MRD Assessment and its Clinical Implications in AML

Transcript:

Harry Erba, MD, PhD: OK, I’ve been talking around MRD [minimal residual disease]. Dan, bring us through, how do we measure MRD if we think it’s important?

Daniel Pollyea, MD, MS: This is such a good conversation, let’s try to put it in context a little bit. The term we should be using, we default to minimal residual disease. Really what we should be talking about is measurable residual disease because really, any degree of disease when a patient goes into a morphologic remission is something that we can potentially be concerned about. Historically, we’ve gotten patients into morphologic remissions, maybe cleared a cytogenetic clone, and that’s about as deep or as good as we could tell with respect to how well a therapy worked. But in the modern era, we have all these tools to detect teeny tiny little bits of disease that are left over. And that’s what we’ve been discussing, what are the prognostic implications or therapeutic or treatment decisions that we make related to this.

But I think it’s worth having a few moments to talk about some of these very sophisticated methods because they’re out there. I think one of the most accessible ways to measure MRD are with flow-based therapies. And then there’s 2 different subsets within that discussion. If you have a baseline sample of the patient’s phenotype, as Harry referred to at the beginning, and one of the very important reasons to get this testing done from the beginning. Then there’s the potential to match a post-remission bone marrow sample, usually, to see if you have any traces of that diagnostic aberrant phenotype. And then within flow, there are also ways to detect disease even if you don’t have a baseline sample with these so-called different than normal approaches where large databases have determined what are the known phenotypes that are associated with this disease and matching up post-treatment marrows to see if there’s any overlap there.

Then I think what we’re all very excited about are some of these genomic techniques, so we take advantage of some of the information that we collect at the time of diagnosis, some of these gene mutations that exist. And then use that as a hook to look down really deep at very low levels, lower probably than is possible with flow-based techniques to see if there’s evidence of disease as defined by the continued existence of detectability of that gene mutation that is carried with the disease. So I think functionally those are the big picture ways to do MRD and all of these interesting discussions I’m sure will be clarified in the coming years about what this all means and what we should be doing about it.

Harry Erba, MD, PhD: But I think there are some very important nuances there. Flow-based, you require a very skilled hematopathologist to be able to pull out that leukemia-associated phenotype, as you mentioned or what’s it called, different from normal, basically, phenotype. So that’s a challenge there and not well standardized. The other thing that people misunderstand is they think next-generation sequencing, oh, that sounds really sensitive, it’s DNA, right? Next-generation sequencing has a limit of detectability of about 1% to 2%, something like that, based on inherent errors that occur during PCR [polymerase chain reaction] amplification. And now there’s going to be advances in that with bar-coding of the samples that will get us down to deeper levels. But this is not routinely available. And even PCR tests, people think that PCR tests will give you this information, but you actually have to have quite a few blasts still being present, I mean morphologically obvious to actually see it. And then we have the problem that Sasha [Alexander] referred to. You talked about these 3 mutations that might actually persist. Tell us more about that.

Alexander E. Perl, MD, MS: People can walk around with mutations in their blood that don’t actually predict for leukemia even though we see leukemias that have the same mutations in their cells. So you could have clonal hematopoiesis that arises as a consequence of aging that will functionally change how your bone marrow cells are working, but it may not progress beyond that. It’s sort of like an MGUS [monoclonal gammopathy of undetermined significance] of the hematopoietic system that predisposes to myeloid neoplasia but may not progress to it. If you have that, those cells can get additional mutations and go on to develop AML [acute myeloid leukemia]. And when you go back into remission, you test these patients and you find the clonal hematopoiesis but not necessarily the leukemia, no additional mutations. So some of the more common mutations that are seen include DNMT3A, TET2, and ASXL1 or DTA mutations. And their persistence is not very strongly predictive of leukemic relapse. But the additional hits that go from clonal hematopoiesis toward myeloid neoplasia are the ones you have to watch out for, not necessarily those ones.

Daniel Pollyea, MD, MS: This makes it so incredibly challenging because there are other genes that are involved in clonal hematopoiesis and can be mutated in an individual person. And for that person, the persistence of detection of that gene is not an MRD marker. But for other patients or the next patient you see, it might be a very important marker of their disease. So this is a real challenge to try to be sorting this out. And not to mention, going back to the original conversation, that even if we had a very predictable and accepted test to test for MRD, the same conversation applies, about what do we do about this and what are the implications, transplant versus nontransplant. We have lots of work to do in this area.

Transcript Edited for Clarity