June 2015, Vol 4, No 3

← Back to Issue

Case Studies: Incorporating Molecular Biomarkers into Therapy for Breast Cancer Is Fraught with Difficulty

PMO Live

Many therapeutic agents target HER2-positive breast cancer. Unfortunately, patient selection for these agents has not been refined, so all HER2-positive patients receive them. In estrogen receptor (ER)-positive disease, new agents improve the response to hormone-targeted therapies, but again, the appropriate patients for targeted agents have not been defined. “Because many agents work for the treatment of breast cancer, it’s a little more difficult to identify the specific subsets for which these targeted agents will be most important,” said Hope S. Rugo, MD, speaking at the Third Annual PMO Live conference.

At this time, selecting agents based on genomic profiling in breast cancer is experimental. There are no data to suggest improvement in outcome by doing profiling before selecting a treatment in patients with breast cancer, said Rugo, Director, Breast Oncology and Clinical Trials Education, University of California, San Francisco (UCFS). “Most of the time, agents are off-label, so it has to be done in the setting of a clinical trial,” she said.

The promise of genomics in the clinic is to enhance prognosis and prediction and to discover new “actionable” targets. Genomic assays for use in breast cancer are the Recurrence Score and MammaPrint.

“One of the big issues that we face in breast cancer is heterogeneity,” she said. Recent data in breast cancer suggest that metastatic sites may have different markers than the primary tumor sites, “so there’s a big question about what we want to be following when we’re treating patients,” she said. “Almost all data to date that have used genetic markers to try to identify a subset of patients that respond to a specific targeted agent have used primarily archived tumor tissue from the primary tumor site. That’s one particular problem that we have.” Heterogeneity within the tumor itself poses the same problems.

Heterogeneity of assays also exists, and standardization of assays, interpretation, and reporting are needed, said Rugo.

Recent studies have identified 4 to 7 different subtypes of triple-negative breast cancer. “The key here is identifying the targets,” she said. The targets may span these subtypes, but clearly these are all great laboratories doing excellent work, and they’re coming up with different information because of different platforms. One of the questions is whether genomic assays aid in determination of treatment or confuse the picture. I think in breast cancer, specifically, that these tests need to be done in the setting of a clinical trial and not as a clinical service.”

Case: Invasive Ductal Cancer

A 43-year-old woman presented with a 1.2-cm, grade 3, invasive ductal cancer that was ER-positive in 60% of cells, progesterone receptor (PR)-positive in 5% to 10% of cells, with a HER2/chromosome 17 ratio of 1.1 (clearly not amplified).

Her oncologist sent a Recurrence Score, which was 31. The high Recurrence Score suggested the addition of chemotherapy to the patient’s hormone therapy. Her ER score was negative at 6.4, her PR score was negative at 5.2, and HER2 was equivocal at 10.8.

Before the patient was referred to us, a MammaPrint assay was ordered and came back showing high-risk, ER-negative, PR–very low positive, HER2-negative, luminal type disease.

Repeat immunohistochemistry (IHC) and fluoresence in situ hybridization (FISH) studies at UCSF showed ER/PR positivity by IHC but a clearly negative bioexpression analysis. The HER2 test was indeterminate. On FISH, the mean ErbB2 signals per nucleus was 6.0, which is positive by the new American Society of Clinical Oncology/College of American Pathologists guidelines. The ratio was clearly negative at 1.1.

The Recurrence Score was repeated and came back identical to the first (31), meaning high risk of recurrence with tamoxifen alone. The ER score was borderline negative, the PR score was negative, and the HER2 score was equivocal.


This case highlights some of the difficulties with the use of large expression panels to determine treatment. “Essentially, we didn’t know what the right answer was for this tumor,” said Rugo. The woman was treated with chemotherapy, trastuzumab, and hormone therapy and is doing very well a number of years later.

Case: Value of Sampling

A 78-year-old woman presented with a 2.3-cm, node-negative, pleomorphic and classic lobular cancer. It was ER/PR–strongly positive, grade 2. The Recurrence Score was 33, and this time, the assay came back as ER/PR negative. Pathology review found 2 clearly different areas within the tumor: one strongly positive for ER with classic lobular histology, and the other pleomorphic and ER negative.

This woman was treated with adjuvant chemother­apy (cyclophosphamide, methotrexate, and fluorouracil for 8 cycles) and has done very well. She continues to take hormone therapy and is about 4 years out from diagnosis.

Mutation-Matched Therapy


Finding a genetic mutation does not necessarily lead to improved outcome with a matched therapy. In a clinical trial randomizing patients to receive hormone therapy with or without an mTOR inhibitor (J Clin Oncol. 2013;31[suppl]. Abstract LBA509), there was no difference in progression-free survival (PFS) based on whether the patient had wild-type or a single mutation in either PI3-kinase or the fibroblast growth factor receptor or PTEN, which are the most frequently altered genes. Fifty percent of patients in the trial had alterations in PI3-kinase, but it didn’t predict PFS, said Rugo.

Prognosis is very poor with multiple mutations. “We don’t know what to do when patients have multiple mutations, because we don’t know which one is the most important in driving the poor outcome,” she said.

In the phase 3 CLEOPATRA biomarker analysis (Cancer Res. 2012;72[24 Suppl]. Abstract S5-1), PIK3CA mutation was associated with a poorer prognosis in both the placebo and pertuzumab arms. Mutations in PIK3CA were not associated with resistance to pertuzumab because patients derived similar additional benefits independent of PIK3CA mutational status. The same was true of other mutations that were evaluated in the trial. “Unfortunately, all the mutation analysis did was give us prognostic information,” said Rugo.

Case: Activating HER2 Mutations

A 58-year-old woman presented with a node-negative, ER-positive, HER2-positive invasive ductal cancer. She received endocrine therapy and radiation per current National Comprehensive Cancer Network guidelines. She then presented with inflammatory disease with extensive involvement of skin, breast, and regional nodes.

On biopsy, the tumor was ER/PR/HER2-negative, and this was repeated. She received multiple lines of therapy over a course of about 2 years with little evidence of response, but with progression of disease, extensive skin involvement, reduced range of motion, and fungal pneumonia.

Her tumor was sent for next-generation sequencing. Two activating
ErbB2 mutations were found that in preclinical data have been shown to be sensitive to both trastuzu­mab and lapatinib in lung cancer. In addition, the patient might qualify for entry into clinical trials of heat shock protein 90–targeted agents.

The woman received HER2-targeted therapy in a clinical trial, using the HER2 tyrosine kinase inhibitor neratinib in combination with chemotherapy. Positron emission tomography showed symptomatic and objective improvement. The improvement was quite durable, with a response duration of 8 months, and it was the only treatment to which her cancer responded.

Unfortunately, finding actionable targets is not true for most breast tumors, as the next case demonstrates.

Case: Driver Mutations in Breast Cancer: Few Have Matched Therapy

A 45-year-old physician presented with a node-positive, high-grade, triple-negative breast cancer treated with adjuvant chemotherapy. She had a recurrence at 12 months with a large mediastinal mass and compression/shift of major vessels and airways.

Pathology showed her to have a triple-negative breast cancer, high grade, with spindle cells. A tumor sample was sent to Caris Target Now, and a list of potentially active chemotherapy agents was generated. They included the drugs that she had already received in the adjuvant setting.


Through genomic sequencing, 6 genomic alterations were discovered (AKT2 amplification, PTEN, PIK3CA, CDKN2A/B loss, MYC amplification, and TP53), most of which have no matched drug at present.

“The role of next-generation sequencing remains unclear in breast cancer; it’s primarily a research tool,” said Rugo. “I don’t think we should be spending a lot of money in our breast cancer patients on it outside the setting of a clinical trial. We need the ability to look at panels of genes correlated to clinical phenotype and outcome.”

The most common driver mutations in breast cancer still only occur in less than 10% of patients. “We’re going to have to have trials where we’re identifying treatments for tiny numbers of patients,” she said. Because of the low frequency of each of the driver mutations in metastatic breast cancer, thousands of patients may need to be screened to find a single patient with a genomic alteration that is actionable.

“We’re going to have to do that as a group,” she said. One trial, SAFIR 01, accrued 423 patients and had 404 biopsy samples. Comparative genomic hybridization (CGH) arrays in 287 found targetable alterations in 194 (67%). “However, most of these were rare, and there were no drugs to target them,” she said. “They gave treatment driven by genomics to only 48 patients (25%) and 4 with CGH-identified ErbB2 amplification.” The objective response rate was 9%, and the rate of overall response plus stable disease at 16 weeks was 28%.

Adaptive trial designs in which patients are randomized based on their interim responses to treatments is one approach to overcome this hurdle. “We may be able to identify specific subtypes of breast cancer within our broader groups that respond to targeted agents,” said Rugo. “Those agents will go into phase 3 trials in the neoadjuvant setting.”

The AURORA trial is an international trial with a planned component of basket trials in which cancer patients are entered based on their mutational status. As part of the trial, patients have access to specific drugs not approved for disease, based on genomic profiling.

Uncategorized - June 15, 2015

BRAF Mutations: An Old Oncogene and a New Target in Non–Small Cell Lung Cancer

Lung carcinoma has been the leading cause of cancer deaths in the United States and worldwide despite advances in chemotherapy.1,2 Management of non–small cell lung cancer (NSCLC) has evolved significantly since 2004, when mutations in epidermal growth factor receptor (EGFR) were found to be the determining factor of response to [ Read More ]

Letter to Our Readers - June 15, 2015

The Increasingly Important Role of Pathology in Oncology Patient Care

Dear Colleague, We are pleased to offer this issue of Personalized Medicine in Oncology (PMO) to you, our reading community. Since the onset of the personalized medicine era, we have repeatedly heard about the importance of the multidisciplinary team to include physicians, nurses, pharmacists, pathologists, social workers, and patients. Of [ Read More ]