Tumor Tell-All

Unraveling complex genetic stories in cancer cells may lead to personalized treatment

Tumors are ugly. But the staff at Massachusetts General Hospital takes a snapshot of almost every one that crosses the doorstep.

Michael Morgenstern

TUMOR TESTING DREAMS | Though some patients do benefit from drugs that target specific cancer-causing mutations, in most cases a tumor’s underlying mutations are unknown. Someday, researchers hope, comprehensive genetic tumor testing (steps depicted below) will become cheap and fast enough to influence patient care, providing every cancer sufferer with personalized treatment. E. Feliciano

A drug targeting a common mutation in patients with spreading melanoma reduced tumor presence (darkness in chest and abdomen) within 15 days. K.T. Flaherty et al/NEJM 2010

POWER OF THE BLUEPRINT | A recent genetic study identified a segment of chromosome 15 that had relocated to chromosome 17 in a patient with a difficult-to-diagnose case of leukemia. The change, which affected the cancer-related RARA gene (bright blue), was not visible via a microscope. J.S. Welch et al/JAMA 2011

These snapshots are not photographs, but are rather whole rap sheets on the genetic deformities that twist normal cells into cancerous ones. In a laboratory tucked within the labyrinth of corridors connecting the hospital’s many buildings, researchers punch tiny cores no bigger than a grain of rice from tumor samples. Those cores are handed off to robots and tested for 110 mutations that commonly strike 15 genes important in cancer.

Across the country, doctors at the Oregon Health & Science University in Portland use another method for testing tumors. The staff there looks for 643 different mutations in 52 genes in solid tumors, such as those of lung and colon cancer, and screens blood or bone marrow from leukemia patients for 370 mutations across 31 genes. Though the tests don’t reveal everything that has gone wrong to lead to a patient’s tumor, they may point to mistakes that drive the cancer. “They’re the original ‘stomp on the gas pedal’ type of mutations,” says Christopher Corless, a pathologist who directs the tests at Oregon.

Efforts under way at these two cancer centers are creating some of the first ripples in what many scientists predict will be a growing wave of genetic testing of tumors. Traditionally doctors order tests to see if just one of a small handful of genes is broken, and such tests have been used widely for the treatment of only breast cancer and a few other cancer types. But that piecemeal approach is giving way to more comprehensive probes of cancer’s molecular workings.

Big cancer centers and clinical labs at university-affiliated hospitals around the country are now adopting tumor-testing programs similar to those in Massachusetts and Oregon. Even though the wave has yet to swell outside of academic centers to smaller community hospitals and doctors’ offices, some clinicians are already finding troublemaker genes and drugs that can counteract them. Scientists are taking the process many steps further in the lab, deciphering a tumor’s complete genetic instruction book. Having such information, at least in the case of one man with a rare cancer, has helped uncover unexpected cancer drivers and tailor patient care.

Such comprehensive testing is beginning to change the way many doctors and researchers think about cancer. Soon tumors may be diagnosed and treated much like an infectious disease. Just as identifying the bacteria or virus responsible for an infection helps doctors prescribe the right medication, finding the mutations behind a tumor could lead to treatments that target and knock out cancer cells while sparing healthy ones.

Where a quick-and-easy cure isn’t possible, cancer could be transformed into something akin to chronic infections like HIV or hepatitis C. “We’re trying to convert all of cancer to what we did for HIV,” says David Ryan, an oncologist at Massachusetts General Hospital, in Boston. With cancer, “You hear this constant, ‘I’m going to beat it, I’m going to beat it, I’m going to beat it.’ Well, HIV never goes away. You still have to take the medicines, but you can live with it.”

Live (longer) with it

Pharmaceutical companies have a couple of success stories that suggest Ryan’s vision may not be far-fetched. A drug called Gleevec has extended by years the lives of people with one form of leukemia. That drug stops the cancer-promoting action of specific proteins that help tell a cell when to grow and divide.

Some cancer-causing mutations switch these proteins, known as tyrosine kinases, to a permanent “on” position, like the accelerator pedal in a car getting stuck to the floor. Gleevec helps pull back on the throttle, slowing or stopping the cancer’s growth. Other drugs (erlotinib, gefitinib, cetuximab) block the action of a tyrosine kinase called the epidermal growth factor receptor or EGFR, which turns on cell growth programs. Genetic mutations that keep the protein permanently active or that cause too much of it to be produced can lead to lung cancer.

Despite these signs that targeting specific genes and their products can help fight cancer, only about 5 percent of all cancer patients have benefited from genetic testing thus far, estimates Daniel Haber, a cancer geneticist at Mass General who counts the development of erlotinib as the happiest experience of his research career. Yet the number of beneficiaries may be about to grow.

Researchers have recently learned that between 40 and 60 percent of mela­nomas and 7 to 8 percent of all cancers have accelerator-sticking mutations in a gene called BRAF that codes for another growth-control molecule. The most common of the mutations changes one link in the amino acid chain that makes up the BRAF protein, a switch from valine to glutamic acid.

A study published in the New England Journal of Medicine last year showed that an experimental drug known as vemurafenib could melt away many mela­noma tumors that had spread throughout the body by inhibiting the activity of the mutant BRAF protein. The drug, now approved by the U.S. Food and Drug Administration, has stopped tumor growth for more than seven months in patients with the mutation, extending their lives.

Other drugs are buying cancer patients even more time. Drugs that combat EGFR abnormalities have given lung cancer patients as long as 30 months from diagnosis. Standard chemotherapy typically offers only 12 months, says William Pao, director of personalized cancer medicine at the Vanderbilt-Ingram Cancer Center in Nashville.

“That’s a big difference if they want to see their kid graduate or another baby being born,” Pao says.

A case of cancer

“Magical” drugs like these are still wishful thinking for most patients, though, says John Iafrate, a molecular pathologist at Mass General and a pioneer of the hospital’s tumor snapshot program. The search for anticancer medications that pinpoint specific genetic defects reminds him of the early days of antiretroviral therapy for HIV.

Attacking one tumor process may not be enough to completely eliminate the cancer, he says, just as the antiretroviral drug AZT wasn’t able to control HIV infections on its own.

To leap forward, scientists have to find out what makes each person’s cancer tick, so they can go after it with the right drug combination. Iafrate and others are testing a few dozen genes known to be important in many different cancers. But that approach offers limited information — and in some cases can even mislead doctors.

That’s what happened in the case of a 78-year-old man with a very rare tumor on his tongue. Doctors treating the British Columbia man had his tumor tested for a small number of mutations, finding that the tumor cells made twice as much EGFR protein as normal cells do. A drug prescribed to combat the change proved futile; the cancer spread to the man’s lungs.

At that point, the clinicians decided they needed help. They turned to Steven Jones and his team at Canada’s Michael Smith Genome Sciences Centre at the British Columbia Cancer Agency in Vancouver. Jones’ team deciphered the complete genetic blueprint of one of the lung tumors. In a study published last year in Genome Biology, the researchers described the genetic mess: 7,629 genes were duplicated, triplicated or more; at least four chromosomes were missing huge chunks of DNA; and four genes contained mutations that would alter protein products. Also, 1,078 genes had higher-than-normal activity, while 1,986 others were less active than normal.

Compiling the data turned out to be the easy part. Figuring out which of these myriad abnormalities were responsible for the man’s cancer required “quite a lot of computational gymnastics,” Jones says. “There is no computer program that you just put your data in and it spits out what’s causing the problem.”

It took “15 people in a room scratching our heads and consulting the literature” to conclude that a gene called RET and its cronies were probably driving the tumor’s runaway growth, Jones says. RET produces a protein that helps cells decide when it is time to grow and differentiate. Other mutations in known cancer genes probably goaded further growth and made the tumors immune to the first drug.

With that info, the doctors put the man on a drug that inhibits the RET protein and other growth-promoting proteins. After about a month of treatment, the lung tumors had shrunk by 22 percent. But after four months on the drug, the tumors began to grow again, so doctors tried two other drugs. Those drugs bought the man another three months before the cancer started progressing yet again.

Researchers examined the genetic blueprints of the treatment-resistant cancer and found nine new mutations that weren’t in the original tumor or the man’s normal DNA. And other bits of DNA continued to be added and lost as well. Jones and colleagues speculate that only a carefully concocted cocktail of drugs could have stopped the cancer, which ultimately led to the man’s death.

Too little, too late

The case highlights how having a tumor’s complete genetic profile — what scientists call a genome sequence — can change patient care. But the study is also an exception. Most studies catalog a tumor’s genetic changes long after it has been removed from the body, when it is far too late to influence treatment, says Elaine Mardis, a genome scientist at Washington University School of Medicine in St. Louis.

Genome sequences aren’t often used during early treatment because it can take months to prepare samples, compile the genetic blueprints and then analyze the results. In a study published in the April 20 Journal of the American Medical Association, Mardis and her colleagues showed that they could find the source of a patient’s leukemia in just seven weeks, a short enough time to affect treatment.

Still, Pao says clinical tests shouldn’t take much longer than a week or two. And besides costing valuable time, the bill for a complete genetic blueprint can be substantial. Some companies estimate that it costs $5,000 to $6,000 to assemble a tumor’s whole genome.

Mark Boguski, a pathologist at Harvard Medical School, thinks it’s a bargain. “If I had cancer and $10,000, I’d have two choices: Take a cruise and check off my bucket list, or get my genome sequenced,” he says. “I’d get my genome sequenced, no question.”

But there are additional costs in the bioinformatics. Cancer databases that might help in interpreting genetic details usually don’t present information in clinically useful ways, Pao says. Even for a doctor who has a patient’s cancer genome in hand and knows all the mutations, it is incredibly hard to extract meaning from the alphabet soup of genetic errors. Pao and his Vanderbilt colleagues have built a new database that may help doctors decide which of the many mutations in a cancer cell are important, and which drugs to prescribe.

Right now, every time scientists compile a person’s genetic information, they have to sift through a mountain of data to find the changes that are driving that person’s cancer. “Every case is a research project,” Boguski says.

Spotting the drivers

Jerry Shay, a cell biologist at the University of Texas Southwestern Medical Center at Dallas, once wondered whether churning out reams of genetic data was even worthwhile.

“I started this thinking that we’d show most of this stuff was rubbish, and we were wasting money sequencing cancer genomes,” he says. “I’m a complete turn-around.”

Shay’s original problem with most cancer genome studies was that they came up with exhaustive lists of all the ways that cancer cells are messed up and then somebody had to make a rather subjective decision about which of those abnormalities was important. Instead of guessing, he and his colleagues decided to ask colon cancer cells.

Previous studies had estimated that 151 genes play a role in colon cancer. Only eight to 15 were thought to really drive the tumor — causing its out-of-control growth. The vast majority of mutations were thought to have happened incidentally and were like passengers on a runaway bus.

Shay’s team grew cells from the lining of the colon in lab dishes and then introduced mutations in two genes frequently involved in cancer, p53 and KRAS. But cells with mutations in either of those genes grew fairly normally, the researchers reported in the July Cancer Research. Then the researchers used a genetic trick to start knocking out each of the 151 genes one by one from colon cells that already carried either the p53 or KRAS abnormalities.

Instead of a bus with a few drivers and lots of passengers, Shay’s team found that 65 of the presumed passengers were anything but backseat drivers. Those mutations had a hand either directly on the wheel or were involved in biological processes with genes that did, encouraging the colon cells to grow like tumors. Of those, 49 sparked cancerlike behavior when paired with either p53 or KRAS mutations.

If the study were a Dr. Seuss book about colon cells, it might be called “Oh, the Places You’ll Go Wrong!” There could be 50 to 100 paths that lead a colon cell to cancer, not just eight to 15 as other researchers had thought, Shay says.

“The idea that cancer cells accumulate a lot of incidental mutations that don’t mean much is not well-founded,” he says. “It’s just not as simple as we’d like to think.”

Finding so many genes steering cancer could be good news for treatment. Right now, there is no way to stop KRAS once it has run amok. But the right drugs might persuade a codriver to hit the brakes, Shay suggests. He says more tests are needed to determine if other cancers also have many drivers.

Personal touch

After compiling genetic blueprints of more than 400 tumors from 20 differ­ent types of cancer, Mardis and her colleagues have discovered that genome sequences can point to more than just a cancer’s driving mutations.

One thing Mardis and her collaborators have learned is that tumors are not monolithic entities. They contain many groups of cells, some with mutations that might render them immune to chemotherapy or targeted drugs. Even if such cells make up only 10 percent or less of a tumor, they could still cause the tumor to spread or cause a recurrence at the original tumor site. Mardis wants to determine just how many cells’ DNA needs to be thoroughly looked at to identify all the problem mutations.

“We’re not going to learn that information until we just go ahead and do it,” she says.

Moving genetic testing forward may also one day help reveal whether precancerous cells are going to turn into cancer, saving some people from unnecessary surgery while allowing others to shut down cancer before a tumor revs up. “People don’t really understand just how personalized cancer care is going to become,” Mardis says.

The personal touch is still on the horizon, though. Despite decades of openly declared war on cancer, researchers are still in the early phase of learning about the genetics behind the disease.

When it comes to cancer, says Ryan, doctors are in the same position they were in when he started working in 1988 at Columbia College of Physicians and Surgeons in New York City. “HIV was rampant. Every night I was on call we’d admit 10 people, and seven to eight of them had HIV.” Once the protease inhibitors came out in the mid-1990s, the university hospital no longer needed a floor for AIDS and tuberculosis. “It’s gone,” he says. “That’s what we want for cancer.”

Tina Hesman Saey

Tina Hesman Saey is the senior staff writer and reports on molecular biology. She has a Ph.D. in molecular genetics from Washington University in St. Louis and a master’s degree in science journalism from Boston University.

From the Nature Index

Paid Content