Before we leave the topic of FLT3 inhibitors, I want to talk about the difference between type 1 and type 2 inhibitors. A type 1 inhibitor is a second-generation drug like gilteritinib that inhibits both the TKD and the ITD mutations, and a type 2 inhibitor such as quizartinib inhibits only the ITD, not the TKD. There has been a lot of discussion about the mechanisms of resistance with those type 2 inhibitors, where you might see TKD mutations coming out, but there are also mutations that occur at the time of progression after gilteritinib with 

Sasha, you presented on a comparison of these 2, and I want your summary of whether you think there is room for both a type 1 and type 2 second-generation FLT3 inhibitor. Sasha?

Just to clarify, I will say something. The presentation we did was a poster that largely looked at the question of how active gilteritinib was in the population studied in the phase 3 quizartinib study. It did not compare gilteritinib with quizartinib, though realistically, if you look at the data sets, you can draw your own conclusions. You have to be careful with cross-study comparisons when you are looking at 2 different data sets. I would not suggest that this is what we were trying to do.

We know that the QuANTUM-R study and the ADMIRAL study, which are phase 3 evaluations of a second-generation FLT3 inhibitor vs salvage chemotherapy, were similar in design. The 2 drugs are similar in terms of how they work, their activity, response rates, and even the survival. At the end of the day, they are more similar than they are different. There are improvements in response, and there are improvements in survival with each drug compared with chemotherapy. The reason we did this analysis was to ask if the differences in terms of the survival benefit seen in the ADMIRAL and QuANTUM-R trials were because of the populations of the studies or in any way the differences in the drugs. I do not think that, at the end of the day, we can say that because it is not a direct comparison of the drugs. Gilteritinib is active in patients who would have met eligibility for the QuANTUM-R trial: who were higher risk than those on the ADMIRAL trial because they had to relapse within 6 months of first remission, and they could not have a TKD mutation as their only 

 ITD mutations, and the response rates were largely overlapping in terms of the 2 drugs. The survival was 10 months in the gilteritinib arm of the ADMIRAL trial; that would have matched that of the QuANTUM-R trial, and the differences in adverse effects were relatively small: more QT prolongation with quizartinib, more LFT [liver function test] abnormalities with gilteritinib. There are some differences in terms of duration of response and time to progression. The full count recovery is more common with gilteritinib, but there is room for both if we could have access to both.

In Japan, both drugs are approved, and I am a bit envious of that scenario. As you know, the ODAC [Oncologic Drugs Advisory Committee] decided not to recommend approval for quizartinib, so its approval process in the relapsed/refractory setting was halted there. The drug has undergone a frontline evaluation, 7+3 chemotherapy plus quizartinib vs placebo, which has completed enrollment. That is the QuANTUM-First study, and when that reads out, we will know whether that study meets its primary end point, which has been modified to be overall survival. It may take a bit longer to get that readout.

Videos

Type 1/2 FLT3 Inhibitors in AML

I want to use that as a quick segue away from 

, and now talk about combinations. Eunice, I am going to turn to you for that. What do you think about the data presented from The University of Texas MD Anderson Cancer Center of gilteritinib with venetoclax, trying to combine both of these targeted approaches?

This trial was also led by Sasha, and it was the dose-expansion cohort of a phase 1 study where they initially looked at the tolerability and safety, and they are now looking at the efficacy in patients with relapsed/refractory disease. A small proportion of those patients who had gone on study had failed either venetoclax alone or gilteritinib alone: about 20% of patients. The fact that you could have response rates well over 80% looking at CR [complete response], Cri [CR with incomplete count recovery], and CRh [complete response with partial hematologic recovery] in those individuals with the combination of venetoclax and gilteritinib was highly promising. This is something. Some of us are so impressed with these results that we are starting to consider using that combination in the off-label setting for patients with 

-mutant disease that has recurred or come back particularly after a venetoclax-azacitidine treatment. Even though Sasha says that would be his preferred regimen at the time of recurrence with 

-mutant disease, instead of just using the gilteritinib alone, we are continuing with the venetoclax as well.

That approach looks very promising for a few reasons. No. 1, it overcomes different mechanisms of resistance. As you know, you can get resistance developing to gilteritinib primarily through 

 mutations, and you can get resistance to venetoclax based on the development of 

 mutations. Combining the 2 seems to make a lot of sense from a biological point of view, and it overcomes cross-mechanisms of resistance. Patients also like it because it is a completely oral regimen. We have to say that the tolerability and dosing to minimize myelosuppression while preserving extremely high response rate is an attractive option, mostly because the patients who have gone on a bunch of therapy with either the VIALE-A trial or azacitidine-based therapy are usually unfit patients. These are not patients whom you would be asking to undergo a salvage treatment with MEC [mitoxantrone, etoposide, cytarabine] or FLAG [fludarabine, high-dose cytarabine, GCSF]. That combination is extremely promising. I know there is discussion about potentially moving that to the up-front setting and exploring that in a triplet regimen. That also suggests that it is feasible to give dual targeted therapy, which is also an important concept. Most of the combinatory studies at 2020 ASH [American Society of Hematology] Annual Meeting were combinations; for example, venetoclax with low-dose chemotherapy, venetoclax with high-dose chemotherapy, and venetoclax with a TKI [tyrosine kinase inhibitor]. The fact that we can combine venetoclax with a targeted TKI is valuable information, and it is very promising.

Role of Dual Targeted Therapy in AML

Before we get to the combination, let’s talk about sequencing because we now have more data looking at sequential use of FLT3 inhibitors or a second-generation drug like gilteritinib following the first-generation drugs, midostaurin/sorafenib. I know you presented on this, Sasha, at the 2020 ASH [American Society of Hematology] Annual Meeting. Do you want to summarize some of that data?

 I presented data that came from the ADMIRAL study and the CHRYSALIS study for the phase 1/2/3 development of gilteritinib, recognizing that the phase 1/2 study was designed for anyone with relapsed and refractory disease, the vast majority of whom were 

 mutated, but it did not have any restrictions in terms of prior therapy. About a quarter of the patients on that study had a prior FLT3 inhibitor. By the time gilteritinib was brought to the phase 3 setting, midostaurin was approved in many places, and sorafenib was commonly used post-transplant, so there were patients who were treated with these drugs, but you could not have had prior salvage therapy when you went on the ADMIRAL study. In general, there were relatively few people who had had a prior FLT3 inhibitor based on when the study enrolled. Only 12% of patients in the ADMIRAL trial had a prior FLT3 inhibitor, and the question comes up: should we be giving gilteritinib in somebody who has progressed after a regimen containing midostaurin with sorafenib now that these are commonly used in frontline therapy, either with induction and consolidation or post-transplant.

We did not have a lot of data to guide that from what we got out of the ADMIRAL trial, so we pooled the analysis of those 2 studies, the phase 1/2 CHRYSALIS trial and the phase 3 ADMIRAL trial, among the patients treated who had had prior FLT3 inhibitor, either midostaurin or sorafenib. We then looked at the outcomes of patients who had not had that on both of the arms for the ADMIRAL study—because there was a chemotherapy arm—what we basically found was this: the response rates to gilteritinib were still quite robust if you had prior midostaurin or sorafenib without much of a difference between the 2 of them. There was not a finding of pan-resistance if you had progressed after prior therapy. The response rates were a bit lower, more like about 40% composite CR [complete response] rate rather than 50%, but both of these were certainly better than the chemotherapy arm comparator. Even when we did a statistical comparison of the survival of those treated with gilteritinib on the gilteritinib TKI [tyrosine kinase inhibitor] arm vs the placebo arm, that was still highly statistically significantly better.

We can at least say that, when we looked at patients who had had prior FLT3 inhibitor midostaurin or sorafenib, responses were fairly comparable, and survival may be a bit lower than that seen in patients who had not had a prior FLT3 inhibitor, but it is still superior to alternatives. I would not recommend anything other than what you would ordinarily do for these patients: Use the data that say that gilteritinib is a superior option than salvage chemotherapy. It does not change the directive to use gilteritinib as the preferred salvage agent in patients with relapsed and refractory disease.

At this year’s meeting, Keith Pratz from your institution, the University of Pennsylvania Perelman School of Medicine, presented data from the phase 1b study of gilteritinib in combination with standard intensive chemotherapy, 7+3 and high-dose cytarabine, with a high response rate of 80%. It was basically a study that was large enough to say that this can be done safely.

What we are getting at here is sequencing. Should we start with a first-generation or a second-generation drug, first? We do not know. The PrECOG study that is being led by Selina Luger and Keith Pratz as well will be looking at intensive chemotherapy with either midostaurin or with gilteritinib in previously untreated patients with 

 mutations. We have to wait for the results of that type of trial. A similar study with crenolanib vs midostaurin is also ongoing at this point. We await the results of those studies that look at the benefit of a type of a second-generation drug in combination with chemotherapy vs the first-generation drugs.

Sequencing of FLT3 Inhibitors in AML

Eunice, why don’t you discuss the data that you presented on gilteritinib?

At the 2020 ASH [American Society of Hematology] Annual Meeting, we at Roswell Park Comprehensive Cancer Center presented the updated safety cohort data from the LACEWING study, which is a phase 3 randomized controlled trial of gilteritinib plus azacitidine vs azacitidine alone in individuals with AML [acute myeloid leukemia], 

 mutant, who are considered unfit for intensive chemotherapy.

The original trial was a 3-arm cohort trial in which patients with FLT3-mutated AML were unfit, and they were supposed to be randomized to receive gilteritinib alone, azacitidine alone, or the combination approach. Patients were to be randomized after a lead-in safety cohort to determine the best dose of gilteritinib, either 80 or 120 mg. Data presented at the meeting included the safety cohort, of which there were 15 patients. The dose of gilteritinib was determined to be 120 mg. Of those 15 patients, some of whom got 80 mg and some of whom got 120 mg, 10 of 15 had a response. There was about a 33% CR [complete response] rate and a total CR/CRc [composite CR] rate of 67%, and the median overall survival with the long follow-up was 10.8 months, with 1 patient making it all the way to 36 months on therapy.

Based on that result, we have moved into the randomized component. There was a protocol amendment in which the company decided not to enroll patients further on the gilteritinib-alone arm, and that was based on data suggesting that there was potentially a low response rate with monotherapy and increased risk of developing further resistance down the line, as well as the data from studies like the VIALE-A trial suggesting that combinatorial approaches might be better. That study is about half accrued; it is a trial in progress, so they are anticipating to finish accrual within the next year or so and hope to have future results at another meeting. What I would like to highlight is the high response rates, which are similar to the response rates they were seeing with the VIALE-A trial with about two-thirds of patients achieving a CR/CRc. It is a well-tolerated drug in the outpatient setting.

Let me get in the middle of this duel between the 2 of you on this topic because I thought it was interesting. Eunice, I agree with you. As Sasha pointed out, the hazard ratio for survival in the VIALE-A study was not significantly different from 1 for patients azacitidine-venetoclax vs azacitidine-placebo. Here is the issue: It is the sequencing issue, and it is also the response rate because when these patients present, they are sick and need a response. What you could say of both studies is that azacitidine-venetoclax and azacitidine-gilteritinib can control the disease in these patients with 

You are on solid ground choosing either of them. I will let you comment about triplet combinations in a second, but what I will mention what I find interesting here: In Courtney DiNardo’s presentation at EHA [the European Hematology Association 2020 Congress] of azacitidine vs azacitidine-enasidenib, there was no difference in overall survival. And the overall survival in those patients with 

 mutations was remarkable: 22 months, median overall survival. I cannot help but believe that it is because the patients who got azacitidine alone or azacitidine with enasidenib could then receive some other therapy once they relapsed: venetoclax-based therapy or another IDH2 inhibitor. It is the sequencing issue that is going to confound some of these survival studies. I am going to turn it back to you to comment on those things.

Most of what we have learned about giving drugs that target 

 mutations potently—the current second-generation drugs—has been done in the relapsed/refractory setting. That is where we enrich for the leukemia that grows most rapidly and is most 

 driven, and there is no test that can say, “This is 

 driven and this is not” that we have so we can discriminate these. That is where we have seen the benefit of a potent selective FLT3 inhibitor: the relapsed and refractory setting.

To me personally, as much as I want to bring these drugs to the frontline setting to suppress those clones early, we do not know that that is critical. That is why we are doing studies in the frontline intensive setting asking whether a drug like midostaurin or a drug like gilteritinib proves to be the better frontline drug, and we do not know the answer. Midostaurin is a weaker, less selective FLT3 inhibitor, but it may be a better FLT3 inhibitor because it hits all these other targets.

When you get to patients not getting intensive chemotherapy where they see less of the backbone treatment that could suppress the growth of all the clones, what is the best approach? I do not think anybody knows. What we need is a study that randomizes patients between 1 of these 2 approaches, either an HMA [hypomethylating agent]–plus–venetoclax approach—you could even argue an HMA only, like the study that Eunice is describing—vs a combination that includes a FLT3 inhibitor. That LACEWING study is important to define this. It is a randomized trial, and I hope we learn from this what the right answer is. It could be that there is no big difference in part because you are treating patients in the frontline setting. You may select for a growth of leukemia that has nothing to do with FLT3 through that approach.

What the data do show in the frontline setting is that, when patients get venetoclax plus an HMA, they do better than a patient who got an HMA alone in a study that was appropriately powered to answer that question. It was not powered to answer the question of the subgroup of 

 mutations, so I have trouble using a subgroup analysis to say that I should not use what has been proven to be a better drug if I had the option of HMA alone. I personally would have trouble enrolling to a study that had an HMA-only backbone based on the trial results of the VIALE-A study. I have been giving venetoclax plus azacitidine, but I freely admit that I do not know if this is the right thing to do, and we should be doing a randomized controlled trial.

I would agree with the points that you made, Harry, that getting a response is the most important first goal: getting a CR or a CRc. There is no doubt that using venetoclax and azacitidine leads to 67% CR/CRc rate as opposed to 20% to 30% with azacitidine alone. In fact, that is the same rationale that I use to treat patients with 

 mutations with venetoclax-azacitidine. Even though the survival data are not necessarily better, I can get a response, and if I can get a response in those patients, whether they are a patient with FLT3 mutations or 

 mutations, you know what I can do? I can take them to transplant. If I can take them to transplant with little treatment-related mortality going into that transplant, that is the best approach for some of these poor prognostic subtypes of AML.

I have also heard the argument that patients should receive venetoclax-azacitidine, get that into that CR, and do what you mentioned briefly: sequencing. Go ahead and give them venetoclax-azacitidine, and then if they relapse, give them gilteritinib subsequently as a salvage maneuver.

The concern I have somewhat with venetoclax-azacitidine is mostly a theoretical 1. I do not disagree with Sasha, and I would not want to disagree with Sasha, but there are a lot of data coming out from Courtney DiNardo and Andrew Wei showing that 

 mutation is a major mechanism of resistance to venetoclax long term. When they looked at patients and looked at their mutational profiles, having a 

 mutation or the development, acquisition, or selection for a 

 mutation was a primary mechanism by which patients failed venetoclax-azacitidine. This again supports the triplet studies that are being done or at least the 2 double-targeted therapies, which we will talk about in a bit, by Naval Daver and Alexander Perl looking at the venetoclax-gilteritinib combination.

We are lucky to have many options. Whether you want to start with the FLT3 inhibitor azacitidine or you want to start with venetoclax-azacitidine and then switch, there are some data suggesting that venetoclax-azacitidine works best when you use it in the up-front setting. We know that venetoclax-azacitidine in the relapsed/refractory setting can only result in response rates as low as 20% to 30%. For that reason, people also do the venetoclax-azacitidine combination first and save the FLT3 inhibitor. We do not know. Either of these is a great topic for discussion, and we will be trying to sort out these data.

It is the same thing with IDH inhibitors. Are they better? Are they worse? Do they add anything? Is there a survival benefit? We do not know, but I do not think that we have to know. If we can use all these drugs in some sort of combination, and patients can make it 22 to 28 months or make it to a transplant, I do not think people will care how we got to that point.

I agree. My patients want to spend as much time in remission as possible.

 It is about getting in remission and staying in remission on drugs they can tolerate. What we do not know is the combinations vs sequencing and what the best way to do that is.

Management of AML: FLT3 Inhibitors and BCL2 Inhibitors

In this next section, want to focus on targeted therapies. More specifically, we are going to be talking about targeted therapies against 

 mutations, and we will talk a bit about some investigational agents that are getting some of our attention. I want to start with an important topic, and that is the use of FLT3 inhibitors not for patients eligible for intensive chemotherapy. We know from the RATIFY trial, from the German study group, that the addition of midostaurin to intensive chemotherapy improved survival and other outcomes.

Sasha, I want to turn to you about what you do for patients who are unfit for intensive chemotherapy. They are not eligible by age or comorbidities but have a 

 mutation, whether that is an ITD or TKD mutation. How do you choose therapy for those patients?

I can tell you what we do at the University of Pennsylvania Perelman School of Medicine, but despite spending a lot of time studying 

 mutations and trying to be as expert as possible in 

, I do not know that what I am doing is the right thing. One problem we have is that there are a bunch of different options, and we do not know what the proper 1 is. We have good data from the VIALE-A study in patients who are unfit for intensive frontline chemotherapy to say that a combination regimen that includes venetoclax is better than a single-agent regimen with either a hypomethylating agent or low-dose cytarabine, which I do not commonly use as a single agent.

Most typically, I use venetoclax with azacitidine the same way it was used in the 

 paper recently published by Courtney DiNardo and colleagues, which showed a survival benefit for venetoclax plus azacitidine over azacitidine plus placebo. In that study, they did a subgroup analysis of a bunch of different molecular mutations, including 

. They did not break out ITD vs TKD in that study, but they showed that both responses were higher in the combination arm and that survival was better in the combination arm. For the question of what is better, azacitidine vs venetoclax plus azacitidine, venetoclax plus azacitidine beats out azacitidine alone in patients with 

 mutations, with all those caveats. There are no alternatives.

I disagree with that, Sasha. The hazard ratio on that VIALE-A trial was overlapping. For patients with 

 mutations, I am not 100% sure that is correct. I know that Marina Konopleva presented data from her group at The University of Texas MD Anderson Cancer Center at the 2020 ASH [American Society of Hematology] Annual Meeting saying that there was no statistical difference between those who got azacitidine alone vs venetoclax-azacitidine at their institute. I would have to throw that out there.

You are taking the words out of my mouth. I was going to say that, if you look at the point estimate, it was 0.66 for patients with 

 mutations. For the whole study, it was like 0.66 for everybody, which looks the same, but the confidence interval exceeded 1. When you look at that, without the caveats that it is a combination of ITD and TKD, and you break out those mutations, there was no difference in the data that Marina presented for the ITD-positive patients, which raises the question: Should we be adding venetoclax to these patients in the frontline setting? The TKD-positive patients did better, but there was not as clear a benefit there, which brings up a study that you presented in which a FLT3 inhibitor was added to azacitidine: gilteritinib plus azacitidine. That trial is in progress, so we need to see how the trial reads out, but it is at least a feasible approach. You can comment more on the data that you presented, but that is 1 alternative that is about giving a FLT3 inhibitor with a hypomethylating agent.

Role of FLT3 Inhibitors in Chemo-Ineligible AML

Sasha, I know you must have just as many thoughts as Eunice does on this. I am going to go to your expertise, and you can then do whatever you want. Here is my question for you: I gave somebody 7+3 chemotherapy and midostaurin, and I then gave them HiDAC [high-dose cytarabine] and midostaurin and sent off 1 of these fancy MRD [minimum residual disease] assays for the 

 ITD, not the usual PCR [polymerase chain reaction] assay, the 

 ITD MRD assay, and it is completely negative. Does that patient need a transplant still? Should I go with the MRD as the marker or the biology of the disease?

I am happy to say that I have absolutely no idea. I do not have data that will tell me what to do with this particular mutation at this particular time point. That is always the issue with MRD: What is the assay, what is the time point, what is the lesion, and how are you determining that there is MRD present or absent? For example, in the patient you mentioned, were they 

 mutated at diagnosis? If so, what happened to that mutation? Did they have another marker that you would want to track and a balanced translocation or something like that? We can see a 

Did they have ELN [European LeukemiaNET recommendations for diagnosis and management of acute myeloid leukemia in adults] adverse-risk disease? No matter what, I want to transplant them because the survival is terrible outside that, even when you see MRD negative disease. MRD is most helpful in telling you what to do in somebody who is on the fence in that intermediate range. For the patients who have poor-risk biology, I have not seen convincing data that MRD negativity is a strong predictor of long-term success with chemotherapy. If you have adverse-risk disease by karyotype, MRD is not all that helpful.

The converse is not necessarily true. In core-binding factor disease, there are pretty good data to say that, if that persists by molecular, then that is something you can ignore. You cannot. It is not something that is not going to go away even though there is some old literature that says it can sometimes persist in remission, similar to the example that Eunice mentioned. I do not want to hang my hat on that in a patient like this with a 

There are newer assays looking at the burden of 

 ITD mutation as something that you can measure: something you can watch go down and something you can potentially change by transplant. We have incorporated that into post-transplant maintenance studies, such as the MORPHO study, the BMT CTN [Blood and Marrow Transplant Clinical Trials Network] 1506 trial, which gave gilteritinib vs placebo posttransplant but enrolled patients pretransplant for this very question. If you have MRD positivity pretransplant and then went on to get a transplant, what happened after that? Do those patients benefit from maintenance? For the converse, if you did not have detectable MRD, is there a benefit of maintenance in that setting?

What I can say is that there is a small amount of pretransplant data that Chris Hourigan recently published from a prior BMT CTN study looking at reduced intensity vs myeloablative transplant, where he did not have the diagnostic specimens. But if he detected a 

 ITD mutation at a cutoff of 10-4 pre-transplant, there were no survivors at 1 year. Presumably, the transplant-related mortality is not that high, so these were essentially all leukemic relapses post-transplant, so that is a bad scenario.

I do not know about MRD negativity and what that predicts in this day and age for adding a FLT3 inhibitor in the frontline setting. We still need to get that off the frontline studies. Those studies are currently enrolling, so let’s wait until we have the data before we make decisions. If you see positivity, that is not a good sign. We should be transplanting those patients. I would say that, outside a trial, you should be giving those patients maintenance.

We have a lot of questions and a lot of work that we have to do there.

MRD Monitoring in FLT3 ITD AML

I want to come back to Adam. We’re running out of time on this, and we can’t do it justice, but what about using these assays, next-generation sequencing and flow cytometry, for looking at minimal residual disease [MRD]? Are we there yet? Can we use these in our clinical practice?

I’ve got 3 hours to answer that question. Of course, it’s the holy grail. We all know that, and there are other diseases where MRD has been useful, like in CML [chronic myelocytic leukemia] and perhaps in ALL [acute lymphocytic leukemia]. There are emerging data on AML for sure, and there are many studies showing its value. The major issue, at least from a flow cytometric point of view—which experts can certainly do and pick up 1 of 104, or 1 of 100,000, leukemic cells perhaps using flow cytometry—is that it is largely done only in a few experts’ hands. It’s not widely available. Even when it does become widely available, there are going to be standardization issues with flow cytometry. It’s not as easy as picking up a single monoclonal CD5-positive lymphocyte in CLL, for example.

By contrast, many of the molecular assays are likely to be more useful, more standardizable, and more sensitive or as sensitive for tracking measurable residual disease. Unfortunately, not 100% of patients are amenable to molecular-based MRD testing. For those who have a known chromosomal translocation in 8;21 or chromosomal inversion 16, there are sensitive molecular assays to track disease. 

 mutations can be tracked quite sensitively, and we’re aware of the papers on that and on the value of finding measurable disease for 

Even in that scenario, there isn’t standardization, and there is more than 1 assay. Do you go for a quantitative RT [reverse transcriptase] RNA-based assay? Do you do a DNA PCR [polymerase chain reaction] assay? Do you do something other than the PCS assay, using droplet digital-based approaches? What about next-generation sequencing [NGS] for measurable residual disease? Unfortunately, although they again published data, most NGS panels are not sufficiently sensitive for MRD testing, but some people use it to track disease. One has to appreciate that they’re not yet quite as sensitive as some of the single-gene assays. In the future, that may change. There will be newer ways to do next-generation sequencing using duplex PCR assays, using unique molecular identifiers, or using error-corrected sequencing. These are all on the horizon, and they may allow us in the future to use NGS platforms to track MRD rather than going for single-gene assays, but we still have a way to go.

I need to let the clinicians weigh in a bit on this, and it’s a huge topic. If you don’t mind, I’m going to ask a specific question, and then you can do whatever you want, Eunice and Sasha. You give a patient with AML without a marker, without 

 mutation, 7+3 chemotherapy. At the end of it, you get back your blood marrow biopsy showing complete remission, but they find 0.1% myeloid blasts with a similar immunophenotype to the leukemia-associated immunophenotype. Eunice, what do you do?

That’s the million-dollar question. If we measure minimal residual disease, what are we going to do about it? In the past, it’s like this: I’m not going to look for it because what if I find it? I’m not sure I would do anything differently. If I picked up that small amount of MRD, I would potentially enroll patients on clinical trials. There are a number of entities, particularly immunotherapeutics, looking at whether they can function as MRD erasers similar to how we use blinatumomab in ALL. In that situation, I would enroll them on a clinical trial to theoretically erase the MRD.

The other option would be if that patient is MRD positive. We all know that’s not a good thing. How many abstracts and presentations have we seen that say if you’re MRD positive, you’re going to do worse than if you’re MRD negative? How many Kaplan-Meier curves do we have to have that? In that situation, the data suggest that it’s not useful to continue to give the same type of chemotherapy. If the patient is MRD positive after 7+3 chemotherapy, would I go on to a 5+2 chemotherapy consolidation or a high-dose cytarabine consolidation? I might not because that suggests that it is an indicator of therapy resistance. If I were to continue the patient on therapy, I might switch to an epigenetic-based therapy. More likely, the better option would be that I would send them to transplant.

There are a lot of studies that say if they patient is MRD positive, then yes, they’re going to do worse with transplant. On the other hand, we’ve all been in that situation, with somebody who has 0.01% MRD. And you decide to give them more therapy to try to get rid of it. The next time you do it, it’s 1% MRD. You say, “That didn’t work. I’m going to switch, and I’m going to try something else.” The next time you do it, it’s 5%. You’re then going back to your transplant person and saying, “I was going to do this transplant a few months ago, but now they have 5% disease.” The transplanter is saying, “What are you doing? You should have sent them to us when they were 0.01% MRD. I would have transplanted them, and they would have had a better option.” We don’t have a good answer. If I see that, I’m going to try to send them to transplant if I can’t get them on a trial. Right now, that would be my way to pursue it.

To go back to what Adam was saying, I tend to believe flow cytometry is done by my expert flow cytometry team at my academic institute, Roswell Park Comprehensive Cancer Center, although that’s not standardized. Your assay getting 0.01% might be 0 at my institute, and it might be 5% at Sasha’s institute at the University of Pennsylvania Perelman School of Medicine. So there is inconsistency.

Some people say, “We’re going to use NGS, and we’re going to use the mutational marker instead. That’s more precise.” Fifty percent of people don’t have a measurable marker. Of the people that have measurable markers, there have been data presented at 2020 ASH [American Society for Hematology Annual Meeting] that for a certain percentage of patients, 

 was the example of a mutation that doesn’t go away. It’s a founder mutation; you can track leukemic disease by doing next-generation sequencing for 

. There are some data presented at this year’s meeting that a certain percentage of patients who have 

 positivity after induction chemotherapy will have that 

-positive disease go away with subsequent therapy or even with no therapy. There are a lot of trials saying, “This person turned positive by 

, so we need to do something.” They’re now saying, “Maybe if you did nothing, then it would go away.” We don’t know anything based on some of these techniques. It varies based on the technique and who does it.

For molecular markers, we also have the question that comes up all the time: People will look at the time of remission, and they’ll do mutational testing if they haven’t done it at diagnosis. They pick up 

 mutation in the patient who has a complete remission. How do we know that’s not clonal hematopoiesis? Is that mutation diagnostic for the disease? Could it be the underlying background marrow in which the AML develops? There are a lot of questions. I’m going to turn it over to Sasha to address what he would do.

Assessing MRD Status in AML

, when mutated, are targets, but there has been a lot of discussion in the literature about the prognostic significance of 

 mutations. At the 2020 ASH [American Society for Hematology Annual Meeting], there was a presentation by Dr Sara Zarnegar-Lumley from Vanderbilt University Medical Center that looked at over 3500 patients with acute myeloid leukemia [AML] who had been registered and treated on COG [Children’s Oncology Group] studies, as well as SWOG, ECOG, the TCGA [The Cancer Genome Atlas], and Beat AML. In these 3500 patients, the age went from less than 1 month to 88 years old. The majority were pediatric patients, somewhere around 2200, so it drove the analysis. In that analysis, only 8.6% had an 

 mutation, which is lower than what we usually think of in adults. As she showed, the incidence went up with advancing age, up to 20% 

 mutations in older individuals. In this data set of 3500 patients, they showed that 

 mutation did not demonstrate any independent prognostic determination on outcome, but the co-occurrence with nucleophosmin mutation and 

One of the difficulties with her study is that the older patients she looked at mostly came from the Beat AML study. As we all know, it did not use intensive chemotherapy; the ECOG, SWOG, and COG studies did. It’s a bit harder to discern the prognostic significance there. It was interesting that they did not have enough patients with the isolated 

-R172 mutation, shown by Elli Papaemmanuil to be a group of patients who seemed to have a better prognosis: those who had 

 mutations but not others. That wasn’t seen here, and again, it was probably because the age was skewed toward younger patients. I wanted to give that 1 little update about 

: It doesn’t appear to be independently prognostically significant.

Prognostic Significance of IDH Mutations in AML

Sasha, 1 of the most common mutations we will see in our patients with AML [acute myeloid leukemia]—as Eunice mentioned, it is targetable in initial therapy and the relapsed/refractory setting—is the 

 mutation. Of course, we know there are several types of mutations, but the most common are the ITD and TKD. Do these have the same frequency and prognostic significance?

 internal tandem duplication mutations are both more common and more sinister, not so much for initial response to chemotherapy, but there is a higher relapse rate. Although things are getting better with the addition of FLT3 inhibitors to frontline chemotherapy, I don’t think we’ve eliminated the risk attributed to 

 mutations in terms of the survival of patients. As you mentioned, patients can have a 

-TKD mutation. They may even have both, and these may be independent clones or even in the same cell as we’ve studied. What also can vary is how much of the 

 mutation is present with a higher aggressiveness being associated with a higher ratio of 

-ITD to wild-type mutations in patients who have that mutation. In general, we want to know a lot of information about the 

 gene, not just if a mutation is present or absent. We want to know which mutation is there and how much mutation is there to a certain extent.

-ITD mutation, is how you know whether this patient needs a transplant. Studies have looked at the allele burden as a predictor of that in the presence of an 

 mutation, but there’s not an international harmonized agreement on how to measure that. We don’t have an IS [international scale] percentage the way we would have for 

 mutations, where we can say that if a patient has a low ratio, don’t transplant in the presence of 

, or if a patient has a high ratio, they should be transplanted. In my practice at the University of Pennsylvania Perelman School of Medicine, I err on the side of referring these patients for transplant when possible. If they have good transplant donor options, and they’re fit patients, the outcomes have generally been favorable.

We are starting to get data comparing our results with other laboratories in the context of clinical trials. We’re enrolling patients to a study comparing midostaurin vs gilteritinib, and I get a study-designed lab tests for allelic ratio on that study on anybody we enroll. We also get an in-house test, and they’re not always the same. In fact, they seem to be consistent from patient to patient within the lab, but they don’t necessarily line up exactly. This is something we’d like to solve in the field, so if we’re using this to make decisions, every lab comes up with the same answer. It would be nice if this were something with a normal value, so you could say it definitely falls in this range rather than out of the range, but we don’t have that yet. I’m cautious to use allelic ratio to recommend transplant.

It’s interesting that you say that, Sasha, yet smart leukemia doctors in Europe have developed the ELN [European LeukemiaNet recommendations for management and diagnosis of acute myeloid leukemia in adults] 2017 and now 2020 risk stratification and guidelines, and that allelic ratio is very important. Although I agree with you, let’s get a third clinician. Eunice, at Roswell Park Comprehensive Cancer Center, how important is allelic burden in determining treatment algorithms for your patients?

We also are on the side of saying that presence of any detectable 

-ITD mutation confers a worse prognosis. We measure the allelic ratio, but much of the data that have been acquired with the FLT3 inhibitors have demonstrated that those inhibitors are equally effective in patients who have high allelic ratio as well as those who have low allelic ratio, and in some cases in patients who have TKD mutation. I routinely refer all my patients who have 

-ITD mutation for transplant. As you know, the RATIFY trial data with midostaurin demonstrated that the benefit of midostaurin could not be disentangled from the benefit of performing an allogeneic stem cell transplantation and first remission for individuals who achieved the most clinical benefit from that FLT3 inhibitor.

Following that evidence-based trial is what we do. I have a question for the group: What about TKD mutations? Certainly ITD mutations can be detected specifically at the time of diagnosis with PCR [polymerase chain reaction] assays and with allelic ratios. Oftentimes, we don’t pick up some rarer or less common tyrosine kinase domain mutation until we get the results of the next-generation sequencing [NGS] panel back, and then the question is that, prognostically, they’re not as impactful as the ITD mutation. If you pick up a TKD mutation later, and maybe you didn’t do the tyrosine kinase inhibitor up front, what do you guys do with that information? Harry, do you continue? Do you still think a TKD mutation is poor prognosis? Would you add a tyrosine kinase? I’m sure there are some individuals in the community who are performing these next-generation sequencing or myeloid panels who aren’t getting results until 2 or 3 weeks later. The question I get is: “What do you do now that you’ve detected something, but it’s already a couple of weeks from 7+3 chemotherapy?” Do you still send them to transplant, or do you add a FLT3 inhibitor with their next cycle? What is your practice?

-TKD mutation, I’m not at all as certain about that single gene as a prognostic event in AML, so it’s the co-mutational pattern that goes with it that may drive my decision-making. Having said that, however, if a 

-TKD mutation is present, I will use midostaurin as part of the induction regimen. In fact, as you remember from the subset analyses, the patients who had the greatest hazard ratio for improvement in survival were the patients with TKD mutations, making some people wonder if the benefit of this first-generation FLT3 inhibitor is not just through inhibition of FLT3 but because of its inhibition of other serine threonine and tyrosine kinases that may also be active and driving the leukemia. I definitely use it there.

 mutations, I’m going to come back to this, so I’m going to save it for later. I want Sasha to specifically discuss whether he thinks the benefit of gilteritinib, a second-generation drug that’s a type 1 inhibitor, is equal as a single agent in ITD and TKD mutations. If I see a TKD mutation, I will use—in its labeled indication—gilteritinib for those patients. Prognostically, it may not be as important, but that’s definitely a therapeutic target that I will take advantage of.

The whole thing about allelic burden is that it’s only 1 of the things that we know is driving prognosis. We don’t talk about the number of 

 mutations and ITD mutations that are prognostically important. The other thing is the length of the 

-ITD mutation as well as the position of the mutation. There were data presented at the 2020 ASH [American Society for Hematology Annual Meeting] from the RATIFY trial showing that, when the insertion is in the beta sheet of the TKD1 domain, those patients have a worse outcome. There is a lot we don’t understand about how to judge the prognostic significance of 

 mutations, but I agree with you, Eunice and Sasha. It’s like being a bit pregnant. We know what the outcome is going to be, but the ITD mutation is there.

I need to turn back to Adam. Adam, I want to get you back in this. What would you say to an investigator or a clinician who says, “I sent off the NGS panel. Why do I need to do the PCR test for 

 mutation? Isn’t that redundant? Are there problems with double charging and things like that?” How would you answer those questions?

That’s a great segue to a question I was ultimately going to raise. Users of these technologies—clinicians and oncologists in the community—need to know what the PCR panel is telling us when it tests for 

 mutation vs what the next-generation sequencing panel is telling us when it looks for 

 mutations. Both TKD mutations, D835 in particular, are standard things we look for by PCR, the results of which you can get in a few days, maybe even less depending when the run is occurring with an allelic ratio. You’re restricted to knowing just what the ITD length is, what the allelic ratio is, and if there is a D835 mutation for most assays. By contrast, if you’re doing it by next-generation sequencing, the turnaround time is a lot slower: a week or more. You still get the information about the length of an ITD if it’s there, but you’ll get added information about where the insertion occurred. Next-generation sequencing can help with that and determining whether it goes into the beta-pleated sheet, which may be prognostically relevant. By sequencing, you’re not restricted to looking only at the D835; you may pick up that and other important gatekeeper mutations, resistance mutations. Both are going to pick up ITDs and mutations, but if you look at the details, there are differences. Although next-generation sequencing is quantitative, people are familiar with VAFs [variant allelic frequencies] of mutant vs wild type; that number does not easily translate into early integrations. You’re getting similar but different information from the 2 technological approaches: PCR on 1 hand, next-generation sequencing on the other.

FLT3 Testing for Treatment of AML

It appears that, although testing rates are improving over time, there is still a need to improve the rate of testing, especially in community practice. My colleague Dan Pollyea has published for the Connect MDS [myelodysplastic syndrome]/AML [acute myeloid leukemia] Disease Registry with a series of patients in that registry, and the registry results are showing that molecular testing is done in only 70% of patients in the community and government sites. I was wondering, Adam, if you would like to comment on the need for testing. Then we’ll move to turnaround times. Adam?

Yes, thanks Dr Erba. The point you brought up about FISH [fluorescence in situ hybridization] analysis and cytogenetics was something I was going to embrace as well, and although we are focusing on the importance of so-called molecular testing, I don’t think we should be talking about molecular testing in isolation. We should talk broadly about genetic testing, turnaround times, and impact on therapy to make sure that people don’t think that molecular testing is going to replace cytogenetics and FISH analysis. Absolutely not; we still need karyotyping in all patients.

To get to your point regarding patients who may benefit from gemtuzumab or who may benefit from liposomal ara-C [cytarabine] and daunorubicin, we routinely screen all patients when they come in for CBF [cerebral blood flow] FISH analysis, so we can tell from the day the leukemia is diagnosed by FISH results whether they have a chromosomal translocation of 8;21 or chromosomal inversion at 16 by FISH analysis more rapidly than we can by cytogenetics, so they can benefit from that form of therapy.

Similarly, as you alluded to, because our cytogenetic chromosomal turnaround time is longer than we would like, in a patient who may benefit from such liposomal therapy—who does not have a history of MDS, who does not have a history of prior therapy—we want to diagnose AML with myelodysplasia-related changes [MRC] by cytogenetic testing, so we do a rapid FISH panel. We’ll look for abnormalities in chromosome 5, abnormalities in chromosome 7, and deletion of 17p, getting these results on the same day to make decisions about how to administer the 7+ 3 conventionally or in a liposomal formulation.

With regard to molecular testing, it’s also important for practitioners to be cognizant of the different technologies that are involved. Everyone talks NGS [next-generation sequencing] these days: next generation sequencing or MPS [massively parallel sequencing]. This technology, although important and used in most myeloid and lymphoid malignances, is time consuming. The results generated by massively parallel sequencing are unlikely to be back in time to affect early therapy.

Therefore, the tests that are likely to impact early therapy are usually done as single gene assays like the 

 assays. If you need a test, and the result is going to be required to make a therapeutic decision within a week, next-generation sequencing is not the test to do to give you that answer in the short term. For that reason, although the mutations I’ve mentioned are covered on most NGS panels, you want a fairly rapid, usually PCR [polymerase chain reaction]–based single gene assay to pick up a 

Especially in the relapsed/refractory setting, where certain drugs have their best efficacy shown with a specific mutational complement, whether it’s a FLT3 inhibitor like gilteritinib or an IDH1/2 inhibitor like ivosidenib or enasidenib. Those mutations can sometimes change from diagnosis to relapse. In that setting, we need a rapid turnaround time. I absolutely agree with Adam: For most of the things we find on a next-generation sequencing panel, we can wait a bit of time. But it has not replaced karyotyping, and it has not replaced getting a full genetic analysis in terms of cytogenetics.

I agree. Sasha, since you have the floor, let me turn to you from the clinical perspective. You’re going to be waiting some amount of time, and we’ve all been trained that you don’t go to sleep until you write for the 7+3 chemotherapy on that newly admitted patient. What should we be doing? How much treatment delay is allowable? Are there any data that help us and inform us?

“Don’t just stand there. Do something” is what we are feeling, but “Don’t just do something. Stand there” is potentially the right answer because you want to know that you’re giving the best regimen for your patient. If the results from some of this testing lead you from 1 regimen to another, it is important to be right. What the data have shown is that, for certain patients, picking the right regimens can lead to a survival benefit, so you don’t want to rush into something too quickly. That being said, I don’t think we can say, “You’ve got AML. Come back in a few weeks, and we’ll treat you” and not expect that a good number of patients will get sicker.

With close monitoring of patients, it is safe to follow them during that time. We have data that Christoph Röllig and colleagues published in the past year that showed a delay of up to a couple of weeks—with close monitoring in the context of patients ultimately treated with intensive therapy—was not associated with a lower remission rate or impaired survival. If you need to choose a regimen based on the genetic data, and you’re closely monitoring your patient, and your patient is not developing potentially life-threatening complications under your watch, then you can delay that without impairing the ability of the new regimen to give you that benefit.

One of the challenges I face is that these patients are often identified in emergency departments in the community. They get shipped into our inpatient service, and there is a lot of “Why aren’t you doing something?” We’re just waiting, and this adds to the length of stay. We’ve started getting in the practice of potentially discharging patients if they are stable enough and waiting for that evaluation to come back.

It’s important to discuss this with the pathologist. When I got to Duke, they talked about their limited resources, and we have to take that into account. Pathology departments also have constraints on what they can do. We set up a program so we have a way of identifying newly diagnosed patients and relapsed/refractory patients for whom a treatment decision needs to be made. We can distinguish them from others where there can be more time from the bone marrow biopsy to the actual report. We get FISH analysis results back quickly as well as the PCR panel for 

Adam, what conversations do you have with Sasha at Penn and our other colleagues, Selina Luger and David Porter and the like?

We have frequent communications regarding new patients. We have open channels of communication. We’ve also worked intently to work up and develop pathways. For each time a new patient with AML comes in, we’ve got a pathway that we follow in terms of what the standard laboratory testing is going to be, along the lines of what I alluded to regarding FISH analysis, cytogenetics, and molecular testing. We didn’t talk about PML [promyelocytic leukemia]/RARA [retinoic acid receptors] or translocations in chromosomes 15;17, but we do that as well. We have standard, quite well-formulated pathways each time a new patient with acute leukemia comes in, AML in particular, more robustly than for patients with ALL [acute lymphoblastic leukemia]. We also recognize early on patients who come in who are relapsed. We communicate rigorously, vigorously, and frequently. Beyond that, we have established pathways and guidelines for how to do the tests we need to do.

Molecular Testing in Community Practices

® titled “Acute Myeloid Leukemia: Evolving Perspectives on Testing, Targeted Therapies, and Transplantation.” I am Dr Harry Erba from Duke University School of Medicine in Durham, North Carolina.

Joining me in this discussion are my colleagues Dr Alexander Perl from the University of Pennsylvania Perelman School of Medicine in Philadelphia, Pennsylvania; Dr Eunice Wang from Roswell Park Comprehensive Cancer Center in Buffalo, New York; Dr Adam Bagg, a hematologic pathologist from the University of Pennsylvania Perelman School of Medicine in Philadelphia, Pennsylvania; and Dr Corey Cutler, a transplant specialist from the Dana-Farber Cancer Institute in Boston, Massachusetts.

We’re going to discuss a number of topics pertaining to contemporary treatment of acute myeloid leukemia [AML]. We’ll discuss the latest research in the field and the impact of recent clinical trials on making decisions around treatment selection. Let’s get started with our first topic.

In the first section, we’re going to discuss molecular testing in AML, which is clearly going to rely heavily on our interactions with our pathologists. Before we get there, it’s important that we help inform our pathologists as clinicians what we need in caring for patients. For this, I’m going to turn to Eunice. Talk about what molecular testing should be done according to NCCN [National Comprehensive Cancer Network] 2021 recommendations and then what you do at Roswell Park. Eunice?

Thanks, Harry. I appreciate the opportunity to speak to the audience. In the past, when we made a diagnosis of acute myeloid leukemia, we relied on morphology and flow cytometry, as well as conventional cytogenetics. In the current era, given the development of many targeted therapies for acute myeloid leukemia, we are required to perform molecular testing, and the NCCN 2021 recommendations have stated explicitly that there are certain actionable mutations that should be expedited at the time of AML diagnosis for therapeutic reasons. These include 

mutations. It is explicitly stated in the guidelines that these are immediately actionable mutations, meaning that we would potentially treat patients differently and offer treatment approaches based on the presence or absence of these mutations.

 mean? That’s a great question, Harry, because we know that some of our colleagues in the community vs academia may have different timelines. At our center, we like to have those results back within 3 to 5 days of obtaining the sample. That sample can be the peripheral blood if there are a significant number of blasts or a marrow sample. So if there is a significant amount of disease in the periphery, there is no need to wait to do a bone marrow biopsy. In fact, we sometimes send peripheral blood first, then do the marrow and wait for those marrow results a couple of days later. In the community, it could take 1 to 2 weeks, and we would advise potentially collaborating with the individuals—Dr Bagg can discuss this later—about how to best facilitate obtaining those results in a timely fashion.

Can you give me a couple of examples of specifically what mutations and changes you’re looking for and how you would select therapy? What are the most urgently required tests that you need to start therapy?

That’s a great question. I’ll give an example. I had a young patient come to see me the other day, a 20-year-old undergraduate student who ran on the cross-country team. She was having progressive anemia, weakness, and symptoms over the last few months right before returning home for the Thanksgiving holiday. She went to employee health and had a CBC [complete blood count] drawn. Unfortunately, that showed about 40% peripheral blasts. We brought her in to see us in our clinic, and we made a diagnosis based on her peripheral blood that she had acute myeloid leukemia. She was admitted, and given that she was 20 years old, we deemed her eligible for intensive chemotherapy. We sent peripheral blood at the time of initial encounter for 

 mutation testing. She was admitted a couple of days later and was receiving 7+3 chemotherapy when we got the results that she was 

In that situation, obtaining that result in the first 3 to 5 days of the diagnosis is crucial because we would be obligated based on that result to start a FLT3 inhibitor like midostaurin or an experimental inhibitor on day 8 of her treatment regimen. In that case, expediting the results in a young patient fit to receive intensive chemotherapy could not only alter the initiation of therapy on time on day 8 but also drive us to consider more seriously performing a bone marrow or allogeneic stem cell transplantation once she achieves a complete remission [CR]. Based on that 

 result in the setting of a normal karyotype, we have also started typing her siblings and are planning to move forward with a transplant in CR1.

I completely agree, but I want to throw in a couple of other examples that may not come from the mutation analysis. Let’s go back to cytogenetics and FISH [fluorescence in situ hybridization] analyses, which are still important. According to the meta-analysis of 5 large, randomized trials of chemotherapy with gemtuzumab ozogamicin, the anti-CD33 antibody, patients with core-binding factor leukemias—those with a translocation between chromosomes 8;21 or an inversion in chromosome 16—had a survival benefit when gemtuzumab was added to chemotherapy in some dose and in some schedules. It varied between the protocols, and it should be done with chemotherapy. You need that result even quicker than the 

 results, with which you’re going to start midostaurin on day 8.

The other place I use it is with a FISH panel that may show me changes associated with AML with myelodysplasia-related changes based on cytogenetics. It appears to be de novo, but it may have cytogenetic changes. So we can have a different discussion about liposomal daunorubicin-cytarabine or, as we’ll talk about later, hypomethylating agents with venetoclax in that population of patients. I agree, it is important to get this information.

Eunice Wang, MD, Roswell Park Comprehensive Cancer Center

Articles