MINDACT Trial Design and Results in Breast Cancer

Transcript:Adam M. Brufsky, MD, PhD: Let’s talk about MINDACT and how it was designed. Mark, do you want to get into that a little bit?

Mark E. Robson, MD: It was essentially designed as a noninferiority trial. It’s a trial that allows you to have a smaller sample size, but perhaps is a little bit less tight in terms of confidence intervals. And they set the lower bar at, I think, 92%. That was thought to be the level at which you wanted to prove that the test segregated there. But the question that we’re talking about now, which is the predictive value in terms of chemotherapy benefit, wasn’t the primary endpoint of the trial. The thing that, in a sense, is of the most interest, both to the oncologist and the patient, is an underpowered analysis. I don’t know if it was a secondary analysis. And so, you can’t draw the conclusion from the results that there was no benefit in the patients who were clinically high risk and genomically low risk. You can draw the conclusion that the benefit is likely to be quite small. What’s small enough to forego it? You wind up needing to have a conversation with patients—which, of course, you always have to have—but some people think that 1% or 2% chemotherapy benefit in that genomically low, clinically high-risk group is worthwhile. Some patients will probably view that that way as well and others will not.

Adam M. Brufsky, MD, PhD: So, just to reiterate what this trial was, I think it took about 6600 women with early stage breast cancer, 0 to 3 nodes positive, and really almost any tumor size. It’s interesting when you go back and look at the data, there probably were, of the 600 women, about several hundred who had tumors that were T3, so they were substantially big tumors. What they did is they gave them a genomic risk, either high risk or low -risk, by the 70-gene assay or Adjuvant! Online, which we all know is not available yet, unfortunately. Adjuvant! Online is offline; it has been that way for a while. And so, I’m going to make a plea to Dr. Ravdin, please give us Adjuvant! Online. We’ve been waiting for it. I know you want to be perfect with it, but please give it to us. But nonetheless, they did a clinical risk based on Adjuvant! Online. And if you were clinically high and genomically high, you got chemotherapy. If you were clinically low and genomically low, you got hormone therapy. There was about 30% in each arm. Of those, just 60% of the patients were done. It was that middle group, obviously, that’s the important one where you were clinically high and genomically low. And there was about 1500 women that were randomized to chemotherapy or not or they randomized to use the clinical option.

And so, the issue in that trial, as you say very well, was this was not powered because it was 750 patients per arm to look at a chemotherapy difference. It was a noninferiority trial where they put their cutoff at 92%. They said that if you were clinically high and genomically low, and got hormone therapy alone, your 5-year distant metastasis-free survival—not disease-free survival, which is different, because it includes a lot of local recurrence—had to be greater than 92%. The trial was powered around that primary endpoint. And so, the issue is in that study: it was 94%. As a secondary endpoint, they went back and they looked at the benefit of chemotherapy. They found that there was no benefit, but it was powered to see at least a 1.5% benefit, which they could not see. So, what does that 1.5% mean? And do you trust a noninferiority trial? These are very good questions. But it is a prospective study that we don’t have much data for.

We have the TAILORx trial, which is a lot bigger. I think it’s like 1150 patients; it’s a big trial. They reported already 1 year ago on the lower-than-11 recurrence score, right? But they haven’t reported the real meat of it, which is what we’re all waiting for. I think that may come out fairly soon from what I understand. The fact is that people who were intermediate risk were randomized to chemotherapy or no chemotherapy. Hopefully, we’ll see those data soon. But the question is, what should we do? I agree that all these trials are roughly the same in what they do. The indications seem to be a little bit greater, and we’re going to talk in a minute about the thing that was presented at San Antonio, but they all kind of do the same thing.

Mark E. Robson, MD: Yes. The way I’ve always approached this, and the way I talk to fellows about it, is you decide ahead of time what it is that you think you want to do with a patient and then you use the test to see if that’s wrong.

Adam M. Brufsky, MD, PhD: And I think that that’s the way we’ve been using it. But the other thing that came out of this that I think is really, really interesting is that it, MINDACT, tells us not to do that. I’d love to hear your guys’ opinions on this because I know I have very strong opinions. And Kim, as usual, has very strong opinions opposite me, which is a good thing; it’s not a bad thing. So, I think things like MINDACT, we’re moving away from that. We’re moving towards if you already know what you want to do; don’t do the test. Come up with some clinical pathologic parameter that tells you, maybe have a low-grade tumor, 3 cm or less, that’s node-negative. Leave those out. Do the MammaPrint. Or if you have a tumor with a Ki-67, if you use it like 50% ER—this may be like 10%—don’t do it. But it’s those middle ones are the ones you want to do.

Kimberly L. Blackwell, MD: So, I don’t agree with you.

Adam M. Brufsky, MD, PhD: I know you don’t. That’s why I’m bringing this up and that’s why we’re talking about it. So, let’s hear it.

Kimberly L. Blackwell, MD: I think women facing breast cancer deserve some type of prognostic genomic assay. We could argue all day long on if is there any predictive value in the MINDACT and the data that have been presented, but prognostically, I think that the genomic predictors help me in the clinic because the tumors that we see probably once a week—the grade 1, HER2-positive or the grade 1 ER-negative—don’t make any sense. We know that the biology of these cancers is probably way more important and is incongruent than the size and the lymph node status. So, I would argue that even in T1—not in an 80-year-old, but in a 40-year-old healthy woman—she deserves to know the IQ test of her tumor. This is where I’ve stopped guessing. Unlike Mark, I don’t have a preconceived notion off of a—let’s take an average breast cancer—2.2 lymph node negative. I send the predictor because I’ve guessed wrong frequently, and high 21-gene assay or high MammaPrint will change my thinking about those. I find them extremely helpful. I think that as long as you understand how the test performs and the information it provides you, patients appreciate it as well. I would argue that everyone needs the assay.

Adam M. Brufsky, MD, PhD: But Aditya, let me bring you into this conversation. At MGH (Massachusetts General Hospital), what do you guys do? It’s a resource allocation question. If you only have so many dollars to spend, shouldn’t we do it to people who need it?

Kimberly L. Blackwell, MD: Giving chemotherapy is expensive. I just want to point that out.

Adam M. Brufsky, MD, PhD: And avoiding that is good.

Kimberly L. Blackwell, MD: Yes, but what if you avoid it in the patient that would have benefitted from it?

Adam M. Brufsky, MD, PhD: We’ll get to that. I agree.

Kimberly L. Blackwell, MD: That’s the oncologist’s dilemma. “How many people do I overtreat every day?”

Adam M. Brufsky, MD, PhD: I agree with that. What is your guys’ thoughts?

Aditya Bardia, MD, MPH: I think it’s a tough question, and we have debated discussions like this at tumor boards. I think there are a couple of things that I would like to highlight. If you look at the validation of Oncotype DX in the original NSABP (National Surgical Adjuvant Breast and Bowel Project) paper, grade was an independent predictor of disease-free survival. And if you look at the hazard ratio, it was actually even more than the Oncotype score. I think grade does have its own value that’s independent of what the Oncotype DX shows. So, I agree with Kim. I think it’s really helpful to get information that we can provide to the patients, as long as we believe that the information is accurate. For patients who are premenopausal, young, or less than 35 years of age, we don’t have the same level of evidence as compared to someone who is postmenopausal. I think we just have to be careful that when we order the assay, it’s ordered in a setting when we get the results, we can trust the results, and we can believe that the data are accurate.

The second point to consider is that all these assays are proliferation assays. There has been interest in just looking at Ki-67, and this IHC4 developed in Europe, which looks at ER, HER2, and Ki-67. And also, a recent paper in JCO from the Hopkins group said that a number of physicians can guestimate what the recurrence score is going to be. And if you look at standard clinical pathological characteristics, that essentially can help you make a decision about who needs chemotherapy versus not. So, I think there is a lot of value in clinical pathological characteristics. But when there is a case, like Kim was mentioning, where you have a grade 1 but you feel that the biology is not reflective in grade 1, I think it’s very reasonable to get additional data so you can make a decision whether you should give chemotherapy versus not.

Transcript Edited for Clarity