New Data on Role of Gene Assays in Breast Cancer Treatment

Patients with ultra-low-risk breast cancer according to genomic assessment had excellent long-term outcomes regardless of clinical risk or whether they received adjuvant therapy, according to new analyses presented at the recent virtual American Society of Clinical Oncology (ASCO) annual meeting.

The MINDACT trial established the 70-gene signature’s ability to risk-stratify patients by a tumor’s genomic signature, while the NRG Oncology/NSABP B-42 trial evaluated the impact of the 70-gene signature on extended aromatase inhibitor (AI) therapy.

In this second of four exclusive episodes, MedPage Today brought together three expert leaders in the field. Moderator Hope S. Rugo, MD, of the University of California San Francisco Helen Diller Family Comprehensive Cancer Center, is joined by Jennifer Litton, MD, of the University of Texas MD Anderson Cancer Center in Houston, and William J. Gradishar, MD, of Northwestern University Feinberg School of Medicine in Chicago, for a virtual roundtable discussion on the impact of these results.

Episode one: PARP Inhibitors for Early, BRCA-Mutant Breast Cancer

Following is a transcript of their remarks:

Rugo: Hello. I’m Hope Rugo, a professor of medicine and director of breast oncology in clinical trials education at the University of California San Francisco’s Comprehensive Cancer Center. We are going to talk about some interesting updates from ASCO 2021. With me today are my excellent colleagues, very knowledgeable and key opinion leaders in breast cancer: Dr. Jennifer Litton is professor of medicine and vice president of clinical research at MD Anderson’s Comprehensive Cancer Center, and Dr. Bill Gradishar, who is professor of medicine and chief of hematology-oncology at Northwestern University. Both of my colleagues are experienced clinical trialists and clinicians who specialize in breast cancer.

Now, we are going to talk about some interesting data that was presented at ASCO 2021 and actually we’ve been waiting for it for a long time. NSABP B-42 published early data and presented it and published it, and then presented updated data with longer follow-up. Now, we are still waiting to see how genomic tests could better help us decide which patients with hormone receptor-positive breast cancer benefit from extending the duration of aromatase inhibitors from 5 to 10 years.

B-42, of course, included a population of patients who switched from tamoxifen to AIs. They just had to have some AI in the first 5 years and then they were randomized to complete 10 years of an AI versus 5 years of placebo. We have that data, but we don’t know who benefits, and there are obvious implications for continuing AIs after 5 years for our patients, who tend to have a lot of side effects that occur every single day with these therapies.

Two studies were presented. One was looking at the Biotheranostics HI test to see whether or not that could help us predict benefit. Previous data have suggested that potentially it helps to predict benefit, if you’ve got 5 years of tamoxifen and then took letrozole or a placebo in MA.17 and in the trials that looked at extending tamoxifen therapy in node-positive disease after 5 years.

The second presentation looked at the 70-gene scoring system to see whether or not your original 70-gene category, high or low, would help to determine the benefit of extending adjuvant endocrine therapy with an AI. Bill, can you tell us a little bit about what this presentation showed and your take on it?

Gradishar: Sure. With respect to the NSABP B-42 trial, as Hope already outlined, this was a trial that you could either have received an AI for 5 years or a sequence of tamoxifen followed by an AI to 5 years, then patients were randomized to letrozole or not, or additional duration. Extended-duration therapy was the question and the trial demonstrated an improvement in disease-free survival. It never showed an overall survival benefit. As outlined already, can we tease out among those patients, who are those that are most likely to benefit and avoid the toxicity that comes with the therapy in some patients and those who aren’t going to benefit?

There have been different molecular assays applied to these datasets that help define hopefully who it is that benefits and does not. One such tool was the Breast Cancer Index [BCI]. As listeners know, that’s been around for some time. It’s been shown to be prognostic over a number of retrospective trials and there’s even a suggestion that it could be predictive in determining the duration of therapy.

But this was an application of that tool to a prospective dataset. They called out from B-42 a translational group that was significant, that had very similar characteristics with the population as a whole.

If you cut to the chase at first flush, the trial is negative, meaning that the application of BCI failed to demonstrate who it was that would benefit from extended durations of therapy. That’s sort of the fundamental takeaway.

But if you look under the hood, so to speak, and start looking at perhaps some of the other details, there are some nuances that I think are important. One of them is that we recognize that many of the higher-risk patients tend to recur early if they are going to do so. If you start looking at the data more when the curve split with respect to disease-free survival at about 4 years and take the analysis from that point forward, there is some suggestion that you can define a group that perhaps would benefit. That is the BCI-high patients compared to the BCI-low patients.

But at the end of the day, this analysis, for what the primary endpoint was, did not show that the BCI could be globally applied to this trial and determine who it is that benefits from longer durations of therapy or not. Whether longer follow-up of this dataset will offer us more insight, we’ll just have to see, but there are some interesting aspects to it.

The second presentation actually was the same trial, but being analyzed with the MammaPrint tool — again, with the recognition that most of the early recurrences, as I pointed out a moment ago — they occur in these higher-risk patients, and where you start to see the curve separate overall in the trial is roughly around 4 years. If you look at it from that standpoint going forward, it does appear that the patients who have a high MammaPrint do not benefit. There is not necessarily a surprise there. They tend to recur earlier compared to the patients with the lower MammaPrint score, where there may be a hint of benefit.

I think what we have to sort of reconcile with all of this data is it’s not an absolute crystal-clear picture of how these tools fit in in making our decision making. I think these tools can be applied for individual patients where there are questions about the toxicity and whether it is worth continuing, but I’m personally not yet convinced that they actually show us the path forward for all patients.

Rugo: It’s really interesting. It’s kind of too bad because we have found these tests to be so incredibly powerful in deciding about chemotherapy or not in prognosis. Jen, how do you decide on which patients should have extended-duration endocrine therapy?

Litton: You know, its really … it’s really hard. The aromatase inhibitors plus or minus ovarian-suppression tamoxifen, I think these just really are some of the biggest mainstays of treating breast cancer and keeping disease away, but they aren’t without toxicity. For many people, they can be really detrimental to quality of life, with hot flashes, vaginal dryness, and joint pains.

As patients are getting close to that fifth year, they are often asking, “How much do I want to give up on my quality of life for that? What’s that for me?” Because risk is so individual. Is it 2% absolute? Is it 3% absolute? I’m just not sure that any of these tools, as I have seen in this data, helps me get to the question that my patients want to know.

For me, I think that the data so far with tamoxifen is very clear, 10 years versus 5, if you can. But right now, unless it’s a very low risk, I’m using ovarian suppression and AI for 5 years based on our SOFT and TEXT data. I don’t think the data is really convincing for 10, though I have had some patients beg me for it because they don’t want to give it up until the data are matured enough. I think there are still the sweet spot, so far around 7 years. I would love to see some more prospective work with these tools, so that we can really apply them.

Rugo: Yeah. Of course, one of the issues is that none of these trials looked at their test prospectively, so that does make a difference because you are always going to have a subset of the total trial population. This was certainly true with these different studies. But I do think it’s an intriguing question about the potential delayed benefit seen with this unplanned analysis of the NSABP B-42 and the HI index, because you expect that the AI you took for 5 years has a carryover effect for 2 to 3 years, and whatever you’re doing after that wouldn’t kick in for a while.

So when they looked at that analysis, they saw the separation of the curves, it could also be invented, but I think that it is really pretty interesting data that does show what we’ve already seen, which is a carryover effect.

Litton: The discussion was just well done by Fabrice André, MD, who brought up, I thought, a really intriguing point. A nuance to B-42 is that the recurrences were very … a very minute number were actual distant recurrences. In fact, most of these recurrences were local recurrences or second primaries, which I would see more of an effect on letrozole 7 years or 8 years.

At the end of the day, it’s the distant that is going to make me want to push this through on someone for 10 years. I think that that was a really interesting nuance of this subset of B-42. I don’t know what you all thought.

Gradishar: Well, the other thing I forgot to mention too is when … was talking about the data. When the BCI was compared to other datasets where it appeared to have some predictive value, the fraction of patients who were getting second breast cancers in B-42 accounted for a lot of the events. Again, to your point, Jen, I don’t know that we are demonstrating that it’s affecting late, distant recurrences. But, again, the question is, are all these datasets similar and are the events that are being tracked the same? I think it’s still confusing.

Rugo: These trials looked at who needs more therapy versus who needs what we think is the standard overall. I think probably we’re still going to be using clinical characteristics largely to direct when we give patients longer-duration endocrine therapy, such as having more positive nodes and higher-risk disease for better or worse. I wish that we had good data to know how to prevent those 10-year metastatic recurrences for 1-cm hormone receptor-positive cancers. That just seems a tragic area we need to address.

But one thing that was interesting, I thought, in that same session was who maybe needs less therapy out of our standard? I don’t think we answered it, but the data from the ultra-low test from MammaPrint or the 70-gene assay is quite intriguing, where you are way far over to the right, so ++, close to +1. That those patients in a previous study that looked at a retrospective dataset by Laura Esserman, MD, from UCSF showed that those patients did extremely well, almost regardless to the endocrine therapy. This time it was applied to the MINDACT population. What did you think, Bill?

Gradishar: I think it basically confirmed what Laura had presented in the past, that there is this population that basically has a remarkably good prognosis. Within MINDACT, which as everyone knows was a huge trial, there were about 3,000 patients that fell into the low category. But distinct from that is the ultra-low, which is another 1,000 patients.

When you look at the characteristics of those ultra-low patients, they predictably tended to be older, lymph node-negative, smaller tumors, hormone receptor-positive. Most of those patients received either endocrine therapy or no systemic therapy, and only a tiny fraction received chemotherapy.

But if you track out how they fared over time, they did remarkably well when you look at the breast cancer endpoints, breast cancer-specific survival. It’s in the high 90s. These are patients who do extremely well and it really lends itself to this question of whether you can even de-escalate further. Since most of these patients had endocrine therapy, it really raises the issue, do all of them even need endocrine therapy? Might they have actually fared just as well in the absence of this. I think it really does identify an exceedingly favorable prognosis in a not-insignificant fraction of patients.

Rugo: Yeah. It’s a really, really interesting dataset. I will say that the follow-up is not long enough in these patients, so we really need to look at their recurrence risks, especially with that carryover effect at 10 and 15 years. Right now, I don’t think that any of us would be comfortable not providing 5 years of endocrine therapy, but we certainly wouldn’t give them longer-duration therapy.

Litton: I completely agree. I just want to make sure no one … I strongly would encourage people not to start to act on this with ultra-low [patients] and withhold endocrine therapy. The study doesn’t support that. It supports the hypothesis to look at it. On Monday morning, this doesn’t change practice.

Gradishar: Yeah, don’t misinterpret my enthusiasm.

[Laughter]

Rugo: No, but it’s still really interesting. I think it does help us. Maybe those people could get tamoxifen instead of an AI, if they don’t tolerate the AI. This was a great discussion and really interesting data. Thank you very much for your excellent comments and reviews.