Gene Expression Assays for Breast Cancer

Joseph Sparano, MD

The Cancer Network recently spoke with Dr. Joseph Sparano, Professor of Medicine and Women's Health at the Albert Einstein School of Medicine and Associate Chairman of the Department of Oncology at Montefiore Medical Center in New York, about the session he will chair at the American Society of Clinical Oncology (ASCO) Breast Symposium to be held September 8-10 in San Francisco.

-Interviewed byRachel Warren

CANCERNETWORK: We'd like to start by asking if you'd give a brief overview of what you hope to cover at the ASCO Breast Symposium.

SPARANO: I will review some of the multi-paramater, or gene expression assays that are currently available and approved for use in clinical decision-making in breast cancer. The two that are the most widely used both inside and outside the United States are the Oncotype DX assay and the MammaPrint assay; both of these are most widely used for women who have ER-positive, HER2-negative breast cancer associated with negative axillary lymph nodes, and are used to help make decisions about potential benefits associated with recommending adjuvant chemotherapy.

CANCERNETWORK: Are there published data on the specificity of these tests for specific subsets beyond ER+ breast cancer patients?

SPARANO: Currently there is very little information about any multi-parameter assays for any breast cancer subtype other than ER+ breast cancer. That certainly is a subject of ongoing research, but right now these tools seem to be restricted to that subtype, and that’s the subtype where we often recommend chemotherapy even in patients who are at low or moderate risk-and these patients do derive some benefit, though the absolute benefit is small. These tools can help us make decisions about who really will benefit from adjuvant chemotherapy.

CANCERNETWORK: We know that these tests often put patients into a kind of “gray zone” where the path of treatment is not all that clear. Do you have a sense of how often that happens?

SPARANO: It isn't all that uncommon. In the Oncotype DX assay, the result is reported as a score that ranges from 0 to 100, with a higher score associated with higher risk of recurrence and a greater benefit from chemotherapy. It's also reported in categories-as low, intermediate, and high-risk. Anywhere from 25% to 50% of patients can fall into the intermediate risk category. It's pretty clear that women whose tumors are associated with a low recurrence score have a low recurrence rate and are very unlikely to benefit from chemotherapy, and those who have a high recurrence score derive great benefit from it, but it's uncertain what the recommendation should be for a woman who has an intermediate or mid-range recurrence score, and that was what the TAILORx trial was intended to address. TAILORx was a large trial in which women who had a mid-range score and who met standard clinical criteria for recommending adjuvant chemotherapy for their ER-positive, HER2-negative, lymph node–negative disease, were randomized to chemo vs no chemo. The results of that study should be out in several years.

The other assay, the MammaPrint assay, reports either a good or a bad signature. So it doesn’t have this issue, where there’s a mid-range or intermediate score.

CANCERNETWORK: So the TAILORx trial results should help clinicians make decisions for how to treat these “gray area” patients.

SPARANO: Absolutely, the TAILOR X trial will clarify what the recommendation should be for women who have a mid-range recurrence score. With regard to the MammaPrint assay, there is a large trial that is almost complete called the MINDACT trial, which was designed to address a similar question. All of the patients who are enrolled in this trial undergo both an evaluation by standard clinical criteria, and by the MammaPrint-the multi-gene expression assay.  Those patients whose gene expression and clinical features are in the poorest category are assigned to chemotherapy. Those who fall into the low-risk category are assigned to endocrine therapy. But the one-third of patients who have discordance between their clinical and genomic features are randomized to treatment by clinical features vs. genomic features. This should address a complimentary question about what the better way is to advise patients-is it based on their clinical features or is it based on their molecular features?-at least as it relates to this assay.

CANCERNETWORK:  They say that when there's this discordance between what the genomic features are and what the clinical features are, the genomic data will prove to be more accurate. But how risky is it to base predictions on genomic tests?

SPARANO:  We don't really have a lot of information to answer that question-but we do have two prospective validation studies that have been done with the Oncotype assay. The first was done with the population of patients who enrolled in the NSABP (National Surgical Adjuvant Breast and Bowel Project) B-20 trial-these are patients who have ER-positive, lymph node–negative breast cancer; they were randomized to tamoxifen vs tamoxifen plus CMF (cyclophosphamide, methotrexate, and fluorouracil 5-FU) chemotherapy. There was a benefit from adding CMF: the risk of recurrence was reduced by about a third, and that translated into about a 3% to 5% absolute benefit. When the Oncotype assay was applied to this population, it was found that only patients who had a high recurrence score clearly benefited from chemo, and rather than having a 3% to 5% absolute benefit, they had about a 20% to 25% absolute benefit.

A similar analysis was done using samples from the SWOG (Southwest Oncology Group) 8814 trial, which included post-menopausal women who had ER-positive, lymph node–positive breast cancer. In this trial patients were randomized to tamoxifen vs tamoxifen plus CAF (cyclophosphamide, doxorubicin, and fluorouracil 5-FU) chemotherapy. When the Oncotype was applied to this population, an analysis showed very similar findings: only patients who had a high recurrence score seemed to benefit from chemotherapy. Those two studies really formed the basis for the TAILORx trial.

CANCERNETWORK:  Would you discuss how you think the FDA approval process for diagnostic tests differs from the approval process for new drugs, and how you think that will affect the roll-out of these tests?

SPARANO: The ONCOTYPE assay was initially approved under the CLIO Act, which is the Clinical Laboratory Improvement Act. This is a set of statutes that regulates the approval of diagnostic tests. Several years ago the FDA issued draft guidance indicating their intention to regulate the approval of these multi-parameter assays under their existing laws that relate to the regulation of medical devices, and the MammaPrint has been approved using that regulatory process. In terms of how medical devices and drugs are approved, it’s a very different process;  the FDA’s views on how to regulate these so-called devices is evolving. They’ve clearly recognized that this is an important area because these assays influence clinical decision-making. The FDA is now also beginning to regulate ‘companion diagnostics,’ or drugs that work only in specific populations of patients, and this is particularly important as we develop new molecular therapies that only work in clearly defined subsets of patients. It will be increasingly important in the development of these drugs that we identify the population of patients that they will work in. It is also important to apply the same rigor to developing the assay that you use in the development of the drug itself.

CANCERNETWORK:  We know that Ireland's National Centre for Pharmacoeconomics declined to cover the Oncotype DX assay for use by patients because of its high cost. Are there issues of cost, and thus issues of insurance and Medicare coverage-that you think will affect the adoption of these tests in the United States?

SPARANO: Well, cost of course is always an issue. In the US, Blue Cross/Blue Shield issued guidance several years ago, stating their approval of the use of Oncotype DX in certain situations because they recognize that this can be cost-saving. The majority of patients who are recommended to receive chemotherapy by clinical criteria are actually not benefiting from chemotherapy. And there have been several pharmacoeconomic evaluations done in various countries. In general, the pharmacoeconomic recommendation is in favor of testing, so I think this is an evolving story. In general the pharmacoeconomic recommendation is in favor of testing, so I think this is an evolving story. Different countries and different regulatory bodies may choose to interpret things differently, and this kind of thing can also change as the cost of certain drugs increases and decreases.