Introduction

Colorectal cancer (CRC) ranks as the second leading cause of cancer-related mortality in western countries and the third most common malignancy in the world, causing some 500,000 annual deaths worldwide, and recently the prevalence of this carcinoma has been increasing [1–4]. CRC incurs an annual expenditure of more than £300 million in surgical, adjuvant, and palliative treatment [5]. Widely accepted is that cancer is a disease caused by accumulation of mutations in specific gene [6]. CRC is believed to develop slowly via a progressive accumulation of genetic mutations [2], and it has been a model for investigating the molecular genetics of cancer development and progression [7]. The identification of the important CRC-related genes may help facilitate the early diagnosis, prevention, and treatment of CRC [8]. Many researchers have been drawn to study the genetic basis of sporadic CRC. And the study of the underlying molecular genetics and biology associated with the development and progression of CRC has led to the significant treatment advances over the past 10 years [9]. In this article, we are highlighting those genetic polymorphisms which may be regarded to genetic predisposition of CRC.

The nucleotide oligomerization domain 2 (NOD2/CARD15) gene situates at chromosome 16q12 within the inflammatory bowel disease (IBD) 1 region [10–12]. NOD2 gene is characterized by a tripartite structure with a C-terminal sensor domain (leucine-rich repeats, LRRs), a central nucleotide binding and oligomerization (NOD or NACHT) domain, and an N-terminal effector domain (CARD) [13]. It is a cytoplasmic molecule involved in sensing microbial cell wall components and regulating inflammatory processes and apoptosis [14–16]. In the past few years, NOD2 has become known as key regulator of chronic inflammatory conditions [17] and polymorphisms in NOD2 have been associated with increased susceptibility to Crohn’s disease (CD) [10–12], a human chronic IBD. Recently, investigation of the susceptibility loci governed by polymorphic alleles, particularly those of the innate immune response, is growing [18]. It was believed that chronic inflammation favors tumorigenesis by stimulating cell proliferation and angiogenesis and by inducing DNA damage [19–21]. Patients with CD have a higher risk of developing CRC [22], and a meta-analysis has also revealed a significantly increased risk of CRC in CD [23]. Researchers began to be interested in whether hereditary susceptibility genes in CD also played roles in CRC, which caused them to investigate a possible influence of NOD2 on the development of CRC.

Four major NOD2 single nucleotide polymorphisms have been described as genetic risk factors of CD: one background polymorphism P268S/SNP5, two missense mutations R702W/SNP8, G908R/SNP12, and one frameshift mutation 3020insC(1,007 fs)/SNP13 [11, 24]. Since Kurzawski et al. described a potential association between 3020insC and the risk of CRC in 2004, possible association of the NOD2 polymorphisms P268S, R702W, G908R, and 3020insC with CRC has been studied among Polish, Greek, Finnish, Hungarian, and New Zealand Caucasian CRC patients [25–32]. However, results are controversial. To clarify the role of NOD2 polymorphisms in the development of CRC, we undertook a systematic review and meta-analysis of published studies.

Materials and methods

Search strategy and selection criteria

The published Quality of Reporting of Meta-analysis (QUOROM statement) was followed [33]. Studies investigating the relationship between CRC and NOD2 polymorphisms were carried out by searching for articles written in English in PubMed, EMBASE, and the Science Citation Index, and limited our search to English papers, from the inception of each database to May 2009. Various combinations of the terms “CRC”, “colon cancer”, “rectal cancer”, “polymorphism,” and “NOD2/CARD15” were used to screen for potentially relevant studies. Inclusion and exclusion criteria: case-control or cohort studies presenting original data on associations between CRC and NOD2 polymorphisms were included.

Data extraction

For all studies, we extracted the following data from original publications: first author and year of publication; genes and relevant polymorphisms; characteristics of the study design and the study population (numbers of cases and controls, matching criteria, study base and host ethnicity). Two reviewers independently extracted data. Disagreements were resolved by discussion. Authors were contacted for further information when necessary.

Statistical analysis

The strength of the associations between CRC and NOD2 polymorphisms was estimated by odds ratios (OR) and 95% confidence intervals (CI). We addressed the association between the presence of at least one high-risk allele and CRC susceptibility and the effect of each high-risk allele was examined separately.

For the rare frequencies of NOD2 P286S, R702W, G908R, and 3020insC homozygous mutants (+/+), we estimated the risk of the homozygous mutants (+/+) and heterozygous (−/+) versus homozygous wild-types (−/−), respectively.

The software HWE (http://linkage.rockefeller.edu/ott/linkutil.htm) was used to evaluate the deviation from Hardy–Weinberg equilibrium in controls. Heterogeneity, evaluated by the Cochrane Q-test among the studies, was considered significant for P < 0.05. The data were combined using both fixed effects and random effects models. Random effects are more appropriate when heterogeneity is present. The analyses were performed with the computer programs Review Manage, version 5.0 (Oxford, England, UK). All P values are two-sided.

Results

Study characteristics

Eighty-five papers relevant to the words searched were retrieved (Fig. 1). Through the step of screening the title, 39 of these articles were excluded (duplicated). Abstracts from 46 articles were reviewed and an additional 20 trials were excluded (clearly not relevant), leaving 26 studies for detailed review. Of these, 18 were excluded (did not meet detailed inclusion criteria); thus, eight papers [25–32], which included 3,524 CRC cases and 2,364 controls, were found to conform to our inclusion criteria. At last, eight case-control studies were included in this meta-analysis. Studies were carried out in Poland, Finland, Greece, New Zealand, and Hungary. A list of details abstracted from the studies included in the meta-analysis is provided through Table 1. The most commonly investigated genotypes were NOD2 P286S, R702W, G908R, and 3020insC, which were reported in two, five, five, and seven studies, respectively. Table 2 showed the allele frequencies and percentage of NOD2 polymorphism carriers among CRC cases and control. Most studies used healthy volunteers or blood donors as control subjects. We assessed deviation from Hardy–Weinberg equilibrium by the HWE program, and the results showed that the genotype distribution of control population in all the eight included studies were in Hardy–Weinberg equilibrium. The publishing year of the included studies ranged from 2004 to 2008. Appropriate molecular methods for genotyping were stated in all studies, all of which was polymerase chain reaction restriction fragment length polymorphism.

Fig. 1
figure 1

Process of study selection

Full size image
Table 1 The baseline characteristics of the studies
Full size table
Table 2 Allele frequencies and percentage of NOD2 polymorphisms carriers among CRC cases and controls
Full size table

Meta-analysis databases

NOD2 P286S polymorphism

Two studies evaluated the NOD2 P286S allele were included in this meta-analysis [29, 31]. Because of the existed heterogeneity (P < 0.05), random model was used. No evidence indicated that individuals carrying the variant genotypes (−/+ · +/+), compared with those carrying the homozygous wild-type (−/−), had an increased risk of CRC (OR 1.27, 95% CI 0.32–5.00, P = 0.73; Fig. 2).

Fig. 2
figure 2

Forest plots of OR with 95% CI for NOD2 P286S polymorphisms and risk of CRC (−/+ ⋅ +/+ versus −/−)

Full size image

NOD2 R702W polymorphism

The dominant model of the NOD2 R702W allele was conducted in this analysis. The results from five studies [27–29, 31, 32] showed that individuals with the variant genotypes (−/+ ⋅ +/+) had significant risk of CRC compared with those with the homozygous wild-type (−/−; OR 1.59, 95% CI 1.09–2.32, P = 0.02), and there was some heterogeneity among these studies (Fig. 3).

Fig. 3
figure 3

Forest plots of OR with 95% CI for NOD2 R702W polymorphisms and risk of CRC (−/+ ⋅ +/+ versus −/−)

Full size image

NOD2 G908R polymorphism

We included five studies [27–29, 31, 32] observing the NOD2 G908R allele, the results showed significant difference between individuals carrying the variant genotypes (−/+ ⋅ +/+) with those carrying the homozygous wild-type (−/−; OR 1.98, 95% CI 1.14–3.44, P = 0.01), which suggested that people carrying the variant genotypes of NOD2 G908R had an increased risk of CRC, compared with those carrying the wild-type. No between-study heterogeneity was found in this analysis (Fig. 4).

Fig. 4
figure 4

Forest plots of OR with 95% CI for NOD2 G908R polymorphisms and risk of CRC (−/+ ⋅ +/+ versus −/−)

Full size image

NOD2 3020insC polymorphism

Seven studies of the NOD2 3020insC were enrolled in this analysis [25–31]. The combined results based on these seven studies showed that, compared with those with the homozygous wild-type (−/−), there was significant risk of CRC of individuals with the variant genotypes (−/+ ⋅ +/+; OR 1.44, 95% CI 1.13–1.84, P = 0.003), and no heterogeneity was indicated (Fig. 5).

Fig. 5
figure 5

Forest plots of OR with 95% CI for NOD2 3020insC polymorphisms and risk of CRC (−/+ ⋅ +/+ versus −/−)

Full size image

People carrying at least one of the variant genotypes

Five of eight studies were eligible for assessing the impact of at least one of NOD2 variant genotypes on the CRC risk [25–28, 30, 31]. Because there was the statistical heterogeneity between studies (P < 0.05), the random effects mode was applied. Among the populations in the included studies, the presence of at least one high-risk allele conferred greater risk for CRC (OR 1.58, 95% CI 103–2.42, P = 0.03; Fig. 6).

Fig. 6
figure 6

Forest plots of OR with 95% CI for at least one high-risk allele of NOD2 polymorphisms and risk of CRC

Full size image

Discussion

NOD2 could be an especially important part of innate immunity for maintenance of the intestinal barrier [34], and it is already proved that the alteration in the NOD2 gene participates in the development of IBD. The NOD family of proteins is mainly expressed in monocytes, macrophages, and B cells [35, 36]. In eukaryotes, nuclear factor κB (NF-κB), whose activity is regulated by, among others, NOD2 protein, plays an essential role in the regulation of basic processes of the organism, including immune response, apoptosis, cell cycle control, and the development of individual cell lines [37]. Mutations in the LRRs of the NOD2 gene could disturb NF-κB activation [12, 36]. Due to the crucial involvement of NF-κB in the regulation of cell division mechanisms, it was attributed with an uncertain role in the process of cancer development, where its activity is significantly elevated. Already in the mid-1990s, the importance of NF-κB was described in relation to cancer of the thyroid, breast, lung, and colorectum [38–41]. The loss of NOD2 gene function is predicted to result in excessive NF-κB activity, leading to an inflammation-dysplasia-carcinoma sequence [18]. The literature includes increasing evidence indicating that NOD2 polymorphisms can be considered predisposition factors to malignancy development, such as gastric, bowel, breast cancer [18, 42], and non-Hodgkin’s lymphoma [43].

This is the first meta-analysis of the association of NOD2 polymorphisms with susceptibility to CRC. We quantified distinct risks for CRC of the four common NOD2 polymorphisms, with narrow confidence intervals. On the basis of eight studies, providing case and control numbers of the NOD2 polymorphisms and CRC risk in Caucasians, including 5,888 subjects (cases:3,524; controls:2,364), our meta-analysis provided good evidence that NOD2 R702W, G908R, and 3020insC polymorphisms were associated with increased risk of CRC. P268S, however, was not shown to have impact on CRC risk; and the effect of 3020insC was most significant. Due to the limited studies (two) involved in the association of P268S with CRC, the pooled results may not be of high reliability. More studies are needed to assess the risk impact of NOD2 P268S.

NOD2 polymorphisms have previously been studied in relation to CD susceptibility [10, 11], and it has already been proved that CD patients are susceptibility to CRC [23]. Kurzawski et al. for the first time associated NOD2 polymorphisms with CRC, and drew a conclusion that NOD2 polymorphisms increased CRC risk, from then seven case-control studies involved in this field, and concluded conflicting results [25–32]. Several factors may contribute to the differences among the researches. First, genetic heterogeneity may be a reason for the conflicting results. The frequency of NOD2 polymorphisms vary considerably between races [44]. Such as in Asia, the frequency of NOD2 polymorphisms is very low, and in a large Japanese cohort of CD patients, none of the R702W, G908R, and 3020insC was found [45]. In Australian population, a lower background allele frequency of 3020insC has been observed [46]. Even in European, a low frequency of the NOD2 polymorphisms in the northern countries, compared with the rest of Europe, has been demonstrated [47]. In this meta-analysis, of the included eight studies, there is only one research on evaluating the NOD2 polymorphisms in relation to CRC outside Europe [29], and the contribution of NOD2 polymorphisms to CRC susceptibility varies in different studies.

Second, variation in patient characteristic (e.g., in terms of age and years from onset) might potentially also contribute to the differences among the researches. The results of Kurzawski et al. indicated that NOD2 may be a predisposing factor to CRC characterized by an older average age (>50 years of age) of disease onset in persons [26]. However, Alhopuro et al. concluded contrary results [25]. Some researchers did not consider age onset stratification in their studies. No definite conclusion about the relation between age and CRC risk is available so far. More careful stratification analysis that takes into account the age of onset is needed. However, due to the restricted sample size such differences were not elaborated.

Furthermore, differences in carcinogenic exposure may modify the inherent risk associated with genetic susceptibility [47]. Modifiable risk factors such as physical activity, weight, and diet, just as genetic risk factors have been implicated in the development of CRC [48]. For instance, the average monthly intake of alcohol may have influence on CRC accident [49, 50]. Some research has further provided strong evidence that smoking is associated with an increased risk of CRC [51, 52]. The interaction between NOD2 polymorphisms and environmental carcinogens may represent one of the mechanisms by which NOD2 polymorphisms increase the susceptibility to CRC.

One of the studies involved in our meta-analysis investigated gene–gene interaction. Suchy et al. found the association of TNFα-1,031 T/T genotype and the NOD2 3020insC polymorphisms may act as low risk modifiers of CRC risk [30]. More researches on NOD2 polymorphisms gene–gene interactions will provide a more comprehensive insight into the associations studied here.

Whereas, inherited susceptibility is responsible for about 35% of all CRC [53], high-risk germline mutations account for < 6% of all cases [54]. In addition to the four newly identified loci, so far, there are only ten identified CRC-associated loci: 14q22.2 (rs4444235, BMP4), 16q22.1 (rs9929218, CDH1), 19q13.1(rs10411210, RHPN2), 20p12.3 (rs961253) [55], 8q24 (rs6983267) [56, 57], 8q23.3 (rs16892766, EIF3H), 10p14 (rs10795668) [58], 11q23(rs3802842) [59], 18q21 (rs4939827, SMAD7) [58, 59], and 15q13 (rs4779584) [60]. If NOD2 is another CRC-associated locus, more studies or larger case-control studies about the association of NOD2 polymorphisms and CRC, genome-wide association studies, replication analyses, and cytogenetic tests should be performed. For the study of the underlying molecular genetics and biology associated with the development and progression of CRC will lead to treatment advances, we may benefit from further study on clarifying the possible roles of NOD2 polymorphisms in CRC. Whereas our results found NOD2 polymorphisms seem to confer considerable risk for CRC, most genetic risk factors for complex diseases have a much smaller impact [61–63]. Clarify the role played by NOD2 polymorphisms in development of CRC still need further research [64].

In conclusion, regardless of age, gender, or presence of symptoms of subjects, deriving data from eight published articles, our meta-analysis suggests that R702W, G908R, and 3020insC increase the susceptibility to CRC in Caucasians, and P268S was not shown to impact on CRC risk. However, due to the small number of studies addressing this association of NOD2 and CRC, whether R702W, G908R, and 3020insC increase the susceptibility to CRC in other ethnicity and other NOD2 polymorphisms provide the same effects to CRC risk in different populations requires further investigation.