iwAL 2019 | Utilizing MRD for AML in the clinical setting

Marion Subklewe:
Hello. I’m here at the second meeting of the iwAL in Barcelona and we just finished the last session of the first day of the iwAL meeting, and was focusing on MRD assessment in AML. My name is Marion Subklewe. I’m from the University Hospital in Munich where I’m heading the MRD program in AML. And I’m joined by Jorge Sierra from Barcelona and Sylvie Freeman from Birmingham. And we’re just going to sum up some of the very interesting discussion we had in the last session on MRD assessment in AML. So, maybe we start with a short summary on the data that you presented from the Spanish group.

Jorge Sierra:
Well, I presented the evolution of our risk therapy in consecutive protocols of intensively treated patients. And this is quite long story already, almost 30 years. But with a better knowledge of the pathophysiology of the disease, and also with the new methods to assess the MRD, we have been able to adapt the therapy more precisely. And, at this moment we are doing MRD by several methods. We do a PCR for CBF AML and also for NPM1. And also we are doing flow cytometry. There is a large division of MRD studies by flow because, at first, the studies here in Spain made by the group in Salamanca were already published early in 2000, and we have followed that.

Jorge Sierra:
And now we are also studying the expression of ND1. We follow that. We have defined three groups of categories based on molecular characteristics and MRD and with different prognosis, and we have been adapted the polar emission therapy based on that.

Marion Subklewe:
Right. Before we go further, maybe you also introduce maybe some of the results that were obtained with the MIC.

Sylvie Freeman:
Yes. So, we’ve had a dual approach, like Jorge is saying as well, using molecular and flow cytometry approaches. And both approaches have been very informative. And what I talked about was our U.K. experience. So, data by David Grimwade and Adam Ivy showed for the NPM1 mutated patients, which is a very specific subgroup, that measurement of mutant transcripts in the blood after induction is highly prognostic. But there’s still what we would call a false negative rate. So, some of the MRD negative patients will relapse, only about sort of 20, 30% at the maximum. And the way of actually reducing these false negatives is to take sequential measurements. So this we continue to do. We sequentially measure patients with NPM1, as well as the core binding factor AMLs by RT-qPCR. So, that was number one.

Sylvie Freeman:
And then, the other approach that we use with flow cytometry, we can look across the genetic subtypes and we can look a little bit more about the context of what the genetic subtypes gives to the interpretation of MRD positivity and MRD negativity, so to make the use of MRD as a prognostic tool more accurate. And I think the other finding that has come up quite recently which builds up on pediatric data, looking at pediatric marrows and using flow cytometry as MRD is, if you look at patients who are classified as refractory, a portion of those are MRD negative by flow. And in fact these patients do as well as patients who are NCR and MRD negative. And we’ve shown similar results in adults after first induction, that up to 30% of patients in two separate trial cohorts from the HOVON and from the U.K. are classified as refractory, and therefore are likely to have intensified treatment, if not transplant. But actually, these patients should be classified as MRD negative, not refractory.

Marion Subklewe:
Right. So, before we maybe go into risk adaptive therapy and how we stratify and what are the problems of stratifying MRD based a therapy, I just want to shortly address methods and technology. I think, also, you two are sort of representatives of standardization and more harmonization approaches. I’d just like you to comment on the strategies you are following. And I just, as introduction, would say at our center we are also following harmonization on the flow based MRD assay. I still think there is a huge heterogeneity in the different trials that make comparison difficult. And I also think that it’s really difficult to have risks stratified therapy on measurement that is still not validated in a lot of trials in a prospective way.

Marion Subklewe:
So, at our center, we clearly propagate and we could show that you should at least integrate two measurements before risk stratification, at least that’s what we propose. And we could show by flow cytometry that there is a certain percentage of patients that, if you do two measurements close, so we do a very early time point, a time point of aplasia, and post induction, that you have patients that are MRD positive at both time points or only at one time point. And, clearly, the worst prognostic group of the patients who are positive at both time points. Whereas, if you’re negative at both time points, they clearly have a favorable outcome, and then you have to intermediate.

Marion Subklewe:
And also, we think that, as you already mentioned, cytomorphology is a difficult topic and can be misleading. And we could also show that flow is a better discriminator on what kind of blast you’re actually looking at through the microscope. So, I think it’s just the difficulty of cytomorphology to discriminate, particularly early after chemotherapy, what kind of blasts you actually are looking at. So, these are sort of two findings we identified in Munich. And so, I would like to have you comment on methods and technology in Spain, I would say, or-

Jorge Sierra:
I fully agree. And what you mentioned, that MRD assessment is different depending on who is doing that, particularly this applies to flow cytometry. And then, I think that the combined approach is a good idea. Also doing in parallel, whenever possible, PCR and flow. Also, eventually in our hands, as we have experienced BD1 expression and MRD by flow, this is another possibility. Also what you mention of the two determinations close in time to define what is in the border. I would say positive negative or negative positive versus consistent results. That’s also very important.

Jorge Sierra:
And I think that really we need more easy algorithms in automatic systems for interpreting flow cytometry. It is too subjective and we need automatic system that make gating and all the studies in the same way. Because, on the other hand, to do centralized studies is not so easy, regarding MRD. It may happen in very well organized comparative groups, but it’s hard to get that academic institution with that revision on doing MRD studies refer the samples to another academic institution. And, at least in our country, they tend to do the analysis at their own institution. So, the trial presented, flow cytometry was not centralized. PCR was centralized but MRD by flow were performed in a reduced number of institution with the most experience. And there is a possibility of some of some heterogeneity or subjectivity.

Jorge Sierra:
I think that there has been a very good step forward that has been worked on in the publication of the ELN guidelines for MRD assessment. This is a very important step forward. There are also recommendation from other flow. And I think that still there is a bit of lack of more definition on how to do it homogeneously. But, for sure, the issue is still open.

Marion Subklewe:
Yes. I completely agree. And I think the ELN is contributing enormously. Hopefully there’s going to be another publication where we’re going to advise in a more concrete way how to assess MRD flow, which is particularly subjective. We in Germany actually have now our first Round Robin assay. And interestingly, a lot of labs said they wanted to participate. Then, actually, the first Round Robin was only done with 12 labs. But it indicates there were over 50 labs who wanted to participate. That there’s a clear interest. A lot of people are doing it, which is somehow dangerous, because they also send out reports. They’re not validated. And, even if it says under the report it’s not validated or something, the treating physicians might take conclusions from this. So, I think it’s also somehow dangerous, and it clearly needs to be validated.

Sylvie Freeman:
Yes. I mean, I completely agree that when you’re opening up the assay to a wide number of labs, that having at least a standardized protocol that they can follow and possibly access to a second check of analysis by a centralized is important. Actually, what surprised me, just being a little bit provocative here, but going the other way, what surprised me when we compared our two cohorts, trial cohorts from the HOVON and from the MRC with these refractory patients, we were doing different MRD assays. The trial cohorts were different, but the survivals were almost identical and the incidence relapse was almost identical.

Sylvie Freeman:
So, I think the lesson in that is, if it’s done well, even though the methods they may not be exactly the same, we’re getting the same results. But what happens is when it’s done less well, which I think your point about the WT1.

Jorge Sierra:
One.

Sylvie Freeman:
Yeah. That’s another assay, which is probably better than poorly done flow cytometry. But, you know Marion, you’ve done some really good work in Germany of harmonizing, and you’re showing that it is possible with a greater number of labs. And you have probably done the same as well, that you can standardize. But I think, if it’s done well, then actually results are very, very similar and comparable.

Jorge Sierra:
Mm-hmm (affirmative).

Marion Subklewe:
Yeah. And I think then the next step, clearly, is going to be an unsupervised analysis. Step by step, hopefully, we’re moving into that direction.

Marion Subklewe:
So, maybe the second round of discussion could be on the risk stratified therapy. We are still very conservative in Germany and currently have no trial running that is based, in a larger randomized fashion, based on MRD, as we are still sort of struggling in harmonization of the MRD assessment. So, you both are more advanced. Maybe you can comment on this.

Jorge Sierra:
Well, yes. Really we combined both the molecular characterization and the MRD. We had previous experience in our trial from 2003 that the MRD by flow was, basically, discriminant. So, this is why. And that there were some patients in the favorable genetic groups that had MRD positivity, and they did poorly. So, this is why we have tried to combine the two methods.

Jorge Sierra:
Well, in one hand it’s logical, but the stratification we did was based on our previous results in non-randomized trials. So, it was, I would say, quite a bit arbitrary, the categories we decided to do. But, particularly for the favorable group, that with favorable genetics and MRD negativity, the results are really outstanding. So, in that population, the combination of the two methods allows to identify particularly 40% of AML patients who have very good outcome and do not need a transplant in first year. So, I think that this is an important finding.

Marion Subklewe:
Yes. So, I think-

Sylvie Freeman:
Yeah. So, the U.K., so we are using MRD for re stratification but in an agnostic way. So, the questions we’ve been asking so far of, you have an MRD result, does different types of treatment make a difference? So, not presuming that intensification is the answer. We don’t know. So, those results will come out over the next few years. So, I think we know that MRD is important. We don’t know how best to treat it, from our point of view.

Marion Subklewe:
To sum up, I think what we’ve talked about is that, despite the Brexit discussions, I think Europe is growing together. And this is going to be really constructive and helpful. And clearly, also, the states. So, it’s going to be a harmonization in the MRD diagnostics, an exchange of data. I think that we are really moving into that direction and a lot of effort is put in there. So, we’ll have a basis that we can also compare trial results and MRD stratified trial results.

Marion Subklewe:
And then, I think, second for the risk stratified therapy, clearly it’s going to be really interesting what comes out of the randomized trials you conducted in the U.K. See how we actually can change the course of the disease by MRD stratified therapy.

Sylvie Freeman:
Yeah. And we’ll see the new therapies coming out. So, immunotherapy and everything, we’ll change that and that’s going to be-

Marion Subklewe:
Yeah. And I think that other questions like how you can use MRD as a surrogate endpoint for clinical trials, which is still a difficult issue, because we can’t use it as a surrogate of a surrogate. So, it has to be a surrogate of overall survival, which is a difficult end point. But I think we first have to do methods and technology completely harmonized, standardized with Round Robin assays for the comparison between the clinical trials and clinical study groups, see what is coming out of clinical trials that are currently recruiting patients where we have a risk based, MRD based treatment algorithms.

Marion Subklewe:
And then the, I think, interesting part will be if we can sort of put in a box for the MRD positive ones and have novel treatment options, see if they can, first of all, do MRD reversal, and if that actually is going to translate into a better, relapse free, or overall survival, or if we are still just identifying patients who have for the worst biology of the disease. This is still somehow unanswered and, clearly, probably there’s also not one answer, because it’s going to be maybe different for different types of disease, different interventions. So, it could be possibly that for one it’s not a predictor, and for other ones, you won’t change the course of it. For example, P53 seems to be something that’s going to be a real challenge. Right?

Sylvie Freeman:
Yeah. So, I mean, I completely agree. And you’ve done this in a way by your genetic stratification, but we just need to take all the added information together, and that’s going to refine things and help us detect signals for what actually works best in smaller subgroups.

Marion Subklewe:
So, you get the final word of the round.

Jorge Sierra:
Well, I fully agree. I think that in all the styles are having performed the last that the MRD persistence is bad news. And that, even therapies like allogeneic transplantation, with all the intensity, and all the immune mediated effect of GBL is affected by the fact that there is MRD before the transplant. So, it’s really, I think, a very strong marker of poor outcome. But we have to delineate a bit more on how to measure that and how to address the situation.