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?

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?

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?

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?

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?

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

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?

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?

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?

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 all answers .