Harry Erba, MD, PhD: The next thing I want to turn to is, OK, we have all of this information right now. I’m going to focus on what we have clinically available, and that is the somatic mutations that occur in acute myeloid leukemia. Now, keep in mind that the people watching are going to be a group of academicians and clinical investigators like ourselves, but a large number of people out there are the people on the frontlines, the clinical physicians who are seeing these patients. And so, what I’d like, Eunice, for you to discuss is, what is clinically important in terms of mutational analysis at the time of diagnosis for a person with acute myeloid leukemia?

Eunice Wang, MD: That is a moving target. When you look at the treatment guidelines, there are different treatment guidelines, there are different molecular classifications, and there are different recommendations on which mutations need to be done by different panels. There’s the American College of Pathologists, the NCCN, and the ELN, and they all recommend different ones.

On a practical basis, I think, when I’m looking at a newly-diagnosed patient with acute myeloid leukemia, “What are the ones I absolutely have to have?” I think it’s very important to know whether they have a core binding factor mutation, that is very important, or FLT3. There are inhibitors for FLT3. There’s IDH1, IDH2—we currently have an inhibitor for IDH2, and there may be one coming for IDH1. NPM1 is also prognostic.

If I had a choice of some of the other ones, there are ones that definitely portend a poor prognosis. I am interested in knowing whether they have AXL1. I’m interested in knowing whether they have TP53. If I could get them, I would also be interested in getting splices on mutations. As you mentioned, Harry, they’re very predictive for patients having a secondary AML. So, I think the basic ones for me right now are NPM1, FLT3, IDH1, and IDH2. But what we typically do at our institute at the time of diagnosis when we collected those samples is send off for a few of those targeted genes.

What we actually do, which I’m fully aware is not available at everybody’s private practice, is biobank. We actually take the cells, we extract the RNA and the DNA, and we put it in our bank. And that allows us to have flexibility when we get additional information, like karyotypic information, to do additional mutation testing. It’s a stepwise approach. We know 50% of patients are going to have a normal karyotype. Those are the ones who I’m really going to want to do a detailed mutational testing on. If you have a complex karyotype, if you have an aneuploidy, those are already bad prognoses, so that already tells me a little bit about what features are there. If I get secondary AML myelodysplastic-related cytogenetic changes, again, that’s going to focus my future mutational testing. That’s the way that we triage so that we don’t do that.

Another way to do it in addition to biobanking is to do the opposite approach, which is just to get full next-generation sequencing and sequence everything in that sample. You have the range of information to not only be making the decisions now, but also down the line. But I think that there is a huge amount of variability. Jorge, what do you do at your facility?

Jorge E. Cortes, MD: We do have an extensive panel that we do for all the patients, but I think we need to differentiate here a little bit between what has direct clinical implications today and what we’re doing in terms of generating knowledge and advancing these to try and help us understand this for the future. If we’re just going to do it clinically, there are panels that can look at 400 different genes. But for a lot of them, we don’t even know what they mean. So, you get a lot of data and it doesn’t generate knowledge when you do it for individual patients. We do it for everybody, but we’re trying to develop that database to do that. With certain mutations, as you described, there are a few that have a direct intervention: TP53, IDH, etc. There is at least a significant prognostic implication that it changes your approach in terms of whether you’re going to use a transplant or not, for example.

Eunice Wang, MD: But sometimes, they can predict resistance. There’s FLT3-ITD, DNMT3A—that’s a pretty bad one to have, right?

Jorge E. Cortes, MD: Absolutely, yes. And actually, I was going to ask you—you mentioned some of the others—what about P53? What do you think of P53?

Alexander E. Perl, MD: Many of us are leaning towards giving hypomethylating agents more and more for a P53-mutated patient, even as frontline therapy in a fit patient, knowing that ultimately, we want to transplant that patient. We’re on the fence of, do we want to start the clock by inducing them? If we achieve a remission, we need to be immediately available to go to transplant, but we may not have a donor right away. So, we might use a hypomethylating agent, which in studies has shown reasonably high response rates and isn’t going to largely make the patient ill in a way that induction might. That buys us a little time, and that information right off the bat can be helpful.

My take on this is it used to be that we would start therapy immediately, almost without regard to many of these diagnostic studies. More and more, we’re trying to get more information even before starting so that we make sure, if there are new drugs that help certain subsets of patients, that we’re basically allocating the right therapy to the right patient based on some of these.

We might use a different induction strategy for a patient with a core binding mutation, a gene fusion of A21 or inversion 16, than what we would for a patient with a normal karyotype or an abnormal karyotype, a high-risk karyotype, secondary AML, which we know is AML with myelodysplasia-related changes only by the karyotype. We might need some karyotypic information, and that can take time. It’s hard to do that. We’re trying to speed up our diagnostic tests to get more information faster. Sometimes, we’re using FISH panels in patients who might be appropriate candidates for therapies that have a proven benefit in AML with myelodysplasia-related changes, where we could make that diagnosis based on karyotypic information.

Transcript Edited for Clarity