Skip to main content
Premium Trial:

Request an Annual Quote

Agendia Study Suggests Subset of ER+ Breast Cancer Patients May Be Mistreated without Molecular Subtyping


Results of a study by researchers from Agendia and The Netherlands Cancer Institute have identified a subgroup of approximately two percent of breast cancer patients who test ER-positive, but also have a basal-like molecular subtype, rather than the luminal subtype normally associated with ER+ tumors.

The study — published last month in the journal Breast Cancer Research and Treatment — showed that this discrepancy may be due to a splice variant that inhibits normal ER function. Though the clinical implications of the presence of this variant have not been studied, the authors wrote that the finding suggests molecular subtypying may be necessary to identify these women, lest they miss out on potentially effective chemotherapy treatment, since basal-like tumors respond well to chemotherapy.

"Normally, standard of care for a woman who is ER-positive by immunohistochemistry, is to give endocrine therapy," Doug Bradley, Agendia's executive vice president of global marketing told PGx Reporter.

"What this study tells us is that there may be a subclass of women out there who, although they are ER-positive, the pathway may be inactive [and] there is no way to know that unless you do a molecular subtyping test," he said. Agendia markets a breast cancer molecular subtyping test, called BluePrint.

Based on what researchers know about the basal subtype, this subset of women should be treated with chemotherapy, in Bradley's view. "It's critical … if they really are a basal-type, because that tumor type has an excellent response to chemotherapy,” he said.

The study, led by Rene Bernards, Agendia's chief scientific officer and a researcher at The Netherlands Cancer Institute, began with a single case tested with Agendia's Symphony suite of tests that includes the MammaPrint, TargetPrint, and BluePrint assays.

The patient was deemed ER-positive by TargetPrint, the company's quantitative gene expression test for ER expression. However, when tested on BluePrint, the results showed the patient to have a basal subtype, a group populated normally by ER-negative, chemotherapy-sensitive cancers.

In addition, when this patient was tested by the company's MammaPrint breast cancer recurrence test, she was found to be at high risk of recurrence.

This seeming discordance between the various tests led the patient's doctor Stephanie Akbari — also an author on the paper — to consult with Agendia.

Bernards told PGx Reporter that after Akbari approached him, the company proposed to try to identify a biologic cause for her unusual test results.

"She called me up as chief scientific officer and asked if I could explain these results and I said, as a scientist, it can only mean one thing— that the estrogen receptor is present but not functional. This would make sense since BluePrint [and MammaPrint] don't look at the estrogen receptor itself, but at downstream genes that are activated by it,” Bernards said. "We then thought we should ask for permission and look into detail at what was happening with this patient."

In the meantime, Bernards said, Akbari had also ordered Genomic Health's risk-stratification assay Oncotype DX. Unlike MammaPrint, Oncotype DX rated the patient low risk for distant recurrence.

In one study conducted by Genomic Health and L’Institut du Sein at the Paris Breast Center and presented at the San Antonio Breast Cancer Symposium last year, MammaPrint and Oncotype DX yielded discordant results in a number of patients. Of 67 samples from patients with ER-positive, HER2-negative early-stage breast cancer, Oncotype DX found only two samples to be at high risk of recurrence, while MammaPrint determined 22 patient samples to be high risk. However, in this head-to-head comparison, study participants weren't followed for outcomes to establish which test was giving the correct risk assessment (PGx Reporter 12/19/2012).

According to Bernards, the Agendia team's analysis of Akbari's patient's expressed ER mRNA revealed a sequence that corresponds to a previously-reported splice variant called ERΔ7, which, according to in vitro biological evidence, inhibits the function of normal (wild-type) ER.

"Our analysis showed that the patient expressed a variant of the estrogen receptor that is clearly defective, but still detectable, which explains why pathology called it ER-positive and it also explains the basal-type result and the difference between the MammaPrint and Oncotype DX [risk scores]," Bernards said.

The Agendia team then set out to try to establish the frequency of ER-positive, yet basal-type women, potentially carrying this ERΔ7 variant, in the larger population that has been studied by the company.

Looking retrospectively at 3,527 cases, the team identified 2,658 ER-positive, HER2-negative tumors for which BluePrint subtype data were also available.

Approximately two percent of these ER-positive patients were basal-type, according to BluePrint, and in 11 of these 55 cases, the group also measured total ER expression and expression levels of the ERΔ7 mutation. Overall, expression of the variant was no different in these basal tumors than it was in a comparison group of luminal tumors, the authors reported. However, expression levels of ERΔ7 relative to overall ER were significantly higher in the basal group.

Agendia's Bradley believes that the main implication of the research is that this two percent of women who test ER-positive, but also fall under the basal subgroup, cannot be identified without molecular subtyping.

Bradley said Agendia's BluePrint is the only commercially-available subtyping assay. However, other subtyping tests have been developed, such as the PAM50 gene signature: the backbone of another breast cancer recurrence risk test — NanoString's Prosigna — which has been available since last year in Europe under a CE-Mark and was granted 510(k) clearance by the US Food and Drug Administration this week (see related story, in this issue).

Bernards reiterated that if a patient is ER-positive by IHC, a physician must offer hormonal therapy according to the standard of care. But, he said, "moving forward, we are planning to track this and advise any physicians that get a reading like this out of the Symphony [suite], that these patients should be given chemotherapy in addition to hormonal therapy."

Akbari took this treatment route for her patient, the first analyzed in the group's study, he added.

Additionally, the study authors also explored how the results might explain discordance between MammaPrint and Genomic Health's Oncotype DX risk classifications.

Bernards explained that while Oncotype DX uses estrogen receptor mRNA levels as part of its recurrence score calculation, MammaPrint doesn’t use the estrogen receptor, but rather relies on downstream targets.

"Since [these tumors are] highly ER-positive at the RNA level, Oncotype would say it's an ER-positive and [therefore] a low-risk tumor … But in this case [of the first ERΔ7 patient] none of the ER target genes were expressed, so MammaPrint classified it as a high-risk," Bernards explained.

"Our test was developed by comparing women who had recurrence in five years and those who didn't," Agendia's Bradley added. "Looking at differentially expressed genes we rank ordered all the genes from one through 5,000 and identified the top 70 that were the most prognostic."

"It was the genes downstream of the [estrogen receptor] that were much more prognostic. The actual ER gene was actually well under 1,000 in terms of its ranking," he said.

According to Bernards, an ERΔ7 tumor behaves like an ER-negative tumor for all intents and purposes. As a result, “all of these women who have this version of the estrogen receptor and get Oncotype … they may be incorrectly assigned to a low-risk category when they are de facto high-risk,” Bernards asserted.

Of course, any clinical implications — whether ER-positive women with a basal subtype will respond well to chemotherapy, as well as whether these women are likely to do poorly on hormonal therapy alone — have not been demonstrated scientifically, Bernards admitted.

However, he also pointed out that as a matter of biology, one cannot accomplish much by trying to inhibit a receptor that is not functional to begin with. “The fact that none of the targets of this estrogen receptor are expressed in these tumors means the receptor is not active – so inhibiting [it] would have no purpose, scientifically,” Bernards said.

Steve Shak, Genomic Health's executive VP of research and development, disagreed with this conclusion.

"We see every day technologies looking at mutations in receptors and splice variants and translocation, [and] hundreds if not thousands of different variants identified, and we need to ask – are they clinically important and should we think about them in the standard of care?" Shak told PGx Reporter this week.

"We are a long ways away from knowing that this particular variant is actually clinically or even biologically relevant," he argued. Citing the fact that the ER-positive, basal-like tumors in Agendia's study showed a high proportion of wild-type ER expression relative to ERΔ7 expression, Shack said it's possible patients with this tumor profile might actually respond well to hormonal therapy. He further criticized the evidence for the activity of ERΔ7 as an inhibitor of wild-type ER, which he argued is based on limited research conducted in non-breast cancer cell lines.

"There are hundreds of articles written every day about what might happen with a mutation. We need to see the evidence in order to see if that's actually true," Shak said.

Additionally, Shak said, the activity of this splice variant, if borne out in clinical evidence would only explain discordance between Oncotype DX and MammaPrint in two percent of ER+ patients, while the French cohort study presented at SABCS last year showed a much larger proportion of patients to have discordant risk results between the two tests.

"This [ERΔ7] variant occurs only in two percent of patients, so there is no way it can explain [this larger discrepancy]," Shak said. "This [study] doesn't address at all … the concern that [these other] patients may be mistreated based on the Agendia assay.”

Indeed, demonstration of discordance between the two tests is not direct evidence that either company's result is correct or incorrect, or which test result would inform more effective treatment.

In the recent Agendia study, the authors wrote that the previously studied biological function of ERΔ7 provides preliminary evidence that, for this small subgroup, molecular subtyping more accurately measures the actual biological nature of a patient's tumor, and that MammaPrint may offer a more accurate risk-assessment than Oncotype DX.

Clinical evidence either supporting or contradicting this conclusion has not been collected.