Posted: February 1, 2011

CASE STUDY: Using TCGA Data to Validate GBM Signaling Pathways

Addison Greenwood

Dr. Cameron Brennan
Image: Dr. Cameron Brennan

“An atlas is not a destination,” said Cameron Brennan, M.D., a neurosurgeon at Memorial Sloan-Kettering Cancer Center (MSKCC) in New York City. Yet the first results from The Cancer Genome Atlas (TCGA) have identified new mutations in glioblastoma multiforme (GBM) and lay out a map of the molecular landscape which might actually lead us to someplace very important.

When not in the operating room or helping patients, Dr. Brennan leads a laboratory studying GBM, which is the most common form of brain cancer and the first cancer type to be analyzed by TCGA. Most patients with this deadly form of cancer live 12-15 months after they are diagnosed, and fewer than 10 percent reach the 5-year survivor’s benchmark.

Mercifully, glioblastoma is a relatively rare cancer, said Dr. Brennan. For decades, however, there hasn't been much progress in treating it. “We do the best we can to safely remove as much tumor as possible,” the surgeon explained, “but with malignant gliomas you usually can’t get all of the tumor cells. We rely on radiation and chemotherapy to reduce the remaining disease. This standard therapy is the common starting point for treating all forms of GBM and is nearly always inadequate.”

The prospects are different with some other cancers. Breast cancer and lung cancer are examples where researchers have identified particular mutations or gene expression patterns that match certain groups of patients to a treatment.

So far, GBM has proven resistant to this approach. Now, the TCGA GBM Research Network has unearthed four distinct “molecular patterns” that may correspond to distinct subtypes of the tumor. “For GBM patients,” said Dr. Brennan, “any movement toward better treatments is urgent and there is a sense that this will involve treating individual tumors based on molecular signatures. This is a huge step in that direction.” For the creators and researchers of TCGA, it is a powerful signal that the information contained in the atlas may provide an integrated picture of the disease and help doctors better match patients to treatments based on the subtype of cancer that they have.

Using TCGA Results to Explore Cancer Biology

Dr. Brennan emphasizes that the molecular patterns found in GBM are not in any sense “the answer we can take directly to the clinic to treat patients.” Not yet. But he makes the point, “Wherever you see cancer genetics following a pattern, such as what we see in these GBM molecular subtypes, there's a chance that patients' tumors may reflect this pattern in response to a particular treatment. This is especially relevant to drugs in clinical trials which target cell signaling pathways.

To follow that lead, Dr. Brennan along with Eric Holland, M.D., Ph.D. and colleagues from MSKCC undertook another study to see whether there might be different ways of defining and perhaps exploiting the cell signaling pathways associated with each GBM subtype to possibly improve treatment strategies for patients.

Cell signaling is a crucial part of the biology of all cancers including gliomas, he explained. Cell signaling, also known as signal transduction, most commonly describes the process by which receptors on the cell’s surface intercept or catch chemical signals coming from outside the cell and transmit them to the inside of the cell, setting off a flow of chemical reactions that promote cell function and growth.

We wanted to see if patterns showed up in the activation of signal transduction pathways by looking at the actual coded proteins that were altered in GBM tumors, said Dr. Brennan. GBM researchers have long been interested in the signaling that occurs in the pathway downstream of tyrosine kinase receptors (TKR) because this pathway is abnormally active in cancerous cells. TKR is a set of enzymes that function as an "on" or "off" switch in many cellular functions and is often targeted by a class of chemotherapy drugs. The study, which measured activity across a number of signaling pathway proteins, was more extensive than any previously done on GBM. Instead of using the TCGA tissue, the researchers collected tumor samples from 27 glioma patients during surgery so that their work would provide a test of the TCGA subtypes and pathways against an independent dataset. Analyzing proteins present in the tumors, they found three distinct signal transduction pathway patterns with little overlap.

One pattern was linked to higher levels of EGFR, a protein that if produced in abnormally high levels can cause cells to divide excessively.  A second was linked to excessive production of PDGFB, one of a family of growth factors that normally plays a significant role in blood vessel formation. Excessive production of PDGFB can drive the cell growth and division associated with cancer and is known to be a potent generator of gliomas in mouse models. The third pattern was notably associated with low levels of Neurofibromin 1 protein (NF1), a gene previously identified as the cause of neurofibromatosis 1, a rare, inherited disorder defined by uncontrolled tissue growth, especially in the nervous system.

When the researchers analyzed the 278 TCGA samples for mutations in the genes that code for those proteins, there was very little overlap among the 163 samples where one of the three was mutated, suggesting three distinct subclasses of the disease. Further analysis showed that the three subclasses defined by the signaling proteins matched up with corresponding subtypes of GBM in the TCGA data that were based on genetic changes.

Finding these patterns of molecular changes across subclasses of GBM is a critical step to the practical development and testing of treatments, explained Dr. Brennan.  If there were no "big picture" of molecular subtypes, we would face a daunting task for each GBM patient that came in the door: literally scores of genes could be part of the problem and tailoring therapy would be reduced to guesswork.

Now we can profile their tumors and just look at a few key features to see if we can place them in one of the new subclasses revealed by TCGA. If so—we're doing this at MSKCC and it appears that we can—we then have some idea of the pathways which are commonly altered in that group and can focus on those that seem to matter.

As scientists now begin to explore each subtype, the brakes are off. Researchers and oncologists are already beginning to present their data in the context of TCGA. In some cases, current and completed clinical trials are being reanalyzed considering the possibility of differing responses among the subtypes. “Conversations throughout the research world are organizing themselves around this new picture,” said Dr. Brennan.  The atlas provides a common reference to help researchers explore and communicate their results.

Expanding the Universe of Cancer Science

There is still much work to do, but studies such as these are clearly on the path to clinical results. As with the TCGA subclasses, this work is laying the foundation for therapies directed at classes of cancers, and clinicians are already leveraging such insights to place newly diagnosed patients into a meaningful context.

It looks to be only a matter of time before researchers focusing on these genes and pathways come up with promising treatments. “In the long run,” says Dr. Brennan, “decoding the cancer genome is the beginning. I think we’re going to see questions being approached that we can’t even imagine today.”