No evidence for association of NOD2 R702W and G908R with colorectal cancer



Polymorphisms in nucleotide oligomerization domain 2 gene (NOD2) have been associated with increased susceptibility to Crohn's disease. Recently, possible association of the NOD2 variants R702W, G908R and 3020insC with colorectal cancer (CRC) has been studied among Polish, Greek, Finnish and New Zealand Caucasian CRC patients, but the results have been controversial. In the Polish study, 3020insC alone, and in the Greek study all the 3 variants, showed association with CRC. In a study from New Zealand, R702W appeared to increase CRC risk. In addition, the combined frequencies of the 3 variants were significantly elevated in CRC patients compared with healthy controls. We have previously shown that NOD2 3020insC is not associated with increased CRC risk in Finland. In the current study, we have genotyped the R702W and G908R in a population-based series of 1,042 CRC patients and in 508 healthy controls to study the possible contribution of these variants to CRC predisposition. Of the CRC patients, 953 were successfully analyzed. R702W and G908R were equally, frequently seen in CRC patients and controls (R702W: 2.2% vs. 2.1%; G908R: 0.3% vs. 0.2%). No associations between NOD2 variants and clinical characteristics were observed. Our results indicate that NOD2 variants R702W, G908R and 3020insC do not predispose to CRC in Finland. Environmental or additional genetic factors may play a role in CRC development in NOD2 variant carriers. Further work is necessary to establish the possible role of NOD2 variants in CRC predisposition. © 2007 Wiley-Liss, Inc.

Three variants R702W, G908R and 3020insC in the nucleotide oligomerization domain 2 gene, (NOD2, also known as CARD15) have been established to contribute to Crohn's disease in Caucasian populations.1, 2, 3, 4, 5 The strongest evidence has been provided for the 3020insC, which has been reported to associate with Crohn's disease with an odds ratio of 4.09 (95% CI: 3.23–5.18) in a meta-analysis combining results from 42 independent studies.6 In the Asian cohorts studied, NOD2 variants are rare and do not contribute to Crohn's disease, suggesting ethnic heterogeneity in disease predisposition.7, 8, 9 In Europe, the disease susceptibility caused by the R702W, G908R and 3020insC variants appears less prominent in Finland, Sweden, Iceland, Scotland and Ireland, than in other populations.1, 3, 4, 10, 11 For instance, in Scotland and Ireland the population-attributable risk of the 3 common NOD2 variants is estimated to be only a third of that in other Caucasian populations.10 Additional genetic as well as environmental factors, therefore, seem to play a role in the development of Crohn's disease. NOD2 variants show no association with purely colonic Crohn's disease, and to our knowledge no reports have described an increased risk of colorectal cancer (CRC) in Crohn's disease patients with NOD2 variants.5, 6, 11

A Polish study described an association of the NOD2 3020insC variant with CRC.12 Most of the CRC patients were from Szczecin area, and an increased allele frequency was observed in a subgroup of patients >50 years at disease diagnosis (n = 250, OR = 2.23, 95% CI: 1.23–4.1).12 Subsequently, the possible contribution of the NOD2 variants to CRC risk has been examined in 3 studies with inconclusive results. A Greek study described an association of all 3 variants with CRC in a consecutive series of 104 CRC cases.13 Another study reported no association of 3020insC with CRC predisposition in a Finnish population-based series of 1042 CRC cases.14 Moreover, a study from New Zealand evaluated the 3 most common NOD2 variants; R702W, G908R and 3020insC, in a series of 133 Caucasian CRC cases, but only R702W was shown to independently contribute to CRC risk (OR = 2.3, 95% CI: 1.1–5).15 Interestingly, evidence for association of the combined frequency of the 3 common NOD2 variants with CRC risk was obtained in the New Zealand population (OR = 2.8, 95% CI: 1.5–5.4).

The study from New Zealand prompted us to extend our previous work14 and to examine, whether occurrence of the 2 NOD2 variants (R702W or G908R) would associate with CRC. An extensive population-based series of Finnish CRC patients as well as 508 healthy blood donors were screened for R702W and G908R to derive a comprehensive picture of NOD2 variants and possible CRC predisposition in Finns.

Material and methods

Study participants

A population-based series of 1,042 Finnish colorectal cancer (CRC) patients was studied for nucleotide oligomerization domain 2 gene (NOD2) variants R702W and G908R. This sample series, including normal and tumor DNA specimens, was collected previously from 9 Finnish central hospitals and is described in more detail elsewhere.16, 17 The normal tissue DNA was extracted from blood or normal colonic epithelium distant from the tumor margins. The detailed clinical data, including age at diagnosis, sex and family history of the patients, pathology reports and microsatellity instability (MSI) status of the tumors, were available from all of the samples. DNA samples from 508 anonymous Finnish cancer-free blood donors were available from the Finnish Red Cross Blood Transfusion Service and were used as controls. Sample size in R702W genotyping was calculated to be sufficient to discern a similar association than in previous studies (80% statistical power, p < 0.05).13, 15

Genotype analyses

The R702W and G908R variants were genotyped in patients and control subjects using a previously described multiplex amplification refractory system.18 Positive controls for both variants were used in each polymerase chain reaction (PCR). All variant carriers were verified by direct sequencing. The fragments were amplified by PCR using primers 5′CCCAGCTCCTCCCTCTTC 3′ and 5′AAGTCTGTAATGTAAAGCCAC 3′ for G908R,13 and 5′ AGATCACAGCAGCCTTCCTG 3′ and 5′CTCTTGGCCTCACCCGGT 3′ for R702W and the PCR products were sequenced directly using Applied Biosystems (Foster City, CA) BigDye v3.1 sequencing chemistry and ABI3730 Automatic DNA sequencer supplied with the sequencing analysis software v5.0.

Statistical methods

The significance of the differences in the allele frequencies between CRC patients and healthy controls as well as association of variants with the clinical and tumor characteristics was tested using Pearson's χ2test. Odds ratios (OR) were determined with 95% confidence intervals (95% CI). P < 0.05 was considered significant. Power calculations were performed using a web-based algorithm (


The frequencies of the NOD2 variants R702W and G908R were determined in 1,042 Finnish CRC patients and in 508 healthy population-matched controls. A total of 954 patients were successfully analyzed for R702W and 960 for G908R, and results from both variants were available from 953 patients. In CRC patients, allele frequency for R702W was 2.2% and for G908R 0.3% and the corresponding frequencies in controls were 2.1% and 0.2%, respectively (Table I). Only 1 patient was homozygote for R702W but none for G908R. We have previously analyzed the 3020insC in the same population-based series.14 All of the 34 mutation positive cases identified in that study were successfully analyzed in this study and in 40 of the 47 R702W or G908R carriers detected in this study, the 3020insC mutation status was known. By combining these 2 studies, only 1 compound heterozygote for 3020insC and R702W was identified. No other compound heterozygote or patients carrying all the 3 variants were found. There was no difference in the clinical picture of the compound heterozygote patient compared with other patients.

Table I. Genotype and Allele Frequencies of the 3 NOD2 Variants in Finnish CRC Patients and Controls
VariantGenotypeAllele frequency (%)P Pearson's χ2OR (95% CI)
  • 1

    This study.

  • 2

    Alhopuro et al.14

 CRC9134012.20.881.04 (0.61 – 1.78)
 CRC954600.30.571.59 (0.32 – 7.91)
 CRC8923311.90.960.98 (0.51 – 1.88)

We next examined the possible association of the R702W and G908R with the age at diagnosis, sex and family history of the patients (Table II). Patients were divided into early (<60 years at diagnosis) and late onset (≥60 years) groups as in previous studies,13, 15 to allow comparison. Familial cases were defined as patients having at least 1 first-degree relative diagnosed with CRC, whereas sporadic cases had no first-degree relatives with CRC. The frequencies of the variants were similar in familial (n = 135) and sporadic (n = 817) groups (p = 0.32; OR 1.46; CI 0.69–3.10). The NOD2 variants were equally frequently detected in both sexes (p = 0.78; OR 1.09; CI 0.60–1.95), as well as in patients diagnosed less than 60 or over 60 years of age (p = 0.38; OR 0.75; CI 0.39–1.43). We also analyzed the correlation between the NOD2 variant frequencies and tumor characteristics. No evidence for association between R702W and G908R and MSI status or Duke's stage of the tumors was obtained (Table II).

Table II. Clinical and tumor characteristics of Finnish CRC patients with or without NOD2 variants R702W and G908R
ParametersNumber of patientsCarriers (%)1NoncarriersP Pearson's χ2OR (95% CI)
  • MSI, microsatellite instable; MSS microsatellite stable.

  • 1

    Patients with R702W or G908R.

 Female46824 (5.1)4440.781.09 (0.60 – 1.95)
 Male48523 (4.7)462  
 ≥6072033 (4.6)6870.380.75 (0.39 – 1.43)
 <6023314 (6.0)219  
Family history     
 Familial CRC1359 (6.7)1260.321.46 (0.69 – 3.10)
 Nonfamilial CRC81738 (4.6)779  
MSI status     
 MSI CRC1205 (4.2)1150.680.82 (0.32 – 2.11)
 MSS CRC83342 (5.0)791  
Duke's stage     
 A+B56832 (5.6)5360.241.45 (0.78 – 2.72)
 C+D38015 (3.9)365  


We have previously shown that NOD2 3020insC is not associated with CRC in the Finnish population.14 In the current study we have extended our efforts to characterize NOD2 as a candidate susceptibility gene for CRC, and screened the R702W and the G908R variants in the same population-based series of 1,042 CRC patients, in addition to 508 healthy controls. On the basis of these studies, no significant differences in the frequencies of R702W, G908R and 3020insC between CRC patients and controls could be detected (Table I). Also the combined frequency of R702W/G908R/3020insC was similar in CRC patients and controls (p = 0.83). This indicates no clear association of any of the variants with increased risk of CRC in Finland.

Because of the low population frequency in the Finns further studies are required to ascertain the lack of association of G908R with CRC. The allele frequency of R702W did not differ notably among the Finnish and New Zealand Caucasian control cohorts (2.1% vs. 3.0%). However, despite the similar allele frequencies and sufficient statistical power to observe similar association than in the New Zealand and Greek studies,13, 15 no association of R702W was detected with Finnish CRC patients. This indicates, that the lack of association of R702W with CRC in Finland is not explained by the rarity of the variant. Very low penetrance effects cannot be excluded without larger materials. If the data from the 4 previous studies12, 13, 14, 15 and this study is combined to increase the sample size, the mutation frequencies of R702W, G908R and 3020insC do not significantly differ between CRC patients and controls; R702W, p = 0.17; G908R, p = 0.12; 3020insC, p = 0.06 (Table III). If data on all variants (R702W/G908R/3020insC) were pooled, a significant difference is observed between cases and controls (p = 0.008). Thus, further work is necessary to establish the role of NOD2 in CRC predisposition and possible differences between populations.

Table III. Combined frequencies of NOD2 variants in 4 different Caucasian populations1
VariantNumber of subjectsMutant Frequency (%)P Pearson's χ2OR (95% CI)
Wild typeMutant
  • 1

    Kurzawski et al.12 Alhopuro et al.14, Papaconstantinou et al.13, Roberts et al.18, this study.

 CRC1,123685.70.171.34 (0.88 – 2.03)
 CRC1,167302.50.121.71 (0.87 – 3.36)
 CRC1,5961237.20.061.39 (0.99 – 1.94)
 CRC1,53021612.40.0081.39 (1.09 – 1.79)

Previous studies have reported associations of NOD2 with some clinical features that have not been reproduced in subsequent studies. In the Polish study, the 3020insC was linked with an increased risk of CRC at an older age at disease onset,12 whereas in the Greek population, the NOD2 variant carriers had more frequently advanced stage tumors.13 In New Zealand, no correlation between age of onset or tumor behavior and NOD2 mutation carrier status was detected, but NOD2 variants were more frequent in male than in female patients.15 In our previous study, unlike cited elsewhere,15NOD2 variants were not associated with any clinicopathologic characteristics. Similarly, our current study also failed to show a correlation between NOD2 genotype and any of the clinical or tumor characteristics.

Our results from the 1,042 population-based CRC samples suggest, that NOD2 variants R702W, G908R and 3020insC are not as such sufficient for CRC predisposition. Some evidence for association of NOD2 variants with CRC was obtained in the 3 previous studies but not in our studies. Multiple explanations for this are possible. One critical factor for obtaining accurate results from association analyses is definitely of sufficient sample size. Evidence for NOD2 as a susceptibility gene for CRC has been obtained from relative small patient cohorts and this might have some impact on the results. In addition, some negative results might not have been published, generating publication bias. However, we acknowledge the possibility that NOD2 might act as a susceptibility gene for CRC in some other populations than Finns. In this case, we suggest that some environmental or genetic factors might contribute to effects of the germline variant, to promote colorectal tumorigenesis.


We thank Sini Marttinen, Tuula Lehtinen and Kirsi Pylvänäinen for collecting the clinical data, and Professor Kimmo Kontula and Maarit Lappalainen for providing positive controls for R702W and G908R. The work was carried out at the Center of Exellence in Translational Genome-Scale Biology of Academy of Finland.