Minimal Residual Disease Testing in AML: Still a Shifting Target

Elisabeth Paietta, PhD, Comments on European LeukemiaNet Working Party Consensus Document

Get Permission

Elisabeth Paietta, PhD

Elisabeth Paietta, PhD

Testing for minimal residual disease (MRD) has become an established part of the management of acute lymphoblastic leukemia (ALL), but in acute myeloid leukemia (AML), the technology still warrants validation. To address issues and set new standards, the European LeukemiaNet Working Party recently published its consensus document on the topic.1 In an accompanying commentary,2 Elisabeth Paietta, PhD, Professor of Medicine at Albert Einstein College of Medicine and Laboratory Director in Cellular Immunology at Montefiore Medical Center, New York, applauded the effort but zeroed in on the questions that remain. In an interview with The ASCO Post, she shared her concerns.

MRD: Reflection of Persistent Disease

What do you think MRD represents in AML?

First of all, we need to understand that MRD is not a mark of the disease; it is the disease. In AML, we initially used to count the blasts under the microscope, and when we saw evidence of disease, we knew the patient was not in remission. We’re doing the same thing now, but we are just looking at a much higher sensitivity level and better accuracy. What we are seeing is that having disease is worse than not having disease, and MRD is a reflection of persistent disease.

The European LeukemiaNet Working Party advocates keeping the commonly used 0.1% threshold level to distinguish patients with AML who are MRD-positive from those who are MRD-negative, with the caveat that MRD levels below 0.1% may still be of prognostic significance (with no controlled clinical study to show that). In ALL, we know the level of MRD positivity that is clinically important is 0.01—which is 10 times lower than what is currently used in AML. There are data from the Children’s Oncology Group (COG) showing that even below that level, there is still a risk for relapse.

‘Important First Step’

What do you consider the most important aspects of the European LeukemiaNet report?

Standardization for measuring MRD has been completely lacking for AML, at least in the United States, and the European LeukemiaNet report tries to establish standards. Although I don’t completely agree with everything the report is proposing, I think it is an important first step.

Once we figure out exactly which mutations are most important—and at what level—we may be able to perform next-generation sequencing on blood and not need as many bone marrow tests.
— Elisabeth Paietta, PhD

Tweet this quote

The other important thing the report addresses is the quality of the sample. That’s because, in the blood, MRD levels are much lower than they are in the bone marrow—a difference of 20- to 200-fold. You need to do an aspirate, and it must be a first pull. This is a big problem with measuring MRD by flow cytometry. If you have dilution of the marrow with blood, your result is not going to be trustworthy. Usually, if you find anything in the marrow, you can say the patient is MRD-positive, but if you have marked hemodilution, your result can be falsely MRD-negative. We are teaching our Eastern Cooperative Oncology Group (ECOG)–American College of Radiology Imaging Network (ACRIN) clinical trials investigators how to do marrow aspiration for MRD correctly. The European LeukemiaNet emphasizes proper sampling, and that’s a good thing.

You mentioned you don’t agree with everything in the European LeukemiaNet report. What might some of these things be?

The authors propose combining the two approaches for MRD detection by flow cytometry: one based on diagnostic leukemia-associated immunophenotypes, which is commonly used in Europe, and the other based on a different-from-normal strategy, which has been implemented by the COG.

One problem with using leukemia-associated immunophenotypes is that it depends completely on the baseline immunophenotype of the leukemia, and this phenotype changes with treatment. Another problem is that in AML, there is frequently more than one clone, so you have more than one phenotype that you now have to monitor after treatment.

I question why the European LeukemiaNet Working Party is trying to merge the two approaches. The COG has looked at thousands of patients with their excellent different-from-normal approach that, I think, we should all follow. Given its advantages, why not agree on the different-from-normal strategy and the creation of fixed antibody panels for all of us to use?

Next-Generation Sequencing vs Flow Cytometry

Some experts think next-generation sequencing may be a better approach for measuring MRD than flow cytometry. What are your thoughts on this?

I think next-generation sequencing is the way of the future. It’s going to have advantages because the sensitivity is greater, and it may be able to detect MRD in the blood [rather than require a marrow aspirate]. But next-generation sequencing has its own complications.

First of all, there are genes that are commonly mutated in patients with AML but that, unfortunately, are also commonly mutated in older healthy individuals: the so-called DTA mutations (DNMT3A, TET2, ASXL1). In the recent study by Jongen-Lavrencic and colleagues,3 when they excluded the DTA mutations, they found that persons with persistent mutations had a significantly higher relapse rate than those without them: 55.4% vs 31.9% (P = .001).

But what about the 32% of patients who lacked mutations but still relapsed? That’s one-third of your patients you still have to worry about. This has been the recurring theme: 25% to 30% of MRD-negative patients relapse irrespective of the MRD testing methodology, and we don’t know the predictive factors. Once we figure out exactly which mutations are most important—and at what level—we may be able to perform next-generation sequencing on blood and not need as many bone marrow tests.

MRD: Not the Only Measure

Are there other reasons why testing for MRD may be unreliable?

Another important point is that MRD depends on treatment. You cannot use genetic mutations to monitor MRD in patients treated with anything that induces differentiation—such as all-trans retinoic acid and IDH2 inhibitors, at least not early on in the treatment. If a cell differentiates, it does not die right away; the genetic variant persists. So, MRD is a shifting target, and changes in treatment can also change the interpretation of MRD results.

This is relevant for clinical trial design. Most experts say that if you are evaluating a new treatment, you have to establish the clinically relevant threshold of MRD retrospectively in the first trial; then you can apply it prospectively.

I think MRD is an important measure in patients, but it shouldn’t be the only one. We still have to look for those patients who are MRD-negative and who relapse and then find out why they relapsed.

Words of Caution

If testing for MRD cannot be trusted outside of a clinical trials framework, should oncologists be testing for MRD in AML at all?

Yes. The results can be hugely variable. You have to be careful, because measuring MRD and not really knowing what you are doing is worse than not measuring it at all. If you get a false-positive or a false-negative result, you may harm the patient. Again, one of the good points about the European LeukemiaNet paper is that the group is trying to establish standards for testing.

I’m testing predominantly in ALL. Since we don’t currently have an ECOG-ACRIN group trial in AML that requires investigators to look for MRD, I’m not using it at this point.
— Elisabeth Paietta, PhD

Tweet this quote

Testing should not be done by a community laboratory unless the staff have been trained by experts. The European LeukemiaNet Working Party, in its report, clearly says that samples should be sent to a central lab. Even in ALL, the testing situation is not ideal. I run the central lab for the ECOG-ACRIN phase III study of blinatumomab (Blincyto) in B-lineage ALL ( identifier NCT02003222) and am using the COG’s approach. You have no idea how often I receive e-mails from pathology labs who have questionable findings. I am fortunate to be able to discuss disparate results with an expert at the COG to arbitrate them.

My suggestion is that oncologists who want to test for MRD should either put these patients on clinical trials, where there are central laboratories to do the testing, or send samples to reputable commercial laboratories that use the COG approach. Importantly, they should also follow the European LeukemiaNet Working Party’s recommendations for sampling. They definitely should get help with interpretation, which is the most critical factor.

Testing in ALL vs AML

Are you currently testing for MRD in AML?

I’m testing predominantly in ALL. Since we don’t currently have an ECOG-ACRIN group trial in AML that requires investigators to look for MRD, I’m not using it at this point.

In ALL, we know for sure that MRD is very important, and there are several labs in this country that are certified by the COG to test for it. Unfortunately, there exists no proficiency testing for MRD by flow cytometry in the United States. I think in ALL, especially B-lineage, it should no longer be a question. But in AML, it’s still a fairly open question in most cases.


  • Created over 2 years and through 4 meetings by a specially designated MRD Working Party of the European LeukemiaNet
  • 24 faculty members with expertise in AML hematopathology, molecular diagnostics, clinical trials, and clinical medicine
  • 19 institutions in Europe and the United States

Source: Schuurhuis et al.1

We have to develop common approaches and common interpretation tools. As long as everybody does his or her own thing, we’re going to have problems comparing data among labs. That’s the advantage of running a clinical trial.

We need to establish an MRD Working Group for AML, as we’ve done for ALL, which is sponsored by the National Cancer Institute. Without question, the clinically most useful MRD test and targets are not yet determined. ■

DISCLOSURE: Dr. Paietta reported no conflicts of interest.


1. Schuurhuis GJ, Heuser M, Freeman S, et al: Minimal/measurable residual disease in AML: A consensus document from the European LeukemiaNet MRD Working Party. Blood 131:1275-1291, 2018.

2. Paietta E: Consensus on MRD in AML? Blood 131:1265-1266, 2018

3. Jongen-Lavrencic M, Grob T, Hanekamp D, et al: Molecular minimal residual disease in acute myeloid leukemia. N Engl J Med 378:1189-1199, 2018.