Abstract
Genetic instability results, in a large majority of solid tumors, in deep chromosomal rearrangements. However, because chromosomal instability produces highly complex caryotypes, rarely showing stereotypic aberrations, it has not been possible to characterize solid cancers according to specific patterns of chromosomal rearrangements. This contrasts with the situation in hematological malignancies, where cytogenetics has allowed to lay out the basis of a renewed classification. New insights have been brought by the development of comparative genomic hybridization (CGH). This molecular cytogenetics approach was originally devised to detect regions in the genome of tumor cells undergoing quantitative changes, i.e. gains or losses of copy numbers. The large body of studies based on CGH show that solid tumors undergo frequent gains and losses and that every chromosomes show at least one region of anomaly. Furthermore, different tumor types present distinct CGH patterns of gains and losses. These observations favor the idea that it may be possible to type human solid cancers according to their patterns of genomic aberrations. However, despite the fact that a number of CGH based studies present data suggesting that different tumor types or cancers at different stages of evolution show distinct patterns of gains and losses, it has proven difficult to be conclusive. This can be mainly attributed to the lack of spatial resolution of CGH. Indeed, CGH uses metaphase chromosomes as hybridization targets and therefore its resolution is at the level of chromosomal banding. The recent adaptation of DNA array technology to CGH will allow to pass this limitation. In DNA array based CGH (array-CGH) metaphase chromosomes have been replaced by spots of cloned DNA. These DNA clones may either be genomic (BACs, YACs or cosmids) or coding (cDNAs). The resolution of array-CGH is therefore determined by the size of the cloned DNA insert (100 Kb for BACs, 1-2 kb for cDNAs). Data corresponding to each of these clones is or will be in a near future linked to DNA sequence data. Hence, in a near future, array-CGH will allow to increase the resolution from a cytogenetic level to a molecular level. Finally, because array technology is highly adaptable to automation, going from classical CGH to array-CGH will produce a quantum leap in throughput.