Combination Approaches Targeting FLT3 and IDH Mutations

Harry Erba, MD, PhD: If they have a FLT3 mutation, Mark, do you start with a FLT3 inhibitor or an IDH? Are you doing both at the same time?

Mark Levis, MD, PhD: I hate that, but yes.

Harry Erba, MD, PhD: We all hate that.

Mark Levis, MD, PhD: If they are at diagnosis, I treat the FLT3, keeping the IDH in my back pocket. But I just had a patient who I gave HMA [hypomethylating agent] and VEN [venetoclax], then FLT3. A year later, she relapsed, and I checked, waited for next-generation [sequencing], and an IDH1 mutation had shown up. Now, she’s on ivosidenib plus gilteritinib, and had a beautiful response. I use the 2 together. It is mix and match, and we adapt to survive.

Courtney DiNardo, MD: I think the one teaching point would be if you have both a FLT3 and an IDH1 or IDH2 mutation, I would not do just an IDH inhibitor. We have done the studies looking at the monotherapy patients, looking at response and resistance profiles, and patients with FLT3-ITD mutations do not respond well to IDH inhibitor monotherapy.

Mark Levis, MD, PhD: Interestingly, I think the presence of an IDH mutation portends a cell that will quickly figure out how to get resistant to FLT3. So I do like using the 2 together.

Harry Erba, MD, PhD: Dan, let’s say at diagnosis, your patient who is unfit for treatment has an IDH1 or IDH2 mutation. Do you start with an IDH inhibitor? Do you give them HMA-venetoclax? Are there patients for whom you would consider giving just IDH inhibitors?

Dan Pollyea, MD, MS: I have not seen that patient yet.

Harry Erba, MD, PhD: Really?

Dan Pollyea, MD, MS: Yes; I do not hesitate to give a venetoclax-based regimen to these patients for a couple of reasons. I think after an initial period of supportive care, I cannot conjure up the patient who really could not tolerate this. That is the first thing. Then the second thing is, having that IDH [mutation] as part of their biology, it gets me optimistic that we are going to have a good outcome from a venetoclax-based regimen. If we have to take a little more extra supportive care to get the patient to that point, that is fine with me because we know that in the IDH setting, the response rates are so high, the duration of responses are quite good, and the survival is quite good. That is a patient to whom I would enthusiastically give a venetoclax-based regimen. Our label-indicated choice would be ivosidenib alone for an IDH1-positive patient. The data there are pretty small, the numbers are small, and the response rates are not that promising. Courtney’s data, examining adding an HMA to an IDH inhibitor, certainly look a lot better. With that not being something ready for prime time in terms of it being an approved regimen, I would continue to use the venetoclax, azacytidine regimen. I can see, at some point, an IDH inhibitor plus an HMA being the preferred therapy for a frail patient, but it would have to be a really frail patient, in my experience.

Harry Erba, MD, PhD: This gets back to Vinod’s point that, for those patients who are unfit, you really do not have to wait. Courtney, I remember a presentation you made at EHA [European Hematology Association meeting] a couple of years ago about the combination of ivosidenib and azacytidine that really caught my interest. You had an 80% overall response rate; I know there were smaller numbers of CRs [complete remissions], but a high overall response rate. You showed data that in the first cycle, neutrophil counts could get better. Do you want to say more about that? Does that influence your choice of that regimen over azacitidine with venetoclax?

Courtney DiNardo, MD: Yes; there was a few presentations, followed by a publication of azacitidine with ivosidenib, the IDH1 inhibitor. There were 23 patients only, so it was a really small subset, although similar in size to the IDH subsets of azacitidine, VEN [venetoclax] that we talked about. The true CR percentage rate was in the 60s, and most of those patients who responded became at least PCR [polymerase chain reaction]-negative for the IDH1. We are seeing deep IDH1 clearance. Yes, those responses happened quickly with count recovery. You do not have quite that same degree of myelosuppression as you do with azacitidine, venetoclax. The other thing that is interesting to think about, that I do not even have a full sense of, is the difference between patients with IDH1 and IDH2 mutations. When you look at the patients treated with HMA, venetoclax, it is really the patients with IDH2 mutations who do so much better, who have CR or CRi [complete response with incomplete hematologic recovery] rates of 85%, with a median survival that is about 2 years and beyond. IDH1 responses were really spot on with the average for VLEA, with a 66% CR/CRi, median survival of just about a year and a half. I do not know if there is a difference in biology or just because the numbers are so small.

Mark Levis, MD, PhD: But you do know because they behave differently clinically, don’t they?

Courtney DiNardo, MD: I suspect there is a difference.

Mark Levis, MD, PhD: An IDH1 inhibitor will often have this magic response that will last a long time, and work when nothing else has worked, including HMA, venetoclax. I agree. Those IDH2s, those are the ones like an NPM1, they sail on forever on HMA, venetoclax.

Vinod Pullarkat, MD: Is the IDH2 R172 [mutation] less responsive to IDH2 inhibitors?

Courtney DiNardo, MD: Interestingly enough, no. You do not see the same degree of 2HG suppression. That on-target activity with the enasidenib isn’t there, yet the response rate is just as good. In the IDHENTIFY trial we were talking about, that phase 3, looking at enasidenib in the patients with R172 variations, those are the patients for whom you could see a statistical benefit with enasidenib. There is activity, for sure, in the R172s.

Harry Erba, MD, PhD: There is something different about the R172. That was the subset that Elli Papaemmanuil, [PhD,] picked out of that 1500 patient data set; 1% of the patients with intensive chemotherapy had wonderful survival data, so they are not all the same. We do not really understand the biology there of all of those.

TRANSCRIPT EDITED FOR CLARITY