Fingerprinting cancer development.

In the world of cancer, early detection is critical for successful treatment, but early diagnostic tools are in woefully short supply. Among ovarian cancer patients, for example, 80% of cancers are diagnosed at a late stage, and, as a result, only 35% of patients live past five years. Now the Clinical Proteomics Program is blending proteomics (the study of proteins in living cells) and clinical cancer research, with the goal of diagnosing cancer earlier, then tailoring drugs to precisely attack tumors with the fewest side effects for patients.

The effort, begun two years ago, is a joint effort of the U.S. Food and Drug Administration and the National Cancer Institute. It is codirected by Emanuel Petricoin, a biochemist at the Food and Drug Administration Center for Biologics Evaluation and Research in Rockville, Maryland, and Lance Liotta, chief of the Laboratory of Pathology at the National Cancer Institute Center for Cancer Research in Bethesda, Maryland.

Whereas genomics deciphers information stored in DNA, proteomics looks at proteins, which carry out the body's complex functions. Program scientists focus on how proteins signal each other and interact in so-called cellular circuits to find key "nodes" to target with drugs. These nodes are critical intervention points in the pathways involved in cancer, spots at which a drug could be used to block a key protein to prevent unwanted events (such as abnormal cell growth) from happening downstream.

The team designs special protein microarrays to detect specific phosphoproteins, which regulate early steps in cancer progression. These protein microarrays detect 50-100 key proteins at a time. "The only way to know if a pathway is [activated or not] is to measure proteomic levels," says Petricoin. Through this proteomics approach, program scientists hope to pinpoint key phosphoproteins that could, for instance, be controlled with a specific drug to block tumor growth with few side effects, as opposed to poisoning both tumors and healthy cells with the harsh chemotherapy drugs used now.

Protein microarrays have helped the team answer a central question in prostate cancer biology: Are rapidly growing tumor cells caused by an increase in growth rate or by a decrease in cell death rate (apoptosis)? In a study published in the September 2002 issue of Nature Reviews Drug Discovery, cell growth and apoptosis pathways were compared in normal, premalignant, and invasive carcinoma cells in prostate tumor samples. The researchers found that apoptosis signals were primarily suppressed in early carcinogenesis. "This gives clues about how to treat early lesions," says Petricoin. For example, depending on the proteins found to be important, drugs could be tailored to stimulate apoptosis to remove abnormal cells, or to block cell proliferation to prevent tumor growth, or to impede the growth of blood vessels that feed tumors.

Target cells for protein microarrays are gathered using laser-capture microdissection, a technology developed in the late 1990s to tease out normal, premalignant, and tumor cells from tissue samples. In this type of microscopy, a plastic film is suspended a few microns from the microscope slide holding stained cells. When activated by a laser, the plastic film melts, forming a bulge that drops and plucks cells out of the tissue section, explains Petricoin. The cells on the slide are viewed on a computer monitor, and scientists home in on them with a joystick "like in a computer game," he says.

Protein microarrays show patterns, or fingerprints, that identify early-stage cancer or monitor toxic effects of drugs. Petricoin and his colleagues are searching for fingerprints in blood that reflect cancerous processes before tumors are visible. In a study published in the 16 February 2002 issue of The Lancet, they used protein microarrays to analyze blood samples from 50 ovarian cancer patients and 66 healthy controls, and correctly singled out all 50 cancer patients and 95% of the controls.

Proteomic fingerprints also detect heart toxicity, a common side effect of the cancer drug doxorubicin. In collaboration with Frank Sistare, acting director of the Office of Testing and Research at the Food and Drug Administration Center for Drug Evaluation and Research, the team obtained preliminary results from a rat model showing that cardiotoxicity was detected 90% of the time before clinical symptoms occurred. "We're trying to find a fingerprint in a drop of blood that identifies early [heart] damage," says Petricoin. Such a fingerprint could identify susceptible patients, who could then be switched to another treatment.

Brian Leyland-Jones, a professor of oncology at McGill University in Montreal, Quebec, agrees that sampling blood would be a much preferable means for early detection of cancer. Right now, he says, all known biomarkers must be obtained by the far more invasive sampling of tumor tissue. He adds that new tests based on blood samples "offer the ability to find tumors early on and choose the best customized treatment for each patient."