Battle of the Breast Tumor Assays

The genetic tests clinicians use to diagnose and stratify breast cancer are getting increasingly complex and studies show that there may be some discordance between generations when assessing a woman’s risk of cancer recurrence. Molecular subtyping may be the only way to gain clarity.

A team of researchers from multiple institutions, including the University of South Florida and Florida Hospital in Tampa and Morton Plant Hospital in Clearwater, Fla.; from Agendia, makers of the MammaPrint and BluePrint genetic assays; and from Amsterdam, The Netherlands, compared multiple assays using the same samples and found curious results. Co-author Peter Blumencranz, MD, medical director of breast health services and a breast surgeon at Morton Plant Hospital, discussed the discord in an exclusive interview with Molecular Imaging Insight. 

Genetic tests for breast cancer

Rather than focusing on genetic tests that tell women about their chance of developing breast cancer, the tests in question peer into the genetic signature of a tumor itself in an effort to gauge risk of progression and metastasis. To be clear, looking at genetic signatures has been a major game changer for treatment selection and predicting cancer recurrence. The information these assays provide goes far beyond structure of tumors into the biology at work, which is a lot more important.

“We used to think going back many years ago that cancer was very much a logical progression: The bigger it grew, the more likely that it would go to the nodes, and what went to the nodes would go to the bloodstream and so on,” Blumencranz says. “As it turns out, that kind of thinking was not good science. We know that now.”

Most staging for breast cancer is entirely anatomic, but that says little about how biologically aggressive even the smallest tumors can be. The biology of the tumor trumps the anatomy, he adds. 

This is where personalized medicine comes in. It used to be that chemotherapy trials would have large groups of women and each group would represent the chemotherapy they received. The end results of these trials would say something like group A showed 20 percent cancer recurrence and group B showed 40 percent. And these simplified analyses were used as the basis of standards of care. The problem is that these categorizations say nothing about an individual woman’s risk of recurrence. That’s when genetic assays started to fill in the blanks.

Pick one, any one

One of the first molecular assays, called the Oncotype DX Recurrence Score, looked at 21 genes in breast cancer and provided a recurrence score. Another assay, Mammostrat, based on a five-antibody immunohistochemistry test, gives clinicians a risk index for women with human epidermal growth factor receptor 2 (HER2)-negative and estrogen receptor (ER)-positive cancer. One more has come along, Mamma-Print, but this one is a 70-gene microassay analysis that provides a recurrence score from low to high, with a high score meaning the patient would definitely benefit from chemotherapy and for a low score, perhaps not. The problem is that multiple comparison studies showed that some patients, when tested with multiple assays, came back with widely contradicting risk scores. One study in particular showed discordance in up to 30 percent of cases (Proceedings of the 2013 San Antonio Breast Cancer Symposium). 

“Among certain samples that might be low risk in one test came back high risk in another,” Blumencranz notes. “That doesn’t help us very much, because how do we know which is right?”

This comparison study by Steven C. Shivers, PhD, and others from the University of South Florida, presented during the 2014 Miami Breast Cancer Conference in March, provided very clearly conflicting results. For those who were deemed high risk by the MammaPrint assay, almost 33 percent were classified as low risk by the Oncotype DX assay. And for patients classified as low risk by the MammaPrint assay, 5.6 percent were classified as high risk by Oncotype DX. The researchers concluded that MammaPrint and Oncotype DX assays revealed a 19 percent rate of “major discordance.” Similarly, for Oncotype DX and Mammostrat, the discordance rate was 12 percent. And between MammaPrint and Mammostrat assays, there was a 26 percent rate of major discordance. A range of Oncotype DX recurrence scores were seen for every MammaPrint risk group. The researchers suggest the blame could be put on differing assay technology, baseline population characteristics and/or tumor biology. 

Subtyping to the rescue

Researchers in this study brought in a different kind of test, the 80-gene BluePrint assay, which also happens to be marketed by Agendia. This one does not look for genetic signatures for recurrence, but instead tumor subtypes—Luminal A and B, HER-2 and Basal breast tumors. When combining MammaPrint and BluePrint, the researchers were able to more appropriately stratify the patients’ risk.

According to this study, which included 148 patients with ER-positive cancer with or without lymph node involvement, molecular subtyping was linked to nuclear grade, mitotic activity and overall histologic grade. Most major discordances were categorized as luminal B by BluePrint and every case categorized as basal type or HER2-type by this test was determined to be moderate to high-risk by all three assays. Still, researchers do not yet know all the downstream outcomes or have major follow-up studies to go on.

“We don’t have a lot of good long-term outcomes yet and that is going to be very important,” stresses Blumencranz. 

Generally, though, patients who have a luminal A subtype and a low risk signature probably do not need chemotherapy, he says. “More patients get chemo than need it.”

Some women with stage II breast cancer, namely luminal A-type, without node involvement, can proceed with just oral endocrine medication, according to results of the RASTER trial, Blumencranz says. Most of the time people with Luminal A-type will respond well to hormone blocking drugs, but that is not always the case with luminal B. Some of these patients will gain additional benefit from chemotherapy.

In the future, the MIND-ACT trial will be providing information from 6,000 patients and the MINT study could provide more data about the benefits of chemo for improved predictive models.

“This is a moving target and more tests will come out—this is something that they can keep refining,” Blumencranz says. “It’s not for the faint of heart. You have to really study this stuff.”