Decoding tumors in search of more effective cancer treatments
Cancer cells are riddled with genetic errors, and each tumor has its own unique set of mistakes. Reading those errors, scientists believe, will help them not only understand how a tumor came to be, but also how best to poison it.
“Every tumor is telling its own story, its own history,” says Kevin White, director of the Institute for Genomics and Systems Biology at the University of Chicago. One by one, he’s reading and analyzing those stories as part of the university’s $5-million Chicago Cancer Genome Project.
The project is one of dozens of similar efforts around the world to decode and catalog the errors that cause cancer. Scientists hope to discover the most important glitches that will help them predict how well a particular treatment will work. In addition, certain errors may represent weak points in the cancer’s biology that doctors could exploit with new kinds of drugs.
When it comes to cancer treatments, “one size doesn’t fit all,” says Dr. Tanguy Seiwert, an oncologist at the University of Chicago who works with White. Seiwert’s focus is analyzing the genomes of head and neck cancers to look for reasons why certain drugs work well on some patients, but not others.
“If we understand the cancer at the molecular level, then we can modify our treatment to fit that,” he says.
Cancer results from a series of DNA-coding mistakes, or mutations. Normal cells can control their growth — and when they’ve produced enough of a certain kind of tissue, they stop.
But cancers accumulate mutations that deactivate the growth-control system. Mutations can also damage the cell’s quality-control processes so it makes more and more mistakes every time it copies its DNA. Knowing which mutations are responsible for a patient’s tumor could allow doctors to select a personalized therapy that would be most effective for that unique cancer, White says.
“At some point, sequencing the genome of a tumor is going to become a standard diagnostic,” he predicts.
In some cases, doctors already match cancer mutations to treatments. According to the Wellcome Trust Sanger Institute in Cambridge, England, which is running its own cancer genome decoding project, scientists already have identified 457 mutations that are linked to cancer.
For example, some breast cancers overactivate a gene called HER2. The drug Herceptin was designed to attack those tumors specifically. Similarly, some leukemia cells carry a well-known mutation that glues one gene to another; the drug Gleevec targets those cells.
Since the publication of the first cancer genome in 2008, several groups have started cancer-reading projects. The International Cancer Genome Consortium is sequencing tumors representing 50 types of cancers, while the National Institutes of Health’s Cancer Genome Atlas will start with brain, lung and ovarian cancers before expanding to other types.
Michael Stratton, co-leader of the Sanger Institute’s Cancer Genome Project, predicted recently in the journal Science that researchers would publish tens of thousands of cancer genomes within the next decade.
When White kicked off the University of Chicago initiative in 2009, the plan was to sequence 1,000 cancer genomes over three years. He has 500 and expects he’ll keep going after 1,000.
Instead of collecting a few each of a broad array of tumors, White and his colleagues in the Chicago project are making pointed inquiries about particular cancers. “The key is to identify specific questions that are meaningful,” Seiwert says.
The approach is “skinny and deep,” adds Dr. Charis Eng, director of the Genomic Medicine Institute at the Cleveland Clinic, “and that’s a good thing.”
White is partnering with several doctors who can use his sequences to advance their research. Among them is Dr. Ernst Lengyel, a gynecologic oncologist at the University of Chicago. He has long been frustrated by the fact that many women with ovarian cancer go into remission, only to return later with drug-resistant tumors. Only 30% survive ovarian cancer.
Lengyel hopes the sequences of different tumors — 28 so far — will help him figure out why some respond to his drugs but others don’t. Then he could give patients individualized advice about whether chemotherapy is worth it.
Another oncologist eagerly awaiting White’s sequences is Dr. Kevin Roggin, also at the University of Chicago. He expects preliminary results on 28 pancreatic tumors this May.
Pancreatic cancer starts with a cyst, but not all cysts become cancerous. Since the surgery to remove one carries a 30% to 40% risk of complications and a 2% to 10% risk of death, he says, it would be very useful to know who has a harmless cyst and who really needs an operation.
“It’s a matter of accurate diagnosis and early diagnosis,” Roggin says.
Roggin also hopes that finding the mutations that make a cyst or cancer especially nasty will give him ideas about new drugs to treat them.
Unlike researchers in many other sequencing projects, White is taking a shortcut. Instead of sequencing each tumor’s entire genome, he’s focusing just on the genes that are turned on in the cell. In so doing, he eliminates 98% of the DNA and can decode more tumors faster and cheaper.
The research is “very valuable,” although the focus on active genes creates a blind spot, says Dr. Cary Presant, an oncologist at the Wilshire Oncology Medical Group in Los Angeles. “It may be that the key to a person’s cancer is not the switches that have been turned on, but the switches that have been turned off.”
The researchers are now sequencing genes that aren’t on, White says, and they will likely switch to whole-genome sequencing when it gets cheaper.
Further complicating things for all cancer sequencing efforts, Presant adds, is that some cancers are caused not by direct changes to the cell’s DNA but by errors in the way the cell reads and uses that sequence. And a tumor’s genetic code may change over time.
“We can’t lose sight of the big picture and how confusing it is,” he says.
That big picture must include detailed analysis. On their own, sequences are just “garbage,” Eng says. It’s the careful identification of the genes — and how they fit together — that will move cancer research forward.
A tumor might have hundreds or thousands of changes to its genetic code, only some of which are actually causing the cancer. To find the most important errors, White and his colleagues compare their sequences to one another and to those in other databases, like the one at the NIH.
Although sequencing a cancer’s genome offers the hope of individually tailored therapies, actually providing them could be difficult, says Dr. Jonathan Licht, chief of hematology and oncology at the Northwestern University Robert H. Lurie Comprehensive Cancer Center in Chicago. Drug companies, he notes, are unlikely to spend millions of dollars developing a treatment that works for only a handful of patients carrying a rare mutation.
Thus, cancer geneticists must look for commonalities between tumors. For example, many different mutations might all boost cell growth in the same way, so drugs that affect that particular element of the growth process might be more practical than drugs that attack each mutation individually.
In time, the research could change cancer medicine from educated guesswork to reasoned decision-making.
“It’s going to be the right drug for the right patient at the right time,” Presant predicts.