Polymorphisms in the Insulin-Like Growth Factor Axis Are Associated with Gastrointestinal Cancer

Numerous factors influence the development of gastrointestinal (GI) cancer. The insulin-like growth factor (IGF) axis plays a role in embryonic and postnatal growth and tissue repair. Elevated levels of IGFs, low levels of IGF binding proteins (IGFBPs) and over-expression of IGF receptor (IGFR-I) were associated with several stages of cancer. Here, the prevalence of the single nucleotide polymorphisms (SNPs) rs6214 in the IGF type I ( IGF-I ) gene and rs6898743 in the growth hormone receptor ( GHR ) gene in patients with GI cancer and controls was studied.

Overall for GI cancer, the ORs for SNPs rs6214 and rs6898743 were approximately 1.0 (p-value>0.05), using the most common genotypes GG as reference. An OR of 1.54 (95% CI, 1.05–2.27) was found for EC for genotype AA of rs6214. The ORs for EAC were 1.45 (95% CI, 1.04–2.01) and 1.71 (95% CI, 1.10–2.68), for genotypes GA and AA, respectively. Genotype GC of rs6898743 showed an OR of 0.47 (95% CI, 0.26–0.86) for ESCC.

Worldwide, cancer is one of the most prevalent causes of death [1] . In 2011, the overall incidence of cancer in the Netherlands was 100,577 cases, approximately 20% did develop gastrointestinal (GI) cancer [2] . The number of new patients with GI cancer is increasing in the Netherlands over the past 20 years [2] . Ten to twenty percent of the GI cancers are hereditary [3] . There are several other factors that play a role in the development of GI cancer, including age, gender, smoking, physical activity and obesity [4] , [5] . Also dietary factors, such as intake of alcohol, consumption of low amounts of vegetables and fruits, excessive consumption of red meat, and many other dietary factors may have an impact on the development of GI cancer [5] . In addition, risk modulating genetic factors with low penetrance, such as numerous single nucleotide polymorphisms (SNPs), may be involved in the development of GI cancer. In this respect the insulin-like growth factor (IGF) axis may be of interest, since the evidence that it is involved in several stages of cancer is increasing [6] – [8] . The IGF axis plays a role in embryonic and postnatal growth, in tissue repair and many other important processes [6] , [9] , [10] . Elevated levels of IGFs, low levels of IGF binding proteins (IGFBPs) in the circulation and over-expression of IGF receptor type I (IGFR-I) were associated with cancer [7] , [11] . In vitro studies indicated that several components of the IGF axis induce mitosis and prevents apoptosis in cancer cells [12] , [13] . IGF is mainly produced by the liver and it is mediated by the growth hormone (GH) family. The predominant isoform of this somatotropin family is growth hormone 1 (GH1). The growth hormone releasing hormone (GHRH), secreted by the hypothalamus, regulates the production of GH in the adenohypophysis of the pituitary gland [14] . The GH binds to the growth hormone receptor (GHR) in the liver and other target tissues. This interaction promotes the synthesis of IGF-I and IGF-II [15] . IGF is able to bind to the members of the IGFBP family [16] . Most IGFs circulating in the blood stream are bound to IGFBPs with a high binding affinity, resulting in an increased half-life of the IGFs. The majority of IGFs in the circulation bind to IGFBP-3, an acid labile subunit (ALS) complex [17] . Numerous factors, including hormones, diet, ethnicity, age and sex, influence the levels of IGFs and IGFBP-3 in serum and tissues [18] . In addition to the IGFBP family members, IGF-I and IGF-II bind to the two types of IGF receptors, IGFR-I and IGFR-II. IGFR-I, a transmembrane tyrosine kinase receptor, is expressed throughout the human body. Both types of IGF are able to interact with this receptor with different binding affinities. Activation of the IGFR-I related pathway may lead to proliferation, differentiation, migration and prevention of a cell from going into apoptosis [8] . The function of IGFR-II, a cation-independent mannose-6-phosphate receptor, is less clear. This receptor interacts with IGF-I as well as with IGF-II. The binding of this receptor with IGF-I is weak, whereas IGFR-II has a high binding affinity for IGF-II. In contrast to IGFR-I, these interactions do not lead to activation of cell signaling pathways. The main function of this receptor is probably the sequestering of IGF-II and eventually degradation of this protein [9] . Several genetic variants in the components of the IGF axis were already investigated in association with cancer risk. McElholm et al. analyzed 102 SNPs in the IGF axis [17] and characterized three genetic variants that appeared to be associated with esophageal adenocarcinoma: rs6214 in the IGF-I gene, rs6898743 in the GHR gene and a CA-repeat in the promoter region (5′-UTR) of the IGF-I gene, the latter variant however being very rare. The association between the SNPs rs6214 and rs6898743 and other GI cancers was not investigated yet. The SNP rs6214 (c.*2716G>A) resides on the IGF-I gene [19] . This gene is located on the long arm of chromosome 12 at position 23.2 and consists of six exons [20] . The SNP rs6214 is located in the three prime untranslated region (3′-UTR), which has an important role in the translation, localization and stability of mRNA [21] – [23] . In Europe, the allele frequencies of the G and A alleles are 55% and 45%, respectively [24] . The SNP rs6898743 (c.71-26648G>C) is an intronic SNP in the GHR gene, localized on chromosome 5 [17] . In the European population, approximately 75% and 25% are G and C alleles, respectively [25] . In this study, the prevalence of the SNPs rs6214 and rs6898743 in patients with GI cancer and controls were investigated in a Dutch population.

In patients, DNA was isolated from blood or healthy tissue, obtained after resection of the tumor. In controls DNA was isolated from blood. For DNA isolation the High Pure PCR Template Preparation Kit (Roche Diagnostics GmbH, Mannheim, Germany) was used, according to the protocol of the manufacturer. DNA samples were stored at 4°C until use.

The results from the RT-PCRs, genotype and allele frequency distributions, were analyzed by using IBM SPSS Statistics software version 20 (International Business Machines Corp., Armonk, New York, USA). The equality of the mean age was tested using the independent-samples t-test. In the analysis, age and gender were considered as confounders. Odds ratios (ORs) with their 95% confidence intervals (95% CIs) were calculated for the genotypes and for the alleles frequencies by performing unconditional as well as conditional logistic regression analyses. Homozygosity for the most common allele was taken as reference. When the mean age between the patient (sub)group and the control group was significantly different, the ORs calculated with the unconditional logistic regression were corrected for this variable. Results with a p -value <0.05 were considered statistically significant.

This population-based case-control study was performed to investigate the role of the IGF associated SNPs rs6214 and rs6898743 in the development of gastrointestinal cancer. The SNPs rs6214 and rs6898743 were not found associated with GI cancer, however in GI cancer subgroups, some interesting associations were noticed. Genotype AA of rs6214 in the IGF-I gene was associated with EC, but this risk modifying effect of the rs6214 A allele was mainly the result of an association with the histological subgroup of EAC; individuals bearing the GA or AA genotypes showed relative risks of 1.45 and 1.71, respectively. The SNP rs6214 has been investigated in several previous studies on GI cancer risk. Feik et al. reported a significantly higher risk of CRC for carriers of genotype AA [26] . A similar trend was seen in our study but no significance was reached. This difference could be explained by the fact that the study population of Feik et al. only contained 178 CRC patients. Moreover, their ORs were adjusted for gender, age and body mass index (BMI), whereas BMI was not involved in our study. McElholm et al. reported a significant decreased risk of Barrett’s esophagus (BE) for the AA genotype of rs6214 in an Irish study population (OR = 0.43, 95% CI 0.24–0.75) [17] . BE is an abnormal replacement in the esophagus of squamous epithelium by specialized columnar-lined epithelium, and is one of the main risk factors for EAC [27] . Therefore it is surprising that we found a significant inverse association with EAC for the rs6214 GA and AA genotypes. Two main explanations may be given: first, the patient group with BE and the control group in the study of McElholm et al. were relatively small (n = 207 and n = 244, respectively). More important however, is the large difference in allele frequencies, being 48% and 52% for the rs6214 G and A alleles as reported by McElholm et al. in their controls, whereas values of 57% and 43% were reported for Europeans in the HAPMAP database [24] . The latter values are in good agreement with our control allele frequency data (60.5% G and 39.5% A allele). The role of the genetic variant of rs6214 in pancreatic cancer in a Japanese population was assessed by Nakao et al. [28] . Using AA as the reference genotype, no association was found between rs6214 and pancreatic cancer in the overall analysis, whereas 20 years old carriers of the GA and GG genotypes with a BMI ≥25 kg/m 2 showed an increased risk for pancreatic cancer. However, the number of patients included in this sub-category was very low and thus reliability is limited. High plasma levels of IGF-I were suggested to be a risk for various cancers, including CRC [29] , [30] . Therefore, Canzian et al., Al-Zahrani et al. and D’Aloisio et al. investigated the role of the rs6214 SNP on the circulating IGF-I levels in Caucasians [20] , [31] , [32] . However, inconsistent results were obtained. In the study of D’Aloisio et al., the AA genotype appeared to be associated with high mean IGF-I plasma levels, whereas the other two studies did not find an association between this SNP and circulating IGF-I levels. The SNP rs6898743 has not been extensively investigated so far. Carriers of the CC genotypes were found inversely associated with EAC by McElholm et al. [17] , whereas our results showed a tendency for the opposite. It was also remarkable that the allele frequencies in the Irish study population as reported by McElholm et al. [17] (19% G and 81% C for controls) were completely different from the HAPMAP frequencies in the European population [25] (77.9% G and 22.1% C). Our data of 76.9% G allele and 23.1% C allele in controls completely fit with the reported HAPMAP data for Europeans [25] . Some interesting gender related differences were found; in women, genotype GA of SNP rs6214 was significantly associated with GI- and HNC, whereas homozygosity for the variant C allele of rs6898743 was inversely associated with HNC. In men, exactly opposite effects were noticed, however results here are not significant. It should be noticed however that the number of female patients in the sub-group of HNC patients was small (n = 93), which made the significance disputable. The strength of the present study is that the patients and controls were successfully matched by sex, race, recruiting area and also largely by age. This enabled an assessment of association between the SNPs and the outcomes, without being influenced by these confounders. Moreover, the overall study population was large. This study also has its limitations. First of all, the matching by age was not succeeded completely, as the mean ages between the GI- and HNC patient and control groups differed significantly. Therefore, unconditional logistic regression was used which theoretically creates a lower power compared to conditional logistic regression. Secondly, the ORs calculated in the sub-analyses were not adjusted for multiple testing. This might be advisable since multiple test