Protein Patterns Identify Cancer and Assess Drug Efficacy

BETHESDA, Maryland—New findings by proteomics researchers at the National Cancer Institute (NCI) and the Food and Drug Administration (FDA) have advanced efforts to enable physicians to monitor the response of cancer patients treated with molecularly targeted drugs and to diagnose ovarian cancer in the early stages of the disease.

The NCI released the outcomes of three studies originally intended for presentation at the 94th Annual Meeting of the American Association for Cancer Research (AACR) in Toronto. [The 2003 AACR meeting was canceled because of concerns about the threat of severe acute respiratory syndrome (SARS) in Toronto and has now been rescheduled for July 11-14 in Washington, DC.]

In a study led by NCI’s Virginia Espina, MS, MT, researchers identified several proteins—particularly one called AKT—that may prove useful in monitoring the progress of women treated for breast and ovarian cancer (abstract 2963). The approach involves measuring changes in the levels of active proteins inside tumor cells as a means of determining early in treatment whether a drug is working.

Molecularly targeted drugs are aimed at specific molecules in cancer cells, which allows monitoring of the signaling pathways the drugs are likely to affect. NCI researchers currently have studies underway to monitor the key pathways influenced by imatinib (Gleevec), trastuzumab (Herceptin), and gefitinib (Iressa).

Ms. Espina and her colleagues isolated breast cancer cells from tumor biopsies, measured protein levels in the signaling pathways targeted by trastuzumab, and determined how much of each protein was active. They measured the proteins before and several times after the women received trastuzumab therapy.

Prior to treatment, patients with poor outcomes tended to have higher levels of AKT, which promotes cell survival, than the women with better outcomes. Trastuzumab resulted in a decrease in active AKT, which may have enabled cell death, presumably by apoptosis. "Treatment with Herceptin appears to alter the level of active AKT in tumors," said senior investigator Lance Liotta, MD, PhD, of NCI. "We may be able to measure the degree of this change in patients who are receiving treatment to determine whether a drug that inhibits this signaling pathway is best for their individual cancer."

The researchers also suggested that the protein caspase 2, part of the molecular cascade that results in apoptosis, might be a key marker of whether a cell will follow the pathway to survival or death.

Two other papers scheduled for presentation at the AACR meeting involved work by scientists in the NCI-FDA Clinical Proteomics Program.

In collaboration with Correlogic Systems, Inc., of Bethesda, Maryland, NCI and FDA researchers improved the sensitivity of the company’s artificial intelligence computer program for diagnosing ovarian cancer (abstract 4846).

The Correlogic method analyzes protein patterns in blood samples to detect the disease, even in its early stages. In a previously published paper, researchers reported that they could successfully distinguish between women with ovarian cancer and those free of the disease.

In the new study, the team reported the discovery of specific protein patterns that accurately identified 100% of women with and without ovarian cancer. The previous study had correctly identified 100% of the ovarian cancer patients and 95% of the unaffected women.

Correlogic’s technique uses mass spectrometry to analyze blood samples. The researchers credited the improvement in specificity primarily to the use of a mass spectrometer with higher resolution.

"The increased resolution allows us to distinguish more features within the patterns generated from the serum samples," explained Timothy Veenstra, PhD, of the Mass Spectrometry Center at NCI’s Frederick, Maryland, facility.

New Tools

An NCI-FDA team also reported the development of new tools for visualizing and analyzing protein patterns. These tools both identify ovarian cancer and other diseases and allow researchers to determine how far the disease has progressed (abstract 1020).

"The new tools improve upon previous methods of identifying discriminatory protein patterns by allowing researchers to visualize the entire set of proteins in a single view, as well as zoom in and out to focus on regions of interest within the data," said Emanuel Petricoin III, PhD, co-director of the NCI’s Clinical Proteomics Program.

The technology also provides greater sensitivity and accuracy in diagnosis, according to lead investigator Donald Johann, Jr., MD. And it reduces the risk of error, increases productivity, and enables the analysis of large sets of protein data, Dr. Johann said.