The evolution of cancer genomics and what it means to you.
Rita Doerner was stunned earlier this year when doctors told her she had stage 2 breast cancer. The 77-year-old from St. Louis did not have any symptoms. Although her sister died from ovarian cancer, Doerner did not think she had any particular risk. She figured genetics were on her side. “When they told me I had cancer, I didn’t believe it,” she says. In addition to standard hormone therapy, Doerner has tried a new investigational drug—one that is not specifically geared to treat breast cancer. Rather, it targets cancers with a mutation in a gene called PIK3CA—which Doerner has. Because the mutation developed in her tumor and was not inherited or present in her normal cells, it created an opportunity to intervene with a treatment that is quite specific.
The human genome is the instruction manual for assembling all the molecules that keep cells alive. It’s a biochemical code, and sometimes the coding ends up with its own bad autocorrect. Often these glitches don’t matter, but sometimes, like writing “one” when you mean “none,” a slight change makes a big difference.
A malformed molecule comes off the factory line, and like paper airplanes, the proteins of life won’t work if they’re not folded just right. When a cell collects too many misshapen molecules from bad instructions, the normal checks and balances on growth can go offline. These errors manifest in two main ways. In one instance, certain genes, called oncogenes, get switched on and can lead to cancer growth. In another situation, genes, called tumor suppressor genes, are switched off. Either way, these errors can be inherited or develop in the body over time. This is why cancer is ultimately a genetic disease, and why, in Doerner’s case, genetics were not on her side.
Researchers have long searched for genetic errors responsible for corrupting a normal cell and sending it down the path to a runaway tumor. But until recently, the hunt for cancer-causing genes was a slow, tedious and expensive process. Obtaining the sequence of a gene (reading its code) had to be done by hand, sifting through DNA bit by bit. When researchers embarked on the U.S. Human Genome Project in 1990, the job of sequencing all human genes took more than a decade to complete and cost almost $3 billion. Some of the first cancer sequencing efforts cost more than $100,000 per tumor.
But just as your car now has more computing power than an Apollo spaceship, so has genetic sequencing become more compact, affordable and nimble than scientists once thought possible. The most modern method of gene sequencing—called next-generation sequencing—analyzes genes so rapidly that scientists almost have the love-struck awe of a person who’s gone from an old-style, dial-up modem to high-speed Internet overnight.
“Before, we had to search one gene at a time, going through 20,000 genes,” says Kenneth W. Kinzler about the hunt for cancer genes. “More than two decades of research can now be repeated in a week,” adds Kinzler, an oncologist who runs a lab focused on the genetics of cancer at The Johns Hopkins University School of Medicine in Baltimore. It is now possible to quickly reveal the code for every gene inside a tumor in a couple of weeks and compare that tumor’s genes to the genetic instructions of normal tissue. All for just a few thousand dollars—bargain prices, as genomes go.
Next-generation sequencing has not only transformed cancer genetics in the laboratory; it may one day directly guide the practice of oncology. The idea is that by taking a sample of a tumor and rapidly figuring out which instructions are wrong, doctors will know where the cancer drug needs to work. (If the cancer is occurring because certain genetic mutations are allowing specific proteins to take over a cell, you know specifically what task a drug needs to accomplish.) A version of this kind of tailored care is available now only for a handful of well-studied mutations, such as HER2 in breast cancer or EGFR in colon and lung cancers. Doctors envision a day when they can expose all a tumor’s genetic tricks. “I think this is going to be standard of care, and not in 10 years, but a few years,” says Michael Snyder, director of the Stanford Center for Genomics and Personalized Medicine in Stanford, Calif. In the not too distant future, he continues, “I can’t imagine getting cancer and not getting your genome sequenced.”
That said, many hurdles—some scientific, some economic—are keeping next-generation sequencing in the laboratory and out of doctors’ offices for the time being. The challenges are substantial enough that most researchers still hesitate to predict when or how rapid gene sequencing will bring dramatic improvements for patients. Think of it this way: Simply knowing all the suspects to the crime won’t necessarily tell you which ones are guilty or how to stop them. Many genetic mutations have no known drug that works against them.
But researchers have gotten quite good at rounding up possible perpetrators. Next-generation sequencing began to evolve around the time the Human Genome Project was wrapping up but did not become widespread until recent years. In its simplest sense, next-generation sequencing is a complex, computerized process that involves slicing up DNA into tiny bits, sending them through a device that can determine the sequence of each piece one base pair at a time in parallel with thousands of pieces simultaneously and then reassembling the pieces to reveal their order. Researchers recently called this process “arguably one of the most significant technological advances in the biological sciences of the last 30 years.”
Next generation sequencing is a complex, computerized process that involves slicing up DNA into tiny bits, sending them though a device that can determine the sequence of each piece one base pair at a time in parallel with thousands of pieces simultaneously and then reassembling the pieces to reveal their order. Researchers recently called it “arguably one of the most significant technological advances in the biological sciences of the last 30 years.”
Its invention has inspired scientists to create an compendia of cancer genomic analyses, much like the Human Genome Project catalogued the genes of normal life. One effort, called The Cancer Genome Atlas (TCGA), was launched in 2006 by the National Institutes of Health with a goal of publishing information about key genetic changes for at least 20 types of cancer. Thousands of tumor samples have been donated for analysis. An international undertaking is the Collaborative Oncological Gene-environment Study, which is focusing on inherited variations of genes that increase the odds of breast, ovarian and prostate cancers.
Taken together, discoveries made in these cancer genome studies are creating a genetic road map for future treatments and detection. So far TCGA has uncovered new genes that are important to certain kinds of brain tumors, colorectal tumors, lung cancers and others. A paper published last year in the journal Nature reported finding genetic alterations important to the second most common form of lung cancer (squamous cell carcinoma), a malignancy that up to now remained a genetic mystery. And in May, two separate teams of researchers from the TCGA Research Network, writing in The New England Journal of Medicine and Nature, described genetic alterations for endometrial cancer and acute myeloid leukemia. One genetic mutation found in endometrial tumors was previously thought to be important only in colon cancer.
In March, five different reports in the journal Nature Genetics revealed new findings from the Collaborative Oncological Gene-environment Study describing about 70 different inherited genetic variations (also known as polymorphisms) that increase the risk of cancer.
In general, of the genes discovered in cancer genome projects so far, “about two-thirds had previously been implicated in cancer,” Kinzler says. “The other third we had no idea were involved.” The research is also, for the first time, giving scientists a sense of just how many genetic errors add up to cancer. Common malignancies of the breast, brain, colon and pancreas can have around 33 to 66 different mutations. But melanomas and lung cancers can have more than 200 genetic errors because carcinogens, such as certain chemicals in tobacco and ultraviolet light, damage genes. Genetic studies also reaffirm that these mutations do not pop up all at once. A precancerous colon polyp may have relatively few genetic alterations, but by the time cancer develops years later, that lump of tissue might have accumulated dozens of genetic mistakes.
But cancer is not a deadly prize that goes to the tissue with the most mutations. Some childhood tumors have fewer than 10 genetic flaws. What matters is what the misguided genes are telling a cell to do. All genetic injuries are not alike, just as a broken transmission in a car is a more serious and necessary repair than a dented bumper. In other words, next-generation sequencing has made it easier to find genes gone wrong, but no drug can target all those mutations, or even needs to. The hard part is figuring out which mutations (and the molecules they affect) need to be targets of treatment. Scientists talk about “driver” and “passenger” mutations. Next-generation sequencing locates everything on the bus.
That’s why many experts are hesitant about becoming too exuberant about sequencing, at least for now. William Foulkes, director of the Program in Cancer Genetics at McGill University in Montreal, Quebec, describes himself as “healthily skeptical” about what rapid genetic sequencing means for patient care today.
“There is more to be discovered about the complexity of the genome,” Foulkes says. Thinking of it another way, technology can search inside tumors and find mutations that doctors never knew existed—but just like a workplace with bosses, employees and opportunists waiting for their big break, each mutation occupies a different hierarchy in the operation. The most common mutations, or even the mutations most important at the moment, may not be the ones that ultimately determine the success or failure of treatment, Foulkes says. “There will always be the possibility that the mutation lurking deep in the tumor is already resistant to the drug you haven’t yet given,” he adds. Or the drug that doesn’t yet exist.
A focus of research now is how to make use of the deluge of genetic information coming out of the lab. Studies of cancer genomics have found that different mutations can have similar effects on cell growth. And, despite intense study, it’s still unclear why some tumors have the genetic programming to spread to other parts of the body, while others smolder silently for years.
Nonetheless, some companies are starting to ease cancer genomics into clinics. Kinzler from Johns Hopkins, for instance, helped found a company called Personal Genome Diagnostics in 2010 to offer genetic analysis of tumors. But for the most part, genome testing still isn’t useful enough yet to be economically feasible for an everyday oncologist’s office. More often, cancer genome analysis can help patients in research trials. An example: Researchers from Washington University in St. Louis are starting a study of women who test negative for HER2 by the standard test, but nonetheless have an “activating” mutation in the gene. They test negative, says Matthew Ellis, director of the breast cancer program at Washington University, because the mechanism behind their cancer does not require a lot of the HER2 protein to be made, which is the basis for a positive diagnostic test. They could be helped by an experimental drug that targets the mutant HER2—but neither they nor their doctors know it. “What we are doing now is finding these patients by sequencing and treating them,” Ellis says.
For too long in clinical cancer investigation, we have taken a “drug first” approach. Instead of “drug first,” what we need is “biology first.”
And even if the technology does not soon become a part of medical care, it stands to change the language of cancer. Instead of talking about malignancies from their origin—a brain tumor or colon cancer—doctors may someday classify cancers according to the genes that give a tumor its power, like Doerner’s PIK3CA mutation. Drugs will be grouped that way, too. “For too long in clinical cancer investigation, we have taken a ‘drug first’ approach,” Ellis says. Oncologists have an array of chemotherapies on hand and prescribe each agent based on assumptions of how similar tumors have responded, rather than details about the tumor’s biology. “Instead of ‘drug first,’” Ellis says, “what we need is ‘biology first.’”