Putting the Genome to Work in Breast Cancer

OncologyLive, September 2012, Volume 13, Issue 9

An interview with Elaine R. Mardis, PhD, discussing the clinical applications of genomic research on the near horizon that will enable more personalized therapy for patients at earlier stages of breast cancer.

Elaine R. Mardis, PhD, discusses genomic insights into resistance mechanisms in endocrine therapy.

Although breast cancer is among the best-characterized tumor types, continuing efforts to refine the sequencing of the human genome are opening the door to a new treatment paradigm in which oncologists will have swift access to a wealth of information that will enable more personalized therapy for patients at earlier stages of disease.

It is a vision of the future that Elaine R. Mardis, PhD, a speaker at the 11th International Congress on the Future of Breast Cancer, not only anticipates but also helps to create.

Mardis is co-director of the Genome Institute, director of Technology Development, and professor of Genetics and Molecular Microbiology at Washington University School of Medicine in St. Louis, Missouri. The institute is among three large-scale sequencing centers that the National Institutes of Health has enlisted nationwide to join in developing The Cancer Genome Atlas.

One of the institute’s endeavors is to help build a database of human reference genome sequences, a more complete library of information than the initial Human Genome Project generated, which can be used to develop targeted therapies and as a comparison point for an individual patient’s genomic signature.

While that work continues, new sequencing technologies now in development are enabling researchers “to anticipate assembling that person’s genome as opposed to aligning it back to the human reference,” thus expanding their ability to understand what is happening with the patient’s chromosomes, said Mardis.

“It’s a brave new world out there right now, and I’m personally very excited about it because I love the technology aspects—it’s really powerful to see what the technology is going to be able to do,” she said.

In this interview, Mardis discussed clinical applications of genomic research on the near horizon.

OncologyLive: Your recent paper in Nature, “Whole-Genome Analysis Informs Breast Cancer Response to Aromatase Inhibition” [2012;486(7403):353-360],is one of the first genomic studies to demonstrate how this research can be applied in the clinical setting. You found 18 significantly mutated genes in the tumor samples. What is the clinical significance of this research? Mardis: I was honored to recently have a publication in Nature with a large number of collaborators that studied sequencing samples from a clinical trial of aromatase inhibitors. Our overarching goal in this study was to begin to characterize the genomes of patients who were known, by virtue of the clinical trial parameters, to respond to the aromatase inhibitors compared with those who did not respond.

In essence, what we wanted to identify in these data were signatures from the genome that will help us predict which patients will or will not respond to aromatase inhibitors so that we can be more intelligent about our choices in deciding which patients to assign to this therapy versus other types of therapy (eg, surgery, chemoradiation, or other new targeted therapies).

We performed whole-genome sequencing of patients who were about equally stratified between those who did and did not respond based on a fundamental immunohistochemistry marker of Ki67, which provides a metric of tumor cell proliferation.

Some of the outcomes of this study, which focused on estrogen receptor-positive tumors, included new insights into genes that were not previously known to be commonly mutated in luminal-type cancers. In particular, we identified five new genes historically thought to be only involved in hematopoietic malignancies that were clearly highly mutated, and were contributing to the disease onset or were “drivers” of the disease. This was an important finding in terms of what we typically refer to as significantly mutated genes. These significantly mutated genes are genes that, by a variety of mathematical tests, are more commonly mutated than would be typically expected.

We also found that patients with TP53 mutation, high proliferation by Ki67, and luminal B-type tumors were the least likely patients to respond to aromatase inhibitors. By comparison, a new gene that was also identified in our study, MAP3K1, is commonly mutated in patients with luminal A-type tumors and a lower Ki67 proliferation index, and these were the patients who most commonly responded to the aromatase inhibitor therapy.

So, in essence, what we have from this study is the very beginning of a genomic signature that’s combined with conventional pathology markers, such as Ki67 and RNA subtyping, to identify the specific intrinsic subtype of the breast tumor, and to identify patients who are likely to respond before they’re assigned to aromatase inhibitor therapy.

Please discuss the so-called “druggable targets” you found through sequencing these tumors, and the implications in identifying agents that are FDA-approved for other indications.

We did an interesting exercise in the Nature paper that showed that for the mutational spectrum for the breast cancers we sequenced, there are many other drugs (typically small-molecule inhibitors) to which these patients were likely to respond. This raises the specter of targeted therapy in the context of breast cancer care, which is a very important consideration to have on the table.

But what was also offered by this exercise was the fact that many of the drugs that were suggested by these patients’ genomes are not FDA-approved for a breast cancer indication. This suggests that there will be fundamental changes needed in the context of how the FDA uses clinical trial information to approve drugs for therapeutic uses, and suggests that, in many cases, off-label use will become the norm until this testing process can be rectified by the FDA.

This, of course, raises huge issues just in terms of the high costs associated with these therapies. If they’re not FDA-approved or on-label for that patient’s use, then the patient is in a situation where her insurance company typically will not pay for these agents. So, patients have two options: (1) pay for the drugs themselves, which is pretty unlikely for most patients, or (2) appeal to the pharmaceutical company for the off-label use in compassionate care.

This is a huge area that needs to be addressed at the regulatory level, at the clinical trials level, and at many other levels. I can assure you that it’s not just specific to breast cancer. We’re seeing this across all cancer types, in that cancer is a disease of the genome. When the genome is altered, it has relatively little to do with the tissue type in which the cancer is occurring, but has much more to do with the genes that are actually driving the onset and progression of that cancer.

This is the fundamental truth that’s coming out of all the genomic studies and is exemplified by our finding in breast cancer: Genes that were thought to be essentially restricted to the hematopoietic malignancies are actually very prominent players in the development of estrogen receptor-positive breast cancer.

How can community oncologists apply genomic information in their practice today?

Breast cancer is one of the predominant tumor types, and so from a discovery standpoint, we’re fortunate in that regard. But we do have a very large amount of information now to coalesce and distill down for the oncologist who is actively seeing patients every day in clinic. In particular, what I think genomics can contribute to that daily practice regimen is a more specific description of the disease from a molecular standpoint. Although breast cancer is already one of the best- characterized tumor types, genomics will now add some richness to that data and distill out much more information that can be used on a day-to-day basis.

I think the role of the community oncologist in treating patients with breast cancer becomes more well-informed. Obviously, it’s a lot of information to take in, and we’re going to have to develop systems that deliver that information in a much more cogent and hands-on, facile way. At the end of the day, the idea is to begin to treat the disease as early as possible with these new methodologies and this new information, so that we begin to alleviate the transition from primary to metastatic disease.

In that regard, by treating primary tumors up front with the best mechanisms possible, whether it be with combinations of chemotherapy, targeted or small-molecule therapies, or immunotherapies, we can begin to understand how to decrease the number of patients who are progressing to metastatic disease.

How do you see the roles of the oncologist, geneticist, and other healthcare professionals changing in light of these genomic advances?

Genomics has the potential to change the spectrum of care from the standpoint of those who provide care on a daily basis. There will be an onus on the community oncologist and the oncology nurse to provide genomic consent to patients who decide to pursue genomic testing because these patients will have either their full genome or a portion of their genome sequenced, and they’re going to need to understand what that return of information might reveal about their breast cancer, and possibly other aspects of their genomic makeup. This will also elicit, at a much higher level than ever before, the need for genetic counselors to be involved to deliver that information. It will require a very concerted effort of all healthcare providers across the spectrum of care to help breast cancer, patients understand the information, and more importantly, to help patients understand how the information is going to help them be more successful in their fight against the disease.

How will genome sequencing affect the way oncologists stage tumors and evaluate patients for therapy?

Genomic information is very interesting in the context of traditional pathology. For example, by looking at RNA gene expression signatures we can already subtype, but we can also confirm what immunohistochemistry from conventional pathology tells us. Namely, is a patient estrogen receptor-positive or -negative, progesterone receptor-positive or -negative? Is that patient HER2- positive or HER2-negative? So, in essence, the genomics recapitulates the pathology.

This is not to suggest that conventional pathology will fall by the wayside, but what genomics brings to the table is additional granularity or resolution to that patient’s mutational spectrum. Genomics identifies targeted therapies or immunotherapies that may be helpful to that patient above and beyond what we can predict from traditional pathology right now.

What we know from sequencing breast cancer is that it’s a complicated disease, and that inherently, based on the complications in the genomes of these patients, we will need to have a more specific regimen for each patient to ensure success in that patient’s disease. While genomics is complicated, we need to embrace the information and try to understand what its base meaning is so that the oncologist will be more successful in treating that patient moving forward.

How can oncologists determine the right sequencing technology to use for evaluating their patients?

There are a variety of next-generation sequencing instruments available from different manufacturers. Most of them produce short-reads that need to be aligned back to the human genome to make sense of them. Most of these instruments are relatively easy to operate, but the bioinformatic analysis of the data is what’s complex. Once you have the data in hand, interpreting it—both from the standpoint of where the mutations are, and then from an aspect that I call “medical interpretation,” that is, deciding which drugs are indicated by that patient’s genome—is a highly complex operation.

In terms of how this can be applied in practice, there are limited gene panels being used for patient diagnosis at some of the larger teaching hospitals. In limiting the scope of the gene panel, you have an easier time in the analysis because you’re looking at a small number of genes in the bioinformatics, which is not as daunting as a whole-genome approach. That said, we know from sequencing whole genomes of many cancer patients that these targeted panels will often miss some information that would be germane to the patient’s therapy. So there’s a delicate balance here, but what will likely get implemented into the clinic first are the targeted panels. And that’s fine because they’re going to help a large number of patients, just not all.

Then what will emerge over time, as the bioinformatics becomes more refined and more plug-and-play, if you will, will be whole-genome tests that will be much more comprehensive and will most likely tie into information from RNA, which we already know from a discovery or a basic science standpoint is important to cancer and cancer care. At the same time, from a technology standpoint, there are new instruments in development that will sequence an entire genome worth of data overnight.

Consider the practical implications of this: If the sequencing can happen overnight and the bioinformatics that I just described takes on the order of several days, you could be returning genomic information to the oncologist and to the patient in a timeframe that would impact the treatment paradigms for most patients. This would give the oncologists time to till through the information, really get a grasp on what’s being suggested, and come up with a very personalized plan for that patient in a three-week period between her initial diagnosis and when she comes back to clinic to discuss treatment plans.

Can you discuss some of the challenges and future implications you see on the horizon for genomics?

Whole-genome sequencing in particular brings on a number of challenges that are unique to whole-genome sequencing, and not so much to just specific panels of genes. In particular, there are a variety of ethical challenges. They include a decision on what information to return to the patient, and this takes on a very important form and flavor in the context of cancer of any type. For example, fundamental or “constitutional” mutations, as we call them (eg, inherited susceptibility to diseases like Alzheimer or Huntington), may not be so important in the context of a patient who’s fighting for their life against cancer, but may become important later if that fight is successful. As a result, there are inherent challenges in terms of the ethics of returning this information to the patient.

Should the return of information be constrained just to the information that is germane to that patient’s cancer care, or should all information across the full spectrum of the genome be returned to the patient? This is a decision that has to be included in the documents that are used in the patient’s consenting process, and then if there are “constitutional defects” that need to be revealed at some point, there has to be a decision about how to do that. And this is really where genetic counselors need to become involved.

We actually published a unique case that described a patient who was consented onto a research protocol for AML [acute myeloid leukemia] therapy [JAMA. 2011;305(15):1568-1576]. Unfortunately, the patient died during the research project, but she had living children. This presented a dilemma because we had identified in her genome a cancer susceptibility defect that had a possibility of being transitioned to her offspring. However, we were able to deal with it by consenting her living relatives, and thereby were able to bring them in for testing.

In returning information on the genome to the patient, there is also the concern about the ramifications of identifying or disclosing a constitutional defect in the genome to private insurers. Because people have children, this genomic data is also informative about the patient’s children’s potential for developing disease later in life. So, there are a multitude of ethical ramifications that don’t only apply to the individual, but extend to the individual’s family members as well.

Ultimately, there will need to be regulation and genome legislation that encompasses the potential for discrimination in the workplace or in insurance situations.

Table. "Druggable" Targets in Luminal Breast Cancers

Presumptive Oncogenic

"Driver" Mutations

Matching Drugs

FDA-Approved

or Planned Use

KIT

Imatinib, masitinib

CML, GIST

PDGFRA

Imatinib, masitinib

CML, GIST

DDR1,2

Imatinib, nilotinib, bafetinib

CML, GIST

MET

Foretinib, tivantinib, MetMAb

Hepatoma, gastric, pancreatic

JAK1

Ruxolitinib

Myeloproliferative disease, psoriasis

CSF1R

Imatinib, nilotinib, sunitinib

CML, GIST, renal

LTK

Crizotinib

NSCLC

BRAF

Vemurafenib

Melanoma

PIK3CA

BKM120, GDC-0941

Prostate, colorectal, glioblastoma, endometrial

AKT1,2

MK2206

Pediatric cancers

CML indicates chronic myeloid leukemia; GIST, gastrointestinal stromal tumor; NSCLC, non-small cell lung cancer.

Mardis ER. Targeting resistance to endocrine therapy—genomic insights. Presented at: 11th International Congress on the Future of Breast Cancer; July 26-28, 2012; Coronado, CA.