 

line

If we make similar assumptions and calculate attributable fractions as for the other study,
AF(homozygotes+compound heterozygotes) = (68/468)(44.4-1)/44.4 = 14%
AF(homozygotes+all heterozygotes) = (201/468)(4.3-1)/4.3 = 33%


Question 9: How important are these NOD2 variants as risk factors for Crohn disease at the population level?

[view answer]

Question 9 - Answer

These three variants accounted for a sizeable proportion of CD cases in this study population.  Nevertheless, most people with CD did not have even one of them.  As in the other study, attributable fraction estimates should be regarded with caution.   Without more information, we cannot assume that the cases are population-based or that the controls are drawn from the same population.

[Close Window]

 

line

Hugot et al. also presented estimates of the absolute risk of Crohn disease by genotype, using the estimated relative risks described above and assuming that the total population risk was 1 per 1000 people.

 

Question 10: What is the estimated absolute risk (incidence) of Crohn disease by genotype?

[view answer]

Question 10 - Answer

Assume that OR ~ RR and total population risk is 0.001.
Let x be absolute risk of CD in people without any of the NOD2 variants (wild type).
Apportion population risk using prevalence and risk data from the table on previous page:            

wt simple homo compound  
(88/103) x
+ (15/103)(3) x
+ (0.22/103)(40) x
+ (0.29/103)(40) x
= 0.001

Solving for x =0.0007, we can estimate an approximate absolute risk for each group:

Genotype (0.7)(OR) absolute risk per 1000
Wild type
ref 
0.7
Simple heterozygous
(0.7)(2.9)
~  2 
Homozygous
(0.7)(42)
~30
Compound heterozygous
(0.7)(45)
~30

[Close Window]

 

line

In June 2001, a third study by Hampe et al. reported an association with Crohn disease for the insertion mutation of NOD2 reported by both Ogura and Hugot. These investigators also found significantly increased odds ratios for insertion variant heterozygotes (OR=2.6) and homozygotes (OR=42.1, assuming Hardy-Weinberg equilibrium and population prevalence of Crohn disease of 4 per 10,000 people).

All three authors highlighted the importance of this finding for improving understanding of the pathogenesis of Crohn disease. The NOD2 gene product is thought to be important in the innate human immune response, functioning as an intracellular receptor for bacterial lipopolysaccharides and activating NF-kB. The frameshift mutation NOD2 3020insC encodes a truncated protein, which Ogura et al. showed in vitro to have diminished ability to confer responsiveness. This observation suggests several ways that a faulty immune response to bacteria could contribute to the Crohn disease process.

 

Question 11: Considered together, what do the findings of these three studies mean for researchers investigating the genetic basis of Crohn disease? For people with one or more of these NOD2 variants? For people with Crohn disease?

[view answer]

Question 11 - Answer

For researchers:
Together, these findings represent a step forward in understanding the genetics of CD.  However, there is more work to do.  Additional population-based studies need to confirm these findings in patients with “sporadic” CD (i.e., not drawn from families with multiple affected members), comparing them with carefully chosen control groups.  Both Ogura and Hugot noted that genetic susceptibility to CD is not limited to chromosome 16; at least five other loci have been implicated and await further elucidation.  Additional studies should be designed to address the interaction of NOD2 (and other potential susceptibility genes) with environmental exposures in the pathogenesis of CD.

For people with one or more of the implicated NOD2 variants:
These studies show that even people homozygous for these variants have a better-than-even chance of remaining free of CD. For now, there is nothing to be gained by testing family members of affected people for susceptibility to CD.

For people with Crohn disease:
Eventually, better understanding of pathogenetic mechanisms is likely to benefit people with CD by permitting more efficient clinical trials and suggesting more specific therapeutic interventions.  Epidemiologic studies of gene-environment interaction may eventually help identify ways to prevent flare-ups in people with CD, and perhaps even prevent disease onset in some susceptible people.

[Close Window]

 

line

View all answers.

 

line

 

 

Contact Us:
  • Centers for Disease Control and Prevention
    1600 Clifton Rd. Atlanta, GA 30333 USA
    800-CDC-INFO (800-232-4636)
  • Additional information for Public Health Genomics is available on our contact page.
USA.gov: The U.S. Government's Official Web PortalDepartment of Health and Human Services
Centers for Disease Control and Prevention   1600 Clifton Rd. Atlanta, GA 30333, USA
800-CDC-INFO (800-232-4636) TTY: (888) 232-6348 - Contact CDC–INFO
A-Z Index
  1. A
  2. B
  3. C
  4. D
  5. E
  6. F
  7. G
  8. H
  9. I
  10. J
  11. K
  12. L
  13. M
  14. N
  15. O
  16. P
  17. Q
  18. R
  19. S
  20. T
  21. U
  22. V
  23. W
  24. X
  25. Y
  26. Z
  27. #