Future of Cancer Diagnosis, Treatment Lies in Tumor "Barcode"
M. D. Anderson News Release 01/09/04
It has been said that a human being is a veritable encyclopedia of proteins. Proteins are the fabric of life – they provide the bricks and mortar of our cells, and run day-to-day operations. When these functions go awry – when too much or too little protein is produced, when a daisy-chain network of proteins working together is disrupted – illness can arise.
While an errant genetic code may underlie a disorder, biologists have estimated that 98% of disease is caused by something wrong in the proteins that genes produce.
So, in this era of molecular biology, oncology researchers are honing in on the proteins involved in the development and progression of cancer. Only by knowing the proteins altered, or responsible, for unrelenting cell growth can researchers control cancer. And turning cancer into a treatable, even preventable, disease is everyone's hope.
The hunt is on at The University of Texas M. D. Anderson Cancer Center to find such "markers" of cancer – a barcode of proteins or genes that identifies cells that have turned cancerous and gives hints as to how a patient's unique tumor can be treated.
These barcodes are expected to revolutionize cancer care, changing it from treatment based on a tumor's location in the body, to one that centers on an individual's unique cancer. A patient will not be just told that she has, say, breast cancer, but will be given a genetic/proteomic profile of her tumor, complete with a list of therapies shown to work with her subtype.
"In the next five to 10 years, proteins in a patient's tumor will tell us what is important to know about the nature of their cancer and how it should be treated, and gene and protein screens will be used to help direct therapy," says Gordon Mills, M.D., Ph.D., chair of the Department of Molecular Therapeutics.
"This truly will bring individualized treatment to oncology," he says, "and that will bring us closer to our ultimate goal – finding cancer early when it is most curable or making the disease one that can be chronically treated, manageable for many years."
The promise – and the problems
The idea that specific genes and proteins can be used to diagnose cancer and dictate treatment is an immensely exciting notion to cancer clinicians and researchers. Markers are "the apple pie and motherhood of molecular cancer research," says M. D. Anderson's top pathologist, Stanley Hamilton, M.D., head of the Division of Pathology and Laboratory Medicine. "And perhaps even more than genes, proteins are the ultimate participants in the molecular processes important to cancer."
But even as scientists work long and hard to find the specific genes and proteins that signify, say, prostate cancer that will spread, or breast tumors that will remain benign, they know they have a long row to hoe.
Although the 45,000 or so genes in a human cell have been catalogued, scientists do not know the function of most of the genes. The proteins that genes produce are even more mysterious: researchers estimate that each gene can give instructions for as many as 100 different proteins.
Worse still, investigators are finding that the biology of a cancer cell is changeable, so that genes and proteins found to be switched on in a cancer cell one day may not be active the next day. For example, when a chemotherapy drug shuts down one crucial molecular "pathway" of protein in a tumor, other redundant pathways may take over.
When Hamilton looked at what happened to genes in colon cancer cells after exposure to the common chemotherapy drug 5FU, he saw "an astounding change in gene response – 500 genes showed alterations, and 15% of all genes in the cell changed their response in a three-day period," he says. "Cancer cells don't sit by and let themselves be killed. They respond with an absolute ballet of protein molecules handing off functions to one another."
Hamilton says that despite the spate of studies being published that identify markers of interest, most have not been validated. "The markers are just not good enough yet," he says. "The studies may show they are highly biologically significant, but not clinically significant."
Scientists want something as effective as a pregnancy test to diagnose cancer, "but that is not happening yet, and it won't be anything we will have in the next two to five years," he says.
Still, the field is buoyed by success stories, such as the drug Gleevec, which is designed to shut down the crucial genetic/protein pathway involved in chronic myeloid leukemia and another rare cancer of the gastrointestinal tract, called gastrointestinal stromal tumor. Herceptin treats metastatic breast cancer that is driven by a gene known as HER2 that produces too much of a protein that tells cells to divide. Proteins found to be markers of these cancers also proved to be the switch that could turn the disease off.
So now researchers are using the most advanced techniques, such as DNA microarrays and proteomic technology, to find those genes and proteins that are specific to cancer, and that may offer a potential target for drug treatment.
At M. D. Anderson, those investigations are under way in many different types of cancer, and progress has been reported in some of the most common, as well as the most dangerous, types.
For example, the potential of proteins to flag disease made international news in 2002 when a team of researchers that included Mills showed that a palette of just five proteins could identify patients with known ovarian cancer.
Using a drop of blood from 116 women, the researchers correctly identified all of the 50 women who had ovarian cancer, as well as the lack of cancer in 63 of 66 healthy women.
The advance, reported in the British journal The Lancet, was exciting to researchers as well as the public because it offered a new future for one of the most deadly of cancers. Like lung cancer, ovarian cancer is unusually deadly because it is most often detected at a late, difficult to treat stage. If diagnosed before symptoms appear, perhaps by a blood test, however, the cancer is much more treatable.
But the researchers, who also hailed from the National Cancer Institute (NCI), the Food and Drug Administration (FDA), and Northwestern University in Chicago, say they don't yet know what the proteins are or how they even function in a tumor.
Still, the work is so suggestive that Mills and researchers at the NCI and Northwestern are using the protein combination to see if it can detect ovarian cancer patients at several institutions. That study, soon to be completed, will likely be followed by a prospective multicenter clinical trial to validate the proteomics approach to diagnosing ovarian cancer.
"Our goal is to find and classify patterns of protein changes that we can use to detect and predict cancer and other diseases," Mills says.
Subtypes of colon, lung cancer
As more and more subtypes of cancer are discovered, the need for biomolecular markers has become clearer and more urgent, M. D. Anderson researchers say.
Colon cancer is a good example of tumor diversity. It is not just one uniform disease; there are many subtypes that behave in biologically different ways, and some are more prone to spread than others.
Figuring out which colon cancer is more or less dangerous has been the research focus of Robert Bresalier, M.D., professor and chairman of the Department of Gastrointestinal Medicine and Nutrition in the Division of Internal Medicine.
Such knowledge, Bresalier believes, may pave the way to improved diagnosis and therapy. So far, the pursuit has resulted in "substantial progress in finding sensitive, clinically useful biomarkers," he says.
Bresalier and his colleagues have found several proteins associated with colon cancer metastasis, and early laboratory, tissue-based studies indicate they can use protein screens to correctly pick out cancers that have metastasized from those that have not.
One protein, galectin-3, known to act as a kind of biological glue that helps cells stick together, "looks quite good as a potential marker for colon cancer," Bresalier says. Another family of proteins, called mucins, are over produced in most colon cancers, but Bresalier has found that tumors in the colon that produce specific mucin subtypes are associated with a worse prognosis.
"Our department is working at a variety of levels to identify markers, including these two, to help us detect colon cancer earlier and to refine our ability to determine prognosis," he says.
"Although the data we have so far are exciting, they are preliminary. There are no magic bullets yet."
Lung cancer may also be "100 different diseases" and the only way to determine which type of tumor a patient has is to use biomarkers, says Li Mao, M.D., an associate professor in Thoracic Head and Neck Medical Oncology Research.
Mao and his colleagues found molecular evidence that former smokers are still at high risk for developing lung cancer. In studying the genetic effects of tobacco smoke on the lung tissue of chronic smokers, researchers found genetic damage in the lungs of 82% of current smokers and 62% of former smokers. Such damage may lead to lung cancer.
Because it is difficult to take tissue from patients' lungs to determine their risk of developing lung cancer, Mao is working at pinpointing proteins in a simple blood sample that signify development of the disease from premalignant lesions.
"We have found that a single protein marker may not do the whole job," says Mao, "so now we are looking at a panel of molecules combined together that can give us more predictive powers." His team also is looking for markers of head and neck cancer, which also is strongly associated with smoking.
Eventually, a drop of blood from a smoker or former smoker could "predict lung and head and neck cancer risk as well as optimal treatment, possibly within a decade," says Mao. "To do a good biomarkers study, you must have a very strong clinical base, and that is the advantage we have at M. D. Anderson."
Proteins and genes involved in breast cancer
Breast cancer researchers at M. D. Anderson are using both proteins and genes to help determine which therapy will work best in patients – before they even begin any treatment.
In one landmark study, researchers demonstrated that they might be able to predict which women with early stage breast cancer could be cured by using a particular chemotherapy treatment before surgery.
The study, led by Lajos Pusztai, M.D., Ph.D., an assistant professor of Breast Medical Oncology, is one of the first to show that a genetic profile of a breast cancer tumor can help direct therapy.
This potentially represents a big step forward toward "personalized medicine," says Pusztai. "If these results continue to hold up in larger validation studies, they can fundamentally change the way that chemotherapy is selected for patients."
Using breast cancer cells taken from 24 early stage breast cancers, Pusztai and his colleagues discovered 74 genes associated with a cure when a common chemotherapy regimen known as Paclitaxel/FAC was used. From those genes, they created markers to predict the outcome in 21 newly diagnosed patients and found the test was 75% accurate in forecasting which women would respond to FAC.
Now the investigators are testing the gene screen in a larger randomized study at M. D. Anderson. If this study validates the utility of the gene screen, a predictive test may be available for widespread use in two to three years, Pusztai says.
He adds that he expects the test to be over 80% accurate.
Eventually, the researchers want to develop a palette of tests that will predict outcomes from a number of different chemotherapy regimens, he says. "We want to be able to select from a large number of seemingly appropriate chemotherapy choices the regimen that is most likely to cure an individual at the time of diagnosis," he says.
A different team of M. D. Anderson researchers is studying the power of a protein to help doctors decide if breast cancer patients even need such chemotherapy – or if they need much more aggressive therapy.
In a study of about 400 patients reported in the New England Journal of Medicine, the scientists found those with high levels of a protein called cyclin E were significantly more likely to have an aggressive, invasive breast cancer. The protein, which is involved in regulation of a cell's living cycle, appears to be a much better predictor of patient outcome than any current predictive marker, says the study leader, Khandan Keyomarsi, Ph.D., associate professor in Experimental Radiation Oncology.
Currently, the prognosis for women diagnosed with breast cancer is determined by assessing whether tumor cells have spread to lymph nodes. But some women who have cancer cells in the lymph nodes never have a recurrence, while others whose cancer has not spread do have a recurrence. Yet many women undergo chemotherapy because of the uncertainty of their prognosis.
If an accurate predictive marker were available, many women could be spared chemotherapy, Keyomarsi says, but adds that the finding must be validated in larger clinical studies. She also is working on methods to prevent generation of excess cyclin E as a way to prevent a cancer from becoming aggressive.
"My hope is that this technique may help ease the burden of chemotherapy among breast cancer patients," she says.
Mills, who helped develop the ovarian protein test, is currently collaborating with Francisco Esteva, M.D., Ph.D., an assistant professor in Breast Medical Oncology, and Ryuji Kobayashi, Ph.D., a professor in Experimental Pathology, and colleagues at a number of other institutions, to look for novel breast cancer markers that could include proteins, lipids, genes and other molecules associated with tumor cells. Early results in an ongoing study suggest that a panel of proteins is at least as effective as mammography in determining whether women have a breast abnormality, but “these results need to be compared with other ongoing studies,” says Mills.
"There is real excitement in the potential of these protein tests," he says. "For a number of diseases, I think we can expect to have powerful diagnostic and predictive tests within a decade."
"We are early in this field, and its history has been a little optimistic," says Experimental Therapeutics Professor Walter Hittelman, Ph.D., a basic researcher who studies the potential of biomarkers to help thwart even the initial development of cancer. "There will be different rates of discovery for different types of cancer, but we are getting there."