ESMO 2014 Report: Reclassifying cancer through genomics

Keynote lecture: Professor Sir Mike Stratton (Cambridge/Hinxton, UK) by Denys Wheatley.  On the thesis that the genetic understanding of cancer is fundamental to classifying and hence treating patients according to their known defects, work continues unabated on their genomics. Professor Stratton appreciated that this is only part of the story because we must also consider non-genomic influences (epigenetics, micro-environmental effects altering tissue-tissue interactions) and intracellular changes that might contribute (chromatin modification). Nevertheless, the received wisdom is that genetic defects are key in our progress towards better diagnosis and prognosis. At least 20,000 genes and their defects can be screened, and the defects need to be identified such that their protein products might become targets for customized drugs to attack them. The data generated from genomics may also give clues as to which genes are primarily affected in the carcinogenesis, the development of metastases and in their inherent instability that leads to subclones (further mutations engendered by mutations already present in cancer cells). Damage to DNA cannot be prevented due to physical agents (uv from the sun and other sources of radiation) and chemicals agents (mutagens, such as benzpyrene from smoking, and moulds that produce aflatoxins, leading the liver cancer), but there are also spontaneous mutations due to defective replication of DNA probably concomitant with defects in repair enzymes. Genomics of 700 skin cancers indicates that many mutations involve C being replaced by T in the base-pairing in the DNA double helix. In lung cancer C is more commonly replaced by A. These changes will probably lead to further mutations within the same genome – matters get worse as time goes on (a domino effect), and subtypes will arise in the individual tumour making it more heterogeneous, hence more difficult for targeted therapy. The essence of this lecture was on further analysis of the genomic substitution process in cancer, with the interesting observation that a C>T substitution can occur where there are 4 possible bases upstream of it and similarly 4 different bases below it.  Cytosine deaminases may be involved in some changes, converting C to U, which might then be converted to a T.

The permutation of all 4 bases in the DNA that might be substituted means that there will be 96 possible arrangements of neighbouring bases, and these can lead to particular “signatures” by which a much broader spectrum of genomic identification could be discerned. This means that it would be possible to reclassify with much greater stratification a population of some particular tumour type (e.g. hepatocellular carcinoma). Tumours will have patterns of these signatures that can be recognised, but they are not consistent in any one type of tumour. There is also clustering of these changes on chromosomes (mutational hotspots, called kataegis), but no recognizable relationship to any particular sites on them. This creates even more diversity, and therefore the issue is whether this reclassification can help direct therapy by the appropriate choice of agent depending on what genes are involved in these signatures. The work continues, but the impression is that things are getting much more complicated by this genomic analysis, and at present we cannot see the wood for the trees. The possibility exists that because the signatures are so different from tumour to tumour, this project will bring us back to the conclusion that all tumours, like their hosts, are unique.