Cancer Atlas Extends Map of Glioblastoma

Cancer Genome Data Go Public as They Are Compiled

Almost 50 years have passed since scientists first connected a genetic abnormality, the Philadelphia chromosome, with cancer. The discovery was the first of many to lead to a targeted cancer therapy.

Matthew Meyerson and Lynda Chin
Graham Ramsay

Matthew Meyerson and Lynda Chin led a study reporting the first Cancer Genome Atlas results. The multidimensional data is freely available at

Now a project called The Cancer Genome Atlas (TCGA) aims to systematize the discovery of these defects and catalog them. The project members expect the resulting atlas to become a valuable reference for scientists that will accelerate the development of new therapies and diagnostics. Eventually, the atlas may contribute to personalized cancer medicine.

Atlas investigators Lynda Chin and Matthew Meyerson recently led an interim data analysis to demonstrate the potential power of the atlas. They based the report, published online Sept. 4 in Nature, on the first wave of pilot results, which included sequencing data for 600 genes and genomewide copy number, expression, and methylation data from 206 biospecimens of human glioblastoma. The ongoing pilot will also screen more genes from glioblastomas as well as ovarian and lung cancer specimens.

The report revealed evidence corroborating several known genetic abnormalities in glioblastoma in addition to novel discoveries. For instance, their data confirmed a near universal deregulation of the RB, P53, and RTK signaling pathways, and it identified novel mutations in PIK3R1, which encodes the regulatory component of the PI(3)K protein complex. Although rare, PIK3RI mutations had been noted in prior research. According to Chin, the atlas data add statistical significance to these reports, which, in turn, helps investigators form more definitive hypotheses about such rare mutations.

Moreover, the interim analysis pointed to a possible mechanism by which glioblastoma cells become resistant to the standard treatment with the drug temozolomide. Earlier work by David Louis, the Benjamin Castleman professor of pathology at HMS, showed that glioblastoma tumors can develop resistance to treatment through a particular mutation of a mismatch-repair gene. The broader and deeper multidimensional data of the atlas enabled the investigators to articulate a molecular mechanism underlying the emergence of resistance. This hypothesis, if proven, could have a “huge clinical impact,” said Chin, HMS associate professor of dermatology at Dana–Farber Cancer Institute.

Redefining Cancer
Such insights into resistance, an increasingly important problem since many cancer patients relapse after an initially successful course of therapy, may not be a one-time lucky strike. Because of the systematic and comprehensive atlas approach, “we should be able to look at primary cancers and posttreatment, resistant cancers and ask, ‘What are the changes?’” said Meyerson, HMS associate professor of pathology at DFCI.

As the atlas expands over time to include more genes and more types of cancer, Chin expects it to drive development of new therapies, new biomarkers, and new ways of categorizing cancer. “This project will change the way we look at cancer genetics,” said Chin. “It is the first step toward a future of personalized medicine, when we no longer lump cancers by a particular organ site, but treat each as genetically distinct disease.”

“This project will change the way we look at cancer genetics.”

More immediately, however, the atlas data may be most useful to scientists as reference material. For instance, Louis recently compared the genetic signature in resistant tumors he had studied earlier to the atlas data, confirming that the mutations conferring resistance were, indeed, the same. “The power of TCGA is it allows you to go online quickly and get a little extra information to support a hypothesis or refute a hypothesis and therefore fast-track your activities,” he said.

While the systematic approach of the atlas may quicken the pace of discovery, the work does not end there. The real insights that will help translate its genomic data into therapies and diagnostics will likely come from the follow-up cancer biology research that will identify the functional roles these defects play. For instance, a mutation may be found in a large percentage of tumor samples, “but it might be like a brake light. If you fix it, it isn’t going to do much to stop the runaway car,” said Chin, who is also the chair of the Glioblastoma Disease Working Group of TCGA.

To make the atlas available to as many researchers as possible so this downstream work can begin without delay, the Atlas Research Network, which also includes Raju Kucherlapati, the Paul C. Cabot professor of genetics, has created a public repository for the data. The atlas investigators post data in real time and provide it freely to the public, making it open to the thousands of scientists in the cancer research community.

The Data in Dollars
Critics, however, worry about the price tag on the atlas. The pilot alone will cost $100 million over three years.

Part of the funding diverted to the atlas might be better spent, for example, “performing screens to identify vulnerabilities of cancer cells in order to find drug targets, something that we can do now, but is ignored by TCGA,” said Stephen Elledge, the Gregor Mendel professor of genetics and medicine. Elledge recommends that the atlas pilot be independently evaluated to determine if the resulting data warrants continuation of the project or revision.

Mutations involved in altering one signaling pathway to produce glioblastoma.
Courtesy Nature

Personalized cancer medicine. Many different mutations may be involved in altering one signaling pathway to produce glioblastoma. Identifying the specific mutations in a given patient’s tumor could guide treatment the same way doctors determine which bug is causing an infection prior to prescribing an antibiotic. Red indicates mutations that activate genes, gray indicates inactivating alterations, and shading reflects the frequency of genetic alteration.

Atlas investigators agree that functional screening is vital. Moreover, such screening will likely benefit from triangulation with the atlas data to help prioritize and filter out the noise, said Chin. “There isn’t any single answer to this complex question of cancer genetics.”

Meanwhile, the funding outlook for projects like the atlas may be changing as sequencing costs fall. New massively parallel sequencing technologies, which are projected to cut costs by 50-fold, are not online yet. But they will be in the next two or three years.

Starting the atlas before the technology is ready allows the Atlas Research Network of over 100 collaborators from 18 institutions to gel, to establish processes, and to work out computational details and data-sharing policies and technologies. “It gives us a chance to look under the hood and make plans for the long run,” said atlas investigator Eric Lander, HMS professor of systems biology and director of the Broad Institute.

Meanwhile, screening of ovarian and lung cancer samples is in progress, and the next wave of genomic data for glioblastoma is coming online. “It looks like we’re finding additional recurrent mutations [in glioblastoma],” said Meyerson. The analysis is not yet complete, but because the atlas data are released in real time, it is already available at the data portal for anyone in the world to see.

Conflict Disclosure: The investigators report no conflicts of interest.

Funding Sources: The National Institutes of Health