[Cathya]

Cancer Diagnostics: Informing the Development of Tailored Cancer Therapy


Reported by Dorie Hightower
May 23, 2006

Advances in genetic research are transforming cancer diagnosis, and treatments that are tailored to the specific molecular changes that drive tumor growth are now being developed. Because most cancers result from several genetic mutations, NCI is focusing on developing diagnostic tests that identify a number of genetic markers that can be used to predict a person's response to specific therapies and inform treatment decisions for these diseases.

BenchMarks discussed those new diagnostic tools with Sheila E. Taube, Ph.D., director of the Cancer Diagnosis Program (CDP) for NCI’s Division of Cancer Treatment and Diagnosis. The Cancer Diagnosis Program is developing new tests using information from the human genome program and powerful new molecular technologies to individualize cancer treatment.

How do clinicians currently make treatment decisions for breast cancer?

In early stage invasive breast cancer, the most important issue is the likelihood of recurrence. The evaluation of the likelihood of recurrence is based on multiple factors, such as nodal status, tumor size, tumor grade, and ER (estrogen receptor), PR (progesterone receptor), and HER2 status, as well as the age of the patient. Currently, treatment decisions are based on the outcomes of many clinical trials, and on observations of large populations, which provide population-based probabilities of recurrence, but not an individual’s own risk. However, there is a significant unmet need: the ability to provide patients with quantitative, clinically validated predictors of the risk disease recurrence in the individual. Such information can help the patient make the best breast cancer treatment decisions.

How would the OncotypeDX™ improve decision making?

Oncotype DX™ is a tool that quantifies the likelihood of disease recurrence in women with early stage breast cancer and assesses the likely benefit from certain types of chemotherapy. With this information, it may be possible for doctors and patients to make more informed decisions about breast cancer treatment options. The test analyzes a specific set of genes within a tumor to determine a Recurrence Score™--a number between 0 and 100 that corresponds to a specific likelihood of breast cancer recurrence within 10 years of the initial diagnosis. The test has been extensively evaluated on tumor samples from patients where the outcome is known, and the predictive ability for an individual appears to be quite consistent.

Would the Oncotype DX™ substitute for all of those factors, such as the tumor size and the patient’s history and age?

Yes, for early-stage, node-negative and hormone-receptor positive breast cancers. Along with staging, grading and other tumor marker analyses, Oncotype DX™ can provide greater insight into the likelihood of disease recurrence. The quantitative assessment that this test offers is important in weighing the benefits vs. the risks of adjuvant chemotherapy, and helps to determine the most appropriate treatment strategy – tailored to the individual patient.

What is the TAILORx trial?

The Trial Assigning IndividuaLized Options for Treatment (Rx) – TAILORx -- will focus on a molecular signature to determine the most effective cancer treatment, with the fewest side effects, for women with early-stage breast cancer. Using the Oncotype DX™ assay, a diagnostic test developed by Genomic Health, Inc., in collaboration with the National Surgical Adjuvant Breast and Bowel Project (NSABP), a network of cancer research professionals, TAILORx will attempt to determine the most appropriate therapy via this molecular test. If successful, patients with breast cancer will no longer receive all available systemic treatments when they are not necessary, but instead get the treatment that is best for them on an individual basis.

What is meant by over-treatment?

Adjuvant therapy for cancer is intended to reduce the risk of recurrence by eliminating malignant cells that may remain in the body after surgical removal of the primary tumor. The current guidelines recommend adjuvant chemotherapy for about 90 percent of women who fall into the category of node-negative, hormone-receptor positive early breast cancer. But we now know that about 70 percent of those patients will be alive and free of disease 15 years after treatment, with surgery alone--or lumpectomy followed by radiation. And if you add tamoxifen, a hormonal therapy, the outcome is even better. The problem has been that we have not been able to identify the 15-30 percent of patients who will recur, so we end up treating women who probably would not have a recurrence. That unnecessary exposure to the toxicities of chemotherapy is what we mean by “over-treatment”.

What is the downside of having chemotherapy?

One always has to weigh the benefits against the risks. There is a small rate of second cancers that are caused by chemo. If the risk of a second cancer is 1 or 2 percent, but the risk of recurrence is 30 percent, a patient still may choose to have the chemotherapy. There’s also some question about long-term effects on cognition caused by chemo; there can be cardiac problems; and with the addition of taxane drugs there can be nerve damage. Also, chemo often causes people to temporarily lose their hair and experience nausea and vomiting during treatment.

What is unique about Genomic Health’s process and why is it considered a step forward?

Most of the tests for expression of genes in a tumor were initially based on evaluation of frozen tumor material, so the ability to take formalin-fixed tissue (the form of tissue most commonly available), where the nucleic acid is damaged, and still get reproducible results, was a real step forward. Genomic Health’s approach is now being picked up by many of the companies who are involved in developing diagnostic tests.

What are some of the problems faced when moving new diagnostic tools into clinical use?

There are many steps along the way, and a lot of things that need to be assessed. There is the actual reproducibility of an assay--if you do it on Monday and you do it again on Tuesday, will you get the same result? There’s the standardization, reproducibility and robustness of the assay. But there’s also the harder question--does it give you clinically useful information? Tests can be reproducible but not help the patient and physician choose the best treatment options. An example is measurement of some serum markers such as carcinoembryonic antigen (CEA) to monitor for disease recurrence. The test is reproducible but does not provide specific enough information about whether there is disease recurrence to guide treatment – it can be positive for reasons other that cancer or negative if there is very little cancer present.

What is the difference between tests that are prognostic vs. those that are predictive?

Prognostic means the course of the disease in the absence of treatment. It tells you the likelihood of a patient’s disease being aggressive or non-aggressive. So then a doctor decides, based on that, I’m going to intervene, I’m going to treat the patient in a particular way. A predictive test helps you define what treatments are most likely to be effective.

For example, the initial observation of HER2 overexpression was that it seemed to be associated with poor prognosis--with worse outcomes for patients. Ultimately, the HER2 test was not a very strong prognostic indicator, particularly in node-negative breast cancer. Nodal status turned out to be a stronger indicator but HER2 was able to tell you within node-positive disease which patients were more likely to benefit from additional treatment. So, when we then wanted to use HER2 to predict the response to Herceptin, the characteristics of the assay had to be changed. It was difficult to take the test that had been used initially, and scale it up to a test that could be used broadly in many hands. One of the challenges in developing diagnostics is that we often begin analyzing the diagnostic in a context that’s different from the way we want to use it in the end.

How are these tests developed?

Developing these tests is an iterative process. You start out perhaps focusing on your standardization, and you have a proposed clinical use, and you measure the ‘marker’ in that context. And then, if the assay performs well, that is, reproducibly, but you want to change the clinical use, you have to go back and adjust the assay to make sure it can inform the decision you want to make. You always have to retest. You have to go through these loops a number of times until you finally go to a prospective trial.

So that’s what TAILORx is going to do?

That’s what TAILORx is going to do. The OncotypeDX™ was evaluated in an appropriately designed prospective/retrospective trial, in that a set of samples that were taken in a prospective way in the context of a clinical treatment trial but analyzed retrospectively, at the end of the trial. This way they did not have to start from scratch, because having the outcomes on those samples allowed the test to be evaluated without waiting many years for the trial to mature. That’s why we have the clinical trials groups collecting specimens on their trials so that we can then go back and ask questions about diagnostics in the context of a full set of specimens. The reason that OncotypeDX™ is as powerful as it appears to be is that the developers had access to specimens from clinical trials where the patients had been treated uniformly, and where there was good follow-up. So they were able to take these specimens and look at the relationship between the score on the test and the outcome. The OncotypeDX™ was clearly demonstrated to provide a reproducible estimate of the risk of recurrence. TAILORx will allow us to determine whether OncotypeDX™ can also accurately predict which patients will benefit from the addition of chemotherapy to hormonal therapy. It was not possible to evaluate this question for women with recurrence scores in the intermediate range because there were not enough samples from the previous trials.

How much does the OncotypeDX™ cost compared to a mammogram?

The OncotypeDX™ costs a patient about $3,450, if there is no insurance reimbursement. A mammogram costs somewhere in the range of $100-200, but tests such as mammography, PSA screening and pap tests are screening tests. And if an abnormality is found in a screening test, that may lead to additional screening and biopsies, which all cost money.

It’s important to point out that there is a difference between a screening test and one that is performed once a diagnosis has been made. The OncotypeDX™ test has the potential for actually saving money for the healthcare system, because it may reduce the number of women who receive chemotherapy, which can cost $20,000 and more per patient.

What are some of the other treatment issues your program is working on?

We’re looking at a similar problem to the breast cancer and chemotherapy issue with early stage colon cancer, but in this case, patients typically do not get adjuvant chemotherapy after surgery, and some patients and physicians believe that they are being under-treated. So there is a need for a test that identifies those patients who will benefit from the addition of chemotherapy.

What is NCI doing to help bring new diagnostic tools into clinical practice?

We meet with companies to try to focus their efforts, and if they have a potentially useful technology, we help them determine what the clinical questions are, and where the technology might be of greatest use. We begin by focusing on the clinical problem, rather than on the technology or the assay, which is a change in the way assay development has been done in the past, where mostly people develop a test based on some interesting biology, and then they look for a place to use it. Our program is saying, let’s home in on a clinical problem and figure out what technologies or markers are available.

###