Using Genomic Profiling to Guide Treatment Decisions for ER-Positive Breast Cancers

Harold J. Burstein, MD, PhD

Ahead of the 2019 American Society of Clinical Oncology (ASCO) Annual Meeting, we are speaking with Harold J. Burstein, MD, PhD, of the Dana-Farber Cancer Institute and Harvard Medical School in Boston, about genomic profiling for estrogen receptor (ER)-positive breast cancer. Dr. Burstein will be speaking at an Education Session titled, “Genomic Profiling in ER-Positive Breast Cancer, From MINDACT to TAILORx,” on Monday, June 3, at the meeting, which is being held May 31–June 4 in Chicago.

-Interviewed by Bryant Furlow

Cancer Network: What prognostic genomic profiling panels are now approved to inform adjuvant therapy decision making for women with ER-positive breast cancer?

Dr. Burstein: There are several commercially available genomic assays that are used to help guide treatment decisions for women with ER-positive breast cancer. The most commonly used test is the Oncotype DX 21-gene recurrence score from Genomic Health. There are other assays as well, including the MammaPrint 70-gene assay and an assay from Nanostream called the intrinsic subtype assay. All of these are multigene assays built around patterns of gene expression within ER-positive breast cancers, and in very broad terms they help to classify tumors. For example, tumors could be scored as lower risk, which means they typically have a better prognosis and usually do not warrant chemotherapy, or as higher risk, meaning typically a less favorable prognosis that warrants chemotherapy in the early-stage setting.

Cancer Network: How many genes are represented in these different panels, and is there much overlap in the genes in the different panels?

Dr. Burstein: Interestingly, each one of the assays uses both different genes and a different number of genes. The total number ranges from 70 in the MammaPrint assay down to other tests, which are less widely used but use somewhere on the order of 5 to 10 genes. As I said, the most commonly used one is the 21-gene recurrence score. What’s interesting is that most of these assays rely on different genes that they analyze and yet they have tremendous overlap in terms of identifying tumors at either lower vs higher risk. The way I explain that to patients is that you can come up with a lot of decision rules to look at different features of something and classify it. So, for instance, if you were to say, “I want a decision rule to distinguish a horse from a cow,” you could say, “Well, if it has horns it’s a cow, if it has an utter it is a cow. And if it has a mane, it’s a horse, longer snout, longer bushier tail, it’s a horse.” The point is you can use all these different ways of analyzing different aspects of the same animal to come up with a very similar characterization and the same is true for classifying ER-positive breast cancer, you can use a lot of different genes to read out different aspects of the tumor that more or less broadly align with this lower, more favorable risk, which we often call luminal A, and the higher grade, higher score, greater risk, which we often call luminal B tumors.

Cancer Network: How has the availability of these genomic tools in these panels changed treatment decision making over, say, the past 10 or 20 years?

Dr. Burstein: Well, these genomic profile assays have really revolutionized our decision making for ER-positive breast cancer over the past 15 to 20 years. In fact, when we teach this to our fellows, you have to go back to a time long before any of them were actively involved in medicine to remind them of where we were at. Through the 1980s and 1990s, it was clear that chemotherapy reduced the risk of breast cancer recurrence and improved overall survival. It became the norm for women who had higher-risk node-positive breast cancers, and this reached the high watermark in 1997 when we saw data from an important trial, NSABP B-20, which took the lowest-risk group of women, women who had smaller node-negative ER-positive breast cancers, and assigned them to either hormone treatments alone or hormone treatments plus chemotherapy and showed that chemotherapy helped reduce the risk of cancer recurrence quite substantially. That was the point where it looked like everybody was going to get chemotherapy, and in fact the National Cancer Institute issued a broadcast alert that women who had tumors 1 cm or greater should all get chemotherapy. That was an important size because it’s hard to find a cancer that’s smaller than a centimeter; mammograms and physical exams have a detection of about a 1-cm threshold. So, this meant a huge number of women were getting chemotherapy.

In fact, if you go back to data from around 2000, something on the order of 75% or 80% of all women diagnosed with breast cancer were getting chemotherapy and it was done indiscriminately. At the time we all knew that this was overkill. But the problem was we didn’t have a tool that was adequately robust that allowed us to tease out which women did need chemotherapy and which women did not need chemotherapy. That logjam was broken by the development of the 21-gene recurrence score, which was the first genomic assay that gave clinicians a lot of confidence to say this kind of cancer doesn’t need chemo and this other kind of cancer really does need chemo. With that, the use of chemotherapy for ER-positive breast cancer has fallen by about half over the past decade.

Cancer Network: Can you tell us a little bit specifically about the TAILORx study and how it clarified optimal adjuvant therapy for women with intermediate-risk group results from the Oncotype test?

Dr. Burstein: Sure. The 21-gene recurrence score, the Oncotype assay, was first demonstrated to be of value in a retrospective analysis. They went back to that same NSABP B-20 trial that I mentioned and they showed that if the tumor had a really low score, that there was clearly no benefit for chemotherapy on top of hormonal manipulations. By contrast, if the tumor was shown to have a high score, then actually chemotherapy was very important, more important than we would have imagined otherwise. That really led the way into transforming our current practice. However, that left a big group in the middle where it was hard to know with more certainty whether or not chemotherapy might be of clinical benefit.

The TAILORx study was a prospective study supported by the National Cancer Institute, which took women who had ER-positive node-negative breast cancers and who had a tumor tested for the recurrent score. If the recurrent score was between 11 and 25, then the patient was randomly assigned to enter group therapy alone or group therapy plus chemotherapy. The idea here was to do a prospective validation of the score and also to see where the breakpoint was to see what number would correlate with beginning to see a signal of benefit for chemotherapy. The top-line results of that study were very clear, published in 2013 in the New England Journal of Medicine. What the investigator showed was that for women whose scores were 25 or less, there really was no benefit to chemotherapy, in addition to the hormone therapy. All the women get hormone manipulation treatments, but there was no benefit for chemotherapy in that cohort. That was a tremendously powerful validation of the approach that had become a widely used practice. I think that was the solidifier for all that had been transpiring over the past decade.

Cancer Network: Is genomic profiling going to one day replace traditional clinical and pathological risk factor–based prognostication, or do you think it will always be used side by side with other factors?

Dr. Burstein: That’s a really interesting question because traditional pathology actually in retrospect is really good at also figuring out which cancers are likely to need chemotherapy and which are not. Around the rest of the world, most societies, even wealthy developed societies, are not using genomic assays the way that oncologists in the United States are, because of the cost of the product. A routine pathology-examining the levels of ER and progesterone receptor, looking at the grade of the tumor, looking at the so-called proliferation indices, measures like Ki67-actually is really good in good hands at doing a very similar thing. There have been multiple studies around the world in which excellent pathologists at good centers can use traditional pathology and figure out which is a very low-risk group of patients and which is a higher-risk group of patients. One of the dilemmas has been that there is an observer variability on that. If you have a really great pathologist at your university hospital, unless that pathologist is seeing every single case, you have less confidence in the readout. One of the things that genomic tests have been able to do is to become a very reproducible, accessible, and robust tool substituting for that expert pathologist, if you will. The other thing we’ve learned, though, is that the genomic scores complement what the traditional pathology review shows. So, even for a given recurrent score number, let’s say a tumor that has a score of 19, tumors that are lower grade still do a little better than tumors that are higher grade, and tumors that are strongly ER-positive still do better than tumors that are not so strongly ER-positive, and so I think in the end these are very complementary approaches, which together help with optimal decision making.

Cancer Network: You mentioned cost. Has the cost of the profiles of the genomic tools come down over the past decade or are they holding steady?

Dr. Burstein: I don’t have exact numbers but my general sense is that the cost, the commercial retail price, has been more or less the same, maybe even crept up a little bit over time. The tests cost somewhere between $4,000 and $5,000 at retail price. There’s been debate about whether that is money well spent. I will tell you personally I think that it is-if you look at the cost of a course of chemotherapy, it’s typically on the order of tens of thousands of dollars. If one can obtain a relatively inexpensive test on the order of a few thousand dollars to spare a patient tens of thousands of dollars’ worth of therapy, the back of the envelope calculations pretty quickly make it clear that that is a good cost-benefit analysis. The other point is that in relationship to some of our more targeted treatments, this is a drop in the bucket. We’re nowadays looking at treatments for breast cancer that cost $10,000 a month. I think there’s a legitimate criticism of the way that dollars are spent in the clinical enterprise, saying that we probably spend too little on high-quality diagnostics and perhaps too much on therapeutics. That’s a larger debate, but for many people this proves to be a good investment. Having said that, what we’ve seen in a different but related set of genetic testing, which is the hereditary cancer testing, the price of that kind of genetic testing has really come down a lot. So, it wouldn’t surprise me if in the years ahead these genomic tests also see price pressure downwards.

Cancer Network: Is there anything else you want to tell readers that we haven’t discussed already?

Dr. Burstein: Well, I think that the important thing to remember is that when you are encountering a patient with ER-positive breast cancer, you’re making several decisions with them. One is whether or not to give chemotherapy. That is informed by the stage of the cancer, by the patient’s age and their general health, by the features of the tumor under the microscope, and by this genomic tool. So, the genomic tool isn’t the whole story; it’s an important part of the story in terms of whether or not it’s important to get chemotherapy. An analogy I sometimes use with patients is, if you’re choosing a restaurant for dinner you’ve got to think about, is it convenient enough, does it have food that you like, is there an ambience that works for you on that given day, can you get a table? These are all things that factor into it, it’s not just one choice or one factor that you use to make a decision. Certainly, with chemotherapy, which is such an important decision in the management of a patient with breast cancer, it’s built on a lot of different clinical pieces coming together and that’s what we spend our time discussing with patients.

The second thing is that you know in very broad terms that these genomic assays seem to overlap in their value and that probably tells us something important about the biology of breast cancer that we hope to be able to move forward on. From a public health point of view, these tests have really revolutionized practice in the United States. We are giving half as much chemotherapy for ER-positive breast cancers as we used to, and we are seeing results that are getting better and better and better over time. So, we’re not doing the wrong thing. We’re not denying people chemotherapy in general who really need it. The final takeaway is, these tests can be a little complicated for patients; sometimes it can take an extra week or two to get the results, but it really is allowing us to tailor treatment to an individual, to the individual features of their cancer, and to make better decisions.

Cancer Network: Thank you so much. We appreciate your time and for speaking with us about this complex issue.

Dr. Burstein: A pleasure to be with you today. Thank you.