The TAILORx clinical trial, which will evaluate a diagnostic test for breast cancer, is the first trial to be launched under an NCI program designed to bring new clinical tests for common cancers into routine use.

Since it began in 2000, the Program for the Assessment of Clinical Cancer Tests (PACCT) has provided guidance for developing and evaluating biological markers for a variety of purposes, such as diagnosing cancer, assessing prognosis, and predicting a patient's response to therapy.

When the program started, genetic tools such as DNA microarrays were changing the field of cancer diagnostics. Suddenly, it was possible to rapidly identify genes associated with cancer, including potential biological markers for clinical tests.

But despite the tools, remarkably few clinical tests were available to help physicians make treatment decisions for cancer patients. To speed the development of more tests, Dr. Sheila E. Taube, director of NCI's Cancer Diagnosis Program, created PACCT.

"We started PACCT to figure out why there were so few tests being used in the clinic when the literature contained so many reports about promising biological markers," says Dr. Taube.

As a first step, a working group identified some of the barriers to developing tests. These included a lack of regulatory guidance and no standardization in the techniques for developing markers, which made it difficult to compare results across studies.

Next, the team mapped out the steps for identifying and validating biological markers and translating the research into clinically useful tests.

"We determined what the ideal process should be for bringing potential markers through to clinical tests," says Dr. Taube. The process is an iterative loop, she says, in which knowledge gained about an experimental test in the clinic helps refine the test.

The approach stresses the importance of starting with a pressing clinical question that needs to be answered. Once the need is identified, promising markers can be evaluated within that context.

Success depends on having appropriate biological samples for developing markers. For instance, tumor tissue from completed clinical trials can be used to identify markers, but the markers need to be validated in prospective clinical trials, the working group said.

Dr. Taube's team decided to test the strategy by addressing clinical needs in early-stage breast cancer. For some breast cancer patients and their physicians, the most pressing need was for a test that could identify women who needed chemotherapy after surgery.

Studies had suggested that for some women the risk of recurrence was so low that they were unlikely to benefit from the addition of chemotherapy to hormonal therapy after surgery. These women might be spared unnecessary treatment if they could be identified.

In 2002, a PACCT working group was evaluating potential biological markers when Dr. Taube learned about a test being developed by a company called Genomic Health, Inc. The test, Oncotype DX, was designed to predict the risk of breast cancer recurring in certain patients.

"Genomic Health came to us and said they had the answer to our prayers," recalls Dr. Taube. "It turned out that the company had essentially independently followed all of the rules we had set for the ideal way of developing the test."

With support from PACCT, Genomic Health used specimens from previous clinical trials to develop the test in collaboration with NCI Clinical Cooperative Groups. The results, announced in 2004, suggested that the test is useful for identifying women whose risk of recurrence is relatively high or relatively low.

But questions remained about how to treat women whose test scores fell in the middle range. To find answers, PACCT and their collaborators developed the TAILORx trial to determine whether the test can be helpful in selecting the most appropriate treatments for these women.

The results of TAILORx are not expected for a decade, but Dr. Taube believes that Oncotype DX can serve as a model for developing other types of clinical tests for cancer.

By Edward R. Winstead