By Dean A. Haycock

Special to BioWorld Today

The widespread use of fax machines, e-mail and cellular phones have many people complaining of "information overload." Some complain that their brains are numbed by seemingly unending streams of information. There are, however, topics about which even the most stressed person undoubtedly would welcome more, not less, information. A highly detailed description of an individual's cancer is an obvious example.

Significant progress has been made in recent decades regarding cancer diagnosis and the therapy - surgery, radiation, chemotherapy or some combination - that is linked to it. But this area of medicine often is still handicapped by lack of information about the details of individual patient's cancer. Many oncologists suspect that there are subtypes of cancers that they can not now recognize. Being able to spot them could lead to more appropriate and effective treatments.

"Most of the current approaches to cancer diagnosis are founded in microscopic examination of the tumor. This is in some cases augmented by a handful of specific molecular tests, but the general approach to cancer diagnosis is based on microscopic examinations," Todd Golub told BioWorld Today. Golub is a research scientist at the Whitehead Institute/Massachusetts Institute of Technology Center for Genome Research, of Cambridge, Mass., and an assistant professor of pediatrics at the Dana-Farber Cancer Institute, of Boston. He is the lead author of an article in the Oct. 15 issue of Science which describes a promising new approach to diagnosing cancer, "Molecular Classification of Cancer: Class Discovery and Class Prediction by Gene Expression Monitoring." Golub and his co-authors based their genetic approach to classifying cancer on the expression of genes monitored in DNA microarrays.

Microarray technology offers a method for simultaneously monitoring the expression of thousands of genes in a sample. In this study, the researchers started with RNA and used gene probes to obtain a picture of the pattern of gene expression in the samples they tested. Mathematical analysis applied to this pattern provides the means of recognizing and classifying important features in multidimensional data.

"The reason this research is exciting is because it suggests that it is feasible to use genetic analysis, in particular, gene expression profiling, as a mainstay for cancer diagnosis. That is, there is sufficient information contained in the genetic patterns in these tumors to be able to establish a diagnosis," Golub said.

The researchers applied their approach for categorizing tumors based on the particular set of genes they expressed to two different types of cancer, acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL). For these "proof of principle" demonstrations, they used a set of 6,817 genes. They analyzed bone marrow derived from an initial set of 38 patients. Twenty-seven were children with ALL and 11 were adults with AML. Independent samples from other sources were used to confirm the results.

The scientists report that they were able to correctly classify patients' tumors as being one or the other type of leukemia. This technique could be developed to provide a general approach for identifying new classes of other types of cancer and for assigning an individual patient's tumor to such classes. Furthermore, it could lead to better understanding of the pathophysiology of cancer subtypes.

"Our suspicion is that what appears to us microscopically to be a single type of cancer, for example prostate cancer or breast cancer, is likely to be a whole family of different subtypes of cancer that ideally we would like to treat in an individual way specific for each subtype," Golub said. "What this allows us to do is get a more in depth molecular fingerprint of different types of cancer in order to identify subgroups of cancer types that appear identical under the microscope. In the development of novel therapeutic strategies in the biotech sector or the pharmaceutical industry, that is the critical first step to refining the diagnosis so that we can optimally match patients with treatment."

While Golub cautions that "there will certainly be some details to be worked out when applying it to other types of tumors," he considers the methodology to be reasonably straightforward.

"You can think of this in three phases," he explained. "The first would identify previously unrecognized subtypes of cancer. The second phase would be using this information to most rationally utilize existing cancer treatments for these particular subtype of patients. The third phase - probably the most exciting - is to use these molecular insights that we have gained to design highly specific treatments that are really targeting the molecular pathways that are aberrant in particular subtypes of cancer."

The scientists now will focus on extending their results to other common types of human cancer, particularly those that appear to have a variable clinical response to current treatments, an indication that there may indeed exist different, currently unrecognized, subtypes of tumors. "We are pursuing this in lymphoma, for example, where it is difficult to predict response to treatment," Golub told BioWorld Today. "And we are doing this in prostate cancer where the best surgical and chemotherapeutic approach is quite problematic because of the wide range in clinical behavior ranging from slow-growing and indolent to very aggressive. We need better tools to get a hook into that. We are also trying to see if we can use this methodology to predict whether particular drugs are likely to be effective or not."

The work was conducted as part of an ongoing, functional genomics collaborative initiative. The researcher received financial support from the Leukemia Society of America, the National Institutes of Health, the Leukemia Clinical Research Foundation, Affymetrix, Millennium Pharmaceuticals and Bristol-Myers Squibb.

"There has been a lot of general enthusiasm about pharmacogenomics and about bringing genomics technology to the bedside, but there have been few examples of that actually being feasible in the foreseeable future," Golub said. "We are excited about this because I think it's bringing us a step closer to realizing the incorporation of genomics approaches into actual patient care."