A number of genetic disorders exhibit inter-and intra-familial variability. Understanding the factors that control the expression of disease genes should provide insight into the fundamental disease processes and will have implications for counselling patients. Different mechanisms can account for this variability, including environmental factors, genotype-phenotype correlations and imprinting. There is also evidence that, in a number of genetic diseases, gene expression is under the control of modifier loci. In cases where the biological basis of the genetic disease is understood, any genes involved in the pathogenic process represent candidate modifier genes which can easily be evaluated. Alternatively, modifiers can be identified through approaches such as mouse models. Since modifier genes will generally be common and because of confounding environmental influences, linkage analyses in humans will generally be based upon affected or discordant sib pairs. Discordant sib pairs represent an attractive option for linkage studies, because recurrence rates are high and the reduced survival characteristics associated with severe phenotypes will make the likelihood of obtaining clinical material from two living cases difficult. Furthermore, the use of discordant siblings will select for those siblings which possess sufficient dissimilarity at the modifier locus to overcome any shared environmental influence.

The roles of the intrinsic mutation rate and genomic instability in tumorigenesis are currently controversial. In most colorectal tumours, it is generally supposed that the first mutations occur at the adenomatous polyposis coli (APC) locus; APC mutations are thought to provide cells with a selective advantage but have no known effect on the mutation rate. It has also been suggested that genomic instability is the initiating event in colorectal tumorigenesis and, if this is true, mutations of DNA mismatch repair (MMR) genes (or at similar loci) are the most likely candidates. If defective MMR precedes APC mutations, the APC mutations of colon tumours with defective MMR and hence replication errors (RER+) should differ from those of RER- tumours, in at least three specific ways: (1) a higher frequency of allele loss at APC in RER- tumours; (2) more frameshift than nonsense mutations in RER+ tumours; and (3) APC mutations in simple repeat sequences [(N)n, (N1N2)n, or (N1N2N3)n] in RER+ tumours. We found no evidence that sporadic RER+ and RER- colon cancers (including cell lines) differ in any of these three ways. Although it remains possible that MMR is abnormal in tumours from HNPCC families before APC mutations occur, it is likely that in sporadic colon tumours, APC mutations, rather than genomic instability, are the initiating events in tumorigenesis.

The model of colorectal tumorigenesis put forward by Fearon and Vogelstein has had great influence on molecular oncology. They proposed that a series of mutations occur in the progression from normal cells to colorectal cancer and that these mutations are associated with the histological features of such tumours. Several postulates of the model appear to be correct, particularly its emphasis on the stepwise accumulation of genetic changes and the inclusion of mutations at the adenomatous polyposis coli (APC) and TP53 loci. Since the publication of the original model, however, mutations at other loci have been identified which may be alternatives or additions. There is also evidence to suggest that some colorectal cancers develop along a different genetic pathway. In this review, we discuss how tumour development can occur as Darwinian evolution through selection of advantageous somatic mutations. The non-random nature of mutation selection gives rise to genetic pathways of tumorigenesis. In addition, we consider the Fearon and Vogelstein model, its shortcomings and possible additions to it. The evidence suggests that not all colorectal cancers follow the same genetic pathway during carcinogenesis.

No abstract