Drs Schwartzberg and Pegram debate the value of having patients undergo more than 1 multigene assay to predict their risk of recurrence.
Lee Schwartzberg, MD, FACP: I want to comment on what you said about doing additional tests. I agree, if you have 1 multigene classifier, I don’t think you need another one. They don’t line up all that well. A few studies have looked at the percentages of those patients who would be defined as low risk or high risk prognostically by Oncotype DX, MammaPrint, or other [tests]. They’re measuring different genes and, to some extent, you’re getting a slightly different result, but they help regardless. If you get 2 different answers, I certainly wouldn’t recommend a third to break the tie.
Mark Pegram, MD: I wouldn’t know which third to pick. I’m reminded of the OPTIMA clinical trial published in the United Kingdom a few years ago. It had 313 patients, all of whom had PAM50, MammaPrint, MammaTyper, IHC4, Oncotype DX, and IHC4-AQUA. They did all those assays on every single patient, and they found a huge discordance among all these techniques. In fact, Oncotype DX predicted a higher percentage as low risk, so if you want your patient to avoid chemotherapy and be low risk, then get an Oncotype DX based on those data. It was about 82% vs about 60% to 70% for all the other assays combined. Sixty percent were assigned to different risk categories by different tests. That tells you that we aren’t in an era where there’s harmonization among multigene assays. There’s still work to be done in this area. Discordant subtyping was observed in about 40% of those assays. We have a long way to go.
There are some newer systems that I hope will come online clinically that predict biological behavior of breast cancer based on copy number variation. That’s probably a leading candidate for a better classification system compared with what we have now. It would be a multigene assay, because it recognizes 10 or 11 subtypes of breast cancer intrinsically. It has been evaluated by validation sets of up to 7500 patients, and each of those 10 or 11 subsets has a different prognosis. It’s very granular at looking at outcomes and prognosis, but it’s also very granular at looking at probability of pathologic complete response [pCR] rate to chemotherapy in the neoadjuvant setting. Each category has a different pCR rate as well. It seems to be both predictive and prognostic, like oncotype is in that regard, but it’s far more granular than high, intermediate, and low risk. It’s everything in between. That’s what we need moving forward.
Lee Schwartzberg, MD, FACP: These tests go back and were locked down over 10 years ago, even before we had TCGA [The Cancer Genome Atlas] develop the results. Now that we know more about genomics and not just expression, copy number, and other characteristics, we can re-sort our patients. It’s also worth mentioning that it isn’t a linear relationship between what we consider luminal A, luminal B, HER2 [human epidermal growth factor receptor 2] enriched, triple negative, and basal between the immunohistochemistry and the way these tests sort patients by intrinsic subtyping or another method, such as Oncotype.
With MammaPrint, you get BluePrint, which gives you a PAM50-like result. When you look at these patients from immunohistochemistry compared with their BluePrint characteristics, a significant percentage of those who are HER2-positive aren’t HER2 enriched. Perhaps more important, some of the HR [hormone receptor]–positive patients are more basal and might not respond to endocrine therapy. We have a lot to learn in terms of refining our molecular characteristics, even in the HR+. We certainly know that’s true for triple negative, which we define by what it isn’t rather than what it is, and we’re moving toward categories where we might be able to use a positive biomarker there. There are a lot of exciting things happening.
Mark Pegram, MD: You mentioned the basal intrinsic subtype. In data from metastatic case series and trials, I’ve seen studies showing that basal seems to be a resistance factor for CDK4/6 inhibitors as well.
Lee Schwartzberg, MD, FACP: Right.
Mark Pegram, MD: For not only the antiestrogen effect but also its CDK4/6 sensitivity, which is lower among the basal, for example. It’s very interesting.
Lee Schwartzberg, MD, FACP: Hopefully in the near future, we’ll be incorporating some of those into clinical practice. They aren’t there yet, but there’s more to learn.
Mark Pegram, MD: Those studies need validation. So far, we have exploratory cohorts showing that finding, and now we need larger validation cohorts to shore it up so that it’s ready for prime time and practice someday soon.
Transcript edited for clarity.