Colorectal carcinoma (CRC) remains a frequent cause of cancer-associated mortality in the UK and still has a relatively poor outcome. Single gene defects account for up to 2-6% of cases, but twin studies suggest a hereditary component in 35%. CRC represents a paradigm for cancer genetics. Almost all the major-gene influences on CRC have been identified, and the identification of the remaining susceptibility alleles is proving troublesome. Only a few low-penetrance alleles, such as methylene tetrahydrofolate reductase C677T, appear convincingly to be associated with CRC risk. To identify the remaining CRC genes, parallel approaches, including strategies based on linkage and association and complementary analyses such as searches for modifier genes, must be employed. To gain sufficient evidence to prove that a gene is involved in CRC predisposition, it is probably necessary for multiple, adequately-powered studies to demonstrate an association with the disease, especially if the allelic variants have only a small differential effect on risk. It may also be possible to show how genes interact with each other and the environment, although this will be even more difficult. Accurate quantitation of the allele-specific risks in different populations will be necessary, but problematic, especially if those risks combine in a fashion which is not of a straightforward additive or multiplicative type. Without any good prior evidence of the nature of the remaining genetic influence on CRC, the possibility remains that this is a truly polygenic trait or that multiple, rare variants contribute to the increased risk; in these cases, identification of the genes involved will be very difficult. Despite these potential problems, the effectiveness of preventive measures for CRC, especially in high-risk individuals, means that the search for new predisposition genes is justified.

We have examined chromosomal-scale mutations in 34 large colorectal adenomas (CRAs). A small number of changes (median = 2, IQR = 0-4) were found by array-comparative genomic hybridization (aCGH) in most tumours. The most common changes were deletions of chromosomes 1p, 9q, 17, 19, and 22, and gains of chromosomes 13 and 21. SNP-LOH analysis and pseudo-digital SNP-PCR analysis detected occasional copy-neutral LOH. Some aCGH changes found frequently in colorectal carcinomas, such as deletions of chromosomes 4q and 18q, were very infrequent in the adenomas. Almost all copy number changes were of small magnitude, far below the predicted levels even for single copy gain/loss; investigation suggested that these changes were either artefactual or occurred in sub-clones within the tumours. In some cases, these sub-clones may have represented progression towards carcinoma, but comparison with aCGH data from carcinomas showed this to be unlikely in most cases. In two adenomas, there was evidence of a large, outlying number of copy number changes, mostly resulting from part-chromosome deletions. Overall, moreover, there was evidence of a tendency towards part-chromosome deletions-consistent with chromosomal instability (CIN)--in about one-sixth of all tumours. However, there was no evidence of CIN in the form of whole-chromosome copy number changes. Our data did not support previous contentions that CRAs tend to show chromosome breakage at fragile sites owing to CIN associated with an elevated DNA damage response. Chromosomal-scale mutations occur in some CRAs; although CIN is not the norm in these lesions, it probably affects a minority of cases.

Patients with multiple (5 -100) colorectal adenomas (MCRAs) often have no germline mutation in known predisposition genes, but probably have a genetic origin. We collected a set of 25 MCRA patients with no detectable germline mutation in APC, MYH/MUTYH or the mismatch repair genes. Extracolonic tumours were absent in these cases. No vertical transmission of the MCRA phenotype was found. Based on the precedent of MYH-associated polyposis (MAP), we searched for a mutational signature in 241 adenomatous polyps from our MCRA cases. Somatic mutation frequencies and spectra at APC, K-ras and BRAF were, however, similar to those in sporadic colorectal adenomas. Our data suggest that the genetic pathway of tumorigenesis in the MCRA patients' tumours is very similar to the classical pathway in sporadic adenomas. In sharp contrast to MAP tumours, we did not find evidence of a specific mutational signature in any individual patient or in the overall set of MCRA cases. These results suggest that hypermutation of APC does not cause our patients' disease and strongly suggests that MAP is not a paradigm for the remaining MCRA patients. Our MCRA patients' colons showed no evidence of microadenomas, unlike in MAP and familial adenomatous polyposis (FAP). However, nuclear b-catenin expression was significantly greater in MCRA patients' tumours than in sporadic adenomas. We suggest that, at least in some cases, the MCRA phenotype results from germline variation that acts subsequent to tumour initiation, perhaps by causing more rapid or more likely progression from microadenoma to macroadenoma.