Janet Rowley noticed something odd about the glowing chromosomes revealed by her microscope. It was the early 1970s, the first years of the so-called "war on cancer," and she was using a new staining technique to examine cells from patients with chronic myelogenous leukemia (CML), a cancer of the blood that was almost always fatal. The technique highlighted bands within the chromosomes, and she could see an extra piece on the end of chromosome 9. That fragment was nearly the same size as a "missing" chunk of chromosome 22 that other researchers had detected a decade earlier. To Rowley, it looked as if the tips of these two chromosomes had swapped places, or translocated. During the next few years she found two other cases of chromosomal translocation in different forms of leukemia. The finds forever changed the way scientists thought about cancer.
Shuffled chromosomes in leukemia established that broken, scrambled and messed-up genes cause cancer. The genetic code details when cells should grow, divide and eventually die. Cancer is a disease of misinformation—cells ignore the rules, growing despite multiple molecular signals telling them to stop and invading other tissues because they no longer respond to biological messages to stay put or even destroy themselves. In the past four decades scientists have identified thousands of genetic mistakes that either cause cancer or boost the risk of developing it. The effects of these typos are sometimes dramatic—the gene variants BRCA1 and BRCA2 can boost women's lifetime risk of developing breast cancer from 12 percent to 60 percent. Some errors are found only in cancer cells themselves; other changes can be passed from generation to generation. The latter are the mistakes that may be passed down and boost the risk of developing cancer—this is the inherited genetic risk, or the reason that people with a familial history of a disease may want to get tested earlier or more often.
As researchers uncover more genetic mistakes and delve deeper into the human genome, it may be possible to pin down the exact probabilities conferred by inherited genetic risk. If clinicians could scan a healthy person's genes for variations that explain their probability of developing cancer, perhaps they could prevent or catch the disease before it became a problem: Spit into this vial and the doctor will tell you what will ail you in 20 years.
Despite the plummeting cost of DNA sequencing technology, much of the information is a jumble of alphabet soup. Science can figure out what gene variants and markers a person has, but they can't tell exactly what it means for his or her health. It will take researchers years to untangle the genetics of cancer. Even large steps, heralded as a major advances, answer few questions and pose many more.
This spring, a massive international collaboration doubled the number of known genetic regions associated with the risk of breast, prostate or ovarian cancers. The genetic markers are signpost that researchers can follow to better understand the biology of these cancers. Only a few of the 74 newly identified markers are shared by more than one type of cancer, underscoring cancer's complexity. Yet exactly how the findings can inform public health recommendations remains to be discovered. Each marker is associated with small modifications of risk, but the effects add up. The findings could lead to more accurate cancer screening and hint at ways cancers could be caught before the disease becomes aggressive. Only further study, however, will show where to draw the lines between risk percentages that tell patients "not to worry" or "get tested now."
Reams of data
The impressive number of hits in the new work stems from the size of the research effort: 160 institutions around the world analyzed a pool of more than 200,000 individuals' genetic sequences. The international project is called the Collaborative Oncological Gene-environment Study (COGS). To find the dozens of new cancer risk regions, researchers combed the pooled genetic information for variations called single-nucleotide polymorphisms (SNPs). A SNP is change in a single letter of the DNA code, likely introduced as a "typo" during gene replication. If such a change happens within a gene, it can affect the structure of proteins. If it falls within a stretch of DNA that regulates genes, it can affect the amount of protein a cell produces.
The COGS researchers put 211,000 SNPs of interest, which were previously identified in other studies, on a custom-made DNA array that looks a bit like a computer chip. Then they used the chip to scan the pooled genetic information to look for differences between people who had cancer and those who did not. If a particular SNP popped up more often in the group of people who had cancer, that SNP could be linked to increased risk for that cancer.
Most of the SNPs the cancer teams identified are specific to one of the three cancers, but 17 are shared risk factors for all three. The new SNPs, combined with 75 previously known markers, explain a proportion of inherited genetic risk for these cancers: 28 percent for breast, 4 percent for ovarian and 30 percent for prostate cancer. The research was published as a collection of 13 papers in April in Nature Genetics, Nature Communications, PLoS Genetics, The American Journal of Human Genetics, and Human Molecular Genetics (Scientific American is part of Nature Publishing Group).
Comparing genomes to uncover SNPs of interest is one way that researchers dig down to find the genetic basis of complex diseases. "We get little bits that we put together," says Stephen Chanock, chief of the Laboratory of Translation Genomics at the National Cancer Institute. Chanock was involved in several of the new studies. Complex diseases such as cancer spring from many gene variants that all contribute to the disease. These studies are helping researchers fill in the list of risky genetic markers. "What is emerging is the complicated genetic architecture of different diseases," he says.
Following the signs
Rowley's swapped chromosomes eventually led researchers to find a way to treat CML. Now patients can take a pill that jams a monkey wrench into a process vital to the cancer's development. The drug, called imatinib (first marketed as Gleevec in the U.S.), often grants patients a normal life expectancy with minimal side effects. Few cancer treatments have met this high bar, but researchers still comb the genome for cancer's fingerprints and clues to what might stop the disease.
Indeed, the COGS findings provide signposts for future research. For example, a handful of SNPs associated with prostate cancer risk fall within genes important for the binding of a cell to a surface. That surface could be another cell to facilitate communication or create a barrier through which pathogens cannot pass. Cell–cell adhesion is important for immune response and is also involved in tumor metastasis—tumor cells use cell adhesion to stick to a new location in the body. Understanding the mechanisms for cell–cell adhesion may offer insights for new treatments.
In a number of regions the SNPs are involved in more than one type of cancer cluster. "In some cases we are beginning to understand why that is," said Doug Easton, a professor at the University of Cambridge and the lead author of the main breast cancer paper at a press conference before the papers were published. One of the COGS papers honed in on a genetic region that helps control the length of telomeres, which are protective caps on the end of DNA strands. The so-called TERT locus harbors SNPs relevant to both breast and ovarian cancer risk, making it a prime candidate for further study.
The next step for the international cancer teams is an even larger study with a new chip called OncoChip, made by Illumina, Inc.* They plan to screen 600,000 SNPs of interest to see if they are involved five malignancies—ovarian, breast and prostate as well as colorectal and lung cancers. The larger numbers will give the researchers more statistical power to uncover less common gene variants. In addition researchers will map the already discovered variants to figure out which genes and biochemical pathways are involved. Studies of gene function are critical to characterize cancer biology, says Mathieu Lupien, a scientist at the Ontario Cancer Institute and assistant professor at the University of Toronto who was not involved in COGS. "Now we can move forward and understand why it is those genetic defects promote cancer," he says.
Weighing the risks
Genetic markers may lead to better treatments, but researchers also hope to catch cancer before it starts.
Clinicians already use cancer-risk calculators to group people into high- and low-risk categories based on lifestyle choices, environment and family history. The new SNPs could be additional indicators that make the stratification more accurate and efficient. High-risk groups could get targeted recommendations for avoiding risky behaviors or whether to get screened for a type of cancer.
Currently, cancer screening is saddled with a lot of false positives, which means people who do not have cancer are told they are positive. Such results lead to unnecessary and even dangerous procedures—not to mention the anxiety felt by those who believe they have a potentially life-threatening disease. For example, screening for prostate, lung, colorectal and ovarian cancers in 68,436 people over a period of three years led to a least one false positive for 60 percent of men and 49 percent of women. Another study found that follow-up procedures (such as a biopsy) after a false positive cost an average of $1,024 for women and $1,171 for men. The challenge is figuring out where to draw the line for high-risk, says Ros Eeles, a professor of oncogenetics at the Institute of Cancer Research in London and one of the principal investigators involved in the main prostate cancer paper. "We could do the test and give a risk profile, but we don't know what you should do when you have the information," she says. Studies that retroactively profile genetic risk markers in patients could reveal where the lines should fall and what interventions are most effective.
"We're not to the point of being able to predict an individual's risk," says Joe Gray, a professor at Oregon Health & Science University's Knight Cancer Institute and not involved in the COGS studies. A more thorough understanding of risk factors and better cancer screening could lead to a future where doctors can "prevent people from having cancer we don't know how to treat," he says.
More information about how different genetic variants contribute to risk of disease could help refine the definition of high-risk groups. A well-tested SNP profile could sort out individuals at the top of the spectrum, where the benefits of screening would outweigh the risks. Once the genetic risk is understood, public health professionals can employ the same communication strategies used to counsel people about heart disease risk. "We do [stratified screening] all the time with cardiovascular risk," says Hilary Burton, director of the PHG Foundation based in England. Right now the evidence on breast cancer screening is "finely balanced between benefits and harm," she says. The newly identified SNPs can help tip the balance for some carefully identified individuals.
The vision the researchers outline could be in the not-too-distant future. People in their 40s today might see safer, stratified risk screening for some cancers within their lifetimes, Cambridge’s Easton said in a press conference. Still, before all patients can receive the benefits of safe screening, researchers will need to address a gap common to current genome-based discoveries: The 200,000 people in the COGS pool are largely of European descent and live Australia, North America and Europe. Whereas the consortium did find some risk markers specific to people of Asian descent, other genetic groups such as African and indigenous populations in the Americas and Australia are underrepresented.
"We are still very much in the discovery mode," Chanock says. Decades after uncovering the genetic basis of cancer, that is a sobering statement.
*Correction (May 23, 2013): This sentence was edited after posting. Due to an editing error, it had erroneously identified Signature Genomics as manufacturer of the OncoChip.