Grappling With the Genome: Cancer Centers Approach Puzzle in Different Ways

Oncology Live®, May 2012, Volume 13, Issue 5

As the cost of sequencing and analyzing genetic data continues to fall, the nation's leading cancer centers keep unveiling ambitious new clinical programs and research projects that will change the way every cancer specialist practices.

Photo courtesy of the Dana-Farber Cancer Institute.

Tumor tissue samples await analysis for cancer-related genes at the Center for Advanced Molecular Diagnostics at Brigham and Women’s Hospital.

As the cost of sequencing and analyzing genetic data continues to fall, the nation’s leading cancer centers keep unveiling ambitious new clinical programs and research projects that will change the way every cancer specialist practices.

Dana-Farber Cancer Institute, along with Brigham and Women’s Hospital, has launched one of the biggest research programs to date, in which the Boston, Massachusetts, centers seek to make the tumors of every patient with cancer a subject of genetic research.

Other cancer centers, while not broadening the research pool so widely, have deepened it by analyzing multiple tumor samples from each patient or conducting full-genome sequencing of both tumor samples and healthy tissues in tested patients.

These research projects supplement rapidly evolving protocols for clinical care. The tumors of most patients at most major facilities now receive some form of genomic testing to guide treatment. And that testing keeps getting more detailed.

For oncologists, hematologists, and other specialists who treat patients with cancer outside academe, these trends portend more than just an avalanche of studies that will have to be read. They portend a whole new age, albeit one that’s still a bit hazy.

Gordon B. Mills, MD, PhD

“Genetic analysis has already changed cancer care at every practice, and we’ve only seen the tiniest tip of the iceberg,” said Gordon B. Mills, MD, PhD, who chairs the Department of Systems Biology and holds the Olga Keith Weiss Distinguished University Chair for Cancer Research at The University of Texas MD Anderson Cancer Center in Houston.

“But I think we have a little work to do before we know how we can best use these new tools,” he said. “Different cancer centers are trying different things, which will help us figure out what works and what doesn’t, and what all cancer specialists should eventually be doing.”

Sequencing Costs Plummeting

Ever since the Human Genome Project resulted in the full sequencing of the first human genome in 2003, technological advances have increasingly enabled researchers to explore the genetic underpinnings of cancer at a decreasing cost.

“The ability to sequence an individual’s entire genome as well as the patient’s tumor genome is now a feasible enterprise at a cost and speed that was unthinkable even five years ago,” noted Boris Pasche, MD, PhD, of the University of Alabama in Birmingham, and Devin Absher, PhD, of the HudsonAlpha Institute for Biotechnology in Alabama, in an editorial in the Journal of the American Medical Association last year.1

Indeed, it took 10 years and more than $2 billion to sequence the first human genome. Today, whole-genome sequencing costs from approximately $10,000 to $35,000 per human genome, and the National Institutes of Health (NIH) $1000 genome target is within the realm of possibility.2

Photo courtesy of Maggie Bartlett/National Human Genome Research Institute

A researcher at the National Human Genome Research Institute uses a pipette to prepare DNA for sequencing.

Meanwhile, the nation’s network of cancer centers has been gearing up to make the most of the new technology, both individually and with federal leadership. After the genomic changes that occur in gliobastoma and ovarian cancers were successfully mapped, the NIH launched The Cancer Genome Atlas project in 2009. More than two dozen institutions throughout the country are working on characterizing, sequencing, and analyzing the genomes of more than 20 cancers.3

As part of that effort, the University of California, Santa Cruz, has established the Cancer Genomics Hub, intended to serve as a repository and portal for an anticipated deluge of data. Research for the atlas currently generates about 10 terabytes of data each month—far more than the 45 terabytes of data the Hubble Space Telescope gathered in its first 20 years, according to a press release announcing the first supercomputer milestone in May.4

Research Reveals Complexities

Amid this bounty of information, those in the forefront of such research are striking notes of caution about how quickly genomic knowledge can translate into clinical advances.

Much of the uncertainty stems from continuing discoveries about cancer’s true complexity. Mills’ colleagues at MD Anderson, for example, recently made a particularly dispiriting realization that has already transformed cancer care at MD Anderson and other hospitals.

Technology Terms in Cancer Genomics+

Term

Description

Whole-Genome Sequencing

Consists of 3 phases: sample preparation, where genome target is broken into fragments; physical sequencing, where bases in each fragment are identified in order; and reconstruction, where bioinformatics software is used to align overlapping reads from each fragment.

Targeted Genome Sequencing

Refers to strategies that enrich the input for DNA regions such as whole exome or cancer genome.

Cancer Genotyping

Involves use of multiplex assays and microarrays to analyze hundreds to millions of germline, somatic mutations, gene copy number variations, and other alterations affecting gene expression. Detects only known variants that have been selected for analysis.

Bioinformatics

Applies statistics and computer science to biology through information management and algorithm development.

+Tran B, Dancey JE, Kamel-Reid S, et al. Cancer genomics: technology, discovery, and translation [published online ahead of print January 23, 2012]. J Clin Oncol. 2012;30(6):647-660. doi:10.1200/JCO.2011.39.2316.

Cancer, they found, mutates so quickly that metastasized tumors often vary genetically from and within primary tumors. In some cases, tumor genes can vary by 40%. Even a single large tumor can vary genetically from one side to another, an unnerving fact in an age when tests of tumor genes taken from a single biopsy are supposed to tell doctors the proper course of treatment, Mills noted.

MD Anderson has responded by performing genetic testing on several different tumor samples on patients with cancers that are likely to mutate. If the tests show several different strains of cancer, the hospital responds by treating them all. Even so, in those tumor types that MD Anderson has studied, changes between the primary tumor and the metastases correlate strongly with bad outcomes, so the center is working to develop new treatment strategies.

“The discovery of these tumor mutations should be a big step forward in making good treatment decisions— a real advance that doctors everywhere should be thinking about—but it will probably delay efforts to do full genome analysis on every patient and every tumor,” Mills said.

“Not only will there be the extra cost of full sequencing for a handful of samples from every patient, the extra data add astronomical complexity to the analysis of what all those genes do,” he said.

Project Focuses on Full Genome

Nevertheless, at least one institution is moving ahead with full genome analysis, not with all patients, but in at least one large research study.

Matthew P. Goetz, MD

Doctors at the Mayo Clinic in Rochester, Minnesota, have initiated the Breast Cancer Genome Guided Therapy Study (BEAUTY) for 200 women with locally advanced breast cancer. The goal: to discover why some women do not respond fully to neoadjuvant chemotherapy and to speed the development of treatments targeted to nonresponders.

Researchers will perform full genome analysis not only on every woman’s tumor, but also on a healthy blood sample from each participant. The information from the normal or germline sample should help illustrate which human genes lead to what tumor genes, and whether variations in patients’ DNA lead to varying responses to treatment, even when tumor genes are similar.

All tumor samples will be kept alive and grown inside mice, so that fully sequenced live human tumors will be forever available for research and testing.

“The ultimate goal here is to speed drug development by determining which genetic variations determine who responds to a given treatment, and to develop novel therapies for women whose tumors are resistant to standard chemotherapy. Currently, every woman receives a standard course of treatment, and it fails to completely eradicate tumors in a substantial number of patients,” said Matthew P. Goetz, MD, associate professor of Oncology and Pharmacology, at Mayo’s College of Medicine.

Judy C. Boughey, MD

“Preserving the human tumor as mouse xenografts will provide the ability for us to partner with pharmaceutical companies in order to test new drugs, with the focus being on those tumors not eradicated with standard chemotherapy. These are fully human tumors but because they’re preserved as xenografts, you can experiment with treatments in a way you obviously cannot when humans are involved.”

Actually, the ultimate goal is far more ambitious than identifying relevant genetic variations and developing new drugs for this one form of cancer.

“This is a flagship study at Mayo. If successful, we would plan to replicate this approach for different types of cancer,” said Judy C. Boughey, MD, associate professor of Surgery, at Mayo’s College of Medicine. “Multiple other solid organ tumors, such as lung or pancreas, would be appropriate for similar studies.”

Taking a Look at Many Tumors

Unlike Mayo’s study, which will provide extraordinarily deep information about 200 women and 200 tumors, the flagship genomics effort at Dana-Farber will only provide a few hundred data points on every tumor examined.

But Profile, as the program is called, will provide that detail on the overwhelming majority of patients at the hospital—an estimated 10,000 new cases of cancer per year.

Barrett J. Rollins, MD, PhD

Every patient who walks in the door at Dana-Farber and at Brigham and Women’s is asked to participate and, to date, about 70% of them have agreed, said Chief Scientific Officer, Barrett J. Rollins, MD, PhD, who is one of the architects of the program and is the Linde Family Professor of Medicine at Harvard Medical School.

Participation requires no extra biopsies or other procedures. Researchers simply perform extra tests on the tumor samples they naturally take.

Investigators also are analyzing the archived tumor samples of patients who give consent. In theory, they could go back decades because the hospital stores samples indefinitely, but financial considerations have stopped the team from going back further than two years.

The analysis, for both new and pre-existing patient samples, is similar. Rather than looking for the handful of genetic variations that are understood well enough to influence treatment of any particular cancer, the lab examines 471 alleles of 41 genes suspected of playing some role in cancer.

All that information goes into a massive database, where sophisticated software will look for relationships between tumor genes, disease progress, and response to treatment.

This automated data mining constitutes just one of several major uses for all the information that is collected.

Investigators will also have access to the data, stripped of any information that identifies the patient. Researchers who become curious about how a certain genetic variation affects the course of a particular cancer will be able to simply look it up in the Profile database, rather than have to set up a study to collect the information from scratch—a process that Rollins said is often the most costly and time-consuming aspect of a particular study.

The Cancer Genome Atlas (TCGA) Research Network*

The following institutions are the major research centers participating in the TCGA. The project is scheduled to receive $175 million through the American Recovery and Reinvestment Act and $50 million each from the National Cancer Institute (NCI) and the National Human Genome Research Institute.

Center Specialization

Institution

Biospecimen Core Resources

International Genomics Consortium, Phoenix, AZ

The Research Institute at Nationwide Children's Hospital, Columbus, OH

Genome Characterization Centers

Baylor College of Medicine, Houston, TX

Brigham & Women's Hospital and Harvard Medical School, Boston, MA

British Columbia Cancer Agency,

Vancouver, Canada

Broad Institute, Cambridge, MA

University of North Carolina, Chapel Hill, NC

University of Southern California, Los Angeles, and Johns Hopkins University, Baltimore, MD

Genome Sequencing Centers

Baylor College of Medicine, Houston, TX

Broad Institute, Cambridge, MA

Washington University School of Medicine,

St. Louis, MO

Genome Data Analysis Centers

Broad Institute, Cambridge, MA

Institute for Systems Biology, Seattle, WA

Lawrence Berkeley National Laboratory,

Berkeley, CA

Memorial Sloan-Kettering Cancer Center,

New York, NY

University of California, Santa Cruz,

Santa Cruz, CA

University of North Carolina at Chapel Hill,

Chapel Hill, NC

University of Texas MD Anderson Cancer Center, Houston, TX

Data Coordinating Center

NCI Center for Bioinformatics, Bethesda, MD

*Quick facts. The Cancer Genome Atlas website. http://cancergenome.nih.gov/newsevents/forthemedia/quickfacts. Accessed May 7, 2012.

Perhaps more importantly, organizations that want to test drugs on people who have rare combinations of genes and tumors will be able to find a large group of potential patients, already identified, rather than having to find a full cohort of patients by themselves. This step alone could cut development times by years for some drugs.

“Even analyzing 10,000 tumors a year, it’s going to take several years for this study to bear fruit, but there are enormous advantages to a study with these numbers. Many relationships are impossible to see in individual patients or dozens of patients or even hundreds of patients, but they will emerge clearly when you’re studying tens of thousands of patients,” said Rollins.

“Such a project would have been impossible even a couple years ago, but the cost of sequencing genetic samples has fallen so rapidly and the computing power available to analyze the data has gotten so fast, that we can finally do this,” he said. “…At some point, every doctor’s office may be participating in studies like this.”

Rollins said the cost per patient for the 471-allele test has already fallen to just $500, and it continues to fall so quickly that Dana-Farber may soon sequence whole exons rather than just the alleles. Still, even with costs falling so fast, the Profile team is looking hard for financing, either research money from outside sources or data that translate some of their effort from research to treatment and allows them to charge insurers.

Treading Cautious Path to Clinic

But the jump from pure research to improved treatment will be a cautious one, no matter how fast the computers start spotting fascinating correlations.

“We won’t be changing treatment protocols based directly on correlations we find with Profile. We think that such a massive observational study will give us countless ideas to test, but observations can never take the place of controlled tests, no matter how compelling they look,” Rollins said. “Aside from those patients who end up in clinical trials because their records are in our database, participation in Profile probably will not end up changing what care patients receive.”

Photo courtesy of the National Cancer Institute.

The Cancer Genome Atlas (TCGA) Research Network includes three large-scale sequencing centers.

Rollins believes that the 70% participation rate—for a research project that probably won’t benefit patients who enroll—demonstrates how far the medical profession has come in convincing patients of both the importance of genetic research and the efficacy of privacy safeguards. Patients are rapidly becoming more comfortable with the underlying issues.

Looking forward, moving beyond single-hospital initiatives like Profile to ones that collect patient data from several, and eventually most, facilities will likely take some time.

Not only would such studies require even greater technology, they’d also need enormous amounts of coordination (and, given the fact that many doctors have yet to switch to electronic records, probably persuasion), either from the government or some industry group, experts said.

And most powerhouse cancer centers say they’re still at least a year or two away from trying a single-building effort like Profile, observers said. Some even believe it may be too early for even limited genetic testing on all cancer patients.

William Pao, MD, PhD

“At Vanderbilt, we are taking a disease-specific approach, with the goal of trying to build a sustainable model for mutational profiling,” said William Pao, MD, PhD, director of Personalized Cancer Medicine at Vanderbilt University Medical Center in Nashville, Tennessee.

Unlike many cancer centers that are trying to perform some genetic analysis as part of clinical care for all cancer patients, Vanderbilt currently has routine screens for three tumor types: lung, breast, and melanoma. They expect to add systematic colon cancer screens later this year, and other tumor types can be profiled by request.

Pao believes that the genetic makeup of individual tumors should be considered to help prioritize treatment for all oncology patients. In order to help facilitate an informed approach to cancer medicine and to identify appropriate genotype-driven clinical trials for patients, he and his colleagues have developed an online tool, MyCancerGenome (www.mycancergenome. org), freely available to all.

“We’ve worked with colleagues around the world to build a knowledge base about tumor mutations and how they affect cancer responses to targeted therapies. We put all the information online in a way that’s easy for every doctor to access, to understand and, when appropriate, to act upon,” said Pao, speaking of MyCancerGenome.org. “All this research needs to be accessible or doctors won’t be able to take full advantage of it.”

References

  1. Pasche B, Absher D. Whole-genome sequencing a step closer to personalized medicine [editorial]. JAMA. 2011;305(15):1596-1597.
  2. Tran B, Dancey JE, Kamel-Reid S, et al. Cancer genomics: technology, discovery, and translation [published online ahead of print January 23, 2012]. J Clin Oncol. 2012;30(6):647-660. doi:10.1200/JCO.2011.39.2316.
  3. Quick facts. The Cancer Genome Atlas website. http://cancergenome. nih.gov/newsevents/forthemedia/quickfacts. Accessed May 7, 2012.
  4. Stephens T. UC Santa Cruz builds national data center for cancer genome research [news release]. Santa Cruz, CA: University of California Santa Cruz Public Information Office; May 1, 2012. http://news.ucsc. edu/2012/05/cancer-genomics.html. Accessed May 10, 2010.