Harry Erba, MD, PhD: Let me turn the conversation about this study and the use of this drug to some practical issues. One of them is testing. There is a companion diagnostic testing for FLT3-ITD and TKD mutations. I’ve had situations where the PCR testing, which is how the companion testing is done, is clearly positive. Remember, it has to be over a 5% allelic frequency to be considered positive, and those are the patients who went on the study. But then you get the next-generation sequencing panel back 2 to 3 weeks later and it’s negative. When I really push the pathologists, they say, “Well, maybe it’s there at a low level, but we just can’t see it.” Can you explain that to me? What should we do?

Alexander E. Perl, MD: Next-generation sequencing for FLT3-ITD is challenging at a technical level. One of the reasons is what happens with the ITD. There are insertions of sequence into the FLT3 gene, right in a region on the protein that is basically functioning as a negative regulator. So, if you insert sequence there, you lose that function and the kinase stays active. But the amount of sequence added is variable. You could add 3-based pairs, 30-based pairs, 150-based pairs, or 400-based pairs. If you add a large sequence into that area, it’s actually large enough so that when they look at the sequence, it doesn’t align to anything in the genome, and they don’t see that it is something that should be considered a mutation. They just don’t know where to put it. It looks like FLT3, but they don’t see that it came from the prior sequence, and it doesn’t get counted as an abnormality.

Harry Erba, MD, PhD: What’s sequenced has to be exactly where the 2 things are joined.

Alexander E. Perl, MD: Right. So, that gets discarded and they don’t see that it’s there. That can be a problem. That’s 1 reason; that’s an excluding issue. The other reason is that the sequence is longer, and so the speed with which the sequencing reaction goes on takes longer in that sequence, and it doesn’t catch up. The allele frequency drops below the recording threshold, especially at 5%. If you have a small amount of the mutation, it might not be picked up by this, unless they have special algorithms to look for FLT3-ITD.

Harry Erba, MD, PhD: That’s a practical lesson here: A PCR-based assay should be done for FLT3 at the time of diagnosis—1, because the next-generation sequencing panel may miss it.

Alexander E. Perl, MD: Yes.

Harry Erba, MD, PhD: And 2, there’s a difference in time, right? You can get the FLT3 PCR test back.

Alexander E. Perl, MD: They’re getting better with the next-generation sequencing. They’re really getting better at matching that to the PCR and having concordance. They’re really pretty close now. It used to be much worse, but you’re still missing cases with it. I agree, the timing is the really critical thing because of what Eunice said, which is on study, we know the benefit happens when you give it for 2 weeks starting on day 8. You don’t want to get the results back on day 20 and say, “Oops, we’re so far out, and we don’t know if there’s a benefit.”

Eunice Wang, MD: Yes, we don’t know if it still would have the benefit of it.

Harry Erba, MD, PhD: Eunice, do we really know that this is targeting FLT3? Or could it have other activity?

Eunice Wang, MD: I think that’s an excellent question. That has been something: Now that we have the success of this agent, how much of it is due to FLT3? When you actually look at the kinome map—the array of kinases that are targeted by midostaurin—it’s pretty impressive. It hits a lot of different things. That might be in part why we get the GI toxicity: diarrhea, nausea, vomiting. Some patients develop rash, which is significant. Over long-term maintenance, the drug interacts with a lot of other drugs.

One of the issues we’ve had practically is that it can cause QTc prolongation. It interacts with azoles. There are many things about it, and part of that may be that it’s such a broad spectrum in pan-kinase inhibition. That, I think, has led to people to say, “Ideally, AML is unlike CML,” which I mentioned previously. CML is driven by BCR-ABL. That’s the main pathogenic mechanism driving the proliferation of CML cells—that FLT3 does drive AML cells, but it certainly is not what we call a founder mutation, as you were talking about with those initial molecular profiling data. It’s a secondary mutation. There are many other signaling pathways that can drive AML: FGF, EHF, RAS, and other things. And so, potentially in the up-front setting where we’re trying to kill as many leukemia cells as possible, having a broad spectrum of agents—chemotherapy, anthracycline, cytarabine, pan-kinase inhibitors—might be really useful. But the question right now is, if we make a more potent, specifically targeted selective FLT3 kinase inhibitor, how does that function? I think we’re going to talk about that a little bit later.

Alexander E. Perl, MD: That’s clearly where the field is moving and what we want to see, and for good reason. When we’ve looked at midostaurin as a single agent in relapsed/refractory patients with FLT3 mutations, the clinical activity was pretty modest. It clears their blast, but it doesn’t really do very much for their bone marrow.

Eunice Wang, MD: It’s not a durable response rate.

Alexander E. Perl, MD: It’s not very durable. It lasts a very short period of time. The newer drugs have much more significant and much more consistent reductions, or elimination, of circulating blasts. They clear out the marrow, and their responses can be more durable.

Eunice Wang, MD: And they can be potent, very potent.

Alexander E. Perl, MD: And those drugs can be added to frontline therapy.

Harry Erba, MD, PhD: We’re going to get to that. Before we finish this up, there are a couple of very practical issues. You mentioned the drug interactions: midostaurin metabolized by CYP3A4, antifungals, or CYP3A4 inhibitors. There’s concern for more toxicity. At this congress, there is a poster presentation showing that in patients who were receiving antifungals. There did not appear to be extra toxicity that was seen. Still, I think it’s prudent to be cautious, but you don’t have to dose reduce. Has that been your experience?

Eunice Wang, MD: Just carefully following the EKG is sometimes enough, I think.

Alexander E. Perl, MD: If you’re running into excess GI toxicity, if the patient is getting prophylaxis, certainly an alternate drug is reasonable in that setting, whether it’s an echinocandin or an alternate agent.

Harry Erba, MD, PhD: Another practical issue is, maintenance or no maintenance? The study was not designed for that. They were not rerandomized, so it’s not in the label.

Alexander E. Perl, MD: The United States label.

Harry Erba, MD, PhD: It is in the European label. So, yes or no? Do we give midostaurin maintenance or not?

Eunice Wang, MD: I think Richard Larsen presented an abstract just yesterday looking at the impact of the maintenance portion. Now, it was not a rerandomization, and they did all the statistical modeling and so forth to look at how the patients who went on to maintenance did if they had received both placebo and midostaurin. And by their predictive modeling, there was no significant difference in terms of relapse or overall survival or selection or enrichment or improvement in outcomes in the patients who received the midostaurin maintenance.

Harry Erba, MD, PhD: I’m going to push back a little bit, because there was a curve there that really gave me great pause as a clinician.

Alexander E. Perl, MD: Yes, yes.

Eunice Wang, MD: It did, it did.

Harry Erba, MD, PhD: You know what I’m going to say, right? It was when they stopped the midostaurin in placebo. In the midostaurin arm there was this very rapid drop, or number of recurrences. It was temporally related to stopping, and we don’t know how to identify those patients. So, I have heard people say, “Not only am I going to give maintenance, maybe I need to continue the maintenance because I don’t know how to identify those patients.”

Eunice Wang, MD: I understand, but the statistical analysis of those data would support the FDA’s decision not to give it.

Harry Erba, MD, PhD: Oh yes, there are statistics. But then you have the patient who stopped the midostaurin and a month later they relapsed, and they say, “Dr. Erba, what did you do?”

Eunice Wang, MD: But Dr. Erba, are you going to pay for that drug for that patient?

Harry Erba, MD, PhD: I’m sorry, I don’t pay for anything.

Eunice Wang, MD: Who’s going to pay for that?

Harry Erba, MD, PhD: Good points.

Eunice Wang, MD: Jorge, would you do it?

Jorge E. Cortes, MD: I have another practical question, since this is ITD. What about the D835 mutation? We do have a small subset of patients, and what they have is the D835 mutation. Do we use midostaurin or not?

Alexander E. Perl, MD: That’s important because we don’t always move patients with a D835 mutation to transplant because it has not been established they have as sinister an effect.

Eunice Wang, MD: But they did better as well. Both ITD and TKD, independent of the initial allelic ratio—they stratified for high or low allelic ratio—both did better.

Jorge E. Cortes, MD: I bring it up because it’s important. Not all the FLT3 inhibitors inhibit…

Eunice Wang, MD: That’s a distinguishing point.

Jorge E. Cortes, MD: Correct. We’ve been using sorafenib forever, because it’s not approved for this indication but it’s been available. That’s the one we’ve used until we had midostaurin. That doesn’t affect the D835 mutation, so we wouldn’t expect that to be the case.

Transcript Edited for Clarity