Colorectal cancer risk and the <i>APC</i> D1822V variant

Adenomatous Polyposis Coli (APC) tumor suppressor gene, identified in 1991,1 plays a key role in early stages of human colorectal tumorigenesis. High-penetrance germline mutations in the APC gene, usually generating a stop codon, are responsible for the autosomal dominant inherited disease Familial Adenomatous Polyposis (FAP). Multiple germline missense mutations in the APC gene have been reported,2, 3, 4, 5, 6 although their contribution to colorectal cancer (CRC) risk is controversial. I1307K, E1317Q and D1822V are the most prevalent APC variants reported so far. D1822V, which results in an aspartate to valine change at codon 1822 due to an A-to-T transversion, is the most common APC variant described. The possibility that this variant may increase CRC risk has been addressed in only one study,6 which showed no main effect and a reduced risk of colon cancer for those individuals with the homozygous variant diagnosed after age 65 or eating a low-fat diet. Here we report the results of a hospital-based case-control study (346 sporadic CRC cases and 297 controls) designed to assess whether D1822V predisposes to CRC and interacts with other risk factors including family history. A detailed description of the population and study methods has been extensively described elsewhere.7 All subjects were informed and gave written consent to participate in the study and to allow their biological samples to be genetically analyzed, according to the Helsinki declaration. The Local Ethical Committee cleared the study protocol. Genotyping of the D1822V variant was performed by means of fluorescent hybridization probe melting curves using the Light Cycler instrument. Statistical analysis was carried out using logistic regression models from which odds ratios (OR) and 95% confidence intervals (CI) were calculated. To increase statistical power, risk was assessed combining the heterozygous and variant homozygous against the reference category of homozygosity for the more frequent allele assuming a dominant model. All analyses were adjusted for age and sex. In our case-control study, family history of CRC (at least 1 first-degree relative harboring a colorectal tumor but not other cancers), increased CRC risk (OR = 3.23; 95% CI 1.77–5.88; p-value = 0.001). We also observed a high risk of CRC associated to a personal history of symptomatic adenomas, assessed by a previous performance of an endoscopic procedure evidencing any type of adenomas (OR = 4.31; 95% CI 1.44–12.9; p-value = 0.004). As expected, long-term nonsteroidal anti-inflammatory drugs (NSAIDs) consumption was a protective factor (OR = 0.39; 95% CI 0.23–0.68; p-value = 0.001). No significant modification of risk was observed for macronutrients other than alcohol, which was associated with a dose-response increased risk. The frequency of the T allele at codon 1822 was 19.5% in our control population and 19.7% in CRC cases (Table I) being genotypes in Hardy-Weinberg equilibrium. There were no statistically significant differences in CRC risk associated with D1822V genotype (OR for AT/TT = 1.06; 95% CI 0.76–1.48; p-value = 0.72). When cases were stratified by tumor site, this result was similar both in cases of colon (OR for AT/TT = 0.91; 95% CI 0.62–1.33; p-value = 0.62) and rectum (OR for AT/TT = 1.35; 95% CI 0.88–2.07; p-value = 0.17). No interaction was observed with sex or age (data not shown). Regarding dietary habits, no interaction of the variant was observed with dietary fat (Table II) or other nutrients analyzed (data not shown). These results are different from those previously reported by Slattery et al.,6 who described in a large study that the TT genotype was associated with a reduced colon cancer risk among individuals diagnosed after age 65 or eating a low-fat diet. Several factors may account for this discrepancy: i) differences in the population analyzed (Slattery's study was performed on colon cancer cases exclusively); ii) geographical variations in dietary habits and iii) lack of statistical power to detect these interactions in our study since we only observed 14 CRC cases and 16 controls with the TT genotype. We tried to minimize the potential recall bias associated with the use of hospital controls by a careful design and the use of validated questionnaires. Also, it is already known that allele frequencies are not significantly altered in hospital populations.8 While in our study family history increased CRC risk, no interaction with the variant was observed (Table II). We decided to further explore this issue by analyzing a large cohort of 244 unrelated familial CRC cases (Table III). Prevalence of the variant did not vary among affected individuals of classical (n = 90) or attenuated (n = 26) FAP when compared to controls or sporadic CRC cases. Moreover, in 2 FAP families analyzed, the variant did not cosegregate with the disease. Prevalence of the variant was not modified in Hereditary Non-Polyposis Colorectal Cancer (HNPCC) affected individuals (n = 35). However, we observed a nonsignificant lower T allele frequency in HNPCC-like9 (n = 93) cases compared to controls. The heterogeneous nature of this latter group of patients precluded identification of a specific subgroup harboring a lower T allele frequency. From our observations, it can be concluded that the T allele is not related to CRC familial aggregation. Personal history of adenomas was a strong risk factor in our case-control study: AA individuals with a personal history of symptomatic adenomas had 2-fold risk of CRC, being this risk further increased in T allele carriers (Table II). While the risk of malignant transformation of adenomas associated with the T allele might be modest, the impact on the overall cancer burden could be significant due to the relatively high prevalence of the variant allele. These observations must be interpreted with caution due to the low number of cases included. The functional significance of the variant remains unknown. D1822V results in an aspartate (a hydrophilic, amide side chain amino acid) to valine (a hydrophobic, aliphatic side chain amino acid) change at codon 1822, between fourth and fifth 20-amino-acid repeat within the β-catenin downregulation domain. In order to expand our knowledge about the potential pathogenesis of this variant, loss of heterozygosity (LOH) at codon 1822 was studied in paired tumor and normal tissues of 92 germline heterozygous (AT) CRC cases. Results obtained in the Light Cycler instrument were confirmed with a MspI RFLP/PCR assay. LOH was observed in 25 (27%) of the 92 tumours. The A allele was lost in 13 cases and the T allele in 12, indicating that no preferential loss of the normal allele was present. Further investigation is certainly needed to ascertain how this substitution may affect APC regulatory functions and which are the molecular mechanisms involved in its putative function. In summary, we have shown that the D1822V variant of the APC gene does not associate with family history of CRC but may increase the risk of adenoma transformation. Only one study had previously evaluated the role of this variant in the pathogenesis of CRC, so further studies in larger series are needed to confirm these observations that may shed some light in colorectal tumorigenesis. Yours sincerely, We thank Olga Campos, Raquel Cuesta and Gloria Julià of the Program of Molecular Diagnosis of Hereditary Colorectal Cancer for excellent technical support. Francisco Rico was of great help with D1822V genotyping with the Light Cycler instrument. M.M. is a recipient of a FPI MCYT fellowship. The research team belongs to the Network of Cooperative Research on Cancer (C03/10) and Epidemiology and Public Health (C03/09), funded by the Instituto Carlos III, Ministerio de Sanidad y Consumo of Spain. Mireia Menéndez, Sara González, Ignacio Blanco, Elisabet Guinó, Mercé Peris, Miquel A. Peinado, Gabriel Capellá, Victor Moreno