From Paths of Progress Spring/Summer 2014
Can casting a wider net for cancer genes improve patient care?
by Richard Saltus
Dana-Farber/Brigham and Women's Cancer Center (DF/BWCC)
is an environment where doctors and laboratory scientists often rub shoulders. A chance encounter can spark a new research collaboration, another way of looking at a thorny problem, or a different approach to treating a patient whose options have narrowed.
A case in point: In late 2012, thoracic oncologist
David Barbie, MD, had run out of treatment options for Colin Steele, a 61-year-old patient from North Andover, Mass. Steele was being treated for male breast cancer when he was also diagnosed with an aggressive
lung cancer. Despite several rounds of chemotherapy, the lung tumors were spreading and colonies of cancer had reached his brain.
The two cancers had occurred simultaneously, but nothing in Steele's family background or genetic testing revealed an unusual risk. Moreover, tests of his lung tumors didn't reveal any of the known DNA mutations that often drive such cancers and which
sometimes can be attacked with molecularly targeted drugs.
"I kept thinking there must be something more," recalls Barbie, of the
Lowe Center for Thoracic Oncology at DF/BWCC.
Later, while attending a scientific talk, Barbie heard
Nikhil Wagle, MD, a researcher with the
Susan F. Smith Center for Women's Cancers at Dana-Farber, describing an ambitious new research project called CanSeq (pronounced "can seek").
In CanSeq, Wagle and other researchers are evaluating an advanced technology called whole-exome sequencing, which conducts a wide sweep for cancer genes in tumors. It's a joint project involving Dana-Farber, the Broad Institute of MIT and Harvard, and
Brigham and Women's Hospital.
Overall, CanSeq "is about studying and developing best practices around the whole process of returning complex tumor and germline genomic information to oncologists and their patients," explains
Levi Garraway, MD, PhD, of Dana-Farber, who heads the project with
Pasi Jänne, MD, PhD, a DF/BWCC thoracic oncologist and researcher.
Testing tumors for the presence of known cancer mutations is standard practice for many patients coming to DF/BWCC. Depending on the type of cancer, a sample of tumor tissue may be tested for a number of specific mutations, which may vary from patient
to patient. The tumor's mutational profile can guide physicians in choosing designer drugs – a practice known as "precision" or "personalized" cancer care.
In addition, all consenting patients have their tumor's DNA scanned by
Profile, a research program that compiles mutations and other abnormalities found in cancers treated at DF/BWCC and
Dana-Farber/Boston Children's Cancer and Blood Disorders Center. Scientists search the samples for a list of known mutations in tumor tissue, and they also sequence – that is,
read out the entire genetic script of – 305 genes that have been linked to cancer.
Whole-exome sequencing goes further. Unlike "targeted sequencing," the whole-exome method scans all 20,000 genes contained in the human genome.
It's a monumental task, but whole-exome sequencing takes a shortcut by reading only the exons in each gene. Exons are the segments of DNA code that contain the recipes for cell proteins, the large molecules that build and operate the body. In
all, the human genome contains some 180,000 exons. Collectively, these exons are called the "exome," which contains about 30 million letters of genetic code. Most mutations and alterations that cause cancer are likely to be found in DNA in the exons
of genes: Thus, sequencing the exome should capture the vast majority of cancer-related DNA changes.
The goal of the four-year study is to sequence the exomes of 200 patients with advanced lung cancer and 200 with advanced colon cancer. "CanSeq sets up a mechanism so we can understand how having this very comprehensive picture of a tumor can have an
impact on patients," says Garraway. "Is it a good thing? Does it help, and if so, a lot, or not so much?"
Hunting for Genetic Suspects
If standard testing is like hunting for crime suspects at specific addresses, whole-exome sequencing can be thought of as unleashing a massive door-to-door search of an entire city. It can scoop up known wrongdoers, but also flush out thousands of others
not previously identified. Whole-exome sequencing is mainly being applied in research settings, but many believe it is the next step in realizing the ambitious goals of precision cancer care.
"Some of these genes we know nothing about – they're not mutated commonly enough," says Garraway. "But they may be relevant. For example, when a doctor says, 'I have a patient with nothing else to try,' whole-exome sequencing may find a mutation that
could lead to a trial of a drug for this patient."
Wagle's talk about CanSeq prompted a new approach to treating Colin Steele's relentless cancer. After the presentation, Barbie described the case and asked whether whole-exome sequencing might unearth mutations in the cancer that could guide treatment.
Wagle thought it might.
In December 2012, Steele became one of the first patients to undergo whole-exome sequencing through the CanSeq program. After he agreed to join the study, researchers extracted and sequenced DNA from cancerous tissue removed during his lung surgery. They
also sequenced his "germline" DNA – the normal genome he was born with – from a blood sample.
"To identify the mutations driving the cancer, we need to see the patient's normal DNA sequence and subtract it from the tumor DNA, leaving it with changes that are relevant to the cancer," explains
Stacy Gray, MD, AM, a Dana-Farber oncologist and investigator in CanSeq.
Automated DNA sequencing instruments at the Broad deciphered the genetic code of the exons in Steele's cancer within a matter of weeks, generating a gigantic amount of data. "The actual sequencing is easier than interpreting all these pieces of data –
you are going to find things that you don't know what they do, whether they are a bad actor or not," explains Jänne.
Raw sequencing data is pumped through two analytic "pipelines" that highlight DNA alterations that might be clinically relevant. That genomic information is then parsed by a team of physician-scientists who study the DNA variants and rank them in terms
of their potential value for guiding treatment.
A major goal of CanSeq is to create the most efficient infrastructure for the process. "We've learned it's incredibly hard to interpret the data and report back with a turnaround time that's rapid enough to be clinically useful," Garraway says.
In Steele's case, the results validated his physician's hunch and offered a glimmer of hope. It revealed at least three mutations missed by standard tests: a rare mutation in a gene called KRAS; a mutation in the ATM gene; and a deleted
tumor-suppressor gene, LKB1.
Mutant KRAS, which drives cells into cancerous growth, is found in many lung cancers. But standard tests available at the time hadn't spotted the mutation in Steele's tumor, because it occurred at a different position on the gene.
"This particular KRAS mutation was only reported a few years ago, in colorectal cancer," Barbie says. "We know it drives the cancer and predicts that it will be resistant to chemotherapy." However, the discovery opened up experimental treatment
options for Steele that hadn't existed prior to the sequencing.
Barbie discussed the case with a colleague, Leena Gandhi, MD, PhD, in DF/BWCC's thoracic oncology program, who was leading a clinical trial of a drug known as a CDK4 inhibitor that had achieved responses in some patients with KRAS mutations.
Steele enrolled in the trial. "Overall, I'm pretty optimistic," he said recently. "Give me something to try and I'll try it."
In early December 2012, Steele underwent the first of four cycles of treatment with the experimental drug. Over the next months, scans showed the cancer's growth had slowed and some small tumors apparently halted.
"It wasn't a home run," says Barbie, "but it stabilized his disease and caused some shrinkage, which hadn't happened with all his previous chemotherapy. We think the CDK4 inhibitor has damaged his tumor."
After four cycles, the drug was stopped because of worsening side effects. But in 2013, Steele started another trial targeting the ATM mutation in his tumor's DNA. The new drug, veliparib, combines a PARP inhibitor and a CDK4 inhibitor.
"The scans keep saying there's slow growth, and I've survived two and a half years with this lung cancer," says Steel. "I'm feeling pretty good, and I'm thankful for that."
His case is an example of how whole-exome sequencing adds a new dimension to the search for cancer genes, potentially giving patients new treatment options. With this technology nearing clinical use, research like CanSeq is helping shape how exome sequencing
can define a model for integrating it into cancer care.
How Should We Use Whole-Exome Data?
In addition to finding tumor genes, whole-exome sequencing may spot red flags in a patient's DNA, signaling an increased risk of diseases and conditions that have nothing to do with cancer.
These byproducts of exome analysis are known as "incidental findings." How this information should be handled has stirred debate and ethical discussion across the oncology field.
"This is something that is entirely new to cancer care," says Stacy Gray, MD, AM, medical oncologist with the CanSeq project at Dana-Farber. She is co-leader of a project that investigates the ethical, legal, and social implications of exome sequencing
Exome sequencing differs from current tumor profiling in that it also scans the patient's normal, inherited DNA. So, for example, a lung cancer patient's DNA might reveal she has an elevated risk of heart disease, or even sudden death, or that she
carries a mutation that predisposes her to another cancer. Depending on the condition, a preventive or early treatment might save a life. If no treatment is available, or the severity of the risk is uncertain, some would argue the knowledge is
more of a burden to the patient than a benefit.
Gray explains that CanSeq's consent process is designed to ensure that patients understand the potential outcomes of exome sequencing. It allows them to state preferences about which results they wish to receive.
Some patients may want to learn any and all DNA results, no matter how uncertain or distressing. Others might shun any findings not relevant to their cancer. But, do doctors have a duty to override these preferences for the patient's good? Are patients
obligated to share with family members test results that could affect their relatives? What if the patient dies before information on a familial risk becomes available?
The American College of Medical Genetics now lists 56 gene mutations it recommends as conditions that should be reported to doctors and patients. Among them: mutations causing hereditary breast and ovarian cancer, Li-Fraumeni familial cancer syndrome,
Marfan syndrome, retinoblastoma, and certain heart disease risks.
Some argue that such a stance threatens patient autonomy.
CanSeq researchers "have thought this through," says Gray. A clause in the consent form allows researchers to inform a patient of a finding that is "immediately actionable and would be clinically beneficial," she says. "These findings may be returned
to patients on a case-by-case basis if withholding them might result in patient harm."
Patients will not be alone in dealing with these issues. CanSeq oncologists will refer patients with incidental findings to genetic counselors and other appropriate specialists for follow-up.
Paths of Progress Spring/Summer 2014 Table of Contents