by ecancer reporter Clare Sansom

Historically, tumours have been classified mainly according to their organ of origin, and it is only recently that their molecular characteristics have been seen to be at least as important.

We now know that the genomes of tumours with origins at the same site may be more different than those with different origins, and therefore that molecular profiles of tumours are valuable for predicting prognosis and selecting an optimum treatment.

Much of the molecular landscape of cancer, however, still remains uncharted.

The Cancer Genome Atlas Research Network (TGCA) [1] is one of a number of organisations carrying out the systematic genetic analysis of multiple tumour types using omics technologies.

TGCA established its Pan-Cancer project in 2012 to assemble complete and consistent molecular datasets for twelve of the tumour types in its Genome Atlas and to compare the data for each.

The first two full papers from this initiative have been published in the October 2013 issue of Nature Genetics with two accompanying commentaries.

Chris Sander and his colleagues at the Memorial Sloan-Kettering Cancer Center, New York, NY, USA integrated genomic data from 3,299 tumours representing these twelve types, with breast, colorectal and endometrial cancers stratified into molecular subtypes [2].

They classified the tumours using a hierarchical system based on the presence or absence of each of 479 potential functional genomic alterations that they termed SFEs.

This set of SFEs included copy number gains and losses, frequently mutated genes and epigenetically silenced genes.

At the top of the hierarchy, the tumours formed two groups, the C class (characterised largely by copy number alterations) and the M class (characterised largely by point mutations).

The M class was further divided into 17 sub-classes, each characterised by a different set of mutations.

Tumours in the C class tended to contain mutations in the tumour suppressor TP53, indicating that the multiple chromosomal gains and losses that also characterised these tumours were often driven by inactivation of this gene.

This stratification of tumours is necessarily incomplete, but it could improve the design of and patient selection for clinical trials of personalised therapies, or even aid their development.

The second paper in this collection describes research by Rameen Beroukhim of the Broad Institute, Boston, MA, USA and his colleagues into non-germline or somatic copy number alterations (SCNAs) in eleven of these tumour types [3].

The researchers determined the copy number at more than a million genomic positions for a total of 4,934 tumour samples, and analysed the overall purity, ploidy and absolute copy number profiles in 3,847 of these.

The copy number profiles of the samples were found to vary widely around the median number of 39 SNCAs per sample.

Doubling of the entire genome was observed in 37% of the cancers, and this was associated with higher rates of mutations in TP53 and of all other types of SNCA.

Copy number alterations that were bounded at one end by a telomere tended to be longer than those in which both ends were internal to the chromosome.

Chromothripsis, in which very many small rearrangements occur in a single chromosomal region driven by a single event, was observed in 5% of all samples, with varying proportions in the different tumour types.

Recurrent amplifications and deletions were observed in a total of 140 chromosomal regions.

Only 38 of these regions included a known oncogene or tumour suppressor gene, but many of the amplified regions without known cancer-related genes did contain genes involved in epigenetic regulation.

Inverse correlations were observed between 7% of pairs of regions containing recurrent SNCAs, and these region pairs often contained genes coding for proteins known or predicted to have related functions.

This was the largest single high resolution study to date of copy number changes in cancer, and its results, taken together, should provide insights into the mechanisms through which these changes are generated and their role in the disease.

Two commentaries were published alongside these papers; one, by Joshua Stuart of the University of California, Santa Cruz, CA, USA and his co-workers, described the Cancer Genome Atlas Pan-Cancer analysis project in detail and explained the potential therapeutic value of its findings [1].

The other commentary, by Larsson Omberg from the not-for-profit company Sage Bionetworks, of Seattle, WA, USA  and his co-workers, described the software platforms used for the analysis [4].

All researchers used Sage Bionetworks’ web-based Synapse software to analyse and share molecular profiling data obtained using a wide variety of techniques while further developing their bioinformatics methodologies.

This software is freely available and all datasets obtained through the analysis have been released into the public domain.

The system proved robust enough to support the even larger collaborative genomic analysis projects that will doubtless arise following the clear success of the initial Pan-Cancer project.

References

1. The Cancer Genome Atlas Research Network, Weinstein, J.N., Collisson, E.A., Mills, G.B., Mills Shaw, K.R., Ozenberger, B.A., Ellrott, K., Shmulevich, I., Sander, C. and Stuart, J.M. (2013). The Cancer Genome Atlas Pan-Cancer analysis project. Nature Genetics, published online ahead of print 26 September 2013. DOI: 10.1038/ng.2764

2. Ciriello, G., Miller, M.L., Aksoy, B.A., Senbabaoglu, Y., Schultz, N. and Sander, C. (2012). Emerging landscape of oncogenic signatures across human cancers. Nature Genetics, published online ahead of print 26 September 2013. DOI: 10.1038/ng.2762

3. Zack, T.I., Schumacher, S.E., Carter, S.L. and 15 others (2013). Pan-cancer patterns of somatic copy number alteration. Nature Genetics, published online ahead of print 26 September 2013. DOI: 10.1038/ng.2760

4. Omberg, L., Ellrott, K., Yuan, Y., Kandoth, C., Wong, C., Kellen, M.R., Friend, S.H., Stuart, J., Liang, H. and Margolin, A.A. (2013). Enabling transparent and collaborative computational analysis of 12 tumor types within The Cancer Genome Atlas. Nature Genetics, published online ahead of print 26 September 2013. DOI: 10.1038/ng.2761