Acute Myeloid Leukemia Podcast

MRD and Relapse in Patients With AML

Gail J. Roboz, MD; Roland B. Walter, MD, PhD


August 17, 2022

This transcript has been edited for clarity.

Gail J. Roboz, MD: Hi, everyone. My name is Dr Gail Roboz. I'm a professor of medicine and director of the leukemia program at Weill Cornell Medicine and the NewYork-Presbyterian Hospital in New York City. Thank you very much for joining this series on AML, acute myeloid leukemia. It is now my true pleasure to welcome my friend and colleague, Roland Walter, who is professor of medicine at the Fred Hutchinson Cancer Center. We are going to be talking today about a difficult area of AML: patients with relapsed disease. Unfortunately, it is still the case that although we're doing a lot better in AML than we used to, the majority of patients do ultimately relapse. How do you approach a patient with relapsed disease?

Roland B. Walter, MD, PhD: Even though we've made some progress over the past few years, how we approach patients has really not changed fundamentally, right? We will look at what people got before they relapse, what kind of therapies they've got and how long those therapies were effective or not effective, what toxicities there were, and factor this into our decision-making, realizing that our standard therapies are still pretty suboptimal. So the first choice, certainly at a larger center as yours or mine, is to look for a clinical trial if one's available before resorting to standard therapies.

Roboz: Absolutely. So the first thing is the concept of taking another look at the disease because, as we constantly see, what comes back is not necessarily the same as what we started out with. And I would say that at our center, it is definitely a practice to relook at everything, including next-generation sequencing, and not assume that kind of leukemia itself is biologically the same as what we started out with. Would you say that's standard for you guys — to redo everything? Immunophenotype, any molecular testing, molecular diagnostic testing — redo it all

Walter: I think that's certainly true for immunophenotype and genetics, and more and more for molecular diagnostics as well, especially with increasing availability of potentially targetable mutations. That becomes more and more of a necessity that we do that.

Roboz: One of the things to give as a clinical pearl is that if there is going to be consideration for a patient to come in to a major center to think about a clinical trial, I often tell my colleagues — and the community — that if they are suspecting a relapse, not to actually prove it with a bone marrow biopsy before sending the patient in, because then the patient might end up having to have another one a week later as part of the eligibility for a clinical trial. So sometimes I think it's useful when docs call in to say, "Hey, if you think that patient is relapsing, you might want to send peripheral blood, but maybe just send them in before even doing a bone marrow biopsy so they don't end up getting stuck twice." Another thing that you gave in your initial answer was to think about what they have had before and what they're going to have — therefore, what they might respond to again. And that is, of course, a fundamental kind of thinking about whether they have originally had or not had a stem cell transplant as part of their first-line therapy. How does that influence things for you? If somebody has already had a stem cell transplant, are you immediately thinking, Would this patient be able to get a second transplant if they were to get back into remission?

Walter: At a center like ours, where transplantation is done very widely, it comes up very quickly. But we also realize that the posttransplant relapse outcomes are really quite dismal, especially if patients relapse relatively early after transplantation. And it puts a lot of limitations on what you can accomplish, just given the toxicities and residual issues from transplant. So it's sort of a small room to maneuver. But the question comes up for a second transplant or augmentation of the first transplant with donor lymphocyte infusions, for example.

Roboz: If there's a loss of chimerism and somebody seems like they're relapsing post transplant, do you use hypomethylating agents to try to rescue that situation? Do transplanters do that or is that not a thing?

Walter: I think it becomes very individualized, what people try. It probably just speaks to the fact that there's nothing that is very robustly backed up by strong data. It becomes very individualized.

Roboz: So maybe we'll unpack the discussion. You have a young patient who was treated initially with a stem cell transplant as part of their first-line therapy, so that patient is going to get re-looked at — re-diagnosed. You're going to check their molecular profile to see if there is anything potentially targetable. If, all of a sudden, they pop up with a FLT3 or with an IDH2, maybe there might be some standard therapy that could be applicable. You're going to talk to the transplanter, see if there is any donor lymphocyte infusion option or any rescue-of-chimerism option for that kind of a patient. But fundamentally, there is going to be a consideration of whether you can actually get that person back into a durable remission and make their first transplant work, or whether there would even be consideration for a second transplant. That seems like a fairly standard approach.

Walter: Yeah, I think that's pretty standard. But the details may differ from person to person, right?

Roboz: Absolutely. You have plenty of patients, though. I mean, the median age at diagnosis here is 67, 68, 69 years old. There are plenty of patients who might not have been transplanted who are older patients. So did they not get transplanted because they're 85 and they're not eligible for a transplant due to comorbidities? Or did they not get transplanted because they didn't feel like it? Is it the case still that, fundamentally in the relapse setting, unless you can get a patient into a transplant, we don't really have anything that is associated with prolonged remission in the in the post-relapse setting? Do you think that's still a true statement?

Walter: Sadly, I think that's still a very true statement.

Roboz: I've been looking increasingly at some of the older patients who might have been making a decision initially — "Well, you know, I'm 72, I'm 73. I don't feel like having a transplant." But if those patients relapse after their initial therapy and they're in good shape, then at both your center and mine, there is going to be consideration for an allogeneic stem cell transplant for that patient, even potentially well into their 70s.

Walter: We transplant otherwise medically healthy people up to the mid- to late 70s. And you're right — sometimes patients have hesitations about transplantation, either just for disease or toxicity consideration, or there are logistical barriers that make it very difficult. And sometimes the reasoning why transplant may be better than no transplant may not be as sound, but once the disease relapses, I think the risk-benefit ratio might change such that it could change the thinking of a patient or physician involved in the care. It's not always unilateral how the decision is made.

Roboz: I think that's a really good point for the whole discussion — that in a patient who has relapsed, you have to kind of look at everything all over again, and the assumptions that might have been at play at the time for the initial therapy planning might be different. So that ranges from differences in the immunophenotype of the disease, the molecular genetics of the disease, and what might have been the calculus for transplant discussion the first time around. I think that's a good way of thinking about it; it's right to keep an open mind and start all over again, with a lot of attention to clinical trials (whenever somebody would be eligible for a clinical trial). And I think that with a lot of the targeted agents, might there be a CD33 target? Might there be a 123 target? Might there be a CAR T-cell option? All of the clinical trials would be the best option likely for any of these patients in the relapsed setting.

Walter: And sometimes it's fundamental things, like the living situation has changed. The caregiver situation may have changed. Donor availability may have changed over a short period of time sometimes, and transplant study availabilities may have changed in a short period of time as well. So there are a lot of factors that could change and influence decision-making.

Roboz: I think those discussions are among the most interesting, even in some of the novel drug approvals, where people say "transplant ineligible," as if it's an obvious thing. But that's a very dynamic discussion. They might have been ineligible at a particular moment in time, but that moment might have changed. But everything that you're saying and everything that we've been talking about sounds pretty dire. Relapsed AML is a dangerous and deadly thing. And obviously what we would love to do is never have a patient relapse. There's a lot of discussion of MRD, or measurable residual disease, monitoring as a way to try to see what we can do to grab on to AML and fix it, if you will, or help it or get rid of it before it ever goes into relapse. So I want to take us backwards in time before the relapse. You're in remission. What can we do using MRD to try to hang on to that remission and never relapse? How do you measure it, first of all? And how do you test for it?

Walter: I think that that's the first challenge in AML. Unlike other blood cancers, I think we now have a good understanding that AML is very heterogeneous in its biology. And it's not just one disease; it's sort of a grab-bag of many different diseases. With this comes the challenge that there's not one single abnormality that defines even a majority of the disease cases. So this makes it a little tricky to identify good markers for small numbers of residual cells. And in AML, the two complementary approaches have been immunophenotypic abnormalities in how to identify cells with those. And on the other hand, looking at the genetic makeup of abnormal cells and associating those with leukemia-relevant cells. And they're both not perfect. They have both strengths and weaknesses, and I think we've gotten to realize what some of these strengths and weaknesses are. But it's not as trivial as, say, for chronic myeloid leukemia or acute promyelocytic leukemia, where there's this one mutation that seems to be characteristic of the disease and involved in the disease biology fundamentally.

Roboz: I think it's a good moment to call the listener's attention to a couple of publications that you and I are specifically involved in trying to make it easier for the clinician to deal with MRD monitoring and testing in clinical practice. One is the publication of the ELN, or European LeukemiaNet Guidelines for MRD, in Blood, and the second is a paper that we worked on together in Leukemia, trying to show what we do and don't know about MRD, with the don't-know list being a lot longer than the do-know list. But to make it easier for clinicians who are trying to figure out what to do for their patients in practice, let's break this down to a couple of easy things, or at least easier things. So if you have patients with core binding factor leukemia where they have a PCR-able or a molecularly monitorable lesion, like an 8;21 or an inversion 16, I'm going to make the statement that it should be considered standard of care to monitor MRD molecularly in those core binding factor patients, with the idea that if there appears to be a rising transcript level, you want to at least think about doing something to try to if possible prevent an overt relapse. What do you think of that statement? Too strong?

Walter: Partially. I mean, I think you're right. It's the core binding factor leukemia as a good subset of leukemia, where there are really solid data that the MRD detection method is quite reliable in giving you a prognostic piece of information. And I think it's very strongly associated with disease recurrence if these transcript levels increase after they've been either gone or unmeasurable or they have been low and they increase. I think what's a little more difficult to answer at this point is whether this early detection really translates into a meaningful improvement in outcome. In the case of APL, there are at least some data that if you intervene at this very low disease burden stage, then perhaps you're minimizing some of the acute bleeding complications that you might encounter if you treat the person with more disease burden. But I think in the non-APL setting, it's still a little bit of a question in my mind: whether you're truly improving outcomes at the end or you're just realizing earlier that trouble is coming up, and it might give you some time to put logistics in place for, let's say, a stem cell transplant. But I'm still sort of unclear about whether this truly translates into better outcome at the end.

Roboz: This is something that makes it so hard to figure out what to do when you're sitting with a patient. There are a couple of important things that you said that I just want to underline. You drew a distinction between undetectable and actually low levels, and I think that's really important for docs to know, because in the same way as when you're doing a CML monitoring of BCR-ABL, both you and the patient can go nuts if you go up again, down again with these tiny, tiny levels of residual transcript. There are data to suggest that both for the core binding factor patients and also for NPM1 patients, there might in fact be low, low transcript levels, which, if they stay stable, might actually be okay. So you don't want to necessarily go crazy and transplant somebody because they have 1 x 10-23 level detectable by a transcript. What you want to be looking for is trends. And then I do think that what you brought up about logistics… There are some data in core binding factor patients that maybe hypomethylation agents could actually zero out some rising transcript levels. We don't know what to do with that. We don't know whether that means they get monthly treatment forever or whether they get eight cycles or 23 cycles. But it might give you something to do. And it might actually have you watch the patient more carefully and think about what your salvage options might be, either with chemotherapy or with transplant. But regarding the point that you made about molecular monitoring with the flow vs PCR, the one thing that we would say is that for NPM1 patients and for core binding factor patients to use PCR, that's something that can be sent out to labs instead of flow cytometry. That I think is pretty clear — that at least we have some sense of a quantitative measurement if we're using PCR and that rising transcripts over time is not a good thing. That, I would say, is uncontroversial. Would you agree?

Walter: I agree.

Roboz: Now the controversial zone, of course, is the flow cytometry. Here, I think that when patients and docs get back a report that says the blasts are at 0.5%, that sounds pretty good because they had 90% blast at the time of their initial diagnosis. But now the report says 0.5%. That's got to be good. I think most docs were surprised, perhaps, to learn that even 0.1% is a dangerous level of residual disease. What do those percentage points mean? And when do you worry if an aberrant phenotype is showing up in the post-remission setting? Is there a cut-off that you use or that you at least start worrying about?

Walter: I think there are some cutoffs, and unfortunately, in 2022, the cutoffs depend very much on the institution that does the testing because we're far away from harmonized testing. And the performance characteristics of these tests that are available just vary very, very widely from place to place. So what means one thing for one place may mean something very different for another place, and I think that's something to really keep in mind as far as MRD testing goes with flow cytometry. You're right — it sounds very minimal to have 0.1% or 0.2%, but I always remind myself of what this actually translates to as total disease burden. And there are these estimates that if you enter the door as a leukemia patient, you have something like 1013 leukemic cells in your body. If you get into a morphologic remission, you're down to 1011 or 1012 cells. That sounds like a lot of cells, but when you say, well, it's 5% less, it doesn't sound like all that much, right? And if you go down to 0.1%, well, this is still 109 or 1010 cells. If you translate this into absolute numbers, then all of a sudden it doesn't sound like that minimal anymore, and it may make it a bit more visible or imaginable why this is not a good thing. With the assay that we're using, we looked at it carefully, and for us, anything that we detect is not good. But there are other assays where it really just seems to be more of a 0.1% threshold and below. It's not that easy anymore to distinguish, which may just speak to the fact that the assays become more difficult to use at these low numbers of cells.

Roboz: I do think that if we can have everybody think about numbers like 0.1% as worrisome, and speak with the pathologist at your institution and kind of figure out that — "Hey, wait a minute, 0.1%. She said on the podcast to worry about that; now I'm worried about that." I think that that's something to drive home because maybe that patient, for example, might be eligible for additional therapy. So that gets into the even harder question of, okay, we're worried; there is a level of disease left over that we are worried about. But now what are we supposed to do about it? So with respect to additional therapy at the time of identified residual disease, how do you approach that and when would you consider giving more chemo? Do you give something else? Do you do something about that MRD? And how do you make that decision?

Walter: Yes, I think you highlight one of the controversial areas in AML care right now. By now, almost everybody has gotten around to realizing that MRD is not good. But what would you make out of this? I think this is where it really becomes more opinion driven rather than data driven, with perhaps some areas where people coalesce around a certain strategy. Coming back to the core binding factor leukemias and transplant: The core binding factor leukemia patients who have good chemotherapy responses, based on really dramatic decrease in MRD burden, seem to have very low risk for relapse even outside of a transplant setting. So, people feel fairly confident that you can have a good shot at cure without a transplant. If these MRD trends are not as favorable — and the relapse rates in several series have been quite high (Liu Yin et al, Hong-Hu et al) and have been used to argue for a transplantation, with some noncontrolled data making an argument that this actually might improve outcomes overall... I think similar principles might apply for NPM1 leukemias, which you mentioned before. It's very similar as far as MRD monitoring goes. In other settings, it's a little less clear. Whether intervening in MRD or eradicating MRD, or with "MRD eraser therapy," where this truly ultimately translates into something measurably good for patients; or whether we're fooling ourselves and just realizing that MRD is nothing else than a marker for bad disease and that pounding on the marker really doesn't ultimately change outcomes — I think this is really where the big open question is in my mind right now.

Roboz: I think that is exactly right, and the summary statement would be that prevention of relapse is something we'd all love to say we know how to do. But it is not a well-worked-out, exact science. For patients who are in remission, learn about ways of looking at MRD and actually consult with AML centers to see if those patients might be (a) eligible for a disease mitigation strategy or (b) what would be the common wisdom for that particular patient? And then, of course, if there is a relapse referral to an AML center for additional potential strategies. I think this is a good way to wrap up the discussion — that both those at high risk for relapse with MRD monitoring strategies that might be challenging, and those patients with relapse, probably really do need to be at least thought about in the context of an AML center. I really want to thank you for joining in this tough discussion today.

Walter: Thanks for having me, Gail.


Bone Marrow Biopsy

Stem Cell Transplant

Donor Lymphocyte Infusions

Hypomethylating Agents


Measurable Residual Disease

2021 Update on MRD in Acute Myeloid Leukemia: A Consensus Document From the European LeukemiaNet MRD Working Party

Measurable Residual Disease as a Biomarker in Acute Myeloid Leukemia: Theoretical and Practical Considerations

Core Binding Factor

Minimal Residual Disease Eradication With Epigenetic Therapy in Core Binding Factor Acute Myeloid Leukemia

Minimal Residual Disease Monitoring by Quantitative RT-PCR in Core Binding Factor AML Allows Risk Stratification and Predicts Relapse: Results of the United Kingdom MRC AML-15 Trial

MRD-Directed Risk Stratification Treatment May Improve Outcomes of t(8;21) AML in the First Complete Remission: Results From the AML05 Multicenter Trial

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.